Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss
electrometer front end for dmm
890 views
Skip to first unread message
bonesi...@gmail.com
May 30, 2014, 4:10:55 PM5/30/14
to
I want to make measurements on pyroelectric crystals but I think the input impedance of a typical dmm is too low (10 Megaohm or so). Any suggestions on a simple electrometer adapter for a dmm? I can't afford a new electrometer so I was hoping I could make a simple circuit to add to a dmm for these measurements.
Scot
Scot
Maynard A. Philbrook Jr.
May 30, 2014, 4:39:30 PM5/30/14
to
In article <78a38025-10ca-4af9...@googlegroups.com>,
bonesi...@gmail.com says...
you can also use a JFET input type op-amp
you need to look around but I think you can
obtain up around 1G ohm input Z and I am sure
you can get more.
P.S.
they make active fet probes.
Look for "high impedance active probe" circuits, I just found a
couple...
Jamie
bonesi...@gmail.com says...
>
> I want to make measurements on pyroelectric crystals but I think the input impedance of a typical dmm is too low (10 Megaohm or so). Any suggestions on a simple electrometer adapter for a dmm? I can't afford a new electrometer so I was hoping I could make a simple circuit to add to a dmm for these measurements.
>
> Scot
I guess you could make a JEFT front end?
> I want to make measurements on pyroelectric crystals but I think the input impedance of a typical dmm is too low (10 Megaohm or so). Any suggestions on a simple electrometer adapter for a dmm? I can't afford a new electrometer so I was hoping I could make a simple circuit to add to a dmm for these measurements.
>
> Scot
you can also use a JFET input type op-amp
you need to look around but I think you can
obtain up around 1G ohm input Z and I am sure
you can get more.
P.S.
they make active fet probes.
Look for "high impedance active probe" circuits, I just found a
couple...
Jamie
Steve Wilson
May 30, 2014, 5:01:26 PM5/30/14
to
20fA input bias, RRIO, 15V VCC, 1 �V/�C Input Offset Tempco, 1.5 MHz GBW
Note 13: Guaranteed limits are dictated by tester limitations and not
device performance. Actual performance is reflected in the typical value.
http://octopart.com/partsearch#!?q=LMC6482
Might add a 10k from output to ground to provide a load for the output
stage.
dagmarg...@yahoo.com
May 30, 2014, 5:09:56 PM5/30/14
to
On Friday, May 30, 2014 5:01:26 PM UTC-4, Steve Wilson wrote:
http://www.ti.com/product/LMP7721
Some articles for our OP:
http://electronicdesign.com/test-amp-measurement/whats-all-femtoampere-stuff-anyhow
http://www.edn.com/design/analog/4327165/Achieve-femtoampere-leakage-in-surface-mount-op-amp-layouts
Cheers,
James Arthur
> bonesis...@gmail.com wrote:
> > I want to make measurements on pyroelectric crystals but I think the
> > input impedance of a typical dmm is too low (10 Megaohm or so). Any
> > suggestions on a simple electrometer adapter for a dmm? I can't
> > afford a new electrometer so I was hoping I could make a simple
> > circuit to add to a dmm for these measurements.
>
> > I want to make measurements on pyroelectric crystals but I think the
> > input impedance of a typical dmm is too low (10 Megaohm or so). Any
> > suggestions on a simple electrometer adapter for a dmm? I can't
> > afford a new electrometer so I was hoping I could make a simple
> > circuit to add to a dmm for these measurements.
>
> Use LMC6482 as a voltage follower. $2.08 Digi-Key, Newark, etc
>
> 20fA input bias, RRIO, 15V VCC, 1 �V/�C Input Offset Tempco, 1.5 MHz GBW
>
> Note 13: Guaranteed limits are dictated by tester limitations and not
> device performance. Actual performance is reflected in the typical value.
>
> http://octopart.com/partsearch#!?q=LMC6482
>
> Might add a 10k from output to ground to provide a load for the output
> stage.
Beat me to it. The LMC662 is another choice (cheap), and the LMP7721 (3pA, expensive).
>
> 20fA input bias, RRIO, 15V VCC, 1 �V/�C Input Offset Tempco, 1.5 MHz GBW
>
> Note 13: Guaranteed limits are dictated by tester limitations and not
> device performance. Actual performance is reflected in the typical value.
>
> http://octopart.com/partsearch#!?q=LMC6482
>
> Might add a 10k from output to ground to provide a load for the output
> stage.
http://www.ti.com/product/LMP7721
Some articles for our OP:
http://electronicdesign.com/test-amp-measurement/whats-all-femtoampere-stuff-anyhow
http://www.edn.com/design/analog/4327165/Achieve-femtoampere-leakage-in-surface-mount-op-amp-layouts
Cheers,
James Arthur
jeroen Belleman
May 30, 2014, 5:46:07 PM5/30/14
to
bias current. Maybe I just got lucky, because that's *much* better
than the datasheet says it is.
Jeroen Belleman
Steve Wilson
May 30, 2014, 6:06:05 PM5/30/14
to
jeroen Belleman <jer...@nospam.please> wrote:
> I've used an LPC661, which turned out to have an amazing 170aA of
> bias current. Maybe I just got lucky, because that's *much* better
> than the datasheet says it is.
> Jeroen Belleman
Apparently the bias current comes from the input protection circuit. Paul
> I've used an LPC661, which turned out to have an amazing 170aA of
> bias current. Maybe I just got lucky, because that's *much* better
> than the datasheet says it is.
> Jeroen Belleman
Grohe mentioned sonewhere they made some devices without input protection
and could not measure the input current. They could not use the devices
however. They blew the inputs whenever they tried to handle them.
Paul has some interesting posts on how to wire the devices for lowest
leakage and how to clean them. You have to sift through the crud to find
them, but they are worth reading. Sorry for the wrap.
https://groups.google.com/forum/#!searchin/sci.electronics.design/%22paul
$20grohe%22$20lmc660
Here's the tinyurl version
http://tinyurl.com/nag6ym4
John Larkin
May 30, 2014, 6:30:01 PM5/30/14
to
On Fri, 30 May 2014 23:46:07 +0200, jeroen Belleman
<jer...@nospam.please> wrote:
>On 30/05/14 23:09, dagmarg...@yahoo.com wrote:
>> On Friday, May 30, 2014 5:01:26 PM UTC-4, Steve Wilson wrote:
>>> bonesis...@gmail.com wrote:
>>
>>>> I want to make measurements on pyroelectric crystals but I think the
>>>> input impedance of a typical dmm is too low (10 Megaohm or so). Any
>>>> suggestions on a simple electrometer adapter for a dmm? I can't
>>>> afford a new electrometer so I was hoping I could make a simple
>>>> circuit to add to a dmm for these measurements.
>>>
>>> Use LMC6482 as a voltage follower. $2.08 Digi-Key, Newark, etc
>>>
>>> 20fA input bias, RRIO, 15V VCC, 1 ?V/?C Input Offset Tempco, 1.5 MHz GBW
<jer...@nospam.please> wrote:
>On 30/05/14 23:09, dagmarg...@yahoo.com wrote:
>> On Friday, May 30, 2014 5:01:26 PM UTC-4, Steve Wilson wrote:
>>> bonesis...@gmail.com wrote:
>>
>>>> I want to make measurements on pyroelectric crystals but I think the
>>>> input impedance of a typical dmm is too low (10 Megaohm or so). Any
>>>> suggestions on a simple electrometer adapter for a dmm? I can't
>>>> afford a new electrometer so I was hoping I could make a simple
>>>> circuit to add to a dmm for these measurements.
>>>
>>> Use LMC6482 as a voltage follower. $2.08 Digi-Key, Newark, etc
>>>
>>>
>>> Note 13: Guaranteed limits are dictated by tester limitations and not
>>> device performance. Actual performance is reflected in the typical value.
>>>
>>> http://octopart.com/partsearch#!?q=LMC6482
>>>
>>> Might add a 10k from output to ground to provide a load for the output
>>> stage.
>>
>> Beat me to it. The LMC662 is another choice (cheap), and the LMP7721 (3pA, expensive).
>>
>> http://www.ti.com/product/LMP7721
>>
>> Some articles for our OP:
>> http://electronicdesign.com/test-amp-measurement/whats-all-femtoampere-stuff-anyhow
>> http://www.edn.com/design/analog/4327165/Achieve-femtoampere-leakage-in-surface-mount-op-amp-layouts
>>
>> Cheers,
>> James Arthur
>>
>
>I've used an LPC661, which turned out to have an amazing 170aA of
>bias current. Maybe I just got lucky, because that's *much* better
>than the datasheet says it is.
>
>Jeroen Belleman
You could get lucky and have the Vcc and Vss ESD diode leakages
>>> Note 13: Guaranteed limits are dictated by tester limitations and not
>>> device performance. Actual performance is reflected in the typical value.
>>>
>>> http://octopart.com/partsearch#!?q=LMC6482
>>>
>>> Might add a 10k from output to ground to provide a load for the output
>>> stage.
>>
>> Beat me to it. The LMC662 is another choice (cheap), and the LMP7721 (3pA, expensive).
>>
>> http://www.ti.com/product/LMP7721
>>
>> Some articles for our OP:
>> http://electronicdesign.com/test-amp-measurement/whats-all-femtoampere-stuff-anyhow
>> http://www.edn.com/design/analog/4327165/Achieve-femtoampere-leakage-in-surface-mount-op-amp-layouts
>>
>> Cheers,
>> James Arthur
>>
>
>I've used an LPC661, which turned out to have an amazing 170aA of
>bias current. Maybe I just got lucky, because that's *much* better
>than the datasheet says it is.
>
>Jeroen Belleman
cancel.
I guess you could trim the supplies to *make* them cancel.
This is the pA parts tester that I built:
https://dl.dropboxusercontent.com/u/53724080/Gear/Z260_pA/99A260A1.JPG
https://dl.dropboxusercontent.com/u/53724080/Gear/Z260_pA/99A260A3.JPG
https://dl.dropboxusercontent.com/u/53724080/Gear/Z260_pA/99S260A.JPG
Its leakage was probably about 10 fA.
A simple 2N7000 source follower should be pretty good. That will stay
halfway on for hours with the gate floating, so leakage must be pretty
low.
https://dl.dropboxusercontent.com/u/53724080/Parts/2N7000.jpg
--
John Larkin Highland Technology, Inc
jlarkin att highlandtechnology dott com
http://www.highlandtechnology.com
Steve Wilson
May 30, 2014, 7:06:52 PM5/30/14
to
John Larkin <jla...@highlandtechnology.com> wrote:
>
> You could get lucky and have the Vcc and Vss ESD diode leakages
> cancel.
>
> I guess you could trim the supplies to *make* them cancel.
A long time ago, someone posted a reason why this won't work. I have been
>
> You could get lucky and have the Vcc and Vss ESD diode leakages
> cancel.
>
> I guess you could trim the supplies to *make* them cancel.
trying to find the post but have had no luck. I forget the explanation, but
it should be easy to test.
Simply set it up to measure the input bias current, then adjust the supply
voltages to see if the bias current can be switched to the opposite
polarity. If so, it has to go through zero somewhere.
Steve Wilson
May 30, 2014, 7:30:04 PM5/30/14
to
Steve Wilson <no...@nospam.com> wrote:
> John Larkin <jla...@highlandtechnology.com> wrote:
>> You could get lucky and have the Vcc and Vss ESD diode leakages
>> cancel.
>> I guess you could trim the supplies to *make* them cancel.
> A long time ago, someone posted a reason why this won't work.
Actually, if the bias current could be trimmed to zero by adjusting the
> John Larkin <jla...@highlandtechnology.com> wrote:
>> You could get lucky and have the Vcc and Vss ESD diode leakages
>> cancel.
>> I guess you could trim the supplies to *make* them cancel.
> A long time ago, someone posted a reason why this won't work.
supply voltages, the required supply voltages for the specified bias
current whould be noted in the datasheet. And if the diode leakages acted
like resistances, the bias current would change as the input voltage
moved through the operating range.
There should also be a plot somewhere of the input bias current vs VCC
and VEE, and another one showing the bias current vs input voltage.
The LMC6482 and LMC660 specification for input bias current do not
mention the supply voltage, and there are no plots showing changes in the
bias current as a function of input voltage or supply voltage. The only
plots involving bias current show the increase with temperature.
So clearly the input bias current leakage does not act like resistances,
and it is not possible to trim the bias current to zero by changing the
supply voltage.
This raises the question of what creates the input bias current, and
where does it go?
dagmarg...@yahoo.com
May 30, 2014, 7:42:11 PM5/30/14
to
On Friday, May 30, 2014 5:46:07 PM UTC-4, jeroen Belleman wrote:
that the ib was lower, caused by the protection diodes, and spec set by
testing throughput.
The LMC6001 was, I think, tested for lower ib.
I goofed on the LMP7721--ib=3fA, not pA (but what's three orders of magnitude
between friends?).
Cheers,
James Arthur
> On 30/05/14 23:09, dagmargoo...@yahoo.com wrote:
> > On Friday, May 30, 2014 5:01:26 PM UTC-4, Steve Wilson wrote:
> >> bonesis...@gmail.com wrote:
> >
> >>> I want to make measurements on pyroelectric crystals but I think the
> >>> input impedance of a typical dmm is too low (10 Megaohm or so). Any
> >>> suggestions on a simple electrometer adapter for a dmm? I can't
> >>> afford a new electrometer so I was hoping I could make a simple
> >>> circuit to add to a dmm for these measurements.
> >>
> >> Use LMC6482 as a voltage follower. $2.08 Digi-Key, Newark, etc
> >>
> >> 20fA input bias, RRIO, 15V VCC, 1 �V/�C Input Offset Tempco, 1.5 MHz GBW
> >>
> >> Note 13: Guaranteed limits are dictated by tester limitations and not
> >> device performance. Actual performance is reflected in the typical value.
> >>
> >> http://octopart.com/partsearch#!?q=LMC6482
> >>
> >> Might add a 10k from output to ground to provide a load for the output
> >> stage.
> >
> > Beat me to it. The LMC662 is another choice (cheap), and the LMP7721 (3pA, expensive).
> >
> > http://www.ti.com/product/LMP7721
> >
> > Some articles for our OP:
> > http://electronicdesign.com/test-amp-measurement/whats-all-femtoampere-stuff-anyhow
> > http://www.edn.com/design/analog/4327165/Achieve-femtoampere-leakage-in-surface-mount-op-amp-layouts
> >
>
>
> > On Friday, May 30, 2014 5:01:26 PM UTC-4, Steve Wilson wrote:
> >> bonesis...@gmail.com wrote:
> >
> >>> I want to make measurements on pyroelectric crystals but I think the
> >>> input impedance of a typical dmm is too low (10 Megaohm or so). Any
> >>> suggestions on a simple electrometer adapter for a dmm? I can't
> >>> afford a new electrometer so I was hoping I could make a simple
> >>> circuit to add to a dmm for these measurements.
> >>
> >> Use LMC6482 as a voltage follower. $2.08 Digi-Key, Newark, etc
> >>
> >> 20fA input bias, RRIO, 15V VCC, 1 �V/�C Input Offset Tempco, 1.5 MHz GBW
> >>
> >> Note 13: Guaranteed limits are dictated by tester limitations and not
> >> device performance. Actual performance is reflected in the typical value.
> >>
> >> http://octopart.com/partsearch#!?q=LMC6482
> >>
> >> Might add a 10k from output to ground to provide a load for the output
> >> stage.
> >
> > Beat me to it. The LMC662 is another choice (cheap), and the LMP7721 (3pA, expensive).
> >
> > http://www.ti.com/product/LMP7721
> >
> > Some articles for our OP:
> > http://electronicdesign.com/test-amp-measurement/whats-all-femtoampere-stuff-anyhow
> > http://www.edn.com/design/analog/4327165/Achieve-femtoampere-leakage-in-surface-mount-op-amp-layouts
> >
>
>
> I've used an LPC661, which turned out to have an amazing 170aA of
> bias current. Maybe I just got lucky, because that's *much* better
> than the datasheet says it is.
Very impressive. I remember a Bob Pease article somewhere about just that,
> bias current. Maybe I just got lucky, because that's *much* better
> than the datasheet says it is.
that the ib was lower, caused by the protection diodes, and spec set by
testing throughput.
The LMC6001 was, I think, tested for lower ib.
I goofed on the LMP7721--ib=3fA, not pA (but what's three orders of magnitude
between friends?).
Cheers,
James Arthur
John Larkin
May 30, 2014, 9:07:39 PM5/30/14
to
On Fri, 30 May 2014 23:30:04 GMT, Steve Wilson <no...@nospam.com>
wrote:
Clearly not possible, because the data sheet doesn't specify it?
wrote:
Steve Wilson
May 31, 2014, 12:09:50 AM5/31/14
to
John Larkin <jla...@highlandtechnology.com> wrote:
> Clearly not possible, because the data sheet doesn't specify it?
No, because there is nothing on the datasheet to indicate the input voltage
> Clearly not possible, because the data sheet doesn't specify it?
or VCC where the input bias current is measured.
Easy to prove. Go ahead and measure it.
Steve Wilson
May 31, 2014, 12:30:10 AM5/31/14
to
Steve Wilson <no...@nospam.com> wrote:
> John Larkin <jla...@highlandtechnology.com> wrote:
>> Clearly not possible, because the data sheet doesn't specify it?
> No, because there is nothing on the datasheet to indicate the input
> voltage or VCC where the input bias current is measured.
If it were possible to set the input bias current to zero by adjusting the
> John Larkin <jla...@highlandtechnology.com> wrote:
>> Clearly not possible, because the data sheet doesn't specify it?
> No, because there is nothing on the datasheet to indicate the input
> voltage or VCC where the input bias current is measured.
suppply voltages, they would certainly make a very big deal of it and show
how to do it. It would be under the title "Adjusting The Input Bias Current
To Zero", or something similar. There would be plots showing the input bias
current vs time and temperature.
SED would have a ball posting articles on low current measurements. There
would be a race to see who could get down to 1 electron per second.
Keithley would go nuts. They would have to redesign all their femtoamp
current meters and rewrite all their low current measurement articles, such
as "Counting Electrons: How to measure currents in the attoampere range",
http://www.keithley.com/data?asset=50390
whit3rd
May 31, 2014, 12:31:24 AM5/31/14
to
On Friday, May 30, 2014 4:06:52 PM UTC-7, Steve Wilson wrote:
> John Larkin <jla...@highlandtechnology.com> wrote:
> > You could get lucky and have the Vcc and Vss ESD diode leakages
> > cancel.
> > I guess you could trim the supplies to *make* them cancel.
...
> John Larkin <jla...@highlandtechnology.com> wrote:
> > You could get lucky and have the Vcc and Vss ESD diode leakages
> > cancel.
> > I guess you could trim the supplies to *make* them cancel.
> Simply set it up to measure the input bias current, then adjust the supply
> voltages to see if the bias current can be switched to the opposite
> polarity. If so, it has to go through zero somewhere.
There are other possible behaviors; if the output is somehow thermally
> voltages to see if the bias current can be switched to the opposite
> polarity. If so, it has to go through zero somewhere.
coupled to the leakage on the inputs, it could have positive feedback.
So, you could trim the leakage near zero, only to have it snap from
one rail to the other. There might be no stable equilibrium.
A dual FET used as preamplifier, with a standard op amp behind it,
was the old solution. If duals weren't at-or-near unavailable, it'd still work.
John Larkin
May 31, 2014, 12:38:43 AM5/31/14
to
On Fri, 30 May 2014 16:42:11 -0700 (PDT), dagmarg...@yahoo.com wrote:
>On Friday, May 30, 2014 5:46:07 PM UTC-4, jeroen Belleman wrote:
>> On 30/05/14 23:09, dagmargoo...@yahoo.com wrote:
>> > On Friday, May 30, 2014 5:01:26 PM UTC-4, Steve Wilson wrote:
>> >> bonesis...@gmail.com wrote:
>> >
>> >>> I want to make measurements on pyroelectric crystals but I think the
>> >>> input impedance of a typical dmm is too low (10 Megaohm or so). Any
>> >>> suggestions on a simple electrometer adapter for a dmm? I can't
>> >>> afford a new electrometer so I was hoping I could make a simple
>> >>> circuit to add to a dmm for these measurements.
>> >>
>> >> Use LMC6482 as a voltage follower. $2.08 Digi-Key, Newark, etc
>> >>
>> >> 20fA input bias, RRIO, 15V VCC, 1 ?V/?C Input Offset Tempco, 1.5 MHz GBW
>On Friday, May 30, 2014 5:46:07 PM UTC-4, jeroen Belleman wrote:
>> On 30/05/14 23:09, dagmargoo...@yahoo.com wrote:
>> > On Friday, May 30, 2014 5:01:26 PM UTC-4, Steve Wilson wrote:
>> >> bonesis...@gmail.com wrote:
>> >
>> >>> I want to make measurements on pyroelectric crystals but I think the
>> >>> input impedance of a typical dmm is too low (10 Megaohm or so). Any
>> >>> suggestions on a simple electrometer adapter for a dmm? I can't
>> >>> afford a new electrometer so I was hoping I could make a simple
>> >>> circuit to add to a dmm for these measurements.
>> >>
>> >> Use LMC6482 as a voltage follower. $2.08 Digi-Key, Newark, etc
>> >>
>> >>
>> >> Note 13: Guaranteed limits are dictated by tester limitations and not
>> >> device performance. Actual performance is reflected in the typical value.
>> >>
>> >> http://octopart.com/partsearch#!?q=LMC6482
>> >>
>> >> Might add a 10k from output to ground to provide a load for the output
>> >> stage.
>> >
>> > Beat me to it. The LMC662 is another choice (cheap), and the LMP7721 (3pA, expensive).
>> >
>> > http://www.ti.com/product/LMP7721
>> >
>> > Some articles for our OP:
>> > http://electronicdesign.com/test-amp-measurement/whats-all-femtoampere-stuff-anyhow
>> > http://www.edn.com/design/analog/4327165/Achieve-femtoampere-leakage-in-surface-mount-op-amp-layouts
>> >
>>
>>
>> I've used an LPC661, which turned out to have an amazing 170aA of
>> bias current. Maybe I just got lucky, because that's *much* better
>> than the datasheet says it is.
>
>Very impressive. I remember a Bob Pease article somewhere about just that,
>that the ib was lower, caused by the protection diodes, and spec set by
>testing throughput.
>
>The LMC6001 was, I think, tested for lower ib.
Think about an eprom floating gate. It must leak well under an electron per
>> >> Note 13: Guaranteed limits are dictated by tester limitations and not
>> >> device performance. Actual performance is reflected in the typical value.
>> >>
>> >> http://octopart.com/partsearch#!?q=LMC6482
>> >>
>> >> Might add a 10k from output to ground to provide a load for the output
>> >> stage.
>> >
>> > Beat me to it. The LMC662 is another choice (cheap), and the LMP7721 (3pA, expensive).
>> >
>> > http://www.ti.com/product/LMP7721
>> >
>> > Some articles for our OP:
>> > http://electronicdesign.com/test-amp-measurement/whats-all-femtoampere-stuff-anyhow
>> > http://www.edn.com/design/analog/4327165/Achieve-femtoampere-leakage-in-surface-mount-op-amp-layouts
>> >
>>
>>
>> I've used an LPC661, which turned out to have an amazing 170aA of
>> bias current. Maybe I just got lucky, because that's *much* better
>> than the datasheet says it is.
>
>Very impressive. I remember a Bob Pease article somewhere about just that,
>that the ib was lower, caused by the protection diodes, and spec set by
>testing throughput.
>
>The LMC6001 was, I think, tested for lower ib.
hour.
--
John Larkin Highland Technology Inc
www.highlandtechnology.com jlarkin at highlandtechnology dot com
Precision electronic instrumentation
John Larkin
May 31, 2014, 12:45:02 AM5/31/14
to
On Fri, 30 May 2014 21:31:24 -0700 (PDT), whit3rd <whi...@gmail.com> wrote:
>On Friday, May 30, 2014 4:06:52 PM UTC-7, Steve Wilson wrote:
>> John Larkin <jla...@highlandtechnology.com> wrote:
>
>> > You could get lucky and have the Vcc and Vss ESD diode leakages
>> > cancel.
>
>> > I guess you could trim the supplies to *make* them cancel.
>...
>> Simply set it up to measure the input bias current, then adjust the supply
>> voltages to see if the bias current can be switched to the opposite
>> polarity. If so, it has to go through zero somewhere.
>
>There are other possible behaviors; if the output is somehow thermally
>coupled to the leakage on the inputs, it could have positive feedback.
>So, you could trim the leakage near zero, only to have it snap from
>one rail to the other. There might be no stable equilibrium.
The leakage currents will be in the fA range, so the diode power dissipations
>On Friday, May 30, 2014 4:06:52 PM UTC-7, Steve Wilson wrote:
>> John Larkin <jla...@highlandtechnology.com> wrote:
>
>> > You could get lucky and have the Vcc and Vss ESD diode leakages
>> > cancel.
>
>> > I guess you could trim the supplies to *make* them cancel.
>...
>> Simply set it up to measure the input bias current, then adjust the supply
>> voltages to see if the bias current can be switched to the opposite
>> polarity. If so, it has to go through zero somewhere.
>
>There are other possible behaviors; if the output is somehow thermally
>coupled to the leakage on the inputs, it could have positive feedback.
>So, you could trim the leakage near zero, only to have it snap from
>one rail to the other. There might be no stable equilibrium.
will be in the fW range. That will raise the diode temperatures by nanokelvins.
Not much chance of thermal runaway.
John Larkin
May 31, 2014, 12:53:16 AM5/31/14
to
On Sat, 31 May 2014 04:30:10 GMT, Steve Wilson <no...@nospam.com> wrote:
>Steve Wilson <no...@nospam.com> wrote:
>
>> John Larkin <jla...@highlandtechnology.com> wrote:
>
>>> Clearly not possible, because the data sheet doesn't specify it?
>
>> No, because there is nothing on the datasheet to indicate the input
>> voltage or VCC where the input bias current is measured.
>
>If it were possible to set the input bias current to zero by adjusting the
>suppply voltages, they would certainly make a very big deal of it and show
>how to do it.
It must be possible. At some positive input voltage, the upper ESD diode
>Steve Wilson <no...@nospam.com> wrote:
>
>> John Larkin <jla...@highlandtechnology.com> wrote:
>
>>> Clearly not possible, because the data sheet doesn't specify it?
>
>> No, because there is nothing on the datasheet to indicate the input
>> voltage or VCC where the input bias current is measured.
>
>If it were possible to set the input bias current to zero by adjusting the
>suppply voltages, they would certainly make a very big deal of it and show
>how to do it.
conducts. At some negative voltage, the lower ESD diode conducts. There is
obviously a zero current point somewhere between. Question is, where?
Of course, nobody would guarantee this for a $1 opamp, and certainly wouldn't
test for it.
One easy test would be to make an opamp follower with the + input floating. Turn
it on and see where the output settles.
Steve Wilson
May 31, 2014, 12:54:11 AM5/31/14
to
whit3rd <whi...@gmail.com> wrote:
> There are other possible behaviors; if the output is somehow
> thermally coupled to the leakage on the inputs, it could have
> positive feedback. So, you could trim the leakage near zero, only to
> have it snap from one rail to the other. There might be no stable
> equilibrium.
Nah,if Bob Pease designed it, he would arrange the thermal coupling to give
> There are other possible behaviors; if the output is somehow
> thermally coupled to the leakage on the inputs, it could have
> positive feedback. So, you could trim the leakage near zero, only to
> have it snap from one rail to the other. There might be no stable
> equilibrium.
negative feedback. He discussed thermal coupling in one of his articles.
Just for reference, www.introni.it has a complete set of Bob Pease articles
including comments. The site is in Italian and I can't find the page that
refers to the pdfs. However the pdf urls are here. They are around 15mB
each:
http://www.introni.it/pdf/Bob%20Pease%20Lab%20Notes%20Part%201.pdf
http://www.introni.it/pdf/Bob%20Pease%20Lab%20Notes%20Part%202.pdf
http://www.introni.it/pdf/Bob%20Pease%20Lab%20Notes%20Part%203.pdf
http://www.introni.it/pdf/Bob%20Pease%20Lab%20Notes%20Part%204.pdf
http://www.introni.it/pdf/Bob%20Pease%20Lab%20Notes%20Part%205.pdf
http://www.introni.it/pdf/Bob%20Pease%20Lab%20Notes%20Part%206.pdf
http://www.introni.it/pdf/Bob%20Pease%20Lab%20Notes%20Part%207.pdf
http://www.introni.it/pdf/Bob%20Pease%20Lab%20Notes%20Part%208.pdf
http://www.introni.it/pdf/Bob%20Pease%20Lab%20Notes%20Part%209.pdf
I haven't had a chance to go through the articles and find the one on
thermal coupling.
Steve Wilson
May 31, 2014, 1:03:49 AM5/31/14
to
John Larkin <jjla...@highNOTlandTHIStechnologyPART.com> wrote:
> On Fri, 30 May 2014 21:31:24 -0700 (PDT), whit3rd <whi...@gmail.com>
> wrote:
>
>>There are other possible behaviors; if the output is somehow
>>thermally coupled to the leakage on the inputs, it could have
>>positive feedback. So, you could trim the leakage near zero, only to
>>have it snap from one rail to the other. There might be no stable
>>equilibrium.
>
> The leakage currents will be in the fA range, so the diode power
> dissipations will be in the fW range. That will raise the diode
> temperatures by nanokelvins. Not much chance of thermal runaway.
whit3rd is referring to the thermal coupling from the output stage back to
> On Fri, 30 May 2014 21:31:24 -0700 (PDT), whit3rd <whi...@gmail.com>
> wrote:
>
>>There are other possible behaviors; if the output is somehow
>>thermally coupled to the leakage on the inputs, it could have
>>positive feedback. So, you could trim the leakage near zero, only to
>>have it snap from one rail to the other. There might be no stable
>>equilibrium.
>
> The leakage currents will be in the fA range, so the diode power
> dissipations will be in the fW range. That will raise the diode
> temperatures by nanokelvins. Not much chance of thermal runaway.
the input circuits. The LMC660 can supply 40mA with 15V Vcc. If it
delivered 40 mA to a 187 ohm load, it would dissipate 300 milliwatts in the
output stage.
(7.5^2) / 187 = 0.300
That would definitely cause the chip to get hot.
John Larkin
May 31, 2014, 1:07:04 AM5/31/14
to
Steve Wilson
May 31, 2014, 1:12:30 AM5/31/14
to
John Larkin <jjla...@highNOTlandTHIStechnologyPART.com> wrote:
> On Sat, 31 May 2014 04:30:10 GMT, Steve Wilson <no...@nospam.com>
> wrote:
>
>>Steve Wilson <no...@nospam.com> wrote:
>>
>>> John Larkin <jla...@highlandtechnology.com> wrote:
>>
>>>> Clearly not possible, because the data sheet doesn't specify it?
>>
>>> No, because there is nothing on the datasheet to indicate the input
>>> voltage or VCC where the input bias current is measured.
>>
>>If it were possible to set the input bias current to zero by
>>adjusting the suppply voltages, they would certainly make a very big
>>deal of it and show how to do it.
>
> It must be possible. At some positive input voltage, the upper ESD
> diode conducts. At some negative voltage, the lower ESD diode
> conducts. There is obviously a zero current point somewhere between.
> Question is, where?
Bit of a flaw in the logic. When the top diode is conducting in the
> On Sat, 31 May 2014 04:30:10 GMT, Steve Wilson <no...@nospam.com>
> wrote:
>
>>Steve Wilson <no...@nospam.com> wrote:
>>
>>> John Larkin <jla...@highlandtechnology.com> wrote:
>>
>>>> Clearly not possible, because the data sheet doesn't specify it?
>>
>>> No, because there is nothing on the datasheet to indicate the input
>>> voltage or VCC where the input bias current is measured.
>>
>>If it were possible to set the input bias current to zero by
>>adjusting the suppply voltages, they would certainly make a very big
>>deal of it and show how to do it.
>
> It must be possible. At some positive input voltage, the upper ESD
> diode conducts. At some negative voltage, the lower ESD diode
> conducts. There is obviously a zero current point somewhere between.
> Question is, where?
forward direction, the bottom diode is conducting leakage current in the
reverse direction, and vice-versa. So there is no zero current point
somewhere between.
You already have the setup to measure leakage current. It is in an
aluminum box you posted to dropbox. Instead of theorizing, why not
measure it?
Steve Wilson
May 31, 2014, 1:15:22 AM5/31/14
to
John Larkin <jjla...@highNOTlandTHIStechnologyPART.com> wrote:
>>whit3rd is referring to the thermal coupling from the output stage
>>back to the input circuits. The LMC660 can supply 40mA with 15V Vcc.
>>If it delivered 40 mA to a 187 ohm load, it would dissipate 300
>>milliwatts in the output stage.
>>
>>(7.5^2) / 187 = 0.300
>>
>>That would definitely cause the chip to get hot.
> Well, if you want low leakage, don't load the opamp hard. Don't load
> it at all.
Obviously, that would be the way to go. But you missed whit3rd's statement
>>whit3rd is referring to the thermal coupling from the output stage
>>back to the input circuits. The LMC660 can supply 40mA with 15V Vcc.
>>If it delivered 40 mA to a 187 ohm load, it would dissipate 300
>>milliwatts in the output stage.
>>
>>(7.5^2) / 187 = 0.300
>>
>>That would definitely cause the chip to get hot.
> Well, if you want low leakage, don't load the opamp hard. Don't load
> it at all.
about thermal coupling from the output stage back to the input.
Steve Wilson
May 31, 2014, 1:43:12 AM5/31/14
to
Steve Wilson <no...@nospam.com> wrote:
> Just for reference, www.introni.it has a complete set of Bob Pease
> articles including comments. The site is in Italian and I can't find
> the page that refers to the pdfs. However the pdf urls are here. They
> are around 15mB each:
>
> http://www.introni.it/pdf/Bob%20Pease%20Lab%20Notes%20Part%201.pdf
> http://www.introni.it/pdf/Bob%20Pease%20Lab%20Notes%20Part%202.pdf
> http://www.introni.it/pdf/Bob%20Pease%20Lab%20Notes%20Part%203.pdf
> http://www.introni.it/pdf/Bob%20Pease%20Lab%20Notes%20Part%204.pdf
> http://www.introni.it/pdf/Bob%20Pease%20Lab%20Notes%20Part%205.pdf
> http://www.introni.it/pdf/Bob%20Pease%20Lab%20Notes%20Part%206.pdf
> http://www.introni.it/pdf/Bob%20Pease%20Lab%20Notes%20Part%207.pdf
> http://www.introni.it/pdf/Bob%20Pease%20Lab%20Notes%20Part%208.pdf
> http://www.introni.it/pdf/Bob%20Pease%20Lab%20Notes%20Part%209.pdf
>
> I haven't had a chance to go through the articles and find the one on
> thermal coupling.
OK, I found the Bob Pease page. It's at
> Just for reference, www.introni.it has a complete set of Bob Pease
> articles including comments. The site is in Italian and I can't find
> the page that refers to the pdfs. However the pdf urls are here. They
> are around 15mB each:
>
> http://www.introni.it/pdf/Bob%20Pease%20Lab%20Notes%20Part%201.pdf
> http://www.introni.it/pdf/Bob%20Pease%20Lab%20Notes%20Part%202.pdf
> http://www.introni.it/pdf/Bob%20Pease%20Lab%20Notes%20Part%203.pdf
> http://www.introni.it/pdf/Bob%20Pease%20Lab%20Notes%20Part%204.pdf
> http://www.introni.it/pdf/Bob%20Pease%20Lab%20Notes%20Part%205.pdf
> http://www.introni.it/pdf/Bob%20Pease%20Lab%20Notes%20Part%206.pdf
> http://www.introni.it/pdf/Bob%20Pease%20Lab%20Notes%20Part%207.pdf
> http://www.introni.it/pdf/Bob%20Pease%20Lab%20Notes%20Part%208.pdf
> http://www.introni.it/pdf/Bob%20Pease%20Lab%20Notes%20Part%209.pdf
>
> I haven't had a chance to go through the articles and find the one on
> thermal coupling.
http://www.introni.it/riviste_bob_pease.html
There is a series of chapters on common mode which is not included in the
above list. It is "Bob Pease Lab Notes 2005", at
http://www.introni.it/pdf/Bob%20Pease%20Lab%20Notes%202005.pdf
The referring page is
http://www.introni.it/riviste.html
It has a wealth of information, including Jim Williams Lab Notes at
http://www.introni.it/riviste_jim_williams.html
Steve Wilson
May 31, 2014, 2:04:46 AM5/31/14
to
Steve Wilson <no...@nospam.com> wrote:
[...]
More Bob Pease stuff. Here's an article from 1980 showing how to make
stable millivoltmeters when the inputs are open, constant voltage drop
picoammeters, and a bunch of other stuff. It is written around the LM11,
but the same techniques could be applied to the LMC660/LMC64482 series
parts. This is the first article I have come across that features Bob
Pease and Bob Widlar as coauthors.
Applying a New Precision Op Amp
National Semiconductor
Application Note 242
Robert J. Widlar
Bob Pease
Mineo Yamatake
April 1980
Abstract: A new bipolar op amp design has advanced the state of the
art by reducing offset voltage and bias current errors. Its
characteristics are described here, indicating an ultimate input
resolution of 10 uV and 1 pA under laboratory conditions. Practical
circuits for making voltmeters, ammeters, differential
instrumentation amplifiers and a variety of other designs that can
benefit from the improved performance are covered in detail.
Methods of coupling the new device to existing fast amplifiers to
take advantage of the best characteristics of both, even in follower
applications, are explored.
http://www.ti.com/lit/an/snoa818/snoa818.pdf
[...]
More Bob Pease stuff. Here's an article from 1980 showing how to make
stable millivoltmeters when the inputs are open, constant voltage drop
picoammeters, and a bunch of other stuff. It is written around the LM11,
but the same techniques could be applied to the LMC660/LMC64482 series
parts. This is the first article I have come across that features Bob
Pease and Bob Widlar as coauthors.
Applying a New Precision Op Amp
National Semiconductor
Application Note 242
Robert J. Widlar
Bob Pease
Mineo Yamatake
April 1980
Abstract: A new bipolar op amp design has advanced the state of the
art by reducing offset voltage and bias current errors. Its
characteristics are described here, indicating an ultimate input
resolution of 10 uV and 1 pA under laboratory conditions. Practical
circuits for making voltmeters, ammeters, differential
instrumentation amplifiers and a variety of other designs that can
benefit from the improved performance are covered in detail.
Methods of coupling the new device to existing fast amplifiers to
take advantage of the best characteristics of both, even in follower
applications, are explored.
http://www.ti.com/lit/an/snoa818/snoa818.pdf
Tim Williams
May 31, 2014, 3:39:46 AM5/31/14
to
"John Larkin" <jjla...@highNOTlandTHIStechnologyPART.com> wrote in
message news:oinio9tep6qd9bt6b...@4ax.com...
Specifically, a linear difference in leakage currents corresponds to an
exponential difference in voltages. So, trying to find the null might
require +1/-30V supplies, or something like that.
All CMOS chips, when left to their own devices, will not actually have
their inputs floating willy-nilly. They'll eventually get near a rail,
because the difference in currents is not quite zero. It may require
elevated temperature for this to become obvious (if the leakages are well
matched already).
(And of course, the practical significance of this is small, because in
the presence of not very much AC electric field, CMOS inputs can be biased
back into the linear region where logic levels are unknown and current
draw spikes.)
The only possible way to get practical supply voltages with balanced
leakage is to thermally control one or both diodes with some sort of
servo. Ironically, that's a hard one to do monolithically, unless you can
cantilever your diode out on a MEMS structure.
Tim
--
Seven Transistor Labs
Electrical Engineering Consultation
Website: http://seventransistorlabs.com
message news:oinio9tep6qd9bt6b...@4ax.com...
> It must be possible. At some positive input voltage, the upper ESD diode
> conducts. At some negative voltage, the lower ESD diode conducts. There
> is
> obviously a zero current point somewhere between. Question is, where?
The statement is true, but the practical significance is lacking.
> conducts. At some negative voltage, the lower ESD diode conducts. There
> is
> obviously a zero current point somewhere between. Question is, where?
Specifically, a linear difference in leakage currents corresponds to an
exponential difference in voltages. So, trying to find the null might
require +1/-30V supplies, or something like that.
All CMOS chips, when left to their own devices, will not actually have
their inputs floating willy-nilly. They'll eventually get near a rail,
because the difference in currents is not quite zero. It may require
elevated temperature for this to become obvious (if the leakages are well
matched already).
(And of course, the practical significance of this is small, because in
the presence of not very much AC electric field, CMOS inputs can be biased
back into the linear region where logic levels are unknown and current
draw spikes.)
The only possible way to get practical supply voltages with balanced
leakage is to thermally control one or both diodes with some sort of
servo. Ironically, that's a hard one to do monolithically, unless you can
cantilever your diode out on a MEMS structure.
Tim
--
Seven Transistor Labs
Electrical Engineering Consultation
Website: http://seventransistorlabs.com
Steve Wilson
May 31, 2014, 8:03:23 AM5/31/14
to
"Tim Williams" <tmor...@charter.net> wrote:
> "John Larkin" <jjla...@highNOTlandTHIStechnologyPART.com> wrote in
> message news:oinio9tep6qd9bt6b...@4ax.com...
>> It must be possible. At some positive input voltage, the upper ESD
>> diode conducts. At some negative voltage, the lower ESD diode
>> conducts. There is obviously a zero current point somewhere between.
>> Question is, where?
>
> The statement is true, but the practical significance is lacking.
The statement ignores diode leakage.
> "John Larkin" <jjla...@highNOTlandTHIStechnologyPART.com> wrote in
> message news:oinio9tep6qd9bt6b...@4ax.com...
>> It must be possible. At some positive input voltage, the upper ESD
>> diode conducts. At some negative voltage, the lower ESD diode
>> conducts. There is obviously a zero current point somewhere between.
>> Question is, where?
>
> The statement is true, but the practical significance is lacking.
> Tim
Tim Williams
May 31, 2014, 8:28:11 AM5/31/14
to
"Steve Wilson" <no...@nospam.com> wrote in message
news:XnsA33E51F4E3...@69.16.179.20...
Diode current is a continuous function of diode voltage. There is no
magical discontinuity in the physical world where current switches over
from what we call "conduction" to what we call "leakage". Those are just
conventional labels for I > 0 and I < 0. Even if there were such a
change, it wouldn't make any difference to the present case (assuming the
laws of thermodynamics are present).
news:XnsA33E51F4E3...@69.16.179.20...
>>> It must be possible. At some positive input voltage, the upper ESD
>>> diode conducts. At some negative voltage, the lower ESD diode
>>> conducts. There is obviously a zero current point somewhere between.
>>> Question is, where?
>>
>> The statement is true, but the practical significance is lacking.
>
> The statement ignores diode leakage.
Reiterating ones' ignorance does not a statement disprove. :-)
>>> diode conducts. At some negative voltage, the lower ESD diode
>>> conducts. There is obviously a zero current point somewhere between.
>>> Question is, where?
>>
>> The statement is true, but the practical significance is lacking.
>
> The statement ignores diode leakage.
Diode current is a continuous function of diode voltage. There is no
magical discontinuity in the physical world where current switches over
from what we call "conduction" to what we call "leakage". Those are just
conventional labels for I > 0 and I < 0. Even if there were such a
change, it wouldn't make any difference to the present case (assuming the
laws of thermodynamics are present).
Steve Wilson
May 31, 2014, 9:07:12 AM5/31/14
to
"Tim Williams" <tmor...@charter.net> wrote:
> "Steve Wilson" <no...@nospam.com> wrote in message
> news:XnsA33E51F4E3...@69.16.179.20...
>>>> It must be possible. At some positive input voltage, the upper ESD
>>>> diode conducts. At some negative voltage, the lower ESD diode
>>>> conducts. There is obviously a zero current point somewhere
>>>> between. Question is, where?
>>>
>>> The statement is true, but the practical significance is lacking.
>>
>> The statement ignores diode leakage.
>
> Reiterating ones' ignorance does not a statement disprove. :-)
>
> Diode current is a continuous function of diode voltage. There is no
> magical discontinuity in the physical world where current switches
> over from what we call "conduction" to what we call "leakage". Those
> are just conventional labels for I > 0 and I < 0. Even if there were
> such a change, it wouldn't make any difference to the present case
> (assuming the laws of thermodynamics are present).
>
> Tim
You exclude zero in your paragraph.
> "Steve Wilson" <no...@nospam.com> wrote in message
> news:XnsA33E51F4E3...@69.16.179.20...
>>>> It must be possible. At some positive input voltage, the upper ESD
>>>> diode conducts. At some negative voltage, the lower ESD diode
>>>> conducts. There is obviously a zero current point somewhere
>>>> between. Question is, where?
>>>
>>> The statement is true, but the practical significance is lacking.
>>
>> The statement ignores diode leakage.
>
> Reiterating ones' ignorance does not a statement disprove. :-)
>
> Diode current is a continuous function of diode voltage. There is no
> magical discontinuity in the physical world where current switches
> over from what we call "conduction" to what we call "leakage". Those
> are just conventional labels for I > 0 and I < 0. Even if there were
> such a change, it wouldn't make any difference to the present case
> (assuming the laws of thermodynamics are present).
>
> Tim
There is a magical discontinuity at zero. In the forward direction, the
current increases exponentially with voltage according to the diode
equation:
http://en.wikipedia.org/wiki/Diode_modelling
In the reverse direction, the leakage current is virtually independent of
voltage up to the point where the diode breaks down:
http://en.wikipedia.org/wiki/Reverse_leakage_current
These are two different mechanisms.
The statement only talks about forward conduction. If the leakage was
zero, the zero current point would be anywhere between VCC and Gnd.
Steve Wilson
May 31, 2014, 9:24:53 AM5/31/14
to
Steve Wilson <no...@nospam.com> wrote:
> There is a magical discontinuity at zero. In the forward direction,
> the current increases exponentially with voltage according to the
> diode equation:
>
> http://en.wikipedia.org/wiki/Diode_modelling
>
> In the reverse direction, the leakage current is virtually independent
> of voltage up to the point where the diode breaks down:
>
> http://en.wikipedia.org/wiki/Reverse_leakage_current
>
> These are two different mechanisms.
And this explains why you cannot set the bias current to zero in the
> There is a magical discontinuity at zero. In the forward direction,
> the current increases exponentially with voltage according to the
> diode equation:
>
> http://en.wikipedia.org/wiki/Diode_modelling
>
> In the reverse direction, the leakage current is virtually independent
> of voltage up to the point where the diode breaks down:
>
> http://en.wikipedia.org/wiki/Reverse_leakage_current
>
> These are two different mechanisms.
LMC660/LMC64482 series parts. The diode leakages for the top and bottom
diodes add together. However, they act as current sources, so they are
independent of voltage.
This means changing VCC and VEE or changing the voltage at the input to
the op amp will have no effect on the input bias current.
The reason this is true is because the actual bias current into the
mosfet junction is close to zero, as Paul Grohe stated. So the only
currents involved are the leakage currents into the protection diodes.
I have answered the questions I asked earlier. The mechanism that creates
the input bias current is diode leakage. The currents flow to VCC and
gnd.
Tim Williams
May 31, 2014, 10:05:30 AM5/31/14
to
"Steve Wilson" <no...@nospam.com> wrote in message
news:XnsA33E5CC636...@69.16.179.21...
> There is a magical discontinuity at zero.
Can you please show me where this occurs in the following equation? :-)
I = Is * (exp(Vf/Vth) + 1)
George Herold
May 31, 2014, 10:32:10 AM5/31/14
to
On Saturday, May 31, 2014 12:53:16 AM UTC-4, John Larkin wrote:
> On Sat, 31 May 2014 04:30:10 GMT, Steve Wilson <no...@nospam.com> wrote:
>
> >Steve Wilson <no...@nospam.com> wrote:
> >> John Larkin <jla...@highlandtechnology.com> wrote:
> >>> Clearly not possible, because the data sheet doesn't specify it?
> >> No, because there is nothing on the datasheet to indicate the input
> >> voltage or VCC where the input bias current is measured.
> >If it were possible to set the input bias current to zero by adjusting the
> >suppply voltages, they would certainly make a very big deal of it and show
> >how to do it.
>
> It must be possible. At some positive input voltage, the upper ESD diode
> conducts. At some negative voltage, the lower ESD diode conducts. There is
> obviously a zero current point somewhere between. Question is, where?
Isn't it just the reverse leakage (saturation) current of a diode?
> On Sat, 31 May 2014 04:30:10 GMT, Steve Wilson <no...@nospam.com> wrote:
>
> >Steve Wilson <no...@nospam.com> wrote:
> >> John Larkin <jla...@highlandtechnology.com> wrote:
> >>> Clearly not possible, because the data sheet doesn't specify it?
> >> No, because there is nothing on the datasheet to indicate the input
> >> voltage or VCC where the input bias current is measured.
> >If it were possible to set the input bias current to zero by adjusting the
> >suppply voltages, they would certainly make a very big deal of it and show
> >how to do it.
>
> It must be possible. At some positive input voltage, the upper ESD diode
> conducts. At some negative voltage, the lower ESD diode conducts. There is
> obviously a zero current point somewhere between. Question is, where?
And they'd want to make that flat over the whole range.
Right till the doide starts to turn on, and for pA that's a
small voltage... less than 100mV at a guess.
Some silly ideas.
What about temperature?
maybe there's a reasonable temp where the current is minimal.
or perhaps a temperature gradient across the inputs?
You could try balancing the current.
How much does a reversed biased LED leak?
What's a low leakage diode?
George H.
Phil Hobbs
May 31, 2014, 10:43:30 AM5/31/14
to
On 5/31/2014 12:53 AM, John Larkin wrote:
> On Sat, 31 May 2014 04:30:10 GMT, Steve Wilson <no...@nospam.com> wrote:
>
>> Steve Wilson <no...@nospam.com> wrote:
>>
>>> John Larkin <jla...@highlandtechnology.com> wrote:
>>
>>>> Clearly not possible, because the data sheet doesn't specify it?
>>
>>> No, because there is nothing on the datasheet to indicate the input
>>> voltage or VCC where the input bias current is measured.
>>
>> If it were possible to set the input bias current to zero by adjusting the
>> suppply voltages, they would certainly make a very big deal of it and show
>> how to do it.
>
> It must be possible. At some positive input voltage, the upper ESD diode
> conducts. At some negative voltage, the lower ESD diode conducts. There is
> obviously a zero current point somewhere between. Question is, where?
>
> Of course, nobody would guarantee this for a $1 opamp, and certainly wouldn't
> test for it.
>
> One easy test would be to make an opamp follower with the + input floating. Turn
> it on and see where the output settles.
>
>
The gate current of some pHEMTs (e.g. the SKY65050) changes sign at a
> On Sat, 31 May 2014 04:30:10 GMT, Steve Wilson <no...@nospam.com> wrote:
>
>> Steve Wilson <no...@nospam.com> wrote:
>>
>>> John Larkin <jla...@highlandtechnology.com> wrote:
>>
>>>> Clearly not possible, because the data sheet doesn't specify it?
>>
>>> No, because there is nothing on the datasheet to indicate the input
>>> voltage or VCC where the input bias current is measured.
>>
>> If it were possible to set the input bias current to zero by adjusting the
>> suppply voltages, they would certainly make a very big deal of it and show
>> how to do it.
>
> It must be possible. At some positive input voltage, the upper ESD diode
> conducts. At some negative voltage, the lower ESD diode conducts. There is
> obviously a zero current point somewhere between. Question is, where?
>
> Of course, nobody would guarantee this for a $1 opamp, and certainly wouldn't
> test for it.
>
> One easy test would be to make an opamp follower with the + input floating. Turn
> it on and see where the output settles.
>
>
small negative bias. A disconnected gate will spontaneously go a couple
of hundred mV below the source, and then stop.
Cheers
Phil Hobbs
--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC
Optics, Electro-optics, Photonics, Analog Electronics
160 North State Road #203
Briarcliff Manor NY 10510
hobbs at electrooptical dot net
http://electrooptical.net
Phil Hobbs
May 31, 2014, 10:52:46 AM5/31/14
to
On 5/31/2014 9:24 AM, Steve Wilson wrote:
> Steve Wilson <no...@nospam.com> wrote:
>> There is a magical discontinuity at zero. In the forward
>> direction, the current increases exponentially with voltage
>> according to the diode equation:
>>
>> http://en.wikipedia.org/wiki/Diode_modelling
>>
>> In the reverse direction, the leakage current is virtually
>> independent of voltage up to the point where the diode breaks
>> down:
>>
>> http://en.wikipedia.org/wiki/Reverse_leakage_current
>>
>> These are two different mechanisms.
>
> And this explains why you cannot set the bias current to zero in the
> LMC660/LMC64482 series parts. The diode leakages for the top and
> bottom diodes add together.
How does that work? One is a positive current, and the other is negative.
> Steve Wilson <no...@nospam.com> wrote:
>> There is a magical discontinuity at zero. In the forward
>> direction, the current increases exponentially with voltage
>> according to the diode equation:
>>
>> http://en.wikipedia.org/wiki/Diode_modelling
>>
>> In the reverse direction, the leakage current is virtually
>> independent of voltage up to the point where the diode breaks
>> down:
>>
>> http://en.wikipedia.org/wiki/Reverse_leakage_current
>>
>> These are two different mechanisms.
>
> And this explains why you cannot set the bias current to zero in the
> LMC660/LMC64482 series parts. The diode leakages for the top and
> bottom diodes add together.
> However, they act as current sources, so they are independent of
> voltage.
Second order effects such as bias voltage might very well be enough to
make them match exactly.
>
> This means changing VCC and VEE or changing the voltage at the input
> to the op amp will have no effect on the input bias current.
>
> The reason this is true is because the actual bias current into the
> mosfet junction is close to zero, as Paul Grohe stated. So the only
> currents involved are the leakage currents into the protection
> diodes.
>
> I have answered the questions I asked earlier. The mechanism that
> creates the input bias current is diode leakage. The currents flow to
> VCC and gnd.
>
>
Phil Hobbs
May 31, 2014, 10:53:32 AM5/31/14
to
On 5/31/2014 10:05 AM, Tim Williams wrote:
> "Steve Wilson" <no...@nospam.com> wrote in message
> news:XnsA33E5CC636...@69.16.179.21...
>> There is a magical discontinuity at zero.
>
> Can you please show me where this occurs in the following equation? :-)
>
> I = Is * (exp(Vf/Vth) + 1)
>
> Tim
>
It's minus one. Otherwise you have a magic battery.
> "Steve Wilson" <no...@nospam.com> wrote in message
> news:XnsA33E5CC636...@69.16.179.21...
>> There is a magical discontinuity at zero.
>
> Can you please show me where this occurs in the following equation? :-)
>
> I = Is * (exp(Vf/Vth) + 1)
>
> Tim
>
haitic...@gmail.com
May 31, 2014, 11:37:17 AM5/31/14
to
On Friday, May 30, 2014 4:10:55 PM UTC-4, bonesi...@gmail.com wrote:
> I want to make measurements on pyroelectric crystals but I think the input impedance of a typical dmm is too low (10 Megaohm or so). Any suggestions on a simple electrometer adapter for a dmm? I can't afford a new electrometer so I was hoping I could make a simple circuit to add to a dmm for these measurements.
>
>
>
> Scot
> I want to make measurements on pyroelectric crystals but I think the input impedance of a typical dmm is too low (10 Megaohm or so). Any suggestions on a simple electrometer adapter for a dmm? I can't afford a new electrometer so I was hoping I could make a simple circuit to add to a dmm for these measurements.
>
>
>
If your currents are in the fA range, you are in another regime than a
pyroelectric crystal that puts out more power. Assuming the former, Paul Grohe
and Bob Pease ("whats all this femtoampere stuff anyhow?") have examples of
what you must do to get a measurement. Here is a list of what they do:
-space wiring if possible (TI datasheet actually suggests this.)
-scrub any insulators with solvent. (Grohe has a video on this.)
-use only polystyrene for insulators - epoxy too conductive.
-teflon has charge storage problems - it's too good.
Pease went to a *lot* of trouble. For an easy solution, space wire an electrometer IC from TI. See if the crystal you are using has enough capacitance to hold a signal on the input of the fA leakage IC.
techlib.com has some interesting circuits using the 4117 jfet (obsolete.) But I
think the overall design idea here is that you use an integrating cap as
feedback on the 6001 op amp to persist a transient signal, if you must. And if
you experimental set-up allows continuous current source, it can just trickle
charge a low leakage cap. Measurements in the attoampere range are done this way.
I wouldn't even think of PCB's or insulators though. Too leaky.
jb
haitic...@gmail.com
May 31, 2014, 11:45:58 AM5/31/14
to
Electrometers will pick up fields in the room.
haitic...@gmail.com
May 31, 2014, 11:58:52 AM5/31/14
to
>
> I wouldn't even think of PCB's or insulators though. Too leaky.
>
>
>
> jb
This is a tube version, and is under $100 usually.
jb
haitic...@gmail.com
May 31, 2014, 12:36:28 PM5/31/14
to
And another thought to make things complete: IF you are in the fA regime,
try to avoid an integrating cap as feedback on an op amp. This is because
switches to zero them are leaky. Pease, who was a better circuit than I'll ever
be, used a mechanical push-rod to zero his feedback cap.
Charles Wenzel at techlib.com has a possible way around this. He puts a 2n4117
jfet as an input stage on a capacitance. (est. 5 pF of a ion chamber.) This is
connected to a simple DC amplifier, and read by a picaxe ADC at intervals.
The 4117 has very low gate leakage.
And here's the cute trick: When he wants to zero the capacitor (could be your
xtal), he just brings the source low using the picaxe, and the inbuilt
diode in the jfet just shorts the cap to ground. If your xtal needs more
capacitance, a 5 pF PS to ground might suffice. If, OTOH, your signal is truly
transient, then maybe his scheme could be used to switch (short out) a feedback
cap o an interator.
I built and ran the circuit, which worked fine, except I should have space-wired more of it, so was beset by leakage.
The whole enchilada is on techlib.com
jb
try to avoid an integrating cap as feedback on an op amp. This is because
switches to zero them are leaky. Pease, who was a better circuit than I'll ever
be, used a mechanical push-rod to zero his feedback cap.
Charles Wenzel at techlib.com has a possible way around this. He puts a 2n4117
jfet as an input stage on a capacitance. (est. 5 pF of a ion chamber.) This is
connected to a simple DC amplifier, and read by a picaxe ADC at intervals.
The 4117 has very low gate leakage.
And here's the cute trick: When he wants to zero the capacitor (could be your
xtal), he just brings the source low using the picaxe, and the inbuilt
diode in the jfet just shorts the cap to ground. If your xtal needs more
capacitance, a 5 pF PS to ground might suffice. If, OTOH, your signal is truly
transient, then maybe his scheme could be used to switch (short out) a feedback
cap o an interator.
I built and ran the circuit, which worked fine, except I should have space-wired more of it, so was beset by leakage.
The whole enchilada is on techlib.com
jb
Phil Hobbs
May 31, 2014, 12:48:28 PM5/31/14
to
On 5/31/2014 12:36 PM, haitic...@gmail.com wrote:
> And another thought to make things complete: IF you are in the fA regime,
> try to avoid an integrating cap as feedback on an op amp. This is because
> switches to zero them are leaky. Pease, who was a better circuit than I'll ever
> be, used a mechanical push-rod to zero his feedback cap.
You can do it by briefly inverting the power supplies with a mux and a
> And another thought to make things complete: IF you are in the fA regime,
> try to avoid an integrating cap as feedback on an op amp. This is because
> switches to zero them are leaky. Pease, who was a better circuit than I'll ever
> be, used a mechanical push-rod to zero his feedback cap.
couple of current-limiting resistors. The protection diodes work like a
sampling bridge. I forget whose idea this was--it's brilliant. It's
similar to the one you give below, but better since it doesn't have the
diode drop uncertainty. You don't get back to exactly zero, of course,
so you have to measure the reset value.
>
> Charles Wenzel at techlib.com has a possible way around this. He puts a 2n4117
> jfet as an input stage on a capacitance. (est. 5 pF of a ion chamber.) This is
> connected to a simple DC amplifier, and read by a picaxe ADC at intervals.
>
> The 4117 has very low gate leakage.
>
> And here's the cute trick: When he wants to zero the capacitor (could be your
> xtal), he just brings the source low using the picaxe, and the inbuilt
> diode in the jfet just shorts the cap to ground. If your xtal needs more
> capacitance, a 5 pF PS to ground might suffice. If, OTOH, your signal is truly
> transient, then maybe his scheme could be used to switch (short out) a feedback
> cap o an interator.
>
> I built and ran the circuit, which worked fine, except I should have space-wired more of it, so was beset by leakage.
>
> The whole enchilada is on techlib.com
>
> jb
>
Phil Hobbs
May 31, 2014, 12:51:08 PM5/31/14
to
John Larkin
May 31, 2014, 1:00:15 PM5/31/14
to
could, rosin flux all over everything, and added a bunch of fingerprints on top
for luck.
https://dl.dropboxusercontent.com/u/53724080/Gear/Keithley/Leak_Gloppy_Flux.JPG
https://dl.dropboxusercontent.com/u/53724080/Gear/Keithley/Leak_PCB.JPG
Pins the needle on the 1e14 ohm range.
https://dl.dropboxusercontent.com/u/53724080/Gear/Keithley/Leak_Test.JPG
The killer is water-wash flux. Nasty stuff.
John Larkin
May 31, 2014, 1:04:08 PM5/31/14
to
https://dl.dropboxusercontent.com/u/53724080/Gear/Keithley/Keithley.JPG
https://dl.dropboxusercontent.com/u/53724080/Gear/Keithley/Keithley_610C.pdf
Fred Bartoli
May 31, 2014, 2:04:58 PM5/31/14
to
Le Fri, 30 May 2014 15:30:01 -0700, John Larkin a écrit:
>>I've used an LPC661, which turned out to have an amazing 170aA of bias
>>current. Maybe I just got lucky, because that's *much* better than the
>>datasheet says it is.
>>
>>Jeroen Belleman
>
> You could get lucky and have the Vcc and Vss ESD diode leakages cancel.
>
> I guess you could trim the supplies to *make* them cancel.
I've done this with some CMOS switches to get to the fA level territory.
The fun thing is to automate trimming.
--
Thanks,
Fred.
>>I've used an LPC661, which turned out to have an amazing 170aA of bias
>>current. Maybe I just got lucky, because that's *much* better than the
>>datasheet says it is.
>>
>>Jeroen Belleman
>
> You could get lucky and have the Vcc and Vss ESD diode leakages cancel.
>
> I guess you could trim the supplies to *make* them cancel.
I've done this with some CMOS switches to get to the fA level territory.
The fun thing is to automate trimming.
--
Thanks,
Fred.
bloggs.fred...@gmail.com
May 31, 2014, 2:15:38 PM5/31/14
to
On Friday, May 30, 2014 4:10:55 PM UTC-4, bonesi...@gmail.com wrote:
> I want to make measurements on pyroelectric crystals but I think the input impedance of a typical dmm is too low (10 Megaohm or so). Any suggestions on a simple electrometer adapter for a dmm? I can't afford a new electrometer so I was hoping I could make a simple circuit to add to a dmm for these measurements.
>
>
>
> Scot
You need to learn A LOT MORE about pyroelectric crystal detectors before you even begin to waste time on an amplifier, which are simple and straightforward for the most part.
> I want to make measurements on pyroelectric crystals but I think the input impedance of a typical dmm is too low (10 Megaohm or so). Any suggestions on a simple electrometer adapter for a dmm? I can't afford a new electrometer so I was hoping I could make a simple circuit to add to a dmm for these measurements.
>
>
>
> Scot
From the introductory literature:
"The spontaneous polarization of the detector material induces a net positive
charge on one face of the crystal and a net negative charge on the other. The
amount of charge on the crystal faces is small and is fairly quickly neutralized by
deposition of ions from the surrounding atmosphere or by electrical conduction.
Pyroelectric detectors, therefore, can only be used to detect modulated radiation,
and the modulation must be fast enough that the neutralization of the surface
charge can't keep up. These detectors can be operated in either the voltage
mode or the current mode, and one mode or the other will be more appropriate
depending on the application. In the voltage mode, it is the voltage generated
across the entire pyroelectric crystal that is detected. The incident radiation is
absorbed on the exposed or top surface of the crystal. The thermal energy is
conducted through the crystal to the back or bottom face bringing the entire
crystal to a roughly uniform temperature. The crystal is more or less uniformly
polarized with net positive charge on one crystal face and net negative charge on
the other. Clearly, voltage mode detection can only be used at modulation
frequencies low enough so that the thermal energy has enough time to be
conducted through the crystal before the incident radiation starts back the other
way. Current mode detection is appropriate for high frequency modulation where
the temperature of only a relatively thin layer of the crystal at the exposed
surface is affected by the incident radiation. The modulation frequency is fast
enough that the thermal energy is not conducted into the bulk of the crystal.
Since the absorbed energy is not distributed over the bulk of the crystal, the
temperature excursions of the crystal material immediately adjacent to the
exposed face are maximized as is the charge generated on the exposed face. In
current mode operation, it is the current flowing on and off the electrode on the
exposed face of the crystal that is detected. The temperature at the back surface of the detector and the charge on that surface doesn't change at the high
modulation frequencies used for current mode detection.
In voltage mode operation, the amount of surface charge generated on a
pyroelectric detector by the incident radiation is typically quite small. The
detector is basically a capacitor with a shunt or leakage resistance. A typical
pyroelectric detector will have a capacitance on the order of 10 pF or more and a
shunt resistance typically greater than 1012
Ω. The responsivity of a typical
pyroelectric detector operated in the voltage mode peaks for modulation
frequencies somewhere in the 0.01-0.1 Hz range. It falls off at lower frequencies
due to discharging or neutralization of the surface charge and at higher
frequencies due to the thermal mass of the crystal. A preamplifier, of course,
necessarily provides an additional parallel shunt path for neutralization of the
surface charge, and the resistance of that path must be greater or at least not
smaller than the shunt resistance of the detector itself if it is not to significantly
reduce the detector signal and degrade performance."
Phil Hobbs
May 31, 2014, 3:12:33 PM5/31/14
to
On 5/31/2014 12:38 AM, John Larkin wrote:
> On Fri, 30 May 2014 16:42:11 -0700 (PDT), dagmarg...@yahoo.com wrote:
>
>> On Friday, May 30, 2014 5:46:07 PM UTC-4, jeroen Belleman wrote:
>>> On 30/05/14 23:09, dagmargoo...@yahoo.com wrote:
>>>> On Friday, May 30, 2014 5:01:26 PM UTC-4, Steve Wilson wrote:
>>>>>
>>>>> 20fA input bias, RRIO, 15V VCC, 1 ?V/?C Input Offset Tempco, 1.5 MHz GBW
>>>>>
>>>>> Note 13: Guaranteed limits are dictated by tester limitations and not
>>>>> device performance. Actual performance is reflected in the typical value.
>>>>>
>>>>> http://octopart.com/partsearch#!?q=LMC6482
>>>>>
>>>>> Might add a 10k from output to ground to provide a load for the output
>>>>> stage.
>>>>
>>>> Beat me to it. The LMC662 is another choice (cheap), and the LMP7721 (3pA, expensive).
>>>>
>>>> http://www.ti.com/product/LMP7721
>>>>
>>>> Some articles for our OP:
>>>> http://electronicdesign.com/test-amp-measurement/whats-all-femtoampere-stuff-anyhow
>>>> http://www.edn.com/design/analog/4327165/Achieve-femtoampere-leakage-in-surface-mount-op-amp-layouts
>> that the ib was lower, caused by the protection diodes, and spec set by
>> testing throughput.
>>
>> The LMC6001 was, I think, tested for lower ib.
>
> Think about an eprom floating gate. It must leak well under an electron per
> hour.
>
>
Right around that value, in very round figures. If you assume 10 fF
capacitance per cell and a 3-V gate voltage shift for reliable
forgetting, a 20-year lifetime implies
dq/dt < (3V)/(6E8 s)*1e-14 F* 6.2e18 (e-/C) = 3E-3 electrons/s, or about
1.2 electrons per hour.
> On Fri, 30 May 2014 16:42:11 -0700 (PDT), dagmarg...@yahoo.com wrote:
>
>> On Friday, May 30, 2014 5:46:07 PM UTC-4, jeroen Belleman wrote:
>>> On 30/05/14 23:09, dagmargoo...@yahoo.com wrote:
>>>> On Friday, May 30, 2014 5:01:26 PM UTC-4, Steve Wilson wrote:
>>>>> bonesis...@gmail.com wrote:
>>>>
>>>>>> I want to make measurements on pyroelectric crystals but I think the
>>>>>> input impedance of a typical dmm is too low (10 Megaohm or so). Any
>>>>>> suggestions on a simple electrometer adapter for a dmm? I can't
>>>>>> afford a new electrometer so I was hoping I could make a simple
>>>>>> circuit to add to a dmm for these measurements.
>>>>>
>>>>> Use LMC6482 as a voltage follower. $2.08 Digi-Key, Newark, etc
>>>>
>>>>>> I want to make measurements on pyroelectric crystals but I think the
>>>>>> input impedance of a typical dmm is too low (10 Megaohm or so). Any
>>>>>> suggestions on a simple electrometer adapter for a dmm? I can't
>>>>>> afford a new electrometer so I was hoping I could make a simple
>>>>>> circuit to add to a dmm for these measurements.
>>>>>
>>>>>
>>>>> 20fA input bias, RRIO, 15V VCC, 1 ?V/?C Input Offset Tempco, 1.5 MHz GBW
>>>>>
>>>>> Note 13: Guaranteed limits are dictated by tester limitations and not
>>>>> device performance. Actual performance is reflected in the typical value.
>>>>>
>>>>> http://octopart.com/partsearch#!?q=LMC6482
>>>>>
>>>>> Might add a 10k from output to ground to provide a load for the output
>>>>> stage.
>>>>
>>>> Beat me to it. The LMC662 is another choice (cheap), and the LMP7721 (3pA, expensive).
>>>>
>>>> http://www.ti.com/product/LMP7721
>>>>
>>>> Some articles for our OP:
>>>> http://electronicdesign.com/test-amp-measurement/whats-all-femtoampere-stuff-anyhow
>>>> http://www.edn.com/design/analog/4327165/Achieve-femtoampere-leakage-in-surface-mount-op-amp-layouts
>>>>
>>>
>>>
>>> I've used an LPC661, which turned out to have an amazing 170aA of
>>> bias current. Maybe I just got lucky, because that's *much* better
>>> than the datasheet says it is.
>>
>> Very impressive. I remember a Bob Pease article somewhere about just that,
>>>
>>>
>>> I've used an LPC661, which turned out to have an amazing 170aA of
>>> bias current. Maybe I just got lucky, because that's *much* better
>>> than the datasheet says it is.
>>
>> that the ib was lower, caused by the protection diodes, and spec set by
>> testing throughput.
>>
>> The LMC6001 was, I think, tested for lower ib.
>
> Think about an eprom floating gate. It must leak well under an electron per
> hour.
>
>
Right around that value, in very round figures. If you assume 10 fF
capacitance per cell and a 3-V gate voltage shift for reliable
forgetting, a 20-year lifetime implies
dq/dt < (3V)/(6E8 s)*1e-14 F* 6.2e18 (e-/C) = 3E-3 electrons/s, or about
1.2 electrons per hour.
John Larkin
May 31, 2014, 3:28:50 PM5/31/14
to
I think a cool science project could be done with an eprom, specifically
resolving single-electron leakage. Maybe.
Super-low-leakage opamps might be package dominated. The TO-can versions have
way more bias current than the DIPs. Sadly, the V- pin is almost always next to
the IN+ pin. I guess there's not enough market for a better pinout.
I should quantify the 2N7000 leakage. You can do some fun stuff with a 2N7000, a
battery, a resistor, and an LED. There must be small-signal mosfets, without
gate protection, that have really low leakages.
Phil Hobbs
May 31, 2014, 3:33:07 PM5/31/14
to
right-hand section is much better isolated.
>
> I should quantify the 2N7000 leakage. You can do some fun stuff with
a 2N7000, a
> battery, a resistor, and an LED. There must be small-signal mosfets,
without
> gate protection, that have really low leakages.
>
springs shorting their leads together. Might be fun to measure.
Tim Williams
May 31, 2014, 4:09:43 PM5/31/14
to
"Phil Hobbs" <ho...@electrooptical.net> wrote in message
news:5389ECEC...@electrooptical.net...
Assuming the absence of illumination, anyway. :)
news:5389ECEC...@electrooptical.net...
>> I = Is * (exp(Vf/Vth) + 1)
>>
>>
> It's minus one. Otherwise you have a magic battery.
Oops, a typo :)
Assuming the absence of illumination, anyway. :)
Tim Williams
May 31, 2014, 4:15:18 PM5/31/14
to
"Phil Hobbs" <ho...@electrooptical.net> wrote in message
news:5389EA92...@electrooptical.net...
> The gate current of some pHEMTs (e.g. the SKY65050) changes sign at a
> small negative bias. A disconnected gate will spontaneously go a couple
> of hundred mV below the source, and then stop.
Just like the olden days ;)
> small negative bias. A disconnected gate will spontaneously go a couple
> of hundred mV below the source, and then stop.
Typically an open circuit grid goes to -2 or -3V. Negative current is
limited by cathode emission (the high-energy tail of thermally freed
electrons -- thus, should be higher in thoriated tungsten types) and
balanced by grid leakage (emissions from the grid wires being hot by
proximity, and dissipation if forward biased).
Isn't there a similar thing with old fashioned JFETs, dependent on Vds?
Tim Williams
May 31, 2014, 4:18:52 PM5/31/14
to
"John Larkin" <jjla...@highNOTlandTHIStechnologyPART.com> wrote in
message news:0tako91t5mqe14k2a...@4ax.com...
> I think a cool science project could be done with an eprom, specifically
> resolving single-electron leakage. Maybe.
A related subject,
> resolving single-electron leakage. Maybe.
https://www.youtube.com/watch?v=eafhFs-ZOgs
Phil Hobbs
May 31, 2014, 4:20:46 PM5/31/14
to
On 5/31/2014 4:09 PM, Tim Williams wrote:
> "Phil Hobbs" <ho...@electrooptical.net> wrote in message
> news:5389ECEC...@electrooptical.net...
>>> I = Is * (exp(Vf/Vth) + 1)
>>>
>> It's minus one. Otherwise you have a magic battery.
>
> Oops, a typo :)
>
> Assuming the absence of illumination, anyway. :)
>
> Tim
>
Hardly matters--diodes don't even come close to obeying the 'diode
> "Phil Hobbs" <ho...@electrooptical.net> wrote in message
> news:5389ECEC...@electrooptical.net...
>>> I = Is * (exp(Vf/Vth) + 1)
>>>
>> It's minus one. Otherwise you have a magic battery.
>
> Oops, a typo :)
>
> Assuming the absence of illumination, anyway. :)
>
> Tim
>
equation' (AoE has a graph somewhere--the exponential constant
frequently differs from kT/e by a factor as large as 2).
What we laughingly call "high level injection" happens a lot sooner in a
real diode than in a diode-connected transistor.
Phil Hobbs
May 31, 2014, 4:28:57 PM5/31/14
to
On 5/31/2014 4:15 PM, Tim Williams wrote:
> "Phil Hobbs" <ho...@electrooptical.net> wrote in message
> news:5389EA92...@electrooptical.net...
>> The gate current of some pHEMTs (e.g. the SKY65050) changes sign at
>> a small negative bias. A disconnected gate will spontaneously go a
>> couple of hundred mV below the source, and then stop.
>
> Just like the olden days ;)
>
> Typically an open circuit grid goes to -2 or -3V. Negative current
> is limited by cathode emission (the high-energy tail of thermally
> freed electrons -- thus, should be higher in thoriated tungsten
> types) and balanced by grid leakage (emissions from the grid wires
> being hot by proximity, and dissipation if forward biased).
>
> Isn't there a similar thing with old fashioned JFETs, dependent on
> Vds?
Dunno, never measured it. The pHEMT thing was for a client gig a couple
> "Phil Hobbs" <ho...@electrooptical.net> wrote in message
> news:5389EA92...@electrooptical.net...
>> The gate current of some pHEMTs (e.g. the SKY65050) changes sign at
>> a small negative bias. A disconnected gate will spontaneously go a
>> couple of hundred mV below the source, and then stop.
>
> Just like the olden days ;)
>
> Typically an open circuit grid goes to -2 or -3V. Negative current
> is limited by cathode emission (the high-energy tail of thermally
> freed electrons -- thus, should be higher in thoriated tungsten
> types) and balanced by grid leakage (emissions from the grid wires
> being hot by proximity, and dissipation if forward biased).
>
> Isn't there a similar thing with old fashioned JFETs, dependent on
> Vds?
of years back, for a research lab in the Far East. They had a
biochemical application where they really wanted to measure 1 nA in a
100 MHz bandwidth. I pointed out that that was 31 electrons per inverse
bandwidth, so that even in an ideal case they'd never have more than 15
dB SNR, but they still wanted to do it.
I got pretty close--within a factor of three, which was better than I
thought I could do, and learned a lot in the process. Since then I've
used quite a few microwave transistors in lower-frequency applications,
because they're magic. SiGe:C transistors are especially good--a BFP640
has an fT of almost 50 GHz, together with a practically infinite Early
voltage, and a 1/f corner below a kilohertz. You can do a lot with
that, if you're careful about the PCB layout.
John Larkin
May 31, 2014, 4:59:15 PM5/31/14
to
On Sat, 31 May 2014 10:43:30 -0400, Phil Hobbs <ho...@electrooptical.net> wrote:
>On 5/31/2014 12:53 AM, John Larkin wrote:
>> On Sat, 31 May 2014 04:30:10 GMT, Steve Wilson <no...@nospam.com> wrote:
>>
>>> Steve Wilson <no...@nospam.com> wrote:
>>>
>>>> John Larkin <jla...@highlandtechnology.com> wrote:
>>>
>>>>> Clearly not possible, because the data sheet doesn't specify it?
>>>
>>>> No, because there is nothing on the datasheet to indicate the input
>>>> voltage or VCC where the input bias current is measured.
>>>
>>> If it were possible to set the input bias current to zero by adjusting the
>>> suppply voltages, they would certainly make a very big deal of it and show
>>> how to do it.
>>
>> It must be possible. At some positive input voltage, the upper ESD diode
>> conducts. At some negative voltage, the lower ESD diode conducts. There is
>> obviously a zero current point somewhere between. Question is, where?
>>
>> Of course, nobody would guarantee this for a $1 opamp, and certainly wouldn't
>> test for it.
>>
>> One easy test would be to make an opamp follower with the + input floating. Turn
>> it on and see where the output settles.
>>
>>
>
>The gate current of some pHEMTs (e.g. the SKY65050) changes sign at a
>small negative bias. A disconnected gate will spontaneously go a couple
>of hundred mV below the source, and then stop.
>
>
So, with grounded source and positive drain, the gate goes negative? Sort of an
>On 5/31/2014 12:53 AM, John Larkin wrote:
>> On Sat, 31 May 2014 04:30:10 GMT, Steve Wilson <no...@nospam.com> wrote:
>>
>>> Steve Wilson <no...@nospam.com> wrote:
>>>
>>>> John Larkin <jla...@highlandtechnology.com> wrote:
>>>
>>>>> Clearly not possible, because the data sheet doesn't specify it?
>>>
>>>> No, because there is nothing on the datasheet to indicate the input
>>>> voltage or VCC where the input bias current is measured.
>>>
>>> If it were possible to set the input bias current to zero by adjusting the
>>> suppply voltages, they would certainly make a very big deal of it and show
>>> how to do it.
>>
>> It must be possible. At some positive input voltage, the upper ESD diode
>> conducts. At some negative voltage, the lower ESD diode conducts. There is
>> obviously a zero current point somewhere between. Question is, where?
>>
>> Of course, nobody would guarantee this for a $1 opamp, and certainly wouldn't
>> test for it.
>>
>> One easy test would be to make an opamp follower with the + input floating. Turn
>> it on and see where the output settles.
>>
>>
>
>The gate current of some pHEMTs (e.g. the SKY65050) changes sign at a
>small negative bias. A disconnected gate will spontaneously go a couple
>of hundred mV below the source, and then stop.
>
>
electron venturi effect?
Phil Hobbs
May 31, 2014, 5:00:38 PM5/31/14
to
John Larkin
May 31, 2014, 5:04:01 PM5/31/14
to
On Sat, 31 May 2014 15:18:52 -0500, "Tim Williams" <tmor...@charter.net>
wrote:
>"John Larkin" <jjla...@highNOTlandTHIStechnologyPART.com> wrote in
>message news:0tako91t5mqe14k2a...@4ax.com...
>> I think a cool science project could be done with an eprom, specifically
>> resolving single-electron leakage. Maybe.
>
>A related subject,
>https://www.youtube.com/watch?v=eafhFs-ZOgs
>
>Tim
Cool. Super-slow UV imager.
Some of the first CMOS imagers were actually DRAM chips, work done at JPL as I
recall.
wrote:
>"John Larkin" <jjla...@highNOTlandTHIStechnologyPART.com> wrote in
>message news:0tako91t5mqe14k2a...@4ax.com...
>> I think a cool science project could be done with an eprom, specifically
>> resolving single-electron leakage. Maybe.
>
>A related subject,
>https://www.youtube.com/watch?v=eafhFs-ZOgs
>
>Tim
Some of the first CMOS imagers were actually DRAM chips, work done at JPL as I
recall.
John Larkin
May 31, 2014, 5:19:01 PM5/31/14
to
A Pease Puzzler was a bipolar NPN transistor: ground the base, pull up the
emitter, and the collector goes negative. Bob explained it as photons generated
by the b-e zenering and photoelectric generation in the c-b junction. But maybe
he was guessing.
Tom Miller
May 31, 2014, 6:17:22 PM5/31/14
to
"John Larkin" <jjla...@highNOTlandTHIStechnologyPART.com> wrote in message
Is there any info on what NPN device he was using?
tm
Steve Wilson
May 31, 2014, 7:48:53 PM5/31/14
to
Phil Hobbs <ho...@electrooptical.net> wrote:
> On 5/31/2014 9:24 AM, Steve Wilson wrote:
>> Steve Wilson <no...@nospam.com> wrote:
>>> There is a magical discontinuity at zero. In the forward
>>> direction, the current increases exponentially with voltage
>>> according to the diode equation:
>>>
>>> http://en.wikipedia.org/wiki/Diode_modelling
>>>
>>> In the reverse direction, the leakage current is virtually
>>> independent of voltage up to the point where the diode breaks down:
>>>
>>> http://en.wikipedia.org/wiki/Reverse_leakage_current
>>>
>>> These are two different mechanisms.
>>
>> And this explains why you cannot set the bias current to zero in the
>> LMC660/LMC64482 series parts. The diode leakages for the top and
>> bottom diodes add together.
>
> How does that work? One is a positive current, and the other is
> negative.
Exactly. The currents cancel. The degree of matching sets the input bias
> On 5/31/2014 9:24 AM, Steve Wilson wrote:
>> Steve Wilson <no...@nospam.com> wrote:
>>> There is a magical discontinuity at zero. In the forward
>>> direction, the current increases exponentially with voltage
>>> according to the diode equation:
>>>
>>> http://en.wikipedia.org/wiki/Diode_modelling
>>>
>>> In the reverse direction, the leakage current is virtually
>>> independent of voltage up to the point where the diode breaks down:
>>>
>>> http://en.wikipedia.org/wiki/Reverse_leakage_current
>>>
>>> These are two different mechanisms.
>>
>> And this explains why you cannot set the bias current to zero in the
>> LMC660/LMC64482 series parts. The diode leakages for the top and
>> bottom diodes add together.
>
> How does that work? One is a positive current, and the other is
> negative.
current.
>> However, they act as current sources, so they are independent of
>> voltage.
>
> If the diodes are closely matched, the leakage should be close as
> well.
> Second order effects such as bias voltage might very well be enough
> to
> make them match exactly.
may destroy the cancellation.
>> This means changing VCC and VEE or changing the voltage at the input
>> to the op amp will have no effect on the input bias current.
>
> To leading order.
>
>>
>> The reason this is true is because the actual bias current into the
>> mosfet junction is close to zero, as Paul Grohe stated. So the only
>> currents involved are the leakage currents into the protection
>> diodes.
>>
>> I have answered the questions I asked earlier. The mechanism that
>> creates the input bias current is diode leakage. The currents flow to
>> VCC and gnd.
> Cheers
> Phil Hobbs
topics. These confirm your observations and add a bit more light on the
subject. The topics are
1. Removing Input Protection Diodes
2. Diode Leakage Cancellation - Setting Bias Current to Zero
3. Cleaning Plastic Package
4. Layout For Lowest Leakage
5. Latchup
6. Input Protection
7. Testing Op Amps
8. Op Amp as Comparator
===================================================================
1. Removing Input Protection Diodes
Paul Grohe
1/26/03
On Fri, 24 Jan 2003 11:35:06 +0100,
in the newsgroup sci.electronics.design,
al <2...@bill.invalid>
from Technical University Berlin, Germany posted:
> Paul,
>
> how well are the input bias currents in a dual amplifier matched?
Within a few percent - but better than you could do with discrete
devices.
> Can I compensate the temperature dependent bias current with the
> other amplifier?
The "Input Bias Current" is actually the difference in the
leakages of the two ESD protection diodes. The actual MOSFET gate
current is much less.
We actually *tried* to measure the bias current of a device with
the ESD diodes cut out - it was in the "noise" (*if* we did not
accidently "zap" it first - geez they were sensitive!).
A rule of thumb is that the bias current will double every 10�C.
Your best bet is to keep the temperature constant - even if you
have to elevate the ambient temp to keep it constant (like an
XTAL oven).
Cheers,
Paul Grohe
http://tinyurl.com/ktdnm7x
===================================================================
2. Diode Leakage Cancellation - Setting Bias Current to Zero
Paul Grohe
12/31/95
On Sun, 31 Dec 1995 16:14:00 GMT,
in sci.electronics
fra...@euronet.nl (Frank W. van Wensveen)
from Euronet Internet thoughtfully posted:
> I'm looking for someone who can help me with some weird problems I've
> been having with a MOSFET opamp measurement amplifier. I'm trying to
> measure electrostatical charges with a CA3130 opamp. This opamp has
> MOSFETs at the inputs, with internal zener diodes which act as an
> over-voltage protection. (You might want to check the CA3130 data
> sheet for details.)
> The first stage of my amplifier consists of a CA3130 with the
> non-inverting input floating (i.e. not connected to the ground by
> *any* component) and the inverting input connected to the output.
> According to the data sheet, the floating input has an imput impedance
> of about 1.5 Tera-ohm. The opamp is powered by a symetric (sp?) power
> supply at + and - 10 V.
BTW, try Nationals LMC660 (quad) or 662 (dual). Input bias current at
room temperature can be less than 1fA. Call the European number in my
sig for a sample. (yes, I'm biased...pun intended)
> In theory the floating input, screened with a grounded shield to
> prevent it from picking up any charge from the environment, should be
> at ground potential or somewhere near it, and the output of the opamp
> should follow this. (Right?)
> But instead, the output voltage tends towards -Uv if I don't connect
> anything to the non-inverting input pin of the opamp. (Note: the
> non-inverting input pin is *not* connected to anything.) If I connect
> a piece of wire with a length of about 2 inches to the pin and leave
> the other end floating, the output voltage is around zero. If I
> connect a 4 to 6 inch wire to the pin, the output voltage tends
> toward +Uv. Weird but true.
In theory: correct. In practice: Hardly ever.
You are simply seeing the results of the leakages of the input ESD
protection diodes. The ESD diodes are designed so that two diodes
leakage's cancel each other out. The leakage you see is actually the
difference between the two diodes. If your output floats towards V-,
then the diode between the input and V- has slightly higher leakage
than the other one. Temperature changes can also cause this, as well
as contamination or other leakage paths. Make sure the package is
clean, any 'boogers' between the pins can collect dirt and moisture
and cause leakages. This is very noticeable on quads, as they have the
non-inverting pin right next to the power supply leads. Don't use
packages that have been overly handled. Use parts from the shipping
tube and hold them by the ends (not he pins). Don't touch the pins.
Tricks to counter the leakages.
Use an external diode pair (similar to the ESD diodes) and apply a
variable bias on each end to balance out the leakages.
If the leakage is high in one direction, use a single external diode
with variable bias (pot) in the other direction to counteract the
leakages.
If the leakages are very small. Place a single wire "probe" close, but
not in contact with, the input. Apply a bias voltage to the probe to
cancel out the leakages.
In other words, use a controlled leakage to counteract a uncontrolled
leakage.
> I have the following questions:
> 1. What causes the change in output voltage of the opamp if I change
> the length of the wire connected to the floating non-inverting input?
Greater surface area = greater leakages or leakage paths. You could
also be picking up stray RF or ionization. Increasing the surface area
also increases ionization pickup.
> 2. If this is not the best way (quite possible!) to measure
> electrostatical charges, then how can I do it better? Needless to say
> that even an input impedance in the 100 Meg range is much too low to
> measure weak electrostatic charges with any degree of accuracy.
Use a newer generation of op-amp. The RCA is kinda' old. Current
op-amps have input currents of 100fA or less and input R's around 1
Teraohm.
> 3. There is equipment to measure static charge, that is used, among
> other things, for workplace ESD-certification. How do these meters
> work? (Yes, I also asked that question another article.)
Not really sure, I'll keep quiet and let someone else answer that.
> I'd be grateful for any kind of information. If there is anyone who is
> familiar with this kind of measurement technology, please contact me.
> I might be unable to check the newsgroups regularly, as I'm in the
> middle of moving to a new address, so if you post a reply to this,
> please send a copy also to me via Email to fra...@euronet.nl.
> Thanks! - FvW
My experience has been from trying to measure the input bias currents
of our CMOS op-amps (~1 fA or 1x10-15).
Keithley is a great resource for low level measurements.
Check: http://www.keithley.com
Contact them and ask for their "Low Level Measurements Techniques"
handbook. It gives some examples of how to measure charge.
Measuring these minute charges and currents *is* Voodoo science. You
have to think in a completely different way than you are used to.
There are a lot of little things to keep track of at these currents.
IT IS NOT EASY AT ALL!
Good Luck,
Paul Grohe
http://tinyurl.com/p25zsjh
===================================================================
3. Cleaning Plastic Package
Paul Grohe
11/5/97
On Tue, 04 Nov 1997 02:06:41 -0500,
in the newsgroup sci.electronics.design,
Phil Maiorana <pmai...@earthlink.net>
from Biomega Engineering thoughtfully posted:
> You can easily achieve pico amp performance by using an electrometer
grade op-amp
> such as Burr Brown's OPA128: you need a bias current that is at least
an order of
> magnitude less than your minimum measurement current. This is a costly
solution.
> Some of the newer CMOS op-amps (e.g National Semiconductor's LMC
series) have
> electrometer class input bias currents, but I've never tried them in an
electrometer
> circuit (plastic packages are "leaky").
Actually, the plastic packages are VERY good, providing THEY ARE
CLEAN!!! They make very good "electrometer" parts. Try `em!
Plastic parts, fresh out of a factory sealed package, are
un-touched by human hands and are VERY clean.
Any "junk" (oils, dust, flux) between the pins will degrade the
leakage performance.
I have measured the leakages of the various packages, and the
plastic ones are sub-femtoamp. The worst packages are the ceramic
types.
If someone has a "Pizza-fest" at lunch, then handles the parts
without washing their hands, leakages will shoot up tremendously.
Worse, the oil will eventually collect dust and create a leakage
path.
Even the oils on "clean" hands can create leakage paths. If not
immediately, then later.
In short - You should avoid handling the parts, if possible.
Handle them by the ends of the leads, or the ends of the package,
as not to get any "gook" on the package between leads.
A bath in pure alcohol (or similar cleaner), then baked in an
oven at 85'C for an hour will clean the package up. Do not use
the isopropyl alcohol from the dime store!
If you can, avoid soldering the input pins. Solder flux can
create a very nasty leakage path. It does not take much!
We always recommend "up-in-the-air" wiring for any low-leakage
application. Lift the input leg and solder the wires to a
single-pin socket. Insert the lead into the socket instead of
soldering directly.
A little known ditty: For dual parts (LMC662), the lowest leakage
input is the non-inverting input of the second amplifier (pin 5).
It is *away* from the supply pin (pin 4 - `round the corner), and
pin 6 (next door) is usually at the same potential. So, use amp
"B" for your low-leakage circuitry. Shield or guard all power
traces and outputs all the way up to the package.
Also, avoid input voltages 0.5 to 1V from the rails, as the input
currents start to creep-up. Mid-supply is the best range. The
input "leakage" is really the miss-match between the two ESD
protection diodes, as the MOSFET inputs themselves are literally
"nothing".
Our lowest leakage part is still the LMC660/662 in the "N"
plastic DIP package. Out of a handful, you will probably find a
few of them less than 1fA, the rest around 2-3fA, and 5-8fA on a
"bad" day (at +/-7.5V supplies at room temp).
Cheers,
Paul Grohe
http://tinyurl.com/o8rt7c7
===================================================================
4. Layout For Lowest Leakage
Paul Grohe
1/27/99
On Mon, 25 Jan 1999 11:17:18 -0800,
in the newsgroup sci.electronics.design,
Gerry Schneider <not_...@sympatico.ca>
from Bell Solutions thoughtfully posted:
> Paul Grohe wrote:
>
> > For lowest leakage currents, use the dual version and use the
> > second channel ("right side", pins 5,6,7) for the critical
> > circuit.
> Why's that, Paul? Is this standard for all NatSem parts?
Nope. It is due to the layout of the pins and not the properties
of devices. The singles, duals and quads are identical to each
other. The same is true for the "other guys" parts, too.
Lemme' explain:
Most "low leakage" circuits apply the critical signal path to the
non-inverting input.
Note the layout of the typical dual:
A Out V+
A -In B Out
A +In B -In
V- B +in
On channel "A" - the non-inverting input is located right next to
the negative supply pin. It does not take much "Gook(tm)" between
the adjacent pins to create some healthy leakage paths.
A little household dust, combined with some summertime moisture,
creates a nice leakage path between the pins that can easily
exceed the leakage of the op-amp itself.
However, channel B's non-inverting input (pin 5) is on the other
side of the package - well away from the power pin. You will need
to get a lot more "gook" on, and across, the package to create an
equivalent leakage path.
It is also easier to guard and shield pin 5 than it is to guard
and shield pin 3.
On a quad, all the non-inverting pins are located around power
supply pins, making guarding tough. Even a single has it's
non-inverting input next to V-.
Hope this clears up the misunderstanding. There is no difference
between the parts.
Cheers,
Paul Grohe
http://tinyurl.com/pgu5ecc
===================================================================
5. Latchup
Paul Grohe
5/5/97
On Sun, 04 May 1997 19:57:48 GMT,
in the newsgroup sci.electronics.design,
Pat...@Orrell.prestel.co.uk (Patrick)
from [not set] thoughtfully posted:
> Hi All
>
> Does anyone have any idea if applying signals to an IC is likely to
> cause more or less damage than when the chip has its supply off.
Depends on the available voltage and current available at the
pins. If it is less than a diode drop at less than 1uA, then it
should not be a big problem.
> I suppose it is difficult to generalise as it might be dependant on
> the IC
Yes. And dependant on the circuit.
Officially, we do not recommend or condone it.
You can get into trouble a few ways:
1. The inoperable circuit can load down any other circuits
connected at the same time.
2. The unconnected power supply lines can "float", causing the
unpowered chip to be at unpredictable levels. This can cause
input protection diodes to conduct. Anti-reversal clamping
Shottky diodes should be applied to the supply lines to make sure
they do not float past ground (ie: make sure V+ never goes lower
than ground or V- never goes above ground).
3. Trigger a latchup condition.
Here is a copy of my previous post on a similar subject of
"latchup". It is VERY simplified (it was for a non-electronics
group).
If you break down the IC into what it really is, it is just a
pile of P-N junctions. To isolate a device (transistor, diode,
resistor) from the "substrate", you create another junction,
basically a reversed bias diode, so that there is no conduction
from the circuit elements to the "substrate" (the silicon
"floor"). Think of this as a circuit with a reversed diode
connected from every circuit connection point to V-.
For everything to work right, it is assumed that the substrate is
at the lowest possible voltage potential (usually V-).
If the substrate voltage is brought *above* any other terminal,
these "junctions" (diodes) that were created to isolate now start
conducting, because they are now forward biased.
Some junctions create a parasitic junction similar to an SCR. If
just the right conditions are created, the SCR "gate" triggers
and the junction conducts until power is removed. This is the
classic "latchup" condition, and is very common in input stages,
and output stages.
This usually occurs if any pin has a stiff voltage applied to it
*before* the IC's power supplies are applied.
The most common "SCR Latchup" in an unpowered IC is where the V+
is terminal 1, the other pins are the "gate", and the V- is
terminal 2. Tripping the "gate" causes the conduction across the
chip, shorting V+ to V- until power is removed. Depending on the
system power supply, this "short circuit" current can be a few
mA, or a few amps. That's when things go "BOOM!". Usually the
chip just gets hot and may be *slightly* damaged (usually a shift
in offset voltage or other parameters).
IC designers try to design out possible "latch-up" conditions,
and also try to beef things up to prevent damage from latchups
should they occur. But there is no way to be completely safe.
CMOS and older bipolar parts are most susceptible.
Because of their very low operating currents, it does not take
much to latch up CMOS IC's, so never apply signals to a CMOS part
w/o power.
So, the rule of thumb is not to bring any of the inputs or
outputs above V+, or below V-, and not to apply any large current
sources to any pin before the power supplies are on.
If you must apply a voltage to an unpowered IC, always add a
resistor in series with the inputs to limit the current to a safe
value, say about 10K or so, to prevent triggering "latchup".
Cheers,
Paul Grohe
http://tinyurl.com/lnc3efq
===================================================================
6. Input Protection
Paul Grohe
6/24/97
On Mon, 23 Jun 1997 14:18:48 GMT,
in the newsgroup sci.electronics.design,
dm...@ois.com.au
from iiNet Technologies thoughtfully posted:
> Can some one please explain how in the Absolute Maximum Ratings for
> the LMC662 opamp, the max current at the input pin is rated to +/-
> 5mA. ??
The inputs of CMOS op-amps are protected by a pair of reversed biased
diodes from the input to each rail. They are there to protect the inputs
from ESD transients and other "abuse".
However, these diodes can be "turned on" by applying voltages to the
inputs
above or below the supply voltages ("rails").
The 5mA is the maximum current that could *safely* be "sourced" or "sunk"
from the diodes without causing damage to the diodes or the op-amp.
In reality, the current can exceed several dozen mA or more.
This is why it is recommended to place a current limiting resistor in
series with the input if there is a chance that the input could be
"accidentally" pulled above or below the supply lines (ie: signals coming
from other boards, signals generated by devices running from a higher
supply voltage, or signals from external jacks).
Normally, The feedback resistors provide the necessary current limiting,
but buffers and non-inverting amplifiers may need the limiting resistor
added to their input.
So, If you have a signal applied *directly* to an input of the op-amp,
there should be a 10K-100K resistor in series with the input.
|----------+
| |\ |
+--|-\ |
Rlimit | \----O-------
---/\/\/\-----|+/
10K |/
Compared to the Teraohms of equivalent CMOS input resistance, the 10K is
hardly noticed.
If you have any other questions about our parts, feel free to contact our
support group at sup...@nsc.com or (39)558-9999.
Cheers,
Paul Grohe
http://tinyurl.com/p4oc7ph
===================================================================
7. Testing Op Amps
Paul Grohe
8/29/00
In the newsgroup sci.electronics.design.
"Aviv Kfir" <ak...@isdn.net.il> from
Bezeq International Ltd. thoughtfully posted:
>Hi,
>
>does anyone know where can I get the details and meaning
>of each parameter in the data sheet of op-amlifier.
>I have already searched the Analog-devices, National, Maxim and Linear
>but I couldn't find any document that explains these parameters.
>
Shhh.....It's a secret...there are no standards - don't tell anybody...we
just make
`em up!! .:)
Appnote AN-24: A Simplified Test Set for Op Amp Characterisation.
http://www.national.com/an/AN/AN-24.pdf
Note - this is moved to
http://www.ti.com/lit/an/snoa637/snoa637.pdf
This describes the basic "op-amp test loop" for DC tests (PSRR, CMRR,
AVOL, Vos). Not
much has changed in the last 25 years since this document was written -
the basic
tests are the same. Most tests are very simple - and a few are require a
bit of
VooDoo.
BTW: This appnote describes the design of a production "test box" that
was used in
the days prior to automated testing when all production tests were done
by hand. An
operator inserted the device, ran the switch through the ranges and made
sure the
meter readings were within specification and that the oscilloscope traces
were within
the lines marked by grease pencils. Now we do all the same tests in
milliseconds with
modern testers.
There will never be a magic circuit that tests all amplifiers - high
speed devices
have different requirements than micropower devices. One size does not
fit all. Test
circuits must be designed for the particular op-amp family.
Here are some more references:
http://www.national.com/an/AN/AN-A.pdf
http://www.national.com/an/AN/AN-4.pdf
http://www.national.com/an/AN/AN-20.pdf
Also see "Appendix F: How to read a datasheet", written by Bob Pease,
located in the
back of our 1995 Op Amp databook (always overlooked). I can't find that
on line - so
you have to look in the book.
Analog Devices, LTC, and sometimes TI will usually put the test circuits
in the
datasheet. Particularly in their "precision" device datasheets - like the
OP-07/27/37
derivatives and the Instrumentation amps.
Cheers,
Paul Grohe
http://tinyurl.com/k987wrb
===================================================================
8. Op Amp as Comparator
Paul Grohe
11/8/01
On Wed, 07 Nov 2001 09:12:43 -0800,
in the newsgroup sci.electronics.design,
Howard Delman <del...@blueneptune.com>
from Posted via Supernews, http://www.supernews.com posted:
> I am considering using an op-amp (TI TLV2783) as a comparator. Simply
> -- there is no negative feedback. There is a bit of positive feedback
> for hysteresis. Over my many years, I've heard different discussions
> about whether or not this was "bad" for the op-amp. Some co-workers
> have suggested that slamming on op-amp's output against the rails was,
> for some ill-defined reasons, not wise. Does anyone here have real
> world experience with this?
I'll warn you: If you intend to design this into a commercial
application - better check with the manufacturer to make sure
that the device:
1. Can handle large *sustained* differential inputs.
2. Nothing "funny" happens when the output is driven hard
against the rail (ie: excessive supply current draw, latching).
3. No recovery time problems. Low power and micropower devices
may have a rough time coming out of hard saturation.
It's possible to use an op-amp, but we do not recommend it. If
you do use an op-amp, better torture your circuit (over expected
operating temperature/voltage range if possible) to make sure it
behaves as you expect.
If your application is power sensitive, comparators (particularly
CMOS ones) will draw much less supply current than comparable
op-amps. Since you don't care about output distortion, there's no
need to burn bias current in the output stage.
Cheers,
Paul Grohe
http://tinyurl.com/l47wtd8
===================================================================
Maynard A. Philbrook Jr.
May 31, 2014, 8:08:37 PM5/31/14
to
In article <0jhko9951q5ktavsf...@4ax.com>,
jjla...@highNOTlandTHIStechnologyPART.com says...
output at the collector?
Jamie
jjla...@highNOTlandTHIStechnologyPART.com says...
> >Phil Hobbs
>
> Oh, I believe you. It's very cool.
>
> A Pease Puzzler was a bipolar NPN transistor: ground the base, pull up the
> emitter, and the collector goes negative. Bob explained it as photons generated
> by the b-e zenering and photoelectric generation in the c-b junction. But maybe
> he was guessing.
>
>
>
Sounds like capacitance internally to me, unless this is a constant
>
> Oh, I believe you. It's very cool.
>
> A Pease Puzzler was a bipolar NPN transistor: ground the base, pull up the
> emitter, and the collector goes negative. Bob explained it as photons generated
> by the b-e zenering and photoelectric generation in the c-b junction. But maybe
> he was guessing.
>
>
>
output at the collector?
Jamie
Phil Hobbs
May 31, 2014, 8:26:47 PM5/31/14
to
On 5/31/2014 7:48 PM, Steve Wilson wrote:
> Phil Hobbs <ho...@electrooptical.net> wrote:
>
>> On 5/31/2014 9:24 AM, Steve Wilson wrote:
>>> Steve Wilson <no...@nospam.com> wrote:
>>>> There is a magical discontinuity at zero. In the forward
>>>> direction, the current increases exponentially with voltage
>>>> according to the diode equation:
>>>>
>>>> http://en.wikipedia.org/wiki/Diode_modelling
>>>>
>>>> In the reverse direction, the leakage current is virtually
>>>> independent of voltage up to the point where the diode breaks down:
>>>>
>>>> http://en.wikipedia.org/wiki/Reverse_leakage_current
>>>>
>>>> These are two different mechanisms.
>>>
>>> And this explains why you cannot set the bias current to zero in the
>>> LMC660/LMC64482 series parts. The diode leakages for the top and
>>> bottom diodes add together.
>>
>> How does that work? One is a positive current, and the other is
>> negative.
>
> Exactly. The currents cancel. The degree of matching sets the input bias
> current.
But that's what you were previously denying, unless I misread your post.
> Phil Hobbs <ho...@electrooptical.net> wrote:
>
>> On 5/31/2014 9:24 AM, Steve Wilson wrote:
>>> Steve Wilson <no...@nospam.com> wrote:
>>>> There is a magical discontinuity at zero. In the forward
>>>> direction, the current increases exponentially with voltage
>>>> according to the diode equation:
>>>>
>>>> http://en.wikipedia.org/wiki/Diode_modelling
>>>>
>>>> In the reverse direction, the leakage current is virtually
>>>> independent of voltage up to the point where the diode breaks down:
>>>>
>>>> http://en.wikipedia.org/wiki/Reverse_leakage_current
>>>>
>>>> These are two different mechanisms.
>>>
>>> And this explains why you cannot set the bias current to zero in the
>>> LMC660/LMC64482 series parts. The diode leakages for the top and
>>> bottom diodes add together.
>>
>> How does that work? One is a positive current, and the other is
>> negative.
>
> Exactly. The currents cancel. The degree of matching sets the input bias
> current.
You did say,
>>> And this explains why you cannot set the bias current to zero in the
>>> LMC660/LMC64482 series parts. The diode leakages for the top and
>>> bottom diodes add together.
tend to cancel when you add them together. Did you change your mind?
>
>>> However, they act as current sources, so they are independent of
>>> voltage.
>>
>> If the diodes are closely matched, the leakage should be close as
>> well.
>
> True.
>
>> Second order effects such as bias voltage might very well be enough
>> to
>> make them match exactly.
>
> True, if they are very close to start with. However, temperature changes
> may destroy the cancellation.
>
>>> This means changing VCC and VEE or changing the voltage at the input
>>> to the op amp will have no effect on the input bias current.
>>
>> To leading order.
>>
>>>
>>> The reason this is true is because the actual bias current into the
>>> mosfet junction is close to zero, as Paul Grohe stated. So the only
>>> currents involved are the leakage currents into the protection
>>> diodes.
>>>
>>> I have answered the questions I asked earlier. The mechanism that
>>> creates the input bias current is diode leakage. The currents flow to
>>> VCC and gnd.
>
>> Cheers
>
>> Phil Hobbs
>
> I collected a number of Paul Grohe's posts on the LMC660 and related
> topics. These confirm your observations and add a bit more light on the
> subject.
Interesting stuff, thanks.
John Larkin
May 31, 2014, 8:27:56 PM5/31/14
to
Capacitance wouldn't make negative DC. Unless the b-e junction was making AC, as
a zener relaxation oscillator, and the c-b junction was rectifying that.
Phil Hobbs
May 31, 2014, 8:36:42 PM5/31/14
to
John Larkin
May 31, 2014, 8:36:56 PM5/31/14
to
On Sat, 31 May 2014 18:17:22 -0400, "Tom Miller" <tmille...@verizon.net>
wrote:
I know that some silicon logic ICs do emit light, but very little of it. The
effect has been used for logic analysis, but needs tons of signal averaging
http://www.everyphotoncounts.com/apps-testing.php
Bipolar transistors can emit light, too.
--
John Larkin Highland Technology Inc
wrote:
effect has been used for logic analysis, but needs tons of signal averaging
http://www.everyphotoncounts.com/apps-testing.php
Bipolar transistors can emit light, too.
--
John Larkin Highland Technology Inc
John Larkin
May 31, 2014, 9:31:14 PM5/31/14
to
On Sat, 31 May 2014 17:27:56 -0700, John Larkin
<jjla...@highNOTlandTHIStechnologyPART.com> wrote:
>On Sat, 31 May 2014 20:08:37 -0400, "Maynard A. Philbrook Jr."
><jamie_...@charter.net> wrote:
>
>>In article <0jhko9951q5ktavsf...@4ax.com>,
>>jjla...@highNOTlandTHIStechnologyPART.com says...
>>> >Phil Hobbs
>>>
>>> Oh, I believe you. It's very cool.
>>>
>>> A Pease Puzzler was a bipolar NPN transistor: ground the base, pull up the
>>> emitter, and the collector goes negative. Bob explained it as photons generated
>>> by the b-e zenering and photoelectric generation in the c-b junction. But maybe
>>> he was guessing.
>>>
>>>
>>>
>>
>> Sounds like capacitance internally to me, unless this is a constant
>>output at the collector?
>>
>>Jamie
>>
>>
>>
>
>Pease claimed negative DC at the collector.
>
>Capacitance wouldn't make negative DC. Unless the b-e junction was making AC, as
>a zener relaxation oscillator, and the c-b junction was rectifying that.
No, I think the polarity is wrong there.
<jjla...@highNOTlandTHIStechnologyPART.com> wrote:
>On Sat, 31 May 2014 20:08:37 -0400, "Maynard A. Philbrook Jr."
><jamie_...@charter.net> wrote:
>
>>In article <0jhko9951q5ktavsf...@4ax.com>,
>>jjla...@highNOTlandTHIStechnologyPART.com says...
>>> >Phil Hobbs
>>>
>>> Oh, I believe you. It's very cool.
>>>
>>> A Pease Puzzler was a bipolar NPN transistor: ground the base, pull up the
>>> emitter, and the collector goes negative. Bob explained it as photons generated
>>> by the b-e zenering and photoelectric generation in the c-b junction. But maybe
>>> he was guessing.
>>>
>>>
>>>
>>
>> Sounds like capacitance internally to me, unless this is a constant
>>output at the collector?
>>
>>Jamie
>>
>>
>>
>
>Pease claimed negative DC at the collector.
>
>Capacitance wouldn't make negative DC. Unless the b-e junction was making AC, as
>a zener relaxation oscillator, and the c-b junction was rectifying that.
whit3rd
May 31, 2014, 9:51:00 PM5/31/14
to
On Friday, May 30, 2014 9:45:02 PM UTC-7, John Larkin wrote:
> On Fri, 30 May 2014 21:31:24 -0700 (PDT), whit3rd <whi...@gmail.com> wrote:
> >There are other possible behaviors; if the output is somehow thermally
> >coupled to the leakage on the inputs, it could have positive feedback.
> >So, you could trim the leakage near zero, only to have it snap from
> >one rail to the other. There might be no stable equilibrium.
> The leakage currents will be in the fA range, so the diode power dissipations
> will be in the fW range. That will raise the diode temperatures by nanokelvins.
The old CA3140 op amp sucked about 9 mA quiescent current, about a
quarter watt. Heating doesn't happen because of the leakage currents,
but because of the output drive stages.
> On Fri, 30 May 2014 21:31:24 -0700 (PDT), whit3rd <whi...@gmail.com> wrote:
> >There are other possible behaviors; if the output is somehow thermally
> >coupled to the leakage on the inputs, it could have positive feedback.
> >So, you could trim the leakage near zero, only to have it snap from
> >one rail to the other. There might be no stable equilibrium.
> The leakage currents will be in the fA range, so the diode power dissipations
> will be in the fW range. That will raise the diode temperatures by nanokelvins.
The old CA3140 op amp sucked about 9 mA quiescent current, about a
quarter watt. Heating doesn't happen because of the leakage currents,
but because of the output drive stages.
John Larkin
May 31, 2014, 10:02:55 PM5/31/14
to
where is the positive feedback?
whit3rd
May 31, 2014, 10:29:53 PM5/31/14
to
On Saturday, May 31, 2014 7:02:55 PM UTC-7, John Larkin wrote:
> On Sat, 31 May 2014 18:51:00 -0700 (PDT), whit3rd <whi...@gmail.com> wrote:
>
>
>
> >On Friday, May 30, 2014 9:45:02 PM UTC-7, John Larkin wrote:
>
> >> On Fri, 30 May 2014 21:31:24 -0700 (PDT), whit3rd <whi...@gmail.com> wrote:
>
> >
>
> >> >There are other possible behaviors; if the output is somehow thermally
>
> >> >coupled to the leakage on the inputs, it could have positive feedback.
> On Sat, 31 May 2014 18:51:00 -0700 (PDT), whit3rd <whi...@gmail.com> wrote:
>
>
>
> >On Friday, May 30, 2014 9:45:02 PM UTC-7, John Larkin wrote:
>
> >> On Fri, 30 May 2014 21:31:24 -0700 (PDT), whit3rd <whi...@gmail.com> wrote:
>
> >
>
> >> >There are other possible behaviors; if the output is somehow thermally
>
> >> >coupled to the leakage on the inputs, it could have positive feedback.
> >The old CA3140 op amp sucked about 9 mA quiescent current, about a
> >quarter watt. Heating doesn't happen because of the leakage currents,
> >but because of the output drive stages.
> Sure, static power dissipation warms up a chip, and should be minimized, but
> where is the positive feedback?
Consider a transistor that sinks a fixed current, as part of the output drive.
> >quarter watt. Heating doesn't happen because of the leakage currents,
> >but because of the output drive stages.
> Sure, static power dissipation warms up a chip, and should be minimized, but
> where is the positive feedback?
Even with zero load current, that transistor sinks 1 mA from +1V when the
output state is above ground, and sinks 1 mA from -1V when the output
state is below ground...
So, the heat dumped at that part of the IC is modulated by the output voltage.
If that part of the chip is near to the (+) protection diode on the (+) terminal,
and distant from the other protection diodes, it will tend to pull the (+) terminal
high when the output is high, by increasing the leakage.
The result (of the positive feedback) is that some operating conditions become
bistable- a near-zero input leaves the output stuck in its old state.
Tim Williams
May 31, 2014, 11:29:05 PM5/31/14
to
"Phil Hobbs" <ho...@electrooptical.net> wrote in message
news:538A399E...@electrooptical.net...
> Hardly matters--diodes don't even come close to obeying the 'diode
> equation' (AoE has a graph somewhere--the exponential constant
> frequently differs from kT/e by a factor as large as 2).
As in N? Then just adjust substitute Vth with Vth' = N k_B T / q_e (er,
> equation' (AoE has a graph somewhere--the exponential constant
> frequently differs from kT/e by a factor as large as 2).
or 1/N?).
Unless you mean N varies with V, I or T, which I've never heard before --
but, N always seemed uselessly dumb to me: if it is just a constant, you
can fit the curve by Is just as well; SPICE models often do!
Or to put it another way, just measure it at a different temperature where
Vth' is whatever intended value.
Tried looking in AoE but didn't find anything rmotely pertinent in the
index. Why doesn't this thing have Google in it?...
Tim Williams
May 31, 2014, 11:34:08 PM5/31/14
to
"John Larkin" <jjla...@highNOTlandTHIStechnologyPART.com> wrote in
message news:0jhko9951q5ktavsf...@4ax.com...
message news:0jhko9951q5ktavsf...@4ax.com...
> A Pease Puzzler was a bipolar NPN transistor: ground the base, pull up
> the
> emitter, and the collector goes negative. Bob explained it as photons
> generated
> by the b-e zenering and photoelectric generation in the c-b junction.
> But maybe
> he was guessing.
Bizarrely roundabout, but believable; crack open a 2N3055 (because why
> the
> emitter, and the collector goes negative. Bob explained it as photons
> generated
> by the b-e zenering and photoelectric generation in the c-b junction.
> But maybe
> he was guessing.
would you want to actually use one?) and reverse an ampere through the
base; in a dim room, you'll be able to see the faint yellow glow. Be sure
to check that you aren't measuring ambient light also.
Steve Wilson
May 31, 2014, 11:41:17 PM5/31/14
to
Phil Hobbs <ho...@electrooptical.net> wrote:
> On 5/31/2014 7:48 PM, Steve Wilson wrote:
>> Phil Hobbs <ho...@electrooptical.net> wrote:
>>>> And this explains why you cannot set the bias current to zero in the
>>>> LMC660/LMC64482 series parts. The diode leakages for the top and
>>>> bottom diodes add together.
>>> How does that work? One is a positive current, and the other is
>>> negative.
>> Exactly. The currents cancel. The degree of matching sets the input
bias
>> current.
> But that's what you were previously denying, unless I misread your
post.
> You did say,
>
> >>> And this explains why you cannot set the bias current to zero in
the
> >>> LMC660/LMC64482 series parts. The diode leakages for the top and
> >>> bottom diodes add together.
> Since they're opposite in polarity and nearly equal in magnitude, they
> tend to cancel when you add them together. Did you change your mind?
Yes. You are perfectly correct. I was thinking of the effect an external
> On 5/31/2014 7:48 PM, Steve Wilson wrote:
>> Phil Hobbs <ho...@electrooptical.net> wrote:
>>>> And this explains why you cannot set the bias current to zero in the
>>>> LMC660/LMC64482 series parts. The diode leakages for the top and
>>>> bottom diodes add together.
>>> How does that work? One is a positive current, and the other is
>>> negative.
>> Exactly. The currents cancel. The degree of matching sets the input
bias
>> current.
> But that's what you were previously denying, unless I misread your
post.
> You did say,
>
> >>> And this explains why you cannot set the bias current to zero in
the
> >>> LMC660/LMC64482 series parts. The diode leakages for the top and
> >>> bottom diodes add together.
> Since they're opposite in polarity and nearly equal in magnitude, they
> tend to cancel when you add them together. Did you change your mind?
source would have on the currents and neglected to observe the polarity.
I came to the conclusion the currents actually subtract after posting,
but was so engrossed in tracking down Paul Grohe's posts on the LMC660
that I didn't think to post a correction. My apologies for the oversight
and I am very sorry to have wasted your time having to correct an obvious
error.
>>>> However, they act as current sources, so they are independent of
>>>> voltage.
>>>
>>> If the diodes are closely matched, the leakage should be close as
well.
>>
>> True.
>>
>>> Second order effects such as bias voltage might very well be
enough
>>> to make them match exactly.
>>
>> True, if they are very close to start with. However, temperature
changes
>> may destroy the cancellation.
>
> Why? They'd affect matched diodes equally.
Since the leakage current is an exponential with temperature, a small
difference could cause significant imbalance.
Paul Grohe emphasizes controlling the temperature by heating the
junctions in several of his posts. I'd be inclined to try Peltier cooling
to reduce the diode temperature and leakage currents. Maybe there would
be some temperature where the leakage currents canceled.
>> I collected a number of Paul Grohe's posts on the LMC660 and related
>> topics. These confirm your observations and add a bit more light on
the
>> subject.
>
> <snip>
>
> Interesting stuff, thanks.
>
> Cheers
>
> Phil Hobbs
might be interesting to some.
Part 1 is at:
http://www.edn.com/design/analog/4368681/Design-femtoampere-circuits-
with-low-leakage-part-one
Part 2 is at
http://www.edn.com/design/analog/4375459/1/Design-femtoampere-circuits-
with-low-leakage---Part-2--Component-selection
Chris Jones
Jun 1, 2014, 2:04:30 AM6/1/14
to
On 31/05/2014 06:10, bonesi...@gmail.com wrote:
> I want to make measurements on pyroelectric crystals but I think the input impedance of a typical dmm is too low (10 Megaohm or so). Any suggestions on a simple electrometer adapter for a dmm? I can't afford a new electrometer so I was hoping I could make a simple circuit to add to a dmm for these measurements.
>
> Scot
> I want to make measurements on pyroelectric crystals but I think the input impedance of a typical dmm is too low (10 Megaohm or so). Any suggestions on a simple electrometer adapter for a dmm? I can't afford a new electrometer so I was hoping I could make a simple circuit to add to a dmm for these measurements.
>
>
Here is one that I made (for a moving coil meter):
http://www.youtube.com/watch?v=MleqbsCWg2M
You might be better off with a charge amplifier for this application,
i.e. connect the non-inverting input of the op-amp to ground or a fixed
voltage, and connect a polystyrene or teflon or polypropylene capacitor
from the output of the op-amp to the inverting input of the op-amp, and
feed the input current to the inverting input of the op-amp. Putting a
small resistor (3k3 or so) in series with the input current may help
with stability. You can reset the capacitor charge with a very clean
reed relay, or with a circuit that uses PN4117 jfet devices as diodes
that normally are configured to have no bias across the diode, but apply
bias when you need to reset the capacitor.
Chris
Jan Panteltje
Jun 1, 2014, 2:57:40 AM6/1/14
to
On a sunny day (Sun, 01 Jun 2014 16:04:30 +1000) it happened Chris Jones
<lugn...@spam.yahoo.com> wrote in <538ac270$0$12413$862e...@ngroups.net>:
Nice!
<lugn...@spam.yahoo.com> wrote in <538ac270$0$12413$862e...@ngroups.net>:
piglet
Jun 1, 2014, 3:08:12 AM6/1/14
to
On 31/05/2014 23:17, Tom Miller wrote:
> Is there any info on what NPN device he was using?
Pease suggested 2N2222 but other jelly bean parts work too. I've used
> Is there any info on what NPN device he was using?
2N3904 and BC547 types. The few hundred millivolts negative in an
otherwise positive only circuit can be used as a handy micropower source
for op-amp offset null trim.
piglet
rickman
Jun 1, 2014, 3:21:52 AM6/1/14
to
On 5/30/2014 4:10 PM, bonesi...@gmail.com wrote:
> I want to make measurements on pyroelectric crystals but I think the input impedance of a typical dmm is too low (10 Megaohm or so). Any suggestions on a simple electrometer adapter for a dmm? I can't afford a new electrometer so I was hoping I could make a simple circuit to add to a dmm for these measurements.
To minimize leakage current from the traces of the input it might be a
> I want to make measurements on pyroelectric crystals but I think the input impedance of a typical dmm is too low (10 Megaohm or so). Any suggestions on a simple electrometer adapter for a dmm? I can't afford a new electrometer so I was hoping I could make a simple circuit to add to a dmm for these measurements.
good idea to add guard traces around the input trace. I understand this
is common in commercial devices.
--
Rick
Phil Hobbs
Jun 1, 2014, 8:38:42 AM6/1/14
to
On 5/31/2014 11:29 PM, Tim Williams wrote:
> "Phil Hobbs" <ho...@electrooptical.net> wrote in message
> news:538A399E...@electrooptical.net...
>> Hardly matters--diodes don't even come close to obeying the 'diode
>> equation' (AoE has a graph somewhere--the exponential constant
>> frequently differs from kT/e by a factor as large as 2).
>
> As in N? Then just adjust substitute Vth with Vth' = N k_B T / q_e (er,
> or 1/N?).
Right. The so-called diode equation is ridiculously inaccurate in
> "Phil Hobbs" <ho...@electrooptical.net> wrote in message
> news:538A399E...@electrooptical.net...
>> Hardly matters--diodes don't even come close to obeying the 'diode
>> equation' (AoE has a graph somewhere--the exponential constant
>> frequently differs from kT/e by a factor as large as 2).
>
> As in N? Then just adjust substitute Vth with Vth' = N k_B T / q_e (er,
> or 1/N?).
modelling real diodes, because what we laughingly call "high-level
injection" effects dominate down to very low current densities. The
equation works great for diode-connected transistors, at least for
current densities where beta remains high. (A diode-connected
transistor is the world's simplest negative feedback amplifier.)
>
> Unless you mean N varies with V, I or T, which I've never heard before --
> but, N always seemed uselessly dumb to me: if it is just a constant, you
> can fit the curve by Is just as well; SPICE models often do!
is pretty different from exp(x). The forward current of a diode
connected transistor goes up by a factor of 10 in about 60 mV, whereas
your average diode diode requires 80-120 mV.
>
> Or to put it another way, just measure it at a different temperature where
> Vth' is whatever intended value.
>
> Tried looking in AoE but didn't find anything rmotely pertinent in the
> index. Why doesn't this thing have Google in it?...
figures.
Phil Hobbs
Jun 1, 2014, 9:06:04 AM6/1/14
to
On 5/31/2014 10:02 PM, John Larkin wrote:
> On Sat, 31 May 2014 18:51:00 -0700 (PDT), whit3rd <whi...@gmail.com> wrote:
>
>> On Friday, May 30, 2014 9:45:02 PM UTC-7, John Larkin wrote:
>>> On Fri, 30 May 2014 21:31:24 -0700 (PDT), whit3rd <whi...@gmail.com> wrote:
>>
>>>> There are other possible behaviors; if the output is somehow thermally
>>>> coupled to the leakage on the inputs, it could have positive feedback.
>>>> So, you could trim the leakage near zero, only to have it snap from
>>>> one rail to the other. There might be no stable equilibrium.
>>
>>> The leakage currents will be in the fA range, so the diode power dissipations
>>> will be in the fW range. That will raise the diode temperatures by nanokelvins.
>>
>> The old CA3140 op amp sucked about 9 mA quiescent current, about a
>> quarter watt. Heating doesn't happen because of the leakage currents,
>> but because of the output drive stages.
>
> Sure, static power dissipation warms up a chip, and should be minimized, but
> where is the positive feedback?
>
>
Even if there were positive thermal feedback, you can get rid of it by
> On Sat, 31 May 2014 18:51:00 -0700 (PDT), whit3rd <whi...@gmail.com> wrote:
>
>> On Friday, May 30, 2014 9:45:02 PM UTC-7, John Larkin wrote:
>>> On Fri, 30 May 2014 21:31:24 -0700 (PDT), whit3rd <whi...@gmail.com> wrote:
>>
>>>> There are other possible behaviors; if the output is somehow thermally
>>>> coupled to the leakage on the inputs, it could have positive feedback.
>>>> So, you could trim the leakage near zero, only to have it snap from
>>>> one rail to the other. There might be no stable equilibrium.
>>
>>> The leakage currents will be in the fA range, so the diode power dissipations
>>> will be in the fW range. That will raise the diode temperatures by nanokelvins.
>>
>> The old CA3140 op amp sucked about 9 mA quiescent current, about a
>> quarter watt. Heating doesn't happen because of the leakage currents,
>> but because of the output drive stages.
>
> Sure, static power dissipation warms up a chip, and should be minimized, but
> where is the positive feedback?
>
>
using a compound amp made from two single op amps: run the first stage
as a follower and the second stage as an integrator, with the feedback
cap wrapped all the way round. That way the first stage's bias
conditions (both input and output) are signal-independent, and the
second stage's input current doesn't matter.
Phil Hobbs
Jun 1, 2014, 9:13:34 AM6/1/14
to
diodes can do a lot of interesting things, e.g. resetting integration
capacitors.
>
>>>>> However, they act as current sources, so they are independent of
>>>>> voltage.
>>>>
>>>> If the diodes are closely matched, the leakage should be close as
> well.
>>>
>>> True.
>>>
>>>> Second order effects such as bias voltage might very well be
> enough
>>>> to make them match exactly.
>>>
>>> True, if they are very close to start with. However, temperature
> changes
>>> may destroy the cancellation.
>>
>> Why? They'd affect matched diodes equally.
>
> The diodes are not matched exactly, which causes the input bias current.
>
> Since the leakage current is an exponential with temperature, a small
> difference could cause significant imbalance.
to be the same, though, so temperature ought to cause the bias current
to have the same temperature dependence as the individual leakage currents.
>
> Paul Grohe emphasizes controlling the temperature by heating the
> junctions in several of his posts. I'd be inclined to try Peltier cooling
> to reduce the diode temperature and leakage currents. Maybe there would
> be some temperature where the leakage currents canceled.
in a DIP package, hang a small polystyrene cap from output to
noninverting input, put it in a shielded box, and let it run for a few
hours to see where it winds up. RF shielding will be key, even with a
CMOS op amp.
I have a few thousand ex-Soviet 39 nF feedthrough caps that are great
for that sort of thing.
>
>>> I collected a number of Paul Grohe's posts on the LMC660 and related
>>> topics. These confirm your observations and add a bit more light on
> the
>>> subject.
>>
>> <snip>
>>
>> Interesting stuff, thanks.
>
> I found another article on femtoamp circuits by Paul Grohe in EDN that
> might be interesting to some.
>
> Part 1 is at:
>
> http://www.edn.com/design/analog/4368681/Design-femtoampere-circuits-
> with-low-leakage-part-one
>
> Part 2 is at
>
> http://www.edn.com/design/analog/4375459/1/Design-femtoampere-circuits-
> with-low-leakage---Part-2--Component-selection
>
More good stuff, thanks. It's inspiring--maybe I'll do that brain
> might be interesting to some.
>
> Part 1 is at:
>
> http://www.edn.com/design/analog/4368681/Design-femtoampere-circuits-
> with-low-leakage-part-one
>
> Part 2 is at
>
> http://www.edn.com/design/analog/4375459/1/Design-femtoampere-circuits-
> with-low-leakage---Part-2--Component-selection
>
transplant on my Keithley 602 this summer after all.
John Larkin
Jun 1, 2014, 1:52:39 PM6/1/14
to
John Larkin
Jun 1, 2014, 1:57:34 PM6/1/14
to
On Sun, 01 Jun 2014 09:06:04 -0400, Phil Hobbs <ho...@electrooptical.net> wrote:
>On 5/31/2014 10:02 PM, John Larkin wrote:
>> On Sat, 31 May 2014 18:51:00 -0700 (PDT), whit3rd <whi...@gmail.com> wrote:
>>
>>> On Friday, May 30, 2014 9:45:02 PM UTC-7, John Larkin wrote:
>>>> On Fri, 30 May 2014 21:31:24 -0700 (PDT), whit3rd <whi...@gmail.com> wrote:
>>>
>>>>> There are other possible behaviors; if the output is somehow thermally
>>>>> coupled to the leakage on the inputs, it could have positive feedback.
>>>>> So, you could trim the leakage near zero, only to have it snap from
>>>>> one rail to the other. There might be no stable equilibrium.
>>>
>>>> The leakage currents will be in the fA range, so the diode power dissipations
>>>> will be in the fW range. That will raise the diode temperatures by nanokelvins.
>>>
>>> The old CA3140 op amp sucked about 9 mA quiescent current, about a
>>> quarter watt. Heating doesn't happen because of the leakage currents,
>>> but because of the output drive stages.
>>
>> Sure, static power dissipation warms up a chip, and should be minimized, but
>> where is the positive feedback?
>>
>>
>
>Even if there were positive thermal feedback, you can get rid of it by
>using a compound amp made from two single op amps: run the first stage
>as a follower and the second stage as an integrator, with the feedback
>cap wrapped all the way round. That way the first stage's bias
>conditions (both input and output) are signal-independent, and the
>second stage's input current doesn't matter.
>
>Cheers
>
>Phil Hobbs
I've used an LT1028 followed by a fast unity-gain follower as a compound opamp.
>On 5/31/2014 10:02 PM, John Larkin wrote:
>> On Sat, 31 May 2014 18:51:00 -0700 (PDT), whit3rd <whi...@gmail.com> wrote:
>>
>>> On Friday, May 30, 2014 9:45:02 PM UTC-7, John Larkin wrote:
>>>> On Fri, 30 May 2014 21:31:24 -0700 (PDT), whit3rd <whi...@gmail.com> wrote:
>>>
>>>>> There are other possible behaviors; if the output is somehow thermally
>>>>> coupled to the leakage on the inputs, it could have positive feedback.
>>>>> So, you could trim the leakage near zero, only to have it snap from
>>>>> one rail to the other. There might be no stable equilibrium.
>>>
>>>> The leakage currents will be in the fA range, so the diode power dissipations
>>>> will be in the fW range. That will raise the diode temperatures by nanokelvins.
>>>
>>> The old CA3140 op amp sucked about 9 mA quiescent current, about a
>>> quarter watt. Heating doesn't happen because of the leakage currents,
>>> but because of the output drive stages.
>>
>> Sure, static power dissipation warms up a chip, and should be minimized, but
>> where is the positive feedback?
>>
>>
>
>Even if there were positive thermal feedback, you can get rid of it by
>using a compound amp made from two single op amps: run the first stage
>as a follower and the second stage as an integrator, with the feedback
>cap wrapped all the way round. That way the first stage's bias
>conditions (both input and output) are signal-independent, and the
>second stage's input current doesn't matter.
>
>Cheers
>
>Phil Hobbs
The 1028 needs really low-value feedback resistors to keep its noise down, and I
didn't want my precision amp to source that much current and potentially make
thermal tails. So the second stage does the heavy lifting.
Phil Hobbs
Jun 1, 2014, 2:31:53 PM6/1/14
to
the LT1028's frequency compensation is squirrelly to begin with, and
you're talking about running it at lowish gains, I gather.
I've used the compound-integrator thing before in temperature control
loops, where I wanted the first stage to be on the cold plate but didn't
want any thermal forcing due to its bias conditions changing. Works great.
John Larkin
Jun 1, 2014, 2:50:14 PM6/1/14
to
NMR stuff, the first stage gain is typically 20, loop bandwidth 30 KHz maybe, so
it wasn't hard to do. I did it by hand, pre-LT Spice!
haitic...@gmail.com
Jun 1, 2014, 2:51:10 PM6/1/14
to
On Saturday, May 31, 2014 3:33:07 PM UTC-4, Phil Hobbs wrote:
> On 5/31/2014 3:28 PM, John Larkin wrote:
>
> > On Sat, 31 May 2014 15:12:33 -0400, Phil Hobbs <hos...@electrooptical.net> wrote:
>
> >
>
> >> On 5/31/2014 12:38 AM, John Larkin wrote:
>
> >>> On Fri, 30 May 2014 16:42:11 -0700 (PDT), dagmarg...@yahoo.com wrote:
>
> >>>
>
> >>>> On Friday, May 30, 2014 5:46:07 PM UTC-4, jeroen Belleman wrote:
>
> >>>>> On 30/05/14 23:09, dagmargoo...@yahoo.com wrote:
>
> >>>>>> On Friday, May 30, 2014 5:01:26 PM UTC-4, Steve Wilson wrote:
>
> >>>>>>>
>
> >>>>>>> 20fA input bias, RRIO, 15V VCC, 1 ?V/?C Input Offset Tempco, 1.5 MHz GBW
>
> >>>>>>>
>
> >>>>>>> Note 13: Guaranteed limits are dictated by tester limitations and not
>
> >>>>>>> device performance. Actual performance is reflected in the typical value.
>
> >>>>>>>
>
> >>>>>>> http://octopart.com/partsearch#!?q=LMC6482
>
> >>>>>>>
>
> >>>>>>> Might add a 10k from output to ground to provide a load for the output
>
> >>>>>>> stage.
>
> >>>>>>
>
> >>>>>> Beat me to it. The LMC662 is another choice (cheap), and the LMP7721 (3pA, expensive).
>
> >>>>>>
>
> >>>>>> http://www.ti.com/product/LMP7721
>
> >>>>>>
>
> >>>>>> Some articles for our OP:
>
> >>>>>> http://electronicdesign.com/test-amp-measurement/whats-all-femtoampere-stuff-anyhow
>
> >>>>>> http://www.edn.com/design/analog/4327165/Achieve-femtoampere-leakage-in-surface-mount-op-amp-layouts
>
> >>>>>>
>
> >>>>>
>
> >>>>>
>
> >>>>> I've used an LPC661, which turned out to have an amazing 170aA of
>
> >>>>> bias current. Maybe I just got lucky, because that's *much* better
>
> >>>>> than the datasheet says it is.
>
> >>>>
>
> >>>> Very impressive. I remember a Bob Pease article somewhere about just that,
>
> >>>> that the ib was lower, caused by the protection diodes, and spec set by
>
> >>>> testing throughput.
>
> >>>>
>
> >>>> The LMC6001 was, I think, tested for lower ib.
>
> >>>
>
> >>> Think about an eprom floating gate. It must leak well under an electron per
>
> >>> hour.
>
> >>>
>
> >>>
>
> >>
>
> >> Right around that value, in very round figures. If you assume 10 fF
>
> >> capacitance per cell and a 3-V gate voltage shift for reliable
>
> >> forgetting, a 20-year lifetime implies
>
> >>
>
> >> dq/dt < (3V)/(6E8 s)*1e-14 F* 6.2e18 (e-/C) = 3E-3 electrons/s, or about
>
> >> 1.2 electrons per hour.
>
> >>
>
> >> Cheers
>
> >>
>
> >> Phil Hobbs
>
> >
>
> > Yikes, pretty good guessing.
>
> >
>
> > I think a cool science project could be done with an eprom, specifically
>
> > resolving single-electron leakage. Maybe.
>
> >
>
> > Super-low-leakage opamps might be package dominated. The TO-can versions have
>
> > way more bias current than the DIPs. Sadly, the V- pin is almost always next to
>
> > the IN+ pin. I guess there's not enough market for a better pinout.
>
>
>
> The SOT23 ones are better, and of course you can always use a dual--the
>
> right-hand section is much better isolated.
>
>
>
>
>
> >
>
> > I should quantify the 2N7000 leakage. You can do some fun stuff with
>
> a 2N7000, a
>
> > battery, a resistor, and an LED. There must be small-signal mosfets,
>
> without
>
> > gate protection, that have really low leakages.
>
> >
>
>
>
> I still have some 3N163s in my drawer, complete with copperclad steel
>
> springs shorting their leads together. Might be fun to measure.
for leakage islanding on the PCB. Pease put his IC's on 'stilts.'
And about flux, he ran pcb through automatic dishwaher.
Space wiring avoids a lot of problems, though. Bugs turned upside down over
a pcb ground plane possible.
jb
Hmmm Both Jim Williams and Bob Pease died at same time, roughly. I can't resist
the urging to avoid that fate..."We get schmart too old." - Swedish saying.
> On 5/31/2014 3:28 PM, John Larkin wrote:
>
> > On Sat, 31 May 2014 15:12:33 -0400, Phil Hobbs <hos...@electrooptical.net> wrote:
>
> >
>
> >> On 5/31/2014 12:38 AM, John Larkin wrote:
>
> >>> On Fri, 30 May 2014 16:42:11 -0700 (PDT), dagmarg...@yahoo.com wrote:
>
> >>>
>
> >>>> On Friday, May 30, 2014 5:46:07 PM UTC-4, jeroen Belleman wrote:
>
> >>>>> On 30/05/14 23:09, dagmargoo...@yahoo.com wrote:
>
> >>>>>> On Friday, May 30, 2014 5:01:26 PM UTC-4, Steve Wilson wrote:
>
> >>>>>>> bonesis...@gmail.com wrote:
>
> >>>>>>
>
> >>>>>>>> I want to make measurements on pyroelectric crystals but I think the
>
> >>>>>>>> input impedance of a typical dmm is too low (10 Megaohm or so). Any
>
> >>>>>>>> suggestions on a simple electrometer adapter for a dmm? I can't
>
> >>>>>>>> afford a new electrometer so I was hoping I could make a simple
>
> >>>>>>>> circuit to add to a dmm for these measurements.
>
> >>>>>>>
>
> >>>>>>> Use LMC6482 as a voltage follower. $2.08 Digi-Key, Newark, etc
> >>>>>>> bonesis...@gmail.com wrote:
>
> >>>>>>
>
> >>>>>>>> I want to make measurements on pyroelectric crystals but I think the
>
> >>>>>>>> input impedance of a typical dmm is too low (10 Megaohm or so). Any
>
> >>>>>>>> suggestions on a simple electrometer adapter for a dmm? I can't
>
> >>>>>>>> afford a new electrometer so I was hoping I could make a simple
>
> >>>>>>>> circuit to add to a dmm for these measurements.
>
> >>>>>>>
>
>
> >>>>>>>
>
> >>>>>>> 20fA input bias, RRIO, 15V VCC, 1 ?V/?C Input Offset Tempco, 1.5 MHz GBW
>
> >>>>>>>
>
> >>>>>>> Note 13: Guaranteed limits are dictated by tester limitations and not
>
> >>>>>>> device performance. Actual performance is reflected in the typical value.
>
> >>>>>>>
>
> >>>>>>> http://octopart.com/partsearch#!?q=LMC6482
>
> >>>>>>>
>
> >>>>>>> Might add a 10k from output to ground to provide a load for the output
>
> >>>>>>> stage.
>
> >>>>>>
>
> >>>>>> Beat me to it. The LMC662 is another choice (cheap), and the LMP7721 (3pA, expensive).
>
> >>>>>>
>
> >>>>>> http://www.ti.com/product/LMP7721
>
> >>>>>>
>
> >>>>>> Some articles for our OP:
>
> >>>>>> http://electronicdesign.com/test-amp-measurement/whats-all-femtoampere-stuff-anyhow
>
> >>>>>> http://www.edn.com/design/analog/4327165/Achieve-femtoampere-leakage-in-surface-mount-op-amp-layouts
>
> >>>>>>
>
> >>>>>
>
> >>>>>
>
> >>>>> I've used an LPC661, which turned out to have an amazing 170aA of
>
> >>>>> bias current. Maybe I just got lucky, because that's *much* better
>
> >>>>> than the datasheet says it is.
>
> >>>>
>
> >>>> Very impressive. I remember a Bob Pease article somewhere about just that,
>
> >>>> that the ib was lower, caused by the protection diodes, and spec set by
>
> >>>> testing throughput.
>
> >>>>
>
> >>>> The LMC6001 was, I think, tested for lower ib.
>
> >>>
>
> >>> Think about an eprom floating gate. It must leak well under an electron per
>
> >>> hour.
>
> >>>
>
> >>>
>
> >>
>
> >> Right around that value, in very round figures. If you assume 10 fF
>
> >> capacitance per cell and a 3-V gate voltage shift for reliable
>
> >> forgetting, a 20-year lifetime implies
>
> >>
>
> >> dq/dt < (3V)/(6E8 s)*1e-14 F* 6.2e18 (e-/C) = 3E-3 electrons/s, or about
>
> >> 1.2 electrons per hour.
>
> >>
>
> >> Cheers
>
> >>
>
> >> Phil Hobbs
>
> >
>
> > Yikes, pretty good guessing.
>
> >
>
> > I think a cool science project could be done with an eprom, specifically
>
> > resolving single-electron leakage. Maybe.
>
> >
>
> > Super-low-leakage opamps might be package dominated. The TO-can versions have
>
> > way more bias current than the DIPs. Sadly, the V- pin is almost always next to
>
> > the IN+ pin. I guess there's not enough market for a better pinout.
>
>
>
> The SOT23 ones are better, and of course you can always use a dual--the
>
> right-hand section is much better isolated.
>
>
>
>
>
> >
>
> > I should quantify the 2N7000 leakage. You can do some fun stuff with
>
> a 2N7000, a
>
> > battery, a resistor, and an LED. There must be small-signal mosfets,
>
> without
>
> > gate protection, that have really low leakages.
>
> >
>
>
>
> I still have some 3N163s in my drawer, complete with copperclad steel
>
> springs shorting their leads together. Might be fun to measure.
>
>
>
> Cheers
>
>
>
> Phil Hobbs
>
>
>
> --
>
> Dr Philip C D Hobbs
>
> Principal Consultant
>
> ElectroOptical Innovations LLC
>
> Optics, Electro-optics, Photonics, Analog Electronics
>
>
>
> 160 North State Road #203
>
> Briarcliff Manor NY 10510
>
>
>
> hobbs at electrooptical dot net
>
> http://electrooptical.net
There is one of the Ti "electrometer" chips which has pins on one end set up
>
>
> Cheers
>
>
>
> Phil Hobbs
>
>
>
> --
>
> Dr Philip C D Hobbs
>
> Principal Consultant
>
> ElectroOptical Innovations LLC
>
> Optics, Electro-optics, Photonics, Analog Electronics
>
>
>
> 160 North State Road #203
>
> Briarcliff Manor NY 10510
>
>
>
> hobbs at electrooptical dot net
>
> http://electrooptical.net
for leakage islanding on the PCB. Pease put his IC's on 'stilts.'
And about flux, he ran pcb through automatic dishwaher.
Space wiring avoids a lot of problems, though. Bugs turned upside down over
a pcb ground plane possible.
jb
Hmmm Both Jim Williams and Bob Pease died at same time, roughly. I can't resist
the urging to avoid that fate..."We get schmart too old." - Swedish saying.
Jan Panteltje
Jun 1, 2014, 2:53:42 PM6/1/14
to
On a sunny day (Sun, 01 Jun 2014 16:04:30 +1000) it happened Chris Jones
<lugn...@spam.yahoo.com> wrote in <538ac270$0$12413$862e...@ngroups.net>:
<lugn...@spam.yahoo.com> wrote in <538ac270$0$12413$862e...@ngroups.net>:
>On 31/05/2014 06:10, bonesi...@gmail.com wrote:
>> I want to make measurements on pyroelectric crystals but I think the input impedance of a typical dmm is too low (10 Megaohm
>> or so). Any suggestions on a simple electrometer adapter for a dmm? I can't afford a new electrometer so I was hoping I could
>> make a simple circuit to add to a dmm for these measurements.
>>
>> I want to make measurements on pyroelectric crystals but I think the input impedance of a typical dmm is too low (10 Megaohm
>> or so). Any suggestions on a simple electrometer adapter for a dmm? I can't afford a new electrometer so I was hoping I could
>> make a simple circuit to add to a dmm for these measurements.
>>
>> Scot
>>
>
>Here is one that I made (for a moving coil meter):
>http://www.youtube.com/watch?v=MleqbsCWg2M
>
>You might be better off with a charge amplifier for this application,
>i.e. connect the non-inverting input of the op-amp to ground or a fixed
>voltage, and connect a polystyrene or teflon or polypropylene capacitor
>from the output of the op-amp to the inverting input of the op-amp, and
>feed the input current to the inverting input of the op-amp. Putting a
>small resistor (3k3 or so) in series with the input current may help
>with stability. You can reset the capacitor charge with a very clean
>reed relay, or with a circuit that uses PN4117 jfet devices as diodes
>that normally are configured to have no bias across the diode, but apply
>bias when you need to reset the capacitor.
>
>Chris
Nice!
PS
>>
>
>Here is one that I made (for a moving coil meter):
>http://www.youtube.com/watch?v=MleqbsCWg2M
>
>You might be better off with a charge amplifier for this application,
>i.e. connect the non-inverting input of the op-amp to ground or a fixed
>voltage, and connect a polystyrene or teflon or polypropylene capacitor
>from the output of the op-amp to the inverting input of the op-amp, and
>feed the input current to the inverting input of the op-amp. Putting a
>small resistor (3k3 or so) in series with the input current may help
>with stability. You can reset the capacitor charge with a very clean
>reed relay, or with a circuit that uses PN4117 jfet devices as diodes
>that normally are configured to have no bias across the diode, but apply
>bias when you need to reset the capacitor.
>
>Chris
Nice!
this is a nice link talking about why the voltage increases
when you move the capacitor plates away from each other:
http://physics.stackexchange.com/questions/82429/why-does-the-voltage-increase-when-capacitor-plates-are-separated
Phil Hobbs
Jun 1, 2014, 2:59:59 PM6/1/14
to
Maynard A. Philbrook Jr.
Jun 1, 2014, 4:30:03 PM6/1/14
to
In article <lme6le$gc5$1...@dont-email.me>, tmor...@charter.net says...
in science..
A small group of us agreed that if you have to put in constants that
were found by trial and error, Boltzmann, Planck etc.. then the subject
matter was not yet fully understood.
It's like making up formulas using well know values and meanings in the
subject matter but still does not resolve! So they throw in some made up
constant! Sort of like winding coils, every one has their own math and
usually includes a made up constant to make it work!
Don't pay no attention to me, I am just disgruntle how many researchers
just buy into it and say, well that's the way it is.
Why can't we just go back and fix some of the major errors in science
so that we can move on. It wouldn't surprise me if there are two sets of
science out there, one we are allowed to practice and not get to far,
and the other, only the elite can know about it.
Jamie
> > Hardly matters--diodes don't even come close to obeying the 'diode
> > equation' (AoE has a graph somewhere--the exponential constant
> > frequently differs from kT/e by a factor as large as 2).
>
> As in N? Then just adjust substitute Vth with Vth' = N k_B T / q_e (er,
> or 1/N?).
>
>
something I made comment to years ago while talking about the math used
> > equation' (AoE has a graph somewhere--the exponential constant
> > frequently differs from kT/e by a factor as large as 2).
>
> As in N? Then just adjust substitute Vth with Vth' = N k_B T / q_e (er,
> or 1/N?).
>
>
in science..
A small group of us agreed that if you have to put in constants that
were found by trial and error, Boltzmann, Planck etc.. then the subject
matter was not yet fully understood.
It's like making up formulas using well know values and meanings in the
subject matter but still does not resolve! So they throw in some made up
constant! Sort of like winding coils, every one has their own math and
usually includes a made up constant to make it work!
Don't pay no attention to me, I am just disgruntle how many researchers
just buy into it and say, well that's the way it is.
Why can't we just go back and fix some of the major errors in science
so that we can move on. It wouldn't surprise me if there are two sets of
science out there, one we are allowed to practice and not get to far,
and the other, only the elite can know about it.
Jamie
whit3rd
Jun 1, 2014, 5:14:28 PM6/1/14
to
On Sunday, June 1, 2014 12:21:52 AM UTC-7, rickman wrote:
> On 5/30/2014 4:10 PM, bonesi...@gmail.com wrote:
>
> > I want to make measurements on pyroelectric crystals but I think the input impedance of a typical dmm is too low
> On 5/30/2014 4:10 PM, bonesi...@gmail.com wrote:
>
> > I want to make measurements on pyroelectric crystals but I think the input impedance of a typical dmm is too low
> To minimize leakage current from the traces of the input it might be a
> good idea to add guard traces around the input trace. I understand this
> is common in commercial devices.
There are other tricks that might help; drill a hole in the PCB where the sensitive
> good idea to add guard traces around the input trace. I understand this
> is common in commercial devices.
lead goes, and insert a small PTFE insulated terminal. Alternately, bend up
the lead wire and make point-to-point wiring from there.
<http://www.keyelco.com/product-pdf.cfm?p=1858>
whit3rd
Jun 1, 2014, 5:31:46 PM6/1/14
to
On Sunday, June 1, 2014 1:30:03 PM UTC-7, Maynard A. Philbrook Jr. wrote:
> something I made comment to years ago while talking about the math used
> in science..
>
> A small group of us agreed that if you have to put in constants that
> were found by trial and error, Boltzmann, Planck etc.. then the subject
> matter was not yet fully understood.
In science, one accepts observations. In math, one does not (definitions
> something I made comment to years ago while talking about the math used
> in science..
>
> A small group of us agreed that if you have to put in constants that
> were found by trial and error, Boltzmann, Planck etc.. then the subject
> matter was not yet fully understood.
and axioms OK; logic OK; yardsticks, NO).
So, science has always to be open to the possibility that the next
(observation, experiment, test) will cause a theory to be invalidated.
Math does not (though Godel looked hard at proof-of-consistency issues,
and the next PROOF might cause a kind of invalidation).
John S
Jun 1, 2014, 5:44:09 PM6/1/14
to
it, do you?
Maynard A. Philbrook Jr.
Jun 1, 2014, 6:08:40 PM6/1/14
to
In article <lmg6r9$31a$2...@dont-email.me>, Sop...@invalid.org says...
yours, that much is evident!
Would you like to try again with your childish comprehension responses?
It better serves to publicize your inability to understand
reality.
They name streets after you "One way" and if that is the way you think?
You will come to a dead end!
Jamie
> > So, science has always to be open to the possibility that the next
> > (observation, experiment, test) will cause a theory to be invalidated.
> > Math does not (though Godel looked hard at proof-of-consistency issues,
> > and the next PROOF might cause a kind of invalidation).
> >
>
> Excellent and tactful answer. You don't really expect him to understand
> it, do you?
>
My understanding about things are obviously far above the plane of
> > (observation, experiment, test) will cause a theory to be invalidated.
> > Math does not (though Godel looked hard at proof-of-consistency issues,
> > and the next PROOF might cause a kind of invalidation).
> >
>
> Excellent and tactful answer. You don't really expect him to understand
> it, do you?
>
yours, that much is evident!
Would you like to try again with your childish comprehension responses?
It better serves to publicize your inability to understand
reality.
They name streets after you "One way" and if that is the way you think?
You will come to a dead end!
Jamie
k...@attt.bizz
Jun 1, 2014, 7:20:55 PM6/1/14
to
On Sun, 1 Jun 2014 16:30:03 -0400, "Maynard A. Philbrook Jr."
<jamie_...@charter.net> wrote:
>In article <lme6le$gc5$1...@dont-email.me>, tmor...@charter.net says...
>> > Hardly matters--diodes don't even come close to obeying the 'diode
>> > equation' (AoE has a graph somewhere--the exponential constant
>> > frequently differs from kT/e by a factor as large as 2).
>>
>> As in N? Then just adjust substitute Vth with Vth' = N k_B T / q_e (er,
>> or 1/N?).
>>
>>
>
> something I made comment to years ago while talking about the math used
>in science..
>
> A small group of us agreed that if you have to put in constants that
>were found by trial and error, Boltzmann, Planck etc.. then the subject
>matter was not yet fully understood.
So we don't fully understand simple electronics because we still use
<jamie_...@charter.net> wrote:
>In article <lme6le$gc5$1...@dont-email.me>, tmor...@charter.net says...
>> > Hardly matters--diodes don't even come close to obeying the 'diode
>> > equation' (AoE has a graph somewhere--the exponential constant
>> > frequently differs from kT/e by a factor as large as 2).
>>
>> As in N? Then just adjust substitute Vth with Vth' = N k_B T / q_e (er,
>> or 1/N?).
>>
>>
>
> something I made comment to years ago while talking about the math used
>in science..
>
> A small group of us agreed that if you have to put in constants that
>were found by trial and error, Boltzmann, Planck etc.. then the subject
>matter was not yet fully understood.
these silly "constants found by trial en error", like e? I guess we
don't understand anything about solid geometry because we still use
that guesswork Pi thing.
Phil Hobbs
Jun 1, 2014, 7:52:34 PM6/1/14
to
mavens is the number of undetermined parameters required to make the
standard model fit experiment. Smaller is definitely better, because of
course the more you understand, the fewer free parameters you need.
These aren't mathematical constants like pi or e, they're things like
the mass of the electron, Planck's constant, and the fine structure
constant. My rather out-of-date recollection is that there are about
eight of them at the moment.
k...@attt.bizz
Jun 1, 2014, 8:18:38 PM6/1/14
to
only so predictive. One of the many problems with the AGW model
religion.
>These aren't mathematical constants like pi or e, they're things like
>the mass of the electron, Planck's constant, and the fine structure
>constant. My rather out-of-date recollection is that there are about
>eight of them at the moment.
conversion factor to fit it to our arbitrary units.
Steve Wilson
Jun 1, 2014, 8:30:12 PM6/1/14
to
[..]
>> I came to the conclusion the currents actually subtract after
>> posting, but was so engrossed in tracking down Paul Grohe's posts on
>> the LMC660 that I didn't think to post a correction. My apologies for
>> the oversight and I am very sorry to have wasted your time having to
>> correct an obvious error.
> Sure, no worries. Just trying to keep the discussion going. Input
> diodes can do a lot of interesting things, e.g. resetting integration
> capacitors.
Paul mentioned that trick and where it came from. I didn't save the
google reference, but it's good to know it has been around for a while.
>>>> True, if they are very close to start with. However, temperature
>>>> changes may destroy the cancellation.
>>>
>>> Why? They'd affect matched diodes equally.
>>
>> The diodes are not matched exactly, which causes the input bias
>> current.
>>
>> Since the leakage current is an exponential with temperature, a small
>> difference could cause significant imbalance.
> You'd expect the logarithmic slopes of the individual leakage currents
> to be the same, though, so temperature ought to cause the bias current
> to have the same temperature dependence as the individual leakage
> currents.
Yes, and the datasheet shows this. But if you go down in temperature,
diode leakage decreases and bulk resistance may start to play a role.
This may become significant with Peltier cooling, which was my next
topic.
>> Paul Grohe emphasizes controlling the temperature by heating the
>> junctions in several of his posts. I'd be inclined to try Peltier
>> cooling to reduce the diode temperature and leakage currents. Maybe
>> there would be some temperature where the leakage currents canceled.
> Maybe. It would be interesting to take a nice clean LMC660 or LMC6482
> in a DIP package, hang a small polystyrene cap from output to
> noninverting input, put it in a shielded box, and let it run for a few
> hours to see where it winds up. RF shielding will be key, even with a
> CMOS op amp.
> I have a few thousand ex-Soviet 39 nF feedthrough caps that are great
> for that sort of thing.
616 with the 6162 I/O. I really want to get rid of the jfet input stage
in these units and get a more stable zero adjust.
The 616 looks like it might be very simple. The input already uses an op
amp. But the 610C looks much more difficult. The differential jfets are
tightly integrated into the zero adjust and the following stages.
I have a LTspice model of a LMC662 input stage with a bootstrap to +/-
200V running. I see three problems:
1. Input protection is not so simple. You can't just slap back to back
diodes across the input. The LMC662 offset could be 3 mV, and one of the
diodes would start to conduct.
2. Two LND150's could protect the input stage, but I don't know what
would happen if you suddenly applied a large positive voltage to the
input. The common mode range is gnd to VCC - 1.5V. I don't know if it
would do a phase reversal or some other strange response. Maybe it would
need a LMC6482, which is RRIO and has no phase reversal. But it has 10
times the input bias current.
3. I don't know how to add a zero adjust to the output so it would tie
into the existing circuitry.
However, there are some improvements I would like to make to these units.
One change would be to add a center scale switch to the 610C. This would
allow viewing small changes around zero. The mirror scale is ideal for
this feature.
Another is to investigate improving the noise. Since temperature is under
a square root, it turns out dropping the temperature has little effect
when you are dealing with thermal noise.
Here is a simple list I made of the effect of temperature on thermal
power (view in fixed width font)
Source Temp Power in Watts dBm
Room Temperature : 290K 4.003566e-21 -173.97
Dry Ice : 194.65K 2.687221e-21 -175.70
Liquid Nitrogen : 77K 1.063015e-21 -179.73
Liquid Hydrogen : 20.28K 2.799735e-22 -185.52
Liquid Helium4 : 4.2K 5.798268e-23 -192.36
You really have to work hard to see much change. For example, all the way
down to liquid nitrogen gives only 6dB improvement in noise. So we are
not going to see much improvement in noise.
Diode leakage is a totally different story. Leakage is exponential with
temperature - a 10C increase doubles the leakage current.
This seems ideally suited to Peltier cooling. Since the LMC662 has such
low power drain, a small unit might give up to 70C change.
You are probably very knowledgeable on this, but here is a list of
example coolers that might be useful. They are not very expensive:
https://tetech.com/peltier-thermoelectric-cooler-modules/micro/
One unit that might work is
https://tetech.com/wp-content/uploads/2013/11/TE-12-0.45-1.3.pdf
There are probably much better choices. This is just for tabletop
discussions.
A 70C change would cut the leakage by 2^7 = 128. Even if we had to go to
a LMC6482, which has 20fA input leakage, the result would be
20e-15 / 128 = 1.5625e-16, or 15 attoamps.
Even if it doesn't quite make 15aA, the end result could be quite
significant.
The LMC6482AI is specd to -40C and is only $2.17 at Newark:
http://octopart.com/partsearch#!?q=LMC6482AI
I would like very much to do this. What kind of problems am I likely to
run into?
> Cheers
> Phil Hobbs
>> I came to the conclusion the currents actually subtract after
>> posting, but was so engrossed in tracking down Paul Grohe's posts on
>> the LMC660 that I didn't think to post a correction. My apologies for
>> the oversight and I am very sorry to have wasted your time having to
>> correct an obvious error.
> Sure, no worries. Just trying to keep the discussion going. Input
> diodes can do a lot of interesting things, e.g. resetting integration
> capacitors.
google reference, but it's good to know it has been around for a while.
>>>> True, if they are very close to start with. However, temperature
>>>> changes may destroy the cancellation.
>>>
>>> Why? They'd affect matched diodes equally.
>>
>> The diodes are not matched exactly, which causes the input bias
>> current.
>>
>> Since the leakage current is an exponential with temperature, a small
>> difference could cause significant imbalance.
> You'd expect the logarithmic slopes of the individual leakage currents
> to be the same, though, so temperature ought to cause the bias current
> to have the same temperature dependence as the individual leakage
> currents.
diode leakage decreases and bulk resistance may start to play a role.
This may become significant with Peltier cooling, which was my next
topic.
>> Paul Grohe emphasizes controlling the temperature by heating the
>> junctions in several of his posts. I'd be inclined to try Peltier
>> cooling to reduce the diode temperature and leakage currents. Maybe
>> there would be some temperature where the leakage currents canceled.
> Maybe. It would be interesting to take a nice clean LMC660 or LMC6482
> in a DIP package, hang a small polystyrene cap from output to
> noninverting input, put it in a shielded box, and let it run for a few
> hours to see where it winds up. RF shielding will be key, even with a
> CMOS op amp.
> I have a few thousand ex-Soviet 39 nF feedthrough caps that are great
> for that sort of thing.
> More good stuff, thanks. It's inspiring--maybe I'll do that brain
> transplant on my Keithley 602 this summer after all.
I would love to hear how you plan on doing it. I have a 610C and also a
> transplant on my Keithley 602 this summer after all.
616 with the 6162 I/O. I really want to get rid of the jfet input stage
in these units and get a more stable zero adjust.
The 616 looks like it might be very simple. The input already uses an op
amp. But the 610C looks much more difficult. The differential jfets are
tightly integrated into the zero adjust and the following stages.
I have a LTspice model of a LMC662 input stage with a bootstrap to +/-
200V running. I see three problems:
1. Input protection is not so simple. You can't just slap back to back
diodes across the input. The LMC662 offset could be 3 mV, and one of the
diodes would start to conduct.
2. Two LND150's could protect the input stage, but I don't know what
would happen if you suddenly applied a large positive voltage to the
input. The common mode range is gnd to VCC - 1.5V. I don't know if it
would do a phase reversal or some other strange response. Maybe it would
need a LMC6482, which is RRIO and has no phase reversal. But it has 10
times the input bias current.
3. I don't know how to add a zero adjust to the output so it would tie
into the existing circuitry.
However, there are some improvements I would like to make to these units.
One change would be to add a center scale switch to the 610C. This would
allow viewing small changes around zero. The mirror scale is ideal for
this feature.
Another is to investigate improving the noise. Since temperature is under
a square root, it turns out dropping the temperature has little effect
when you are dealing with thermal noise.
Here is a simple list I made of the effect of temperature on thermal
power (view in fixed width font)
Source Temp Power in Watts dBm
Room Temperature : 290K 4.003566e-21 -173.97
Dry Ice : 194.65K 2.687221e-21 -175.70
Liquid Nitrogen : 77K 1.063015e-21 -179.73
Liquid Hydrogen : 20.28K 2.799735e-22 -185.52
Liquid Helium4 : 4.2K 5.798268e-23 -192.36
You really have to work hard to see much change. For example, all the way
down to liquid nitrogen gives only 6dB improvement in noise. So we are
not going to see much improvement in noise.
Diode leakage is a totally different story. Leakage is exponential with
temperature - a 10C increase doubles the leakage current.
This seems ideally suited to Peltier cooling. Since the LMC662 has such
low power drain, a small unit might give up to 70C change.
You are probably very knowledgeable on this, but here is a list of
example coolers that might be useful. They are not very expensive:
https://tetech.com/peltier-thermoelectric-cooler-modules/micro/
One unit that might work is
https://tetech.com/wp-content/uploads/2013/11/TE-12-0.45-1.3.pdf
There are probably much better choices. This is just for tabletop
discussions.
A 70C change would cut the leakage by 2^7 = 128. Even if we had to go to
a LMC6482, which has 20fA input leakage, the result would be
20e-15 / 128 = 1.5625e-16, or 15 attoamps.
Even if it doesn't quite make 15aA, the end result could be quite
significant.
The LMC6482AI is specd to -40C and is only $2.17 at Newark:
http://octopart.com/partsearch#!?q=LMC6482AI
I would like very much to do this. What kind of problems am I likely to
run into?
> Cheers
> Phil Hobbs
Steve Wilson
Jun 1, 2014, 8:44:25 PM6/1/14
to
Steve Wilson <no...@nospam.com> wrote:
> However, there are some improvements I would like to make to these
> units.
> One change would be to add a center scale switch to the 610C. This
> would allow viewing small changes around zero. The mirror scale is
> ideal for this feature.
Scratch that. The 610C already has it!
> However, there are some improvements I would like to make to these
> units.
> One change would be to add a center scale switch to the 610C. This
> would allow viewing small changes around zero. The mirror scale is
> ideal for this feature.
Maynard A. Philbrook Jr.
Jun 1, 2014, 9:15:37 PM6/1/14
to
In article <r1dno9d38hfjp66vi...@4ax.com>, k...@attt.bizz
says...
Lets not be ridicules.. We are talking about constants that have
no direct relations but are used to force an
out come that matches a real live situation instead of using something
like PI for example, which would make since.
There are many examples, and I'll give one. A very commonly used
coil formula to calculate L..
L = 0.012*N^*u*A/l
Ok, so we have this, "u" being a permeability value which is ok, you
need that of course, but then you have 0.012? This only tells me that
something is unknown and fudge! Maybe a correct "u" table should be
created to you don't need that magic number?
another one.
L = u*N^*A/i that's easy and there is no magic numbers put in place,
but then you have this.
L = u*N^*A/I+0.45*D
The first one does not care about the diameter but the second does and
inserts a magic number.
I've seen Boltzmann and Planck constants being use in area's where It
really has no need, yet be corrected with a magic constant down the
line.
I used coils for a simple example for the simple minded.
Jamie
says...
> >matter was not yet fully understood.
>
> So we don't fully understand simple electronics because we still use
> these silly "constants found by trial en error", like e? I guess we
> don't understand anything about solid geometry because we still use
> that guesswork Pi thing.
>
>
No, PI is a constant that has real meaning.. that is obvious.
>
> So we don't fully understand simple electronics because we still use
> these silly "constants found by trial en error", like e? I guess we
> don't understand anything about solid geometry because we still use
> that guesswork Pi thing.
>
>
Lets not be ridicules.. We are talking about constants that have
no direct relations but are used to force an
out come that matches a real live situation instead of using something
like PI for example, which would make since.
There are many examples, and I'll give one. A very commonly used
coil formula to calculate L..
L = 0.012*N^*u*A/l
Ok, so we have this, "u" being a permeability value which is ok, you
need that of course, but then you have 0.012? This only tells me that
something is unknown and fudge! Maybe a correct "u" table should be
created to you don't need that magic number?
another one.
L = u*N^*A/i that's easy and there is no magic numbers put in place,
but then you have this.
L = u*N^*A/I+0.45*D
The first one does not care about the diameter but the second does and
inserts a magic number.
I've seen Boltzmann and Planck constants being use in area's where It
really has no need, yet be corrected with a magic constant down the
line.
I used coils for a simple example for the simple minded.
Jamie
Maynard A. Philbrook Jr.
Jun 1, 2014, 9:16:48 PM6/1/14
to
In article <538BBCC2...@electrooptical.net>,
ho...@electrooptical.net says...
Jamie
ho...@electrooptical.net says...
> In Jamie's defense, one of the metrics used by quantum field theory
> mavens is the number of undetermined parameters required to make the
> standard model fit experiment. Smaller is definitely better, because of
> course the more you understand, the fewer free parameters you need.
>
> These aren't mathematical constants like pi or e, they're things like
> the mass of the electron, Planck's constant, and the fine structure
> constant. My rather out-of-date recollection is that there are about
> eight of them at the moment.
>
> Cheers
>
>
>
Thanks Phil, you understand exactly where I am coming from!
> mavens is the number of undetermined parameters required to make the
> standard model fit experiment. Smaller is definitely better, because of
> course the more you understand, the fewer free parameters you need.
>
> These aren't mathematical constants like pi or e, they're things like
> the mass of the electron, Planck's constant, and the fine structure
> constant. My rather out-of-date recollection is that there are about
> eight of them at the moment.
>
> Cheers
>
>
>
Jamie
Steve Wilson
Jun 2, 2014, 12:07:26 AM6/2/14
to
Steve Wilson <no...@nospam.com> wrote:
[...]
> A 70C change would cut the leakage by 2^7 = 128. Even if we had to go
> to a LMC6482, which has 20fA input leakage, the result would be
> 20e-15 / 128 = 1.5625e-16, or 15 attoamps.
Sorry, got everything right until the end.
The LMC662 might do that. But the LMC6482 would give 150 aA. Still not bad.
Both would probably be buried in the noise.
[...]
> A 70C change would cut the leakage by 2^7 = 128. Even if we had to go
> to a LMC6482, which has 20fA input leakage, the result would be
> 20e-15 / 128 = 1.5625e-16, or 15 attoamps.
The LMC662 might do that. But the LMC6482 would give 150 aA. Still not bad.
Both would probably be buried in the noise.
dagmarg...@yahoo.com
Jun 2, 2014, 12:47:56 AM6/2/14
to
On Sunday, June 1, 2014 8:30:12 PM UTC-4, Steve Wilson wrote:
Condensation.
Cheers,
James Arthur
> Phil Hobbs wrote:
> > On 5/31/2014 11:41 PM, Steve Wilson wrote:
>
> [..]
> > On 5/31/2014 11:41 PM, Steve Wilson wrote:
>
> [..]
Cheers,
James Arthur
josephkk
Jun 2, 2014, 1:14:00 AM6/2/14
to
structure constant isn't all that constant. We have done experiments that
have noticeably bent it. (~5%)
?-)
Reinhardt Behm
Jun 2, 2014, 1:23:24 AM6/2/14
to
josephkk wrote:
> And now for some really annoying news they are finding out that the fine
> structure constant isn't all that constant. We have done experiments that
> have noticeably bent it. (~5%)
>
> ?-)
But these measurements still are waiting for independent approval.
> And now for some really annoying news they are finding out that the fine
> structure constant isn't all that constant. We have done experiments that
> have noticeably bent it. (~5%)
>
> ?-)
As they say: "Strong claims need strong arguments."
--
Reinhardt Behm
It is loading more messages.
0 new messages