On capacitor self-timeconstants
Tom Bruhns
pretty long time constants from plastic film capacitors, especially when
you operate them well below their rated voltage. If I recall correctly,
the context was for a peak-hold or sample-and-hold capacitor. I thought
it would be fun to run a little experiment, so yesterday, I mounted two
0.1uF 200V polyester ("Mylar") caps and two 0.1uF 160V polypropylene
caps on a board, just to hold them. One lead is soldered to the board,
and one lead is "flying," just left in the air. It would be better to
put them in an electrostatically shielding box for this experiment,
but...this is intended to be a simple experiment. :-) I left them to
charge on a reasonably stable power supply at 10.6890V for half an
hour. I can measure them with an HP3456A voltmeter, which, on its 10V
range with autozero done recently but turned off for the measurement,
represents a very low leakage current, around 1pA. I can make a
measurement with the voltmeter pre-charged to close to the capacitor
voltage, then connected to the capacitor for about 1 second, so the
voltmeter should affect the capacitor voltage during that time by around
10 microvolts. So, one day after charging, with the time between at
just over 20C and probably about 50% rel.humidity, I measured them
all...the first measurement after charging. Warning: this should be
considered a preliminary measurement; I'll re-measure sometime next
week, and that should yield a more realistic characterization. I would
not be surprised if the next measurement yields somewhat different
results, and I'd trust the data taken over the longer time to be more
trustworthy. (Note that a small change in capacitance can cause a
change in voltage which is larger than some of these measured values.)
But here are the first measurements, with time constants extrapolated
from them:
delta-t for all: 87000 seconds
initial V for all: 10.6890
apparent apparent
time leakage
Cap V2 delta-V constant current
volts volts fA
polyester 1 10.5928 -0.0962 112 days 111
polyester 2 10.5442 -0.1448 74 days 116
polyprop 1 10.6805 -0.0085 3.45 years 9.8
polyprop 2 10.6842 -0.0048 6.15 years 5.5
Note that these are rather longer times than predicted by a graph you
can find in "Reference Data for Engineers" (fig. 9, pg. 5-19 in the
seventh edition), which suggests about 8 days for polypropylene and 1
day for polyester. Clearly the measurements are accurate enough to
disprove that graph.
Note also that the typical input bias current of a National
Semiconductor LMC6001 op amp (to pick just one example of a modern low
input bias current amplifier, not to particularly recommend this part)
is about 10fA. So you can presumably make an integrator with one of
these polypropylene caps and such an op amp which, at room temperature,
would have a drift rate with zero input current of about 20fA/0.1uF =
200nV/second = 1.3 hours/millivolt. Be really careful about leakage
paths if you try this at home!
Cheers,
Tom
Mark Rehorst
The capacitances will change significantly with temp and since the charge on
the cap is approximately fixed,
the voltage will rise or fall...
MR
Roy McCammon
I'm looking forward to hearing more.
James Meyer
Ionizing radiation, alpha, beta, gamma, and X-ray, all present
in the inevitable background and cosmic radiation, will also cause
leakage in insulators. A lead shield should improve leakage
specifications.
Jim
Spehro Pefhany
> Ionizing radiation, alpha, beta, gamma, and X-ray, all present
> in the inevitable background and cosmic radiation, will also cause
> leakage in insulators. A lead shield should improve leakage
> specifications.
> Jim
So, a charged "floating plate" PP capacitor could make a solid state
dosimeter? ;-)
--
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
Spehro Pefhany "The Journey is the reward"
sp...@interlog.com
Fax:(905) 271-9838 (small micro system devt hw/sw + mfg)
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
dang...@earthlink.net
>On Thu, 29 Jul 1999 16:18:49 -0500, "Mark Rehorst"
><mreh...@fmi.fujitsu.com> wrote:
>
>>Did you control the temperature?
>>
>>The capacitances will change significantly with temp and since the charge on
>>the cap is approximately fixed,
>>the voltage will rise or fall...
>>
>>MR
>>
> Ionizing radiation, alpha, beta, gamma, and X-ray, all present
>in the inevitable background and cosmic radiation, will also cause
>leakage in insulators. A lead shield should improve leakage
>specifications.
>
> Jim
I think the approx. 10-20X measured *relative* improvement in leakage
current, polypropylene vs. polyester, is probably real. Seems to
agree with the dielectric quality of polypropylene vs polyester as
determined by independent studies, e.g.; Jung's capacitor comparison
test.
As to *absolute* leakage values though, you are probably right. All
the environmental effects you mention, and temperature, humidity,
surface contamination as well, would change the leakdown rate.
I ran some dielectric leakage tests once, pin-to-pin, for TO-8
headers, used to package low leakage op-amp chips. We had to use a
dry N2 box to make the measurements, and thouroughly decontaminate the
packages with solvent washes followed by oven drying. Otherwise, the
surface adsorbed humidity and surface contaminants dominated the
leakage measurements.
Neat experiment. Thanks for the data.
Regards,
Dangerdave
James Meyer
wrote:
>the renowned James Meyer <notj...@worldnet.att.net> wrote:
>
>> Ionizing radiation, alpha, beta, gamma, and X-ray, all present
>> in the inevitable background and cosmic radiation, will also cause
>> leakage in insulators. A lead shield should improve leakage
>> specifications.
>> Jim
>
>So, a charged "floating plate" PP capacitor could make a solid state
>dosimeter? ;-)
The theory is there. A regular electroscope type dosimeter is
just a charged cap that measures its own voltage.
Jim
Tom Bruhns
Mark Rehorst wrote:
>
> Did you control the temperature?
Just in a temperature-controlled lab environment. Variation up to about
1 degree C is possible.
>
> The capacitances will change significantly with temp and since the charge on
> the cap is approximately fixed,
> the voltage will rise or fall...
Right. See the caveats in the original posting. That's specifically
why I suggested tuning back in for the second set of measurements next
week. Till then, the caps will be undisturbed except by the fairly
benign lab environment. But the very small measured change at least
suggests a much longer time constant than the reference I gave.
Cheers,
Tom
martin griffith
scribbled:
snip marvellous stuff
>. Be really careful about leakage
>paths if you try this at home!
>
>Tom
Nice bit of home-brew experimenting, and you didn't mention Spice
once!
I think its worth mentioning that Bob Pease has acouple of articles
about this sort of thing at the national semiconductor site, which
may interest you, especially the one about teflon as an insulator
martin
What happens if you play blues music backwards?
Your wife returns to you, your dog comes back to life, and you get out of prison.
Bill Sloman
wave...@SPAMbigfoot.com (martin griffith) wrote:
> On Thu, 29 Jul 1999 13:39:05 -0700, Tom Bruhns <to...@lsid.hp.com>
> scribbled:
> snip marvellous stuff
> >. Be really careful about leakage
> >paths if you try this at home!
> >
> >Cheers,
> >Tom
>
> Nice bit of home-brew experimenting, and you didn't mention Spice
> once!
>
> about this sort of thing at the national semiconductor site, which
> may interest you, especially the one about teflon as an insulator
I must say that Tom Bruhns has done a nice bit of work here, and I'll be
following the thread with interest.
On the teflon insulators, my posting that Tom Bruhns seems to be
following up referred to the "PTFE press-fit terminals" you can buy from
Farnell, if you can find them in the catalogue. "PTFE"is just British
English for Teflon (which anybody with a chemical education knows to be
poly-tetra-fluor-ethylene, or poly-1,2,3,4 fluoro-ethene). We used them
at Cambridge Instruments when we were building really low current
measuring gear, or standing off highish voltages.
--
Bill Sloman, Nijmegen
Sent via Deja.com http://www.deja.com/
Share what you know. Learn what you don't.
Tom Bruhns
if I was really serious about getting to the absolute lowest possible
external leakages. It's just a "quick and dirty" experiment to get an
idea of what modern film caps can do, with no extreme efforts to keep
down the external leakages. By the way, I've been bit by surface
leakage on DIP-8 packages. There's a chopper-stabilized op amp from
Harris that has quite low input bias current, if you are careful. I
used it as an amplifier for a Schottky-diode RF detector which could
"see" RF voltages down to about 30 microvolts. Mind you, you don't get
a whole lot of signal out when you put 30 microvolts of RF in. But I
was plagued by drift which seemed related to humidity. I finally
tracked it down to leakage from the power supply pin (pin 7) on the DIP,
to one of the stabilization cap pins, pin 8. I was longing for the
vapor cleaning tank we used with mass spectrometers at HP's California
Analytical Division, as I tried to clean it sufficiently at home. The
cap ended up tied to pin 8 in the air, with no circuit board contact,
and a couple coats of baked-on clear acrylic plastic spray (after
thorough and careful cleaning) helped out a lot; it's now quite
useable. Trying to get to extremely low leakage currents in home
electronics construction is an interesting challenge.
By the way, I'm doing the capacitor experiment at work. To do it at
home would require wiring up some very low bias current amplifier, and
here I can just use my bench voltmeter, which is high enough input
impedance.
And I agree, Bob Pease has documented some good things about working
with low bias and leakage currents. A lot of this has appeared in his
"Pease Porridge" columns.
I'll do the next measurement on the caps in about a week, unless I get
impatient and try one at the end of this week. I want to avoid making
measurements very often because every 1 second measurement does disturb
the charge some.
Cheers,
Tom
Roy McCammon
>
> But I
> was plagued by drift which seemed related to humidity. I finally
> tracked it down to leakage from the power supply pin (pin 7) on the DIP,
> to one of the stabilization cap pins, pin 8.
Sounds like a job for guard rings.
Tom Bruhns
Roy McCammon wrote:
Yeah, but they are mighty hard to apply to the plastic IC package. The
problem was leakage actually across the package between pins 7 and 8, as
nearly as I could figure. (Had I been too sloppy with flux from
soldering?? Mebbe, but the aerosol flux remover seemed to have a really
hard time cleaning it up completely.) Putting pin 8 (and its associated
capacitor lead) up in the air "dead bug" style didn't solve the problem
directly. It's also a bit of a problem knowing what to tie the guard
to: with op amp inputs, one is generally driven and you can use that
potential to drive the guard around the other one; but there didn't seem
to be any way to know for sure what voltage the chopper "hold"
capacitors would go to. Even the input offset voltage of most op amps
can be a problem for guarding. If you achieve 10^9 ohms from guarded
trace to the guard and there is a 1 millivolt difference between them,
then you've got 1pA of leakage, and ruined any chance of getting full
performance from an op amp capable of 10fA input leakage current. It's
kind of a fun mental/practical challenge to see just how good you can do
with homemade electronics and very low currents.
Cheers,
Tom
dang...@earthlink.net
wrote:
> There's a chopper-stabilized op amp from
>Harris that has quite low input bias current, if you are careful.
Small world. That's where we did the package leakage tests :-)
BTW: I just read that Harris Semiconductor will become Intersil.
Regards,
Dangerdave
Tom Bruhns
measurements, very preliminary after only one day of discharge:
>
> delta-t for all: 87000 seconds
> initial V for all: 10.6890
> apparent apparent
> time leakage
> Cap V2 delta-V constant current
> volts volts fA
> polyester 1 10.5928 -0.0962 112 days 111
> polyester 2 10.5442 -0.1448 74 days 116
> polyprop 1 10.6805 -0.0085 3.45 years 9.8
> polyprop 2 10.6842 -0.0048 6.15 years 5.5
>
Here's this week's update. I've put them back in a dark,
room-temperature place, to be hauled out again in a while (a few weeks?
Someone wanna send me email in September if I forget?) for another
measurement. If it's a couple weeks out, I'll probably post as a new
thread.
delta-t for all: 694500 seconds (total from the start at
990728@1230PDT to 990805@1325PDT)
initial V for all: 10.6890 volts
apparent apparent
time leakage
Cap V2 delta-V constant current
volts volts fA
polyester 1 10.2767 -0.4123 204 days 59
polyester 2 10.3930 -0.2960 286 days 42
polyprop 1 10.6773 -0.0117 20.1 years 1.7
polyprop 2 10.6749 -0.0141 16.7 years 2.0
Things would look even better if you took the delta from the previous
measurement rather than from the start.
It shouldn't be too much of a surprise that they look better this time,
since they were probably subject to some "dielectric soakage" effect in
the first hours after being charged. I did leave them charging for half
an hour before disconnecting the power, but that may well not be enough
to get to within better than 0.01% of final value.
Adrian Jansen
time constant !
On the same subject, but not the same type of caps, does anyone know the
typical TC of those supercaps ( ie the 5 volt, .5 to 1 farad things made by
National Panasonic etc )
--
--
Regards,
Adrian Jansen aja...@ozemail.com.au
Design Engineer J & K Micro Systems
Ph (07) 4639 4676 Microcomputer solutions for industrial control
Tom Bruhns wrote in message <37A9FAAE...@lsid.hp.com>...
Holger Petersen
>Good work, if only one could actually connect to a cap and get that sort of
>time constant !
I once (early-mid 70´ties) used a OTA (LM 3900 ?) from National Semi.
They had an application-note to use two (of 4 OTA´s in the package)
with a potentiometer to adjust the (un-) loading of a capacitor.
It was good to about 30 minutes or more in darkroom-use.
Sorry, I regret to have more details as the timer is gone as well as the
National-AN.
Yours, Holger
Ray S. Bowen
K7ITM
>Good work, if only one could actually connect to a cap and get that sort of
>time constant !
I was thinking about that today. If you can qualify one of those National
LMC6001 amps at, say, 10fA input bias current, you could achieve some really
long time constants with 1uF of polypropylene capacitor. The ones I'm
measuring for my posted results are all 0.1uF caps.
>On the same subject, but not the same type of caps, does anyone know the
>typical TC of those supercaps ( ie the 5 volt, .5 to 1 farad things made by
>National Panasonic etc )
Quite long. It turns out that I charged one of those a few months ago to a bit
over 5 volts, and measure it every once in a while. It's down to around 2.5
volts now, I think. It's right next to where I put the polyester and
polypropylene caps the first week. I haven't taken accurately timed and
recorded measurements on it, because I didn't realize that it would have such a
long self-discharge timeconstant. It's a 1.0F 5.5V unit. Maybe on Monday I'll
charge a couple of them up for a day or so and toss them in the drawer with the
polyester and polypropylene caps and make measurements on all of them every
couple weeks.
(Posting from home...)
Cheers,
Tom
Cheers,
Tom
James Meyer
>ajansen wrote:
>
>>Good work, if only one could actually connect to a cap and get that sort of
>>time constant !
>
>I was thinking about that today. If you can qualify one of those National
>LMC6001 amps at, say, 10fA input bias current, you could achieve some really
>long time constants with 1uF of polypropylene capacitor. The ones I'm
>measuring for my posted results are all 0.1uF caps.
I was thinking that before you finish testing things, you
should completely discharge a set of caps and make a series of
measurements that confirm that the caps stay at zero volts as well as
they hold a charge.
Jim
K7ITM
> I was thinking that before you finish testing things, you
>should completely discharge a set of caps and make a series of
>measurements that confirm that the caps stay at zero volts as well as
>they hold a charge.
>
I trust that it would be ok to run that in parallel with the others. I can
easily make up a jig to test some more of the same brands as I'm testing at
~10V.
Cheers,
Tom
Cheers,
Tom
James Meyer
That would be the best way. Make the tests the same way, but
with zero volts on some of them.
The leakage currents you're estimating are so small that if I
were measuring them, I'd be concerned that there might be some sort of
mechanism involved that was replacing, somehow, part of the charge
that was leaking.
If the caps that started off at zero volts in the same sort of
test setup stayed at zero volts, that would mean that the other
measurements were truly trustworthy.
A story to illustrate potential pitfalls.... (pun intended)
I took apart a defective transducer for a 50 KHz marine depth
sounder once. The transducer element was a disk of piezo ceramic
about 1.5" in diameter and 0.5" tall with silver plated electrodes on
each face. After I extracted the element from the housing, I washed
it thoroughly with soap and water to remove the salt water and bits of
potting material. Then I laid the element on the workbench and
pointed a hot-air gun at it from a foot or two away to dry it off.
After a half hour I went over to pick it up to see if it was dry. It
was warm, but not too hot. As soon as my fingers got close to making
contact with both sides of the disk, there was a sharp CRACK! and I
got a damn good shock. I later measured the capacitance of the disk
at 0.1 uF and the spark must have been in the kilovolt range.
I don't know whether it was the heat or the stream of air that
did it, but that sucker went from zero to near lethal voltages just
sitting there. I tried to duplicate that later without success.
Jim
John Woodgate
<notj...@worldnet.att.net> inimitably wrote:
> I don't know whether it was the heat or the stream of air that
>did it, but that sucker went from zero to near lethal voltages just
>sitting there. I tried to duplicate that later without success.
It's the heat that does that. There's a method of dating pottery that
archaeologists use that is based on something similar. When you heat up
piezo-electric material, 'soaked-in' charge energy that has been trapped
by discontinuities inside the material, in the form of internal
mechanical stresses, can escape and appear as electric charge again at
the surfaces.
--
Regards, John Woodgate, OOO - Own Opinions Only.
Phone +44 (0)1268 747839 Fax +44 (0)1268 777124.
Did you hear about the hungry genetic engineer who made a pig of himself?
PLEASE DO ****NOT**** MAIL COPIES OF NEWSGROUP POSTS TO ME!!!!
Tony Williams
John Woodgate <j...@jmwa.demon.co.uk> wrote:
> It's the heat that does that. There's a method of dating pottery that
> archaeologists use that is based on something similar. When you heat up
> piezo-electric material, 'soaked-in' charge energy that has been trapped
> by discontinuities inside the material, in the form of internal
> mechanical stresses, can escape and appear as electric charge again at
> the surfaces.
Sounds somewhat similar to what happened to the
graphite moderator inside Windscale.
--
Tony Williams.
Roy McCammon
>
> <37b01675...@netnews.worldnet.att.net>, James Meyer
> <notj...@worldnet.att.net> inimitably wrote:
> > I don't know whether it was the heat or the stream of air that
> >did it, but that sucker went from zero to near lethal voltages just
> >sitting there. I tried to duplicate that later without success.
>
> archaeologists use that is based on something similar. When you heat up
> piezo-electric material, 'soaked-in' charge energy that has been trapped
> by discontinuities inside the material, in the form of internal
> mechanical stresses, can escape and appear as electric charge again at
> the surfaces.
How do you put it back?
dang...@earthlink.net
>>On the same subject, but not the same type of caps, does anyone know the
>>typical TC of those supercaps ( ie the 5 volt, .5 to 1 farad things made by
>>National Panasonic etc )
>
>Quite long. It turns out that I charged one of those a few months ago to a bit
>over 5 volts, and measure it every once in a while. It's down to around 2.5
>volts now, I think.
That's remarkable. You could replace many battery back-up circuits
with a simple cap, and eliminate battery replacements.
Regards,
Dangerdave
dang...@earthlink.net
phenomenon. Once the charge in the discontinuities was unbound by the
heat, there was none left to repeat the "shocking surprise".
Regards,
Dangerdave
dang...@earthlink.net
Clifton T. Sharp Jr.
> John Woodgate wrote:
> > <37b01675...@netnews.worldnet.att.net>, James Meyer
> > <notj...@worldnet.att.net> inimitably wrote:
> > > I don't know whether it was the heat or the stream of air that
> > >did it, but that sucker went from zero to near lethal voltages just
> > >sitting there. I tried to duplicate that later without success.
> >
> > It's the heat that does that. There's a method of dating pottery that
> > archaeologists use that is based on something similar. When you heat up
> > piezo-electric material, 'soaked-in' charge energy that has been trapped
> > by discontinuities inside the material, in the form of internal
> > mechanical stresses, can escape and appear as electric charge again at
> > the surfaces.
>
> How do you put it back?
Why, a charge pump, of course.
--
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Cliff Sharp | "Speech isn't free when it comes postage-due." |
| WA9PDM | -- Jim Nitchals, founder, FREE |
+-+-+-+-+-+-+-+-+-+- http://www.spamfree.org/ +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
Tom Bruhns
months. It was _probably_ during April that I charged it up to around
5.25V. I think I'll re-charge it and another one, though they do take
quite a while to fully charge, and track them more quantitatively for a
while.
Of course, the other thing you should ask is how they do at elevated
temperatures. Same with the plastic film ones.
Cheers,
Tom
K7ITM wrote:
>
> ajansen wrote:
> >On the same subject, but not the same type of caps, does anyone know the
> >typical TC of those supercaps ( ie the 5 volt, .5 to 1 farad things made by
> >National Panasonic etc )
>
> Quite long. It turns out that I charged one of those a few months ago to a bit
> over 5 volts, and measure it every once in a while. It's down to around 2.5
> volts now, I think.
...