[EE] Voltage Applied to non-powered LDO Regulator?
Jon Chandler
USB or a 5 volt supply. The complication is that the circuit may also be
powered by 3.3 volts directly from the ICSP connector, with with or without
the other USB/5V source being connected. If the circuit is only powered by
3.3v on the ICSP connector, the LDO has 3.3 volts on its output, which is
3.3 volts more the Vdd. This situation seems to be frowned on when it
happens to be addressed in LDO data sheets.
Is power on only the output side of an LDO a concern?
Are any parts specifically rated to handle this?
Thanks, Jon
--
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist
Olin Lathrop
> If the circuit is only powered by
> 3.3v on the ICSP connector, the LDO has 3.3 volts on its output,
> which is 3.3 volts more the Vdd.
That makes no sense. What is ICSP power connected to if not Vdd?
> Is power on only the output side of an LDO a concern?
Yes.
> Are any parts specifically rated to handle this?
Yes.
********************************************************************
Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products
(978) 742-9014. Gold level PIC consultants since 2000.
Jon Chandler
but here goes.
Essentially, the circuit may be powered from either or both sources:
Source 1: external supply via a 3.3V LDO
Source 2: a regulated source connected to the output side of the LDO.
If the circuit is only powered by source 2, the LDO has 3.3V on its output
compared to 0 on its input.
This appears to be beyond the maximum ratings for the regulators I have
looked at where it is
actually specified.
Any wisdom you could share on voltage regulators that can handle this
situation would
be welcome.
Thanks,
Jon
ivp
For some it could be. The advice for families like the 78xx is to
include a diode from output to input in case the input is shorted,
causing Vout > Vin. The diode discharges Vout capacitors
> Are any parts specifically rated to handle this?
Don't know. When I have dual sources of Vcc (eg mains and
back-up battery) I protect the output of the regulator with a
diode. The voltage drop across that diode is compensated for
by jacking up the reference (commonly called Ground) pin
with a similar diode, thus raising Vout by Vdiode
Jon Chandler
compensating for the voltage drop with an additional diode. I had
considered a blocking diode but I've never seen this trick to
compensate for the drop.
Jon
Michael Watterson
> Thanks, That's a good idea about adding a blocking diode and
> compensating for the voltage drop with an additional diode. I had
> considered a blocking diode but I've never seen this trick to
> compensate for the drop.
>
> Jon
>
>
>
then a 1N4148 with a bias resistor from supply rail as the compensation
diode, otherwise one diode could be 0.5 to 0.6V on output (depending on
Schottky diode type and current) but only 0.15V on the "compensation".
ivp
> compensating for the voltage drop with an additional diode. I had
> considered a blocking diode but I've never seen this trick to
> compensate for the drop.
The same can be accomplished with a couple of resistors too. For
example I use LM317 as 3V3 supplies, as they can be set to any
output voltage from 1.2V upwards. Typical linear regulators can
also be set to any voltage, with their nominal Vout as the baseline
Vdiode can vary with current so you might see loose regulation
if the circuit demand changes. Typically Vdiode increases as the
through current increases. The diode on the reference pin conducts
only a reference current, perhaps a few mA, whereas the output
diode could be conducting a hundred times that. It's a quick test
with a couple of load resistors to test the regulation at extremes
ivp
> output and then a 1N4148 with a bias resistor from supply rail as
> the compensation diode, otherwise one diode could be 0.5 to
> 0.6V on output (depending on Schottky diode type and current)
> but only 0.15V on the "compensation"
That's right. The system can be somewhat dynamic as the load
changes, so it's a technique best used on a circuit about which
something is known. It would probably be safe to allow Vreg to
droop a few percent under heavy load, say 3V3 down to 3V2
for example, unless there's something that relies on tight regulation,
such as an A/D, but that could (perhaps should) be covered with
a separate voltage reference
William "Chops" Westfield
On Jun 19, 2010, at 3:36 PM, Jon Chandler wrote:
> If the circuit is only powered by source 2, the LDO has 3.3V on its
> output
> compared to 0 on its input.
Having the input disconnected is not the same as have 0V on the input...
It's pretty common in lm317/etc datasheets to see a (normally) reverse
biased diode from regulator output to regulator input to provide a
reverse path for current (ie from the output filter capacitance.)
Something like that might work.
BillW
Josh Koffman
> For some it could be. The advice for families like the 78xx is to
> include a diode from output to input in case the input is shorted,
> causing Vout > Vin. The diode discharges Vout capacitors
Would this be a diode going from Vin to Vout with the cathode at Vin?
> Don't know. When I have dual sources of Vcc (eg mains and
> back-up battery) I protect the output of the regulator with a
> diode. The voltage drop across that diode is compensated for
> by jacking up the reference (commonly called Ground) pin
> with a similar diode, thus raising Vout by Vdiode
Interesting! Is the second diode just from Vgnd to ground, with the
cathode to ground?
Josh
--
A common mistake that people make when trying to design something
completely foolproof is to underestimate the ingenuity of complete
fools.
-Douglas Adams
ivp
> at Vin?
Yes. Such a circuit is generally in the datasheet
> Is the second diode just from Vgnd to ground, with the cathode
> to ground?
Yes
If you were to use a zener diode, it would be the other way around,
ie anode to 0V, cathode to the reference pin, with a bias current of
a few mA (eg 1k resistor) from Vout to zener cathode, to get the
zener conducting, as it's in a reverse-biased orientation, unlike a plain
diode which is forward-biased by the reference pin current
RussellMc
>> at Vin?
> Yes. Such a circuit is generally in the datasheet
Yes - but if there is no Vin the output will now be loaded via the
diode path.Depending what else is connected to Vin this may cause
problems. You could then feed Vin via another diode so that direct
application of 3.3V to the regulator output then 'floats' the whole
regulator. Some thought as to secondary effects would be wise.
Note that a diode in the ground pin may cause decrease in regulation quality.
Using a reverse bias protected part is probably the "purest" solution.
- see my next post.
Russell
RussellMc
>> 3.3v on the ICSP connector, the LDO has 3.3 volts on its output,
>> which is 3.3 volts more the Vdd.
> That makes no sense. What is ICSP power connected to if not Vdd?
His description makes complete sense when read through.
If necessary you could draw a diagram to make it clear.
There are two sources of voltage supply to ICSP.
When one is powered and the other unpowered it results in a regulator
being reverse biased, as described.
>> Is power on only the output side of an LDO a concern?
> Yes.
:-)
>> Are any parts specifically rated to handle this?
> Yes.
:-).
You can search for eg "reverse bias protected".
Here's an example of regulator that will draw ~10 uA when reverse
biased by 5V in the 5 volt version. It has a range of Vout versions. .
http://www.onsemi.com/pub_link/Collateral/NCP3335A.PDF
Family & others:
http://www.onsemi.com/PowerSolutions/parametrics.do?id=366&blockId=2024
It's not what you'd probably use BUT is an example of something that
may do exactly what you wan.t
500 mA max, 260 mV drop at 500 mA, stable for all output capos >= 1 uF,... .
Quite a good example of a capable modern LDO.
Note that there are 108 LDOs in the above selector guide BUT only 18
of them are reverse bias protected. It's not a common characteristic,
but available when needed.
With more work you can do similar with a MOSFET (or other) output
switch on the LDO that only turns on hard when Vin is present.
You could do this with as little as 2 x jellybean bipolar transistors
and 4 resistors. Using a MOSFET as the high side switch will allow a
very low voltage drop when suitably designed. If using jellybean
bipolars it will depend on current and drive level. Vsat down to
around 0.1V should be achievable.
Russell
Jon Chandler
This is exactly what I was looking for. I couldn't think of the correct
search term. I figured that my application wasn't totally unique.
Jon
> You can search for eg "reverse bias protected".
> Here's an example of regulator that will draw ~10 uA when reverse
> biased by 5V in the 5 volt version. It has a range of Vout versions. .
> http://www.onsemi.com/pub_link/Collateral/NCP3335A.PDF
>
> Family & others:
> http://www.onsemi.com/PowerSolutions/parametrics.do?id=366&blockId=2024
>
>
Michael Watterson
> O
>
> Interesting! Is the second diode just from Vgnd to ground, with the
> cathode to ground?
>
> Josh
>
. It's common to simulate Car Battery volts using 7812 regulator.
In that case you use two diodes in series (1N4001 or 1N4148) and bias
resistor from supply to adjust the current. I usually decouple with
100nF cap also.
Dario Greggio
"changes" or "upgrade" in Microchip forum?
:)
--
Ciao, Dario
--
Cyberdyne
Xiaofan Chen
> ...anybody feels like commenting or anyway spending some words about the
> "changes" or "upgrade" in Microchip forum?
>
> :)
Hopefully the next MPLAB Update will not like this. ;-)
For those who are interested to comment, please go here.
http://www.microchip.com/forums/m507263.aspx
Some other relevant threads:
http://www.microchip.com/forums/m506722.aspx
http://www.microchip.com/forums/m507152.aspx
--
Xiaofan http://mcuee.blogspot.com
Olin Lathrop
> Any wisdom you could share on voltage regulators that can handle this
> situation would be welcome.
Find one with a diode from output to input internally, or add such a diode
externally. Note that most LDOs have a P FET from input to output, so this
diode is inherent to the design. The MCP1700 are like that, and so are many
others. I've used MCP1700 as you described many times. The diode is shown
explicitly in the datasheet.
Olin Lathrop
> Thanks, That's a good idea about adding a blocking diode and
> compensating for the voltage drop with an additional diode.
It can be useful, but note that you lose voltage accuracy. If you're using
Vdd as your A/D reference making absolute measurements, this scheme may add
too much error.
********************************************************************
Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products
(978) 742-9014. Gold level PIC consultants since 2000.
Olin Lathrop
> Thanks, That's a good idea about adding a blocking diode and
> compensating for the voltage drop with an additional diode.
Another point I forgot to mention in my previous message. If you do that,
you have added significant voltage drop to your "LDO". That may defeat the
purpose of using a LDO in the first place.
********************************************************************
Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products
(978) 742-9014. Gold level PIC consultants since 2000.
Olin Lathrop
>>> If the circuit is only powered by
>>> 3.3v on the ICSP connector, the LDO has 3.3 volts on its output,
>>> which is 3.3 volts more the Vdd.
>
>> That makes no sense. What is ICSP power connected to if not Vdd?
>
> His description makes complete sense when read through.
I could guess what he probably meant, but I shouldn't have to do that. Yes,
this is in part to point out a stupidly worded question and to hopefully
teach him to spend 2 seconds and proofread what he wrote next time.
>>> Is power on only the output side of an LDO a concern?
>
>> Yes.
>
> :-)
>
>>> Are any parts specifically rated to handle this?
>
>> Yes.
>
> :-).
Same thing. This guy needs to learn to ask what he really wants to know,
and to formulate his questions clearly. Because he didn't I was hoping he'd
waste one go around of questions. Unfortuately people like you jump in and
help him anyway, so in the end he gets away with it. Shame on you.
Note that I did give him a decent answer once he finally asked a reasonable
question. I almost didn't reply because his reply to me was top posted, but
I let it go to stick with the point that he fixed his question that time.
********************************************************************
Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products
(978) 742-9014. Gold level PIC consultants since 2000.
Jon Chandler
It must be really tough to know everything and have the world refuse to
follow the standards of one with as much knowledge as you have.
In the future, please feel free to ignore my postings, as I'm certain you
shall do, with my top-posted reply. I hate to cause such a knowledgeable
being as yourself undue duress trying to decipher my incoherent
blathering.
To all others, thanks for the helpful replies.
Jon
Marechiare
> words about the "changes" or "upgrade" in Microchip forum?
>
> :)
That upgrade to Microsoft's software on some site, how is this
connected to [PIC] tag? Anyway, bring them all here and there will be
no need in the discussion at all :-)