Hi Mario,

One thing which is usually done in cases where power supply
malfunction can damage things is to use a crowbar circuit. This
consists of an SCR and a triggering circuit which turns on the SCR to
short the power supply to ground in the event of the voltage going
over a certain amount. If you google for crowbar circuit, you'll find
it. I'm not sure but I think that this can fairly easily be made to
draw only a few uA. You would also want to add a real fuse to the
input of the regulator so that the SCR would actually cause it to
blow.

Sean


On Wed, May 28, 2008 at 9:48 AM, Electron <electron2k4@infinito.it> wrote:
> At 13.16 2008.05.28, you wrote:
>>At 07:02 AM 5/28/2008, you wrote:
>>
>>>Hello,
>>>
>>>It looks like there's no suitable 5V Transil/Tranzorb when you want to protect
>>>a circuit that has to sleep most of the time (say draw no more than
>>>few uA's on
>>>its own when sleeping).
>>
>>You need to protect what (input, output, power supply?) , against
>>what (voltage, available current) , exactly?
>
> Power to several expensive and delicate +5V IC's.
>
> I already had a voltage regulator fail because of a stupid hardware bug of one
> of the above IC's, which made it draw around 190mA in some particular moments
> (bug then acknoledged in the errata): the TPS71550 voltage regulator fried and
> passed +20V to the rest of the board, frying everything.
>
> Because of this experience, I thought it would be wise to put a simple zener-like
> device to protect the expensive chips from possible future failures of the voltage
> regulator. But even if I knew zeners wouldn't do it, now I discover that even
> transils <10V have a lot of leakage current (those >=10V have indeed very little
> leakage current, and in fact having worked with them I had the memory that they
> had very little leakage current and would be a good solution to my problem). I
> cannot offer more than few uA of leakage current due to the circuit's very low
> consumption necessity. But it would still be nice to protect the precious chips.
>
>
>>>A 5.6V Zener already draws too much current at 3.0V, go figure at 5.0V!
>>>
>>>5V Transils/Tranzorbs rarely are seen, and now I know why: if reverse leakage
>>>current is in the urder of uA's on common transils/tranzorbs, on
>>>very low voltage
>>>ones instead it's about 800uA! So they aren't usable for micropower circuits:
>>>http://www.farnell.com/datasheets/35080.pdf
>>
>>Zeners in general have softer "knees" at lower voltages. 3.3 and 2.4V zeners
>>are practically useless, so we usually use more complex circuits, but sometimes
>>you can use the Vf of an LED.
>
> To reach 5V would require a white LED, and I'm afraid I'd have to select the
> right diode among many, since LEDs aren't characterized for this application.
>
>>
>>>In short, there's no way to protect with a diode a micropower 5V line?
>>
>>Is it an input, an output, a power supply, what?
>>
>>>Ok, "good designs never break", but my experience is different.
>>>
>>>I am now using a 50mA polyswitch as a solution (cannot see smaller
>>>ones), I just
>>>hope it reacts quick enough to eventual overcurrents.
>>
>>They are thermal in nature so they don't react particularly quickly, the
>>trip current is temperature sensitive, so you need to allow a lot of margin
>>for higher temperatures and tolerance,  and their capacity for interrupting
>>current, and voltage rating, are often unimpressive. Voltage drop might be
>>an issue too. Other than that, they're great. ;-)
>
> i.e. almost nothing remains ;)
>
> Cheers,
> Mario
>
>
>>
>>>Best regards,
>>
>>Spehro Pefhany --"it's the network..."            "The Journey is the reward"
>>speff@interlog.com             Info for manufacturers: http://www.trexon.com
>>Embedded software/hardware/analog  Info for designers:  http://www.speff.com
>>
>>
>>
>>--
>>http://www.piclist.com PIC/SX FAQ & list archive
>>View/change your membership options at
>>http://mailman.mit.edu/mailman/listinfo/piclist
>
> --
> http://www.piclist.com PIC/SX FAQ & list archive
> View/change your membership options at
> http://mailman.mit.edu/mailman/listinfo/piclist
>
-- 
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist