Negative voltage protection

[mrkev]

Hi.
I am designing new device at work, it is gonna be powered from 12V and there is an switch mode regulator simmilar to TPS54426 at the input. One of the requirements was that it should be able to withstand when someone swaps the power (idealy up to -20V on the input).
I used one shotky diode ( SK16 equivalent ) in series with the regulator and figured out that this should be enough, even tho it has quite big reverse leakage current (500uA at 25°C and up to 5mA at 75°C). But one of my coworkers insisted that since the leakage is significant, there should be another shotky cathode to VIN and anode to ground.

My idea is, that even tho there is nothing about allowed negative currents to VIN pins in the datasheet, they should be able to withstand at least 1mA current each (there are two). And that there is always protective diode between positive voltage and ground inherent to integrated circuit technology.

Any ideas?

[mrkev]

Well, thanks evb149. It might be just my hurt ego, but it was presented to me as a big mistake, which I don't think it is.

The point is, to which degree would you consider using something that is inherent to the technology (like that there is always a reverse diode on practically every pin of every chip) and where you add (perhaps unnecessary) protection. My collegue tends to add zener diodes everywhere, which is IMHO not the best practice.

[mclemens1969]

Depends on use and cost point.  If it's going into space for example to orbit the Earth, then 10,000 additional discrete components is cheap.

[phliar]

Design reviews are great, and professionalism means that you don't take the feedback personally. It does take a little time to get to that point... 

I regard review feedback as a discussion. If I feel strongly about something it is my job to convince the reviewer that he is wrongly-wrong but sometimes I turns out to be mistaken!

I personally tend to put more checking/validation in my designs than is optimal so I can understand your reviewer's point of view. But I'm willing to be convinced; and sometimes I realize that it doesn't make much difference either way so I just go with what the reviewer wants.

(Professionally I do software, but these principles are the same for any kind of engineering.)

[thanasisk]

My feeling is that design choices should be made after the design team has agreed on the precise specifications for the product. What are the expected operation temperature ranges for your equipment? What are the stress factors and extreme operation conditions/scenarios that you have to take into account  for your design?

By the way, if your equipment reaches 75 degrees occasionally, then perhaps the regulator would not behave as you had hoped and the reviewer would be justified (>>1mA current).
BUT this has to be agreed by the team by standardizing specifications within the designs!


Anyway, for reverse polarity protection there is always the fancier mosfet path (maybe it fits your circuits):
http://www.geofex.com/article_folders/mosswitch/mosswitch.htm

[mrkev]

Design reviews are great, and professionalism means that you don't take the feedback personally. It does take a little time to get to that point... 

I regard review feedback as a discussion. If I feel strongly about something it is my job to convince the reviewer that he is wrongly-wrong but sometimes I turns out to be mistaken!
...

My feeling is that design choices should be made after the design team has agreed on the precise specifications for the product.

I feel the same way, but my coworker is one of those people that are really hard to have debate with  , usually he just talks over you and don't let you finish any point.

What are the expected operation temperature ranges for your equipment? What are the stress factors and extreme operation conditions/scenarios that you have to take into account  for your design?

By the way, if your equipment reaches 75 degrees occasionally, then perhaps the regulator would not behave as you had hoped and the reviewer would be justified (>>1mA current).
BUT this has to be agreed by the team by standardizing specifications within the designs!

It should't reach 75 degrees in any case, the expected ranges are from 0°C to 50°C and the whole purpouse is to protect device damage in time of installation (so technician won't blow it up by swapping the + and - power wire)... This is why I don't think it was necessary. I am well OK with adding one diode, just to be safe, I just don't think it's as big of a deal. Especially because I worked as a on-chip-layouter while at university and I know that those clamps are not something that you just add (you'd have to make special measures to make device without them), the same way FET body diode is not separate diode...

[thanasisk]

Well good luck with your energetic coworker, perhaps next time you can tell him to let you finish your sentence,  take a breath, and explain.. 

Adding a diode wouldnt hurt and since it is cheap, just do it and go on with enjoying your life if it were a more complex situation/debate, it would be efficient you both ask for a third opinion

Aby the way, have you considered a glass passivated GP diode for your application?   It has lower leakage.

[Zero999]

Choose your battles wisely. Add the darn diode. It's not worth the time spent having the discussion.

I remember having a similar debate with an electrical goon who wouldn't sign off the safety test for a cabinet I made, because part of it wasn't earthed, which didn't need to be earthed because it was all separated, extra low voltage. I tried to tell him that it didn't need earthing but it was quicker just to spent 5 minutes just making an earth strap. If it was difficult/costly/time consuming to do, then I would have continued the debate, possibly bypassing him and going to his manager but it wasn't worth the trouble.

[DBecker]

You can't cover every scenario.  Someone will always be able to pick a condition or coincidence that will result in failure.

This is a perfect example.

The reverse voltage diode is there to protect against an installation error.  Will installation be happening at 75C?  No.

We don't even need to point out that a reverse voltage won't be at the voltage limit.  In real life, diodes will be far better than spec in this parameter.  And that leakage current will likely not be damaging.

Pick your battles in life.  Sometimes that means winning easy ones early rather than fighting close ones later.