IC damage due to reversed polarity supply (or how to be utterly inept)

[JohnnyMalaria]

   

Okay, so in a skillful display of ineptitude, I managed to install some DIP quad op-amps and some other expensive ICs the wrong way around. Please don't ask how. It has to do with trying to troubleshoot USB backfeed which I had never heard of. WTF. Thankfully, I used sockets.

Anyway, I have read that ICs generally have diode protection against such misfortune. Is that really true? All these ICs were operating at +/-15VDC. "Were" is probably the operative word here.

[james_s]

Been there, done that. In my experience the ICs are usually damaged but it can't hurt to check.

[JS]

When you plug yours IC backwards is wise to power the circuit from a lab power supply and go up slowly with a low current set limiting it from blowing things. If that's done you can catch the thing fast enough not to burn anything. At a volt or two the thing will be drawing as much current as it should at full voltage, so you know there's something wrong.

That said I still cook stuff and smell smoke now days, happens to everybody, nothing like the smell of popped caps in the morning.

JS

[JohnnyMalaria]

Thanks.

The ICs are on a board that is one of a handful that make up an instrument. When I was building and testing this particular board in isolation, I did indeed use a bench supply for the very reason you say  But at some point you have to pull everything together. To minimize the risk of ESD damage I took the ICs out of their sockets before manhandling the board into the enclosure, connecting all the cables etc etc. The orientation I put it in was the opposite that I had been using during development and testing. I said to myself that I should turn it around but I couldn't be bothered. I put the ICs back in in the orientation I was used to visually. Hence...

[Whales]

Anyway, I have read that ICs generally have diode protection against such misfortune.

"Diode protection" typically means that all pins on the device are tied together (often through the internal power rails) via diodes that only turn on when you exceed the voltage ratings of the part.  Designed to eat high-voltage ESD zaps, these diodes also tend to act like shorts if you power the chip backwards or with a voltage that's too high.

Unless the diode protection you talk about comes with a schematic I'd presume it's of this type, which is the opposite of what you would have wanted 

[Rerouter]

Ive never found a reverse polarity protection diode on the chips i have worked with, only the esd diodes for digialt IO and the clipping diodes across the inputs of some op amps.

Generally you would have reverse polarity protection on a board level, sounds like this was ommitted

[amyk]

"Diode protection" typically means that all pins on the device are tied together (often through the internal power rails) via diodes that only turn on when you exceed the voltage ratings of the part.  Designed to eat high-voltage ESD zaps, these diodes also tend to act like shorts if you power the chip backwards or with a voltage that's too high.

This. If you study the basics of how ICs are made, especially CMOS (highly recommended even if you never design ICs), you'll find that, to a rough approximation, everything is either N or P-doped silicon and what replaces "insulator" in a lot of places is a reverse-biased, "parasitic diode". It's also why MOSFETs have a "body diode". See this illustration:

https://en.wikipedia.org/wiki/File:Cmos_impurity_profile.PNG

If you reverse the polarity of the supply, the huge parasitic diode between the N and P body material becomes forward-biased and essentially appears as a diode directly across the supply, which will conduct as much current as it can until something melts or breaks.

[JohnnyMalaria]

these diodes also tend to act like shorts if you power the chip backwards or with a voltage that's too high.

That's very interesting. I saw a symptom of that which first alerted me to having done something wrong. Specifically, I was applying an input signal to one of the op-amps but got no output. When I measured the resistance of my input, it turned out to be a short which really puzzled me. The short only occurred when the (wrong) power was on. Without power there was no short.

[Siwastaja]

With current limited to very low levels, they can survive, because of the intrinsic diode structures. But these diode structures can't stand much current. A few mA tops, typical. Of course, when the whole power is reversed, these diodes act in parallel and can take a bit more. Still, usually, if you just have a constant voltage supply that can supply amperes, the chips are almost 100% surely damaged.

Sometimes, the damage is invisible: some operating parameter changes, possibly only slightly outside the spec. So if it looks operational, you can later spend weeks debugging some strange issue due to this! I have learned the hard way always to replace the ICs when I suspect damage (for example, FET drivers when the FET blows short, even if the driver looks operational).

[NiHaoMike]

Julian Ilett once connected a 74HCT138 backwards, let the smoke out of it, and it somehow seemed to still work!

[JS]

Generally you would have reverse polarity protection on a board level, sounds like this was ommitted

Maybe/probably/most likely not, but the board polarity was correct, unless the diode in the board is looking how the operator has put the ICs it can't do nothing about it but to  the world around burn.

JS

[glarsson]

Many years ago I inserted an Intel 8080 the wrong way. Expensive chip that used +5V, -5V and +12V. Got a bit warm but survived.

[Zero999]

Julian Ilett once connected a 74HCT138 backwards, let the smoke out of it, and it somehow seemed to still work!

If it was in a plastic package, it's possible the die got hot enough to damage the packaging, but not hot enough to destroy the IC.

[amyk]

Many years ago I inserted an Intel 8080 the wrong way. Expensive chip that used +5V, -5V and +12V. Got a bit warm but survived.

Looking at the pinout, that would put +5 on an address bus output, -5 on another, +12 on a data bus input, and ground one of the clock inputs. Not as bad as reversing the polarity completely.

[JohnnyMalaria]

So I've ordered and received replacement ICs (all the 741s were dead, the high precision quad op-amp is functional but who knows what might happen with time, and the two 4-quadrant multiplier/dividers haven't been tested but I'm assuming they are damaged). Go, Mouser! I ordered them at 7pm last night.

Anyway, I'm going to connect my bench supply to the board and I want to make sure I don't backfeed the down converters I'm using. I have a bag full of diodes that I got about 15 years ago that I'll put between the board and the down converters. I decided to test them with my multimeter first. I didn't expect what I saw. I tested the diodes by connecting them to the high side of the power supply (+15V) and measured the voltage. Of course, ideally it should be zero. The graph tells a different story. The y-axis is logarithmic and the horizontal reference line is 15V. All I know about the diodes is that there are different classes:

DPR - unknown
PR - fast recovery
SPR - super fast recovery
UFR - ultra fast recovery
SR - unknown

The PR and UFR diodes behave pretty much as I'd expect (0.15 and 0.05V mean, respectively). The SPR and SR appear to do nothing.

I've read about bidirectional diodes (two zeners) but I can't find any references to the above classes.

Obviously, I don't think it's a good idea to use the SPR or SR but do you think the PR or UFR diodes will be okay? Is it "normal" to have a voltage of ~0.1V when the diode is reversed or should I stay clear of these, too. I don't have any 1Nxxxx diodes on hand (and I'm kicking myself for not ordering them yesterday  )

Thanks.

[jmelson]

. To minimize the risk of ESD damage I took the ICs out of their sockets before manhandling the board into the enclosure, connecting all the cables etc etc.

Well, so instead of a small risk of ESD, you had a high risk of EMD (That's Electro-Massive Damage.)

Really, unless you are building this in a nice warm hut in Siberia in the winter, this ESD thing is just WAY overrated.
Yes, it can happen, and yes, it can also cause delayed failures that are hard to track down.  But, in general, with just the SIMPLEST of tricks, it is a great rarity.  I use NO ESD prevention gear at all (wrist straps, mats, etc.)  I do have a paper-covered metal table that is weakly grounded, and touch the table before picking up any boards.  We do keep the humidity up in winter, but mostly for health reasons.

Jon

[JohnnyMalaria]

Well it's pretty humid here in the North Carolina late spring. 25C/75%RH in my garage right now.

[ebastler]

Well it's pretty humid here in the North Carolina late spring. 25C/75%RH in my garage right now.

That should do for ESD protection.

[james_s]

That makes me feel sticky just thinking about it. Yeah I don't think ESD is something you'll have to worry about right now.

[JohnnyMalaria]

It's all up and running now  Turns out the expensive chips survived but all the quad 741 did not. Well, at least I now have spares.

Couldn't believe my bench supply went screwy yesterday. Thanks to the wonders of Amazon, I'll have a new one tomorrow (for next time....)

I think I'll have Vistaprint make me a banner poster that will read: "If it's not one thing, it's another" with an expletive or two thrown in.

[JS]

Well it's pretty humid here in the North Carolina late spring. 25C/75%RH in my garage right now.

That should do for ESD protection.

Pretty rare here to get under 60%, that might explain why I haven't seen any components damaged by ESD even when components are sold and shipped in plastic bags...

JS

[GadgetBoy]

I spent *TWO HOURS* troubleshooting a simple triple-nickel blinker circuit. Turned out I was trying to use the pins as:

15
26
37
48

instead of

18
27
36
45

I felt *REALLY* dumb when I figured it out.

[Zero999]

Well it's pretty humid here in the North Carolina late spring. 25C/75%RH in my garage right now.

That should do for ESD protection.

Pretty rare here to get under 60%, that might explain why I haven't seen any components damaged by ESD even when components are sold and shipped in plastic bags...

JS

Temperature also makes a difference. Colder air holds less moisture, so even if the relative humidity is near 100% outside, it will be much lower inside a heated building. I notice your location is Argentina which varies considerably from one part of the country to the other.

[JS]

Well it's pretty humid here in the North Carolina late spring. 25C/75%RH in my garage right now.

That should do for ESD protection.

Pretty rare here to get under 60%, that might explain why I haven't seen any components damaged by ESD even when components are sold and shipped in plastic bags...

JS

Temperature also makes a difference. Colder air holds less moisture, so even if the relative humidity is near 100% outside, it will be much lower inside a heated building. I notice your location is Argentina which varies considerably from one part of the country to the other.

I'm from Entre Ríos, 40+°C on summer to 0-°C on winter, even in winter is usually condensing, even higher rel humidity than in summer. I have more trouble with too much humidity than low, I think humidity killed one of my meters last winter. Unused for a while would read negative resistance and drift like a boss while you blink.

Granted it depends on the type of heating used, today I was talking about it, also I use very little heat as I don't feel the cold, I'd stand side to side short sleeved with a body from the South (-20°C on winter) wearing 4 layers of clothing.

 I'm looking for a dryer enviroment to setup the lab as it's way too wet where I have it now. Heater on and condensation even on the floor.

JS

Enviado desde mi LG-M250 mediante Tapatalk