Many Op-amps suddenly dead

[Dannyx]

Good day folks. I'd call this one pretty sick and it drove me nuts for the past couple of hours now: I'm working on a small project and testing things out on the breadboard. The project involves an op-amp, a comparator to be exact, the LM393...as basic as it gets. I didn't have one when I started, so I used a LM833N from my parts bin until my order of 3 LM393s arrived. Since the application is extremely basic, effectively comparing 2 voltages in an open loop setup, I thought it wouldn't matter which op-amp I used and sure enough, it worked as I expected: I tied the output to a NPN to drive a relay and tested by connecting the Inverting input to GND and the N/I input to VCC (12v). The output swings high, relay pulls in, as expected. Switch the inputs around and the output goes low.

I set the project aside for a few weeks. The new LM393 arrive in the meantime. I come back to it today, power up the breadboard, LM833 is stuck ON all the time......

Of course, I proceeded with the usual test: disconnect everything from the LM833, connect both inputs to GND and measure the output: output is still high at 12v ! After a long process in which I tried swapping wires around and doing everything I know, I eventually concluded that it must've died somehow, so I just said screw it and replaced with another LM833....same exact behavior ! Output is high on this one as well ! I even tried a THIRD device, same thing ! These things were new, just sitting there on a shelf !

Still perplexed, I threw in the towel on the LM388 and decided to move on to my 393s which I was going to use anyway: I plonk one on there, same exact setup, but get this: these ones don't turn ON ! That's right: the complete opposite

I took a picture just to make sure I'm not crazy and I'm not making some completely idiotic mistake. You can see the 393 just chilling there. Green wire is VCC, yellow wire jumps said VCC to N/I input, grey wire in bottom-right is GND and orange wire jumps GND to the inverting input. I measure on pin 1 and it's stuck low...not even a twitch on my meter...nothing ! So at this point, I'm just dead on the inside

Do brand new op-amps just randomly die ALL AT ONCE, in different ways like this ? Sitting in the parts bin, unused, doing nothing ? EMP pulse anyone No violent storms or volcanic activity nearby What do you make of this Only solution is to buy more.....

[CaptDon]

Are you sure grounding both inputs produces a valid output? I have never seen a bunch of disconnected solid state devices die just sitting on the shelf with one exception, a 12KW pulse power RCA AVQ-46 radar on my bench unterminated with just the open output waveguide was accidentally commanded to transmit for a few seconds. That indeed fried a few devices laying on the benchtop! I lost a couple of 1N23E/R mixer diodes that were not in their protective foil wrapper and a couple of other devices seemed to sustain damage. In your case I can't see how these devices are all damaged. Something flakey with your breadboard? Power on wrong pins or hooked up backwards? They don't normally die sitting on a shelf. Operator error?

[kripton2035]

just to clarify. where did you buy these opamps ?
today fake world of even simple electronics components is wider as you may think...

[Andy Watson]

It would help if you provided a schematic.

... drive a relay ...

Did you use a diode to catch the back-EMF of the relay coil.

..., but get this: these ones don't turn ON !

The LM383 are open-collector - they don't actively turn on. You need to add a pull-up resistor to Vcc to make it work.

[tooki]

Good day folks. I'd call this one pretty sick and it drove me nuts for the past couple of hours now: I'm working on a small project and testing things out on the breadboard. The project involves an op-amp, a comparator to be exact, the LM393...as basic as it gets. I didn't have one when I started, so I used a LM388N from my parts bin until my order of 3 LM393s arrived. Since the application is extremely basic, effectively comparing 2 voltages in an open loop setup, I thought it wouldn't matter which op-amp I used and sure enough, it worked as I expected: I tied the output to a NPN to drive a relay and tested by connecting the Inverting input to GND and the N/I input to VCC (12v). The output swings high, relay pulls in, as expected. Switch the inputs around and the output goes low.

I set the project aside for a few weeks. The new LM393 arrive in the meantime. I come back to it today, power up the breadboard, LM388 is stuck ON all the time......

Of course, I proceeded with the usual test: disconnect everything from the LM388, connect both inputs to GND and measure the output: output is still high at 12v ! After a long process in which I tried swapping wires around and doing everything I know, I eventually concluded that it must've died somehow, so I just said screw it and replaced with another LM388....same exact behavior ! Output is high on this one as well ! I even tried a THIRD device, same thing ! These things were new, just sitting there on a shelf !

Still perplexed, I threw in the towel on the LM388 and decided to move on to my 393s which I was going to use anyway: I plonk one on there, same exact setup, but get this: these ones don't turn ON ! That's right: the complete opposite

I took a picture just to make sure I'm not crazy and I'm not making some completely idiotic mistake. You can see the 393 just chilling there. Green wire is VCC, yellow wire jumps said VCC to N/I input, grey wire in bottom-right is GND and orange wire jumps GND to the inverting input. I measure on pin 1 and it's stuck low...not even a twitch on my meter...nothing ! So at this point, I'm just dead on the inside

Do brand new op-amps just randomly die ALL AT ONCE, in different ways like this ? Sitting in the parts bin, unused, doing nothing ? EMP pulse anyone No violent storms or volcanic activity nearby What do you make of this Only solution is to buy more.....

Are you even sure this is the problem?

I see you’ve tried to jam stranded wire into the breadboard. They’re not designed for that, so you may have some disconnected wires that you don’t notice. Use proper solid wire (0.5-0.6mm diameter (not cross section!!!), =22-24AWG), breadboard jumpers, or at the absolute minimum, strip your wire cleanly and tin it with solder before insertion.

[Dannyx]

Ok, let's answer some questions, though at this point I'm pretty sure I saw the answer I was looking for, so I AM indeed an amateur who doesn't know his op-amps (an idiot, basically)...

Are you sure grounding both inputs produces a valid output? Something flakey with your breadboard? Power on wrong pins or hooked up backwards? They don't normally die sitting on a shelf. Operator error?

Unfortunately I don't have a scope to see exactly what the output is doing, but with my DMM, it just sits there at 0v.....mocking me....
I went as far as to measure continuity directly on the Op-amp itself to make sure the pins are actually making contact all the way to the end of the corresponding wires and they DO, plus it doesn't explain why it worked the first time.
I THOUGHT I might have done something wrong with the first chip, indeed but why the others too ? I power the breadboard from my bench supply through a DC-jack, so polarity can't possibly switch, unless I move jumpers around, which I didn't, so that's not it either...

Then:

just to clarify. where did you buy these opamps ?
today fake world of even simple electronics components is wider as you may think...

We buy parts from TME, so there's no doubt in my mind they're genuine TI parts, although for such a basic and low power application I'd expect even the cheapest, worst, disgusting knock-offs off Aliexpress to at least swing from one goddamn rail to the other ! I wouldn't trust power electronics or precision stuff indeed, but this is too basic an application not to work !

Then:

Did you use a diode to catch the back-EMF of the relay coil.

Absolutely, plus that's the very first thing I did when I noticed it's no longer working: disconnect the output from anything else, so that's not it.

However, THIS:

The LM383 are open-collector - they don't actively turn on. You need to add a pull-up resistor to Vcc to make it work.

This indeed is the answer I mentioned and it made me go oooooohhhh......so THAT explains it ! Ok, THIS actually makes sense and I had a look at the diagram and the 393 indeed can't swing positive, only to GND, so this is definitely it and I was measuring it all wrong

However, it doesn't quite solve the whole mystery and doesn't explain what happened to the bunch of 833s and whey they seemed to work the first time....I couldn't have killed them one by one as I tried them on the breadboard...............couldn't I ? I even made sure to discharge myself

[jwet]

The mystery of why the 388's burning up is this.  The 388 is an audio amp.  It biases itself between the rails and is designed to be AC coupled to a speaker.  When you use it as a comparator, you are just pushing it hard high or low against the rails- this draws a lot of current- you can measure the supply current- I don't know how the transistor/relay is connected but if there is no current limiting, it won't be good. On a breadboard, there is also no heat sinking either, its own quiescent is .25W (20ma x 12v).  Overall, the amp just burns up- gross overstress.  Its not designed to be a comparator and operate in this DC coupled, rail to rail mode.  You could add a resistor to the output to limit the output current but overall, its just a bad pick, an LM393 is the correct part and has an open collector output as noted.  Look at the '393 data sheet for output drive circuits.

[wiresaho]

If it was working the first time, you have to have got the circuit right, so it cant possibly be that,   but are u sure ur powering the opamps without burning them out?

[Dannyx]

The mystery of why the 388's burning up is this.  The 388 is an audio amp.

Are you sure we're talking about the same part ? I used the LM833, not 388 - no, it's not a typo. It's THIS. Yes, it's audio-related still, but shouldn't "burn", let alone 3 of them in a row and with no load attached! I must've stressed that circuit on and off hundreds of times the first time it worked and it seemed to work perfectly - I assume it would've failed a lot later, if ever.

are u sure ur powering the opamps without burning them out?

I supply12v from my bench supply (as stable as a garden variety adjustable supply can provide in terms of noise, but still). It IS true that I MIGHT've overvolted the first 833 when I increased the supply to like 24v to test the stability of a 431 circuit I was playing with on the same breadboard, while the 833 was still connected to the same supply, but that doesn't explain what happened to the other 2 I tested, making sure the voltage is correct....who knows. Out of curiosity, I'll try sourcing a fourth 833 to see if it behaves differently...

[Doctorandus_P]

Opamps & comparators simply do not work that way. Most need some headroom both on the inputs and the outputs. Usually the inputs voltage range includes one of the power rails, but not both. Sometimes inputs are specified to a bit outside of one of the power rails. The output can need an headroom as big as 1.5V from either of the power rails.

Also when you tie both inputs of an opamp together, the output is undefined. The inputs always have some offset voltage, and it's polarity may varies (both in sign and amplitude) This small offset gets amplified by 100.000 or so, and then saturates the output towards one of the power supply rails.

If you want to test an opamp, build a real circuit with it. For example an amplifier with negative feedback, or an oscillator.

There are many different opamps, and each has it's own peculiarities. So also be aware that not all circuits work with all opamps.

In addition, always add decoupling capacitors to your IC's even on a breadboard. You can get oscillations or other weird behavior without them. Using stranded wire on a breadboard is also annoying. Solid core wire works much better on breadboards.

[ebastler]

Are you sure we're talking about the same part ? I used the LM833, not 388 - no, it's not a typo.

It was you who described the part as am LM388, in your first post. You consistently named it as an LM388, all three times you mentioned it.

Did you use a diode to catch the back-EMF of the relay coil.

Absolutely, plus that's the very first thing I did when I noticed it's no longer working: disconnect the output from anything else, so that's not it.

You missed the point there. If you had the relay connected without a free-wheeling diode, it may have killed the semiconductor parts via the coil's inductance spike, the first times you used the circuit. Disconnecting stuff later may have been to late then.

Are you sure grounding both inputs produces a valid output? Something flakey with your breadboard? Power on wrong pins or hooked up backwards? They don't normally die sitting on a shelf. Operator error?

I went as far as to measure continuity directly on the Op-amp itself to make sure the pins are actually making contact all the way to the end of the corresponding wires and they DO, plus it doesn't explain why it worked the first time.
I THOUGHT I might have done something wrong with the first chip, indeed but why the others too ? I power the breadboard from my bench supply through a DC-jack, so polarity can't possibly switch, unless I move jumpers around, which I didn't, so that's not it either...

Missed that point too, I'm afraid. The question was: If you properly ground both inputs, is that a well-defined state for the op-amp?

[Dannyx]

Ok, let me fix some stuff

Just got done replacing the f-ups in the original post too, so it was indeed a LM833 all along...sure hope I didn't miss any places where I referred to it as 338....so goddamn annoying so I hope that clears it once and for all.....even for me

Yes, the relay had a diode connected from the very start, but once it started acting up, I removed the connection between the base of the transistor and the op-amp anyway and never connected it back on any of the "new" ones.

I answered only the part about the physical aspects of the breadboard and forgot to deal with the first part of that last question: what happens with both inputs grounded (fig. 1). The output of all the 833s I tried is still stuck high - doesn't budge as far as I can measure with a DMM. DC resistance is also high between the output and Vcc/GND (Mohms). That's how I read you'd perform a crude test on an op-amp and I have done it before on a few occasions, though what I usually do is ground one input, while pulling the other one high and observing if the output goes to GND or close to the supply rail respectively (Fig. 2/3). Ok, it makes sense that the 393 can't pull up to Vcc - that's all clear now.

Also, I DO have like a 22uF electrolytic stuck in the breadboard right next to the chip but I pulled it to make the picture look clearer.

Stranded wire is what I have most on hand, all over the shop, and I used to tin the ends to make it easier to work with indeed, but I stopped bothering after a while, TBH. Quick'n dirty until I assemble the final PCB, you know . It's all for fun and not for production. I use tweezers to stick it in the breadboard and it works relatively well if I twist it really well so it doesn't come undone when it meets the springs down inside the hole.

[shapirus]

Not sure about LM833, but with LM393 you don't connect inputs to low impedance nodes like shown in your last post: read https://www.ti.com/lit/an/snoaa35d/snoaa35d.pdf starting from page 19.

[xvr]

what happens with both inputs grounded (fig. 1).

Output could be any. There is not a proper way to test opamp or comparator.

[Dannyx]

Ok, that's useful info, especially since I did PRECISELY that: tied both inputs together to the VCC rail or GND with no resistors. Now that I think of it, I don't know why I kept thinking connecting an op-amp's input straight to VCC would be ok(-ish). I guess I somehow got stuck with the idea that op amp inputs draw virtually "no current", so I can just connect them willy-nilly.

I doubt the 393s got damaged, but you never know. I'll have to test again properly. Still not sure what's going on with the 833s though...

Can I use my DMM to test the output of an open-collector op-amp without actually using a pull-up resistor there to measure voltage ? I assume if I measure DC resistance between the output and GND, it should read low when it turns "on" (N/I more positive than I).

[shapirus]

Can I use my DMM to test the output of an open-collector op-amp without actually using a pull-up resistor there to measure voltage ? I assume if I measure DC resistance between the output and GND, it should read low when it turns "on" (N/I more positive than I).

I'd expect that it would work. Try it and tell us. Verify the results with a pull-up resistor.

edit: I assumed that you meant comparators, not op amps (there are no open collector op amps). They are different devices. You need to call them by their proper names if you want people to understand you correctly.

[RoGeorge]

Not any opamp can be used as a comparator.  Some opamp models can only swing both inputs at once, but if you connect one input to plus and and the other one to minus, that will destroy the opamp.

In the datasheet of https://www.ti.com/lit/ds/symlink/lm833.pdf page 4 of 36 it is specified in remark (4) that LM833 does not support more than 0.6V differential input.  You can swing both inputs at once between anywhere between V+ and V-, but only if you keep them both at less than 0.6V apart, or if you add a protection resistor in series with each input, to limit the max current through the inputs of the opamp.

(4) Excessive input current will flow if a differential input voltage in excess of approximately 0.6 V is applied between the inputs, unless some limiting resistance is used.

[wiresaho]

I don't know if this helps or not,  and I've only used transistors not actually opamps,    but it could help if u dont know,  (but disregard this its only simple, if u already know.)

That when you power the base of a transistor, the base current has to travel back to where its power originated to work,   not just the collector and emitter,  both have to be a closed loop with their respective power that ran them.

Maybe op-amps have a similar situation cause they are made from transistors that maybe your not connecting the circuit right?     

sorry if this is unhelpful information,  just trying to help but might not be.

[jwet]

In your first write up, you had 388 and 833, I saw the 388.

The 833 has a limited common mode input range +-13-14 on +-15v, or about 1.5 to 10.5v on +12.  Outside that range, all bets are off.  The parts can reverse polarity, latchup, burnup, etc.

Op-Amps don't like exceeding their CM.

What's the load- a 1k resistor to a transistor?

Good luck.

[Dannyx]

What's the load- a 1k resistor to a transistor?

Yes, it's a 2n2222a (that annoying kind where you never know which pinout is the correct variant >_>) though I'm not sure about the exact value of the resistor. I wanted to keep the current quite light on the op-amp's output, so I think I went higher, to like 2.2k, since that's what I had in my junk box. It seemed to work wonderfully until now.

I think what happened is: I messed around with the wires as I was testing the 431 project on a different part of the bredboard and the op-amp circuit stopped working because some jumper was disconnected or moved out of place. Thinking there must be something wrong with the op-amp, I began messing with it until I eventually ended up connecting the inputs directly to the supply in an attempt to check its operation. With a 12v supply, this MAY have killed the original 833....then the other two I tried in the same exact setup...I sure hope the 393s fared better, though I may just have one left that I didn't try

I HAVE done this to op-amps a couple of times, but used a lower 5v for brief periods, so they survived.

[shapirus]

Yes, it's a 2n2222a (that annoying kind where you never know which pinout is the correct variant >_>)

Get one of those transistor testers. They're a blessing indeed.

[Dannyx]

I have one, the kind that tests ESR as well, but I was too lazy to go get it, so I turned the transistor around until it worked. I did this separate from the op-amp, of course.

[magic]

National's original LM833 is rated to survive any differential input voltage, but the schematic looks like it wouldn't work correctly with the inputs near ground. If any input is close to ground, the output unconditionally goes high.

The TI datasheet linked above is for TI's "second source" which actually uses a different internal circuit. I'm surprised to see the 0.6V differential input rating (aren't the input transistors lateral PNPs?), but it may be true and it may be possible that this chip can be blown by your tests.

It should be possible to tell these chips by the logo - either NS or TI. Even the NS version made by TI these days should still have NS logo on it, unless they changed something very recently.

If it's some other manufacturer altogether, find the datasheet. If it's from an auction site, it's pretty much guaranteed fake and might even be LM358 inside.

Speaking of which, LM358 should be a good choice of opamp for this kind of application. It doesn't have issues with inputs near ground and it should pass your tests.

[Dannyx]

In the final application, the inputs won't actually be pulled all the way high/low to the supply rail, or GND respectively - this was a crude test I did and it wasn't my original plan. I'm sure if I used some high-ish resistors on the inputs, it would've worked even like this.

I don't want to cram another project in this discussion, because it would deviate from the thread's topic, but very briefly, and I don't claim this is a good idea or a great project, but whatever: I'll use a TL431 as a reference on one input and a second TL431 on the other input. One will stay perfectly stable, while the other one can be altered by, let's say a potentiometer, for simplicity's sake (a liquid detector in reality). You can see where this is going: when the voltage difference between the two becomes too great, the op-amp turns on, turning on my relay. I added a high-ish value resistor between the output and back at the N/I input for some hysteresis, since it's a VERY slow transition from "dry" to "wet" and it would oscillate like crazy when it's sitting in that sweet-spot right in-between states. I tested this and it appeared to work satisfactorily enough.....until the op-amp quit......or not so much quit, as "I probably killed it and its siblings"

[ArdWar]

...a second TL431 on the other input ... can be altered by, let's say a potentiometer....

I'm curious, why would you do it this way instead of using the potentiometer output directly?