Did I cook an op amp?

[paulca]

Should I ever see a voltage being generated by an op amp input when it's disconnected?  No current but a persistent 1.2V. 

Even when it's connected to a sense resistor which without the opamp reads correctly, connect it to the opamp input and it reads 1.2V.

So the opamp is 'producing' that potential on it's input pin.

I'm assuming I've cooked it or something else is afoot. 

It's a dual package and the other amp works.

TLE2142IP

[AndyC_772]

As a matter of principle, don't ever waste time on failed components. If you think it might be dead, swap it for a spare and see if the behaviour changes.

Always have a spare!

Many op-amps do have internal protection components on their inputs, including back-to-back diodes between + and -. What does the other input connect to, and what's its potential?

Incidentally, this protection feature is one of the major differences between op-amps and comparators. A comparator never has them, but an op-amp is usually designed to be used with feedback, so the potentials at the inputs are equal and the back-to-back diodes never conduct.

[jmelson]

Should I ever see a voltage being generated by an op amp input when it's disconnected?  No current but a persistent 1.2V. 


TLE2142IP

This op amp is a bipolar one, and the datasheet shows it uses PNP transistors.  That sheet also shows a bias current of up to 2 uA, so it will definitely get a reading on a DVM.
And, of course, leaving any input pin on any device floating (unless specifically advised in the datasheet) will lead to bad behavior.

Jon

[paulca]

Thanks.  The circuit was working previously.  It's a dummy load.  I have to disconnect the high current side to put it away.  So temp sensor, fan, sense and mosfet gate all come out.

Reflecting on what happened, I think it is cooked.

On wiring it back up I failed to spot the bottom half of a 10:1 divider had come out.  I was using it to test a panel ammeter I got off fleabay and the panel meter was reading very rapidly up to 1A with barely turning the 10 turn pot.

I figured the ammeter was rubbish, so I connected the DMM in series to measure current from a LiPo.  I had just connected the LiPo and the meter started beeping at me, I checked it and it read 16.xx and I rapidly pulled the plug out of the meter as it's 10Amp fused.

What I only realised later was this was caused by the sense resistor wire coming off the pin!  Solid core wires and moving things about all the time had worked it loose.  I put it back on better this time by making it into a coil I could slide over the pin and soldering that.

Rechecking everything and the load never worked again and I have traced it to this 1.2V on the op amp inverting input pin.  It can't read the 200mV of the sense resistor because it, itself is putting 1.2V across it.

I think the wire coming off the sense resistor might have then touched the mosfet pin and created a shorter route to ground through the opamp and the breadboard for all 16Amps.  Ouch.

I will double check tomorrow evening when I get home that nothing else is amiss.

It's odd that the other amp in teh same package survived.

I don't have a spare 2142 and I didn't have time to check the amps I do have for suitable low input voltages.  I ordered two spares from RS.

[paulca]

Eventually traced this to a dead current sense resistor.  It's a 4 pin PowerTron, cost a small fortune £10.  Somehow when the pin fell off or when I soldered it back on the inner "sense" output fried. 

So, when running a 1A load I get

Pins
1-4: 100mV drop, continuity beeper, close to 0.1Ohms
1-3: 100mV drop, continuity beeper, close to 0.1Ohms
1-2: 0V, no continuity, OL Ohms
2-1, 2-3, 2-4: No continuity.

Pin 2 is toast. 

Not a big deal, I just soldered it to pin 1.  It's a Vishay powertron, no mention of fusing or anything in the DS.  Also I never took it higher than 16A or close to 15W heat, unless the LiPo sume how surge currented it enough to have melted the solder on pin 2, but I expect that's unlikely, seemed more like the sense pin falling off was what drove the Mosfet control to slam the gate wide open in the first place.

http://www.vishaypg.com/docs/64238/64238_FPR_4_T220_T221.pdf