Help on over-heating Op-Amp LM2904

[Rat_Patrol]

I took a failed 2904 and put it on a breakout board, and tested it on the bench.

I tied both inverting and non-inverting inputs to ground, gave it 5 volts on Vcc.

Side A worked fine, 4.9 something volts. Side B only gave me .6 volts, so side B is what has the issue.

That means its somewhere on the first schematic, I would think.

[fzabkar]

I took a failed 2904 and put it on a breakout board, and tested it on the bench.

I tied both inverting and non-inverting inputs to ground, gave it 5 volts on Vcc.

Side A worked fine, 4.9 something volts. Side B only gave me .6 volts, ...

I don't think that's a valid test.

You are amplifying the input offset voltage in open loop mode. The output could hit either rail, even if the op amp were working correctly.

[james_s]

I suspect you're frying the IC due to transients on the power supply. The first thing I'd do is use a DSO to look at the power rail, set it to trigger on a rising edge just a little above the normal supply rail voltage and then activate the solenoids. Do the solenoids have internal freewheel diodes or are those part of the controller? I'd probably check those to make sure none of them have gone open. Then look at the power supply and see if you've got bad electrolytics, the power supply in my mom's TV blew up a while back because the bulk filter capacitor had gone open circuit, I repaired it and powered it on the bench and it seemed fine but blew up immediately when I tried it in the TV. I had missed that capacitor being bad and under load that was causing the input rail to bounce all over the place.

[Rat_Patrol]

I was measuring voltage on the outputs. Since I got a difference, why would that not be a valid test?

I should say that my normal job is assembly and PCB layout. This PCB came in from one of our good customers, and he can't get another (obsolete product), and its about $500k to replace this machine that runs fine, sans a PCB fault. Push come to shove it would be worth his money and time to have us design a new PCB and just take the micro off this faulty one and drop it on the new one full of new general components (the micro is an obsolete Motorola). A new PCB may be the only solution if there is some internal trace short anyway (but I don't think there is a trace short).

[fzabkar]

I was measuring voltage on the outputs. Since I got a difference, why would that not be a valid test?

The datasheet states that the typical input offset voltage is 2mV, and the typical open loop gain is 100V/mV.

This means that, even when the inputs are tied together, there is still a voltage difference of +/- 2mV which, when amplified, becomes +/- 200V. That is, Vout could hit either rail.

[Rat_Patrol]

When I tie the non-inverting pin on side A to Vcc, I get 4.98v out of output A.

When I tie the non-inverting pin on side B to Vcc, I get 4.2v out of output B.

[fzabkar]

When I tie the non-inverting pin on side A to Vcc, I get 4.98v out of output A.

When I tie the non-inverting pin on side B to Vcc, I get 4.2v out of output B.

That isn't really a valid test, either. This op amp is not a "rail-to-rail IO" type. It has an input common mode voltage range of 0V to Vcc - 1.7V (for Vcc = 30V). That is, the op amp doesn't operate like a real op amp if the input voltage is within 1.7V of the positive supply rail.

I would configure the op amp as a simple voltage follower and set the input to Vcc/2 by using a 1:1 potential divider.

[Rat_Patrol]

Setup the op-amp as a voltage follower, both sides simultaneously on the same Vcc and GND, and 1/2 Vcc .

Used a pair 22k resistors for the voltage divider (first ones I found of THT resistors), gave it independent  2.2k dummy loads on the outputs.

Side A: 1.8 volts on the output. Nearly no oscillation on my DMM.
Side B: 2.3 volts on the output. Few hundreds (3-7) of a volt oscillation on my DMM.

I did have to grab another blown 2904, as I accidentally blew this one (accidentally cranked the voltage knob on my power supply  ) and I didn't want to skew results.

[Rat_Patrol]

Just took a look with the thermal, and the IC is running at ambient room temp. Been powered for about 10 minutes. Power supply shows zero amps draw (not very precise meter), so that is interesting.

[Rat_Patrol]

I took off multiple resistors on both sides of the op-amp one at a time, even when my 5 volt and 80 volt supplies were killed off, the IC was still hot.

This one is very puzzling to me.

[fzabkar]

I would be worried that a failure of the op amp might lead to the Vcc and V_Sensor voltages rising rather falling, resulting in failures to other components.

I would follow the advice regarding the possibility of open freewheel diodes across the solenoid coils. I would also consider a TVS diode across the op amp's supply rail to protect it from spikes.