Puzzling op amp bug

Jay Kickliter · 2010-09-22 11:33

I made the exact circuit in the attached pdf. It's from TI's datasheet for the OPA2344 op amp.

The first stage seems to be working, but all I'm getting is my supply rail voltage from the output of the second stage. I double and triple checked all the resistors and capacitors, and my buddy checked them. Also swapped out with an identical op amp and the problem persists.

Can anyone see something about the circuit that would be especially tricky? In the second stage I have about a 0.5 delta v between the +&- inputs.

Thanks

bill190 · 2010-09-22 11:45

I don't see a pdf link???

Jay Kickliter · 2010-09-22 11:49

bill190 wrote: »

I don't see a pdf link???

Oops. Thanks for pointing that out.

Leon · 2010-09-22 12:18

Check the DC levels on the op amp inputs, they should be 1/2 the supply voltage.

davejames · 2010-09-22 14:36

Jay,

With the 2344 out of the socket, measure each socket pin with a voltmeter and report back.

Do you have an o'scope available?

Jay Kickliter · 2010-09-22 15:47

davejames wrote: »

Jay,

With the 2344 out of the socket, measure each socket pin with a voltmeter and report back.

Do you have an o'scope available?

Dave and Leon, i had to put it away and go home, since it was driving me nuts. When I get back to school tomorrow I'll take the readings.

Leon · 2010-09-22 16:05

I often do a SPICE simulation with circuits like that (the PCB software I use includes an excellent SPICE implementation). TI probably got it right, but it's reassuring if things work in the simulator.

davejames · 2010-09-22 22:18

...ah, so you reached the stage I call the "throwing things" stage.

Had to walk away? I completely understand!!!

(smiles)

Jay Kickliter · 2010-09-23 07:01

Here's the voltages with the op amp removed and insterted, by pin and function:

1. output stage  1, 1.634, 1.461
2. - input stage 1, 1.650, 1.635
3. + input stage 1, 1.558, 1.558
4. GND
5. + input stage 2, 1.558, 1.558
6. - input stage 2, 1.605, 1.598
7. output stage  2, 1.548, 3.297
8. 3.3

I'm hoping we just made a stupid mistake. Thanks for looking.

bill190 · 2010-09-23 08:13

If any soldering, look for shorts/solder bridges.

Double check the number on the chip with the data sheet?

Double check the pin numbers. 1, 2, 3, 4, 5, etc.? with the diagram on the data sheet.

If you have two of these chips and this is on a breadboard and easily rewired...

Try using 2 chips and the 1st op amp from each chip. (The theory being that the datasheet is wrong as to the drawing of the 2nd op amp)

Or try using the 2nd op amp as the 1st and the 1st as the second. See if the problem moves.

Try disconnecting things one at a time and see what happens.

davejames · 2010-09-23 09:14

DC-wise, open socket, I don't see anything strange. This would imply that the values of the 1M and 20k resistors acting as voltage dividers are similar values respectively.

Do you have an o'scope available? I'd check the 2nd output for oscillation when the op-amp is inserted. A large frequency oscillation can fool DC meters into thinking that there is a DC level.

Is the op-amp power supply pin bypassed according to the data sheet? Lack of local decoupling can cause bizarro problems.

Is the ADC socketed? Observe the 2nd ouput of the op-amp with and without the ADC inserted.

Are the components soldered? Make sure the board is clean of all flux residue.

Regards,

Jay Kickliter · 2010-09-23 16:42

bill190 wrote: »

If any soldering, look for shorts/solder bridges.

Double check the number on the chip with the data sheet?

Double check the pin numbers. 1, 2, 3, 4, 5, etc.? with the diagram on the data sheet.

If you have two of these chips and this is on a breadboard and easily rewired...

Try using 2 chips and the 1st op amp from each chip. (The theory being that the datasheet is wrong as to the drawing of the 2nd op amp)

Or try using the 2nd op amp as the 1st and the 1st as the second. See if the problem moves.

Try disconnecting things one at a time and see what happens.

Thanks for the advice. I'll try using two separate chips to see if the problem moves.

Jay Kickliter · 2010-09-23 16:58

davejames wrote: »

DC-wise, open socket, I don't see anything strange. This would imply that the values of the 1M and 20k resistors acting as voltage dividers are similar values respectively.

Do you have an o'scope available? I'd check the 2nd output for oscillation when the op-amp is inserted. A large frequency oscillation can fool DC meters into thinking that there is a DC level.

Is the op-amp power supply pin bypassed according to the data sheet? Lack of local decoupling can cause bizarro problems.

Is the ADC socketed? Observe the 2nd ouput of the op-amp with and without the ADC inserted.

Are the components soldered? Make sure the board is clean of all flux residue.

Regards,

I did look at everything with an o-scope, it was all pretty clean. I forgot to mention that don't have the ADC, but I did try to simulate it's input Z to see if that had any effect.

You know, like an idiot, I didn't add any decoupling caps. It's on a solderless breadboard, so I know that can cause troubles, but I don't feel that's the case here. Or am I wrong?

Thanks a lot for the help.

davejames · 2010-09-24 08:45

Hi Jay,

Not decoupling a circuit caused me to spend 2 frustrating weeks in Bangkok attempting to install a test package for Philips Semi. After giving up and enlisting the help of a contract apps engineer, I watched him - in less than 30 minutes - determine that the problem was due to an un-bypassed 12V on my load board.

Also, I've seen instances in op-amp based audio circuits where the op-amps will oscillate, "just because", especially when using relatively fast slew-rate devices.

It boils down to "good practice" to decouple any device supply line.

All that said...when you observed the output of the 2nd op-amp, did you see any oscillation?

Later

davejames · 2010-09-28 08:59

Jay - any resolve?

Curious people want to know!

Puzzling op amp bug

Comments