op-amp square waves

[ledtester]

Edit: MC33202 image redone with a higher V/div.

I am testing op-amps I've bought from ebay and Mouser with the classic square wave circuit:



The op-amps I tested are:

- TL082      - ebay
- MC33202 - Mouser
- LM2904   - Mouser
- TL072      - ebay

I've set up the circuit for a nominal 50 KHz frequency, R1 = 8.2K, C = 1 nF. The supply voltage was set to 9V.

The other op-amp in the package was configured as an identical oscillator (same R1 and C values, separate V+/2 rail.)

Here are the results:

TL082:



MC33202:



(old image:

)

LM2904:



TL072:



Observations:

The TL082 output is very clean. The TL082 has a high slew rate, and we're getting that here. The frequency is even higher than the programmed 50KHz.

The MC33202 output is clearly slew-rate limited. The datasheet specifies a typical slew rate of 1V/us, and that's about what we're seeing.

The LM2904 output is even more slew-rate limited (datasheet lists 0.3V/us typical), and so we get more of a triangle wave with a reduced frequency of 20 KHz. There is also some distortion on the downwards slope and bottom of the wave form.

The TL072 output is confusing, because according to the datasheet it should have a "high" slew rate.

Questions:

It looks like the TL082 chips are good, and also the MC33202s seem to be performing within spec.

I would like to understand why the LM2904 wave form has the kinks in it and the "inversion" at the bottom.

And what can be said of the "TL072" chips from ebay? Did I get sold some bogus chips?

[AustinTxBob]

Interesting.  Can you post a pic of your setup?  I'd like to see how you tested them.  Guess I should order a few of the TL082s.

[T3sl4co1l]

What is the unused op-amp section wired as?

I'm surprised there's a difference between '72 and '82.

Tim

[krivx]

Kinks or inversions are common with some opamps when hitting a rail.

I'm also surprised to see a difference between the TL072 and TL082. Looks like the TL072 is doing ~4V in ~10us, which seems well under spec.

Do you have any capacitance across the supply pins? Could be limit by current slew instead of voltage...

[tautech]

MC33202:
Well you will get that waveform IF the waveform V/div setting is too low to display the waveform within the vertical confines of a DSO display.

Do that one again please.

[ledtester]

MC33202:
Well you will get that waveform IF the waveform V/div setting is too low to display the waveform within the vertical confines of a DSO display.

Do that one again please.

You're right, and that's one thing I don't like about how the DS1054Z renders off-screen traces. Initially I used my analog scope to observe the wave forms, and so when the DS1054Z show the "same" trace it didn't occur to me that it was actually being clipped.

Image has been updated in the original post.

[tautech]

MC33202:
Well you will get that waveform IF the waveform V/div setting is too low to display the waveform within the vertical confines of a DSO display.

Do that one again please.

You're right, and that's one thing I don't like about how the DS1054Z renders off-screen traces. Initially I used my analog scope to observe the wave forms, and so when the DS1054Z show the "same" trace it didn't occur to me that it was actually being clipped.

Image has been updated in the original post.

Most DSO's clip waveforms in this manner, 1054's don't have this to themselves. 

Anyways now you know to watch out for this. 

BTW, nice edit.

[ledtester]

Kinks or inversions are common with some opamps when hitting a rail.

I'm also surprised to see a difference between the TL072 and TL082. Looks like the TL072 is doing ~4V in ~10us, which seems well under spec.

Do you have any capacitance across the supply pins? Could be limit by current slew instead of voltage...

Power is coming directly from a linear power supply (a Tenma 72-8335A.) Current draw is < 10 mA.

[krivx]

Kinks or inversions are common with some opamps when hitting a rail.

I'm also surprised to see a difference between the TL072 and TL082. Looks like the TL072 is doing ~4V in ~10us, which seems well under spec.

Do you have any capacitance across the supply pins? Could be limit by current slew instead of voltage...

Power is coming directly from a linear power supply (a Tenma 72-8335A.) Current draw is < 10 mA.

I would throw 100nF across the supply pins of the opamp and see if things change. You are not just testing voltage slew in this case, as the opamp is also charging/discharging a cap from it's output. It's a very small current, but hey - it's a quick thing to check.

An interesting case might be to configure one of the "bad" TL072s as a follower and see if it can produce the output signal of the "good" TL082

[ledtester]

What is the unused op-amp section wired as?

I'm surprised there's a difference between '72 and '82.

Tim

The other op amp is wired in an exact copy of the oscillator circuit. I even duplicate the V+/2 rail. You can probe the output of the other op amp using the "test loop" that's situated between the postive and GND loops.

In all of the cases I got the same results for either op amp.

[T3sl4co1l]

What is the unused op-amp section wired as?

I'm surprised there's a difference between '72 and '82.

Tim

The other op amp is wired in an exact copy of the oscillator circuit. I even duplicate the V+/2 rail. You can probe the output of the other op amp using the "test loop" that's situated between the postive and GND loops.

In all of the cases I got the same results for either op amp.

Ah.

Ahhh...but are they in phase, too?

It's possible, actually..

I forget if any of these amps are the kinds that do bad things when you push them towards a rail (or cut off the input stage, or..), and therefore you might be getting funny business where one side saturating causes the other side to misbehave, and all.

The underlying cause is common bias supply networks, e.g. using one current mirror to supply both op-amp sections, but not making it saturation-proof, so as soon as one stage saturates, bias drops and the whole chip kind of falls over.

This kind of business could easily make them phase-lock, which could explain coherent lumpiness on the edges and stuff.  But really, I'm expecting that they are being well-behaved after all, and, if they are phase-locked, it's just from proximity and supply noise, which is reasonable to expect anyway.

Or they might be too far apart (you're only using ceramic caps, so I'm guessing the tolerance kind of stinks) and free-run independently.

Tim

[mikerj]

You should definitely have a decoupling cap across the op-amp supply pins, it would only take a minor supply glitch from one op-amp switching to affect the threshold on the other one.  I'd stick a 100n across the pins on the back of the PCB and see if that helps.

[Audioguru]

A TL07x is a TL08x selected for low noise so they should perform the same in your oscillator circuit  If you want cheap fakes then buy from ebay.
The TL07x and TL08x opamps have a problem called "Opamp Phase Inversion" which happens when an input voltage becomes within a few volts from the negative rail then the output suddenly goes high. Since your power supply is close to the minimum (7V) of a TL072 then maybe it is having the phase inversion problem.
Your schematic shows no power supply but you say the power supply is 9V. You also say there is a V+/2 rail but do not show how it is made.