TI degrading specs of existing parts without changing part numbers

Regular opamps have about 1pF of input capacitance to ground. Dealing with that isn't complicated. Dealing with the 10pF when it messed me about wasn't complicated either.

Having to do it after I'd laid out the board, and got it made and loaded, wasn't complicated either, but it was definitely irritating.

Do you use opamps with a ground pin?

My favorite gumdrop opamp, OPA197, has a typ "common mode" capacitance of 6.4 pF, whatever that means.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

Don't be silly. The ground reference is through the negative rail

Explain it to me. I'm not sure what the commmon-mode capacitance of an opamp is.

Any explaination why the peaking in your circuit could be blamed on differential capacitance and not care about common-mode capacitance?

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

As we’ve discussed before, the differential input capacitance of an op amp under bias is hard to measure, and the data sheet values are unreliable.

The best way I know of is to put a very small pot directly in series with the inverting input and look at what happens to the phase margin.

The CM capacitance you can get by doing the same on the noninverting input and looking at the bandwidth. Feedback bootstraps out the differential capacitance at low frequency.

Cheers

Phil Hobbs

I've seen the "common mode" capacitance defined as the C between one pin and the rails, and sometimes as the c of both pins in parellel to the world.

A diagram of the several caps, with values, would help.

One can Spice a standard inverting amp and see what it does, assumuing the Spice model is accurate.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

It’s the measurement that’s difficult—if the amp is powered off, none of the junction capacitances will be right.

Cheers

Phil Hobbs

I have a Spice model of the OPA197, basically TI's Pspice model converted for LT Spice.

I made a unity-gain follower sim and drove that with a 1 volt step, through a 1M resistor. The output is a nice exponential with tau =

The TI data sheet says that cm capacitance is 6.4!

The differential input c disappeared because it's bootstrapped in a follower.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

It doesn't. There's a collector/base or gate/drain capacitance that has to be charged up if the input moves vis-a-vis local ground (in this case the local rail voltages inside the chip). There are heroic measures that can minimise this - essentially floating the power rails though they move too slowly to be of all that much help - but regular followers don't use them.

Imagine three caps.

C1 is inverting input to all the virtual grounds.

C2 is n.i. input cap to ground

C3 is between the two inputs.

TI calls C1 the common-mode capacitance Zic

They also call C2 the same thing.

They call C3 the differential capacitance Zid.

My sim measured only C2, because both C1 and C3 followed the input so were bootstrapped out of existence.

C1 and C2 are each spec'd at 6.4 pF. I estmated 6.45 from my sim, eyeballing tau.

OPA197 has a lots of protections and integral EMI filtering, which adds capacitance. Nice safe gumdrop. One can drive the inputs modestly past the rails and it doesn't go crazy. It makes a decent RRIO comparator too.

Some people think the "common mode capacitance" is C1 + C2. TI doesn't.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

Probably not a useful exercise, judging from the unhelpful observations that followed, which I've snipped.

Thanks, that makes sense, beforehand I too would have said it was C1+C2

Of course observations like this are not helpful to you.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

It's interesting, but one can just Spice actual circuits and see what happens. Understanding the capacitances helps make the initial opamp selections to sim.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

I have a parallel story. The problem was to measure the Differential and Common-Mode current passing between a central power system and the power system within a major subsystem on some heavy conductors. There were debates as to just what it all meant.

The solution was to use a current transformer. If both conductors passed in parallel through the current transformer's aperture, the common-mode current was measured. If those two conductors passed in opposite directions, twice the differential current was measured.

Joe

TI managed it much better with the 324 and 339. They brought out the B versions with their own datasheets and kept making the old ones.

I wonder if that process has gone away too. It must be a drag keeping creaky old litho tools and 5- inch wafer processes alive. Probably duct tape on top of JB Weld.

Cheers

Phil Hobbs

In the original National parts, all four channels would go bonkers if any input went slightly below ground.

And the test sets.

A support guy said "That's an old Burr-Brown part. We don't know much about those."

Speaking of TI, this is incredible:

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

And saturated the tail source transistor, which then hogged all the base drive from the other sections.

They fixed that in the 324A and the 358, iirc. Doesn’t seem to happen anymore, at least.

Nice. We’ll have to find a use for it.

Cheers

Phil Hobbs

A frustrating thing is that the opamp manufacturer has an exceedingly accurate spice model of the part that their designers created in order to optimise and verify the design, but the company won't give that one to the public. The schematic in the datasheet is often drawn by marketing people who have never discussed it with the design engineers, let alone seen the real schematic, and is sometimes a source of great amusement to the actual designers. The public spice model is quite possibly created in a similar manner, but possibly corrected based on some measurements by apps engineers or based on complaints from customers. Meanwhile the really accurate model sits on some internal server being ignored while the design engineers have moved on to some other product.