Should I trust simulated supply current of op amp?

[KD494]

Hey all,

I was just running a quick simulation in LTSpice with a model for an op amp, AD8038, that I was hoping to use. Everything looked good other than the quiescent current draw which was nearly 6 times the maximum listed on the datasheet. If this is a more accurate representation of how the part will behave I can't justify using it. Is it common for something as simple as current draw to be modeled so poorly in a macromodel? It's my understanding that these models are not transistor level models but rather something to give you an idea of how it may behave. Any insight you may have from your experiences would be helpful. Thanks.

 

[crutschow]

Usually models give you a nominal current.
Was there any load on the op amp output when you measured the current?
Did you get that model from Analog Devices?

 

[BradtheRad]

I tested a 741 with different load resistances. A schematic of its internals shows the output as a half-bridge, with 25 and 50 ohm resistors in the legs.

I was inclined to think this means it can accurately drive a load of, say, 100 ohms or more. This is not correct. In my tests, a real 741 produced about 5% less output voltage, when I attached a load of 1k ohm. I was surprised because this means it is easy to get distorted response, if I am not careful about component values which I connect to the output.

A typical op amp output spec is 10mA. Example, this is current through a 1k load at 10V, so I should not be surprised at my test results.

- - - Updated - - -

Hmmm, now I see the question is about quiescent current draw. Many times I have observed real devices which draw less than the spec amount. So it is not always predictable, or consistent.

 

[KD494]

Was there any load on the op amp output when you measured the current?
Did you get that model from Analog Devices?

The model did come straight from AD but I did import it into LT Spice so maybe something didn't translate well? After I noticed the current draw I set the part up as a simple buffer and removed any load and saw the same supply current.

 

[FvM]

It's unusual that a SPICE OP model reproduces all properties of a component correctly. In this case, the model obviously doesn't care for correct supply current. I presume it's not a LTSPICE problem.

You should always trust the datasheet, not the model.

 

[KD494]

You should always trust the datasheet, not the model.

Thats what I was thinking, I'll update if anything surprising comes from testing the part.

 

[crutschow]

I downloaded the AD8038 model from AD and it did indeed draw a simulated 6.2mA from the power supply whereas the data sheet states the nominal current is 1mA.
So the model is indeed way off on the current.