Question on input noise current

[SY]

Dumb question- the 797's diff input is cascoded, current mirrored, and bootstrapped six ways from Thursday. Is that possibly the reason for the low bias current?

[Fred Dieckmann]

The simplified schematic shows a diff pair with a folded cascode and a current mirror load. This is followed by two stages of emitter followers. This is simple schematic showing only the basic outline of the signal path. But as op amps go this is a pretty simple topology and not near as convoluted as many I have looked at.

Bias compensation:
http://www.analog.com/library/analog...versary/6.html
Someone will read it eventually I suppose..

[jcarr]

Hi Sy:

>the 797's diff input is cascoded, current mirrored, and bootstrapped six ways from Thursday. Is that possibly the reason for the low bias current?<

The cascode part of the 797 is of the inverted/folded type, which is good for stability and bandwidth, but AFAIK, means little for the base or bias currents of the input diff pair. The second-stage current mirror just sums the plus and minus portions of the signal into a pushpull output, with probably very little effect on the input base or bias currents.

Finally, the major bootstrap mechanism in the AD797 maintains a constant voltage across the summing current mirror, which will kick up the OL gain a lot, and I think will also help the PSRR on the minus voltage rail. But again, I doubt if this represents any direct advantage for input base or bias currents.

Of course, there is always the possibility that the real AD797 circuit is far more intricate than what Analog Devices chose to publish.

best, jonathan carr

[PMA]

Ohayo gozaimas Jonathan,
ogenki desuka?

I have visited your www and there is a greeting in czech language. Does your company have activities in the Czech republic?

Regards,
Pavel

[jcarr]

Pavel: My some-times engineering design partner, Petr Mares, is originally from Czech, and he is also the person who designed and programmed most of the Connoisseur website.

We haven't done anything in the Czech Republic yet, but who knows what the future may bring?

regards, jonathan carr

[Fred Dieckmann]

"Of course, there is always the possibility that the real AD797 circuit is far more intricate than what Analog Devices chose to publish."

That is why it says simple schematic on the drawing from the data sheet. I feel quite sure this is nowhere near the actual circuit and would image that the bias compensation network is one of the bits not shown. Op amps are very proprietary. The reason Spice models are grossly simplified macromodels is for both the protection of the design and lengthy simulation time for a model that represented the real complexity of the design.

http://www.linear.com/pub/document.h...pp&document=51

[andy_c]

Originally posted by Fred Dieckmann
(...)The reason Spice models are grossly simplified macromodels is for both the protection of the design and lengthy simulation time for a model that represented the real complexity of the design.

http://www.linear.com/pub/document.h...pp&document=51

This is a bit off-topic, but to add to what Fred has mentioned, some of the SPICE op-amp models are really poor. In LTSpice for example, all of the Linear Technology op-amp models have the property that the simulated supply current does not change as a function of the output current delivered! I believe this is a property of the Boyle model. The Analog Devices op-amp models are much better. There's an application note describing them here:

http://www.analog.com/UploadedFiles/...01016AN138.pdf

If I ever need an op-amp model for SPICE, I try to grab the Analog Devices model if I can find one.

[fdegrove]

Hi,

That is why it says simple schematic on the drawing from the data sheet. I feel quite sure this is nowhere near the actual circuit and would image that the bias compensation network is one of the bits not shown.

Exactly.

Cheers,

[SY]

AD confirms that there is bias compensation circuitry on the chip.

[jcarr]

Thank you for the clarification, SY.

regards, jonathan carr