ADA4522 new lowish noise chopper

[alanambrose]

Ah just noticed this new '5.8 nV/?Hz typical 117 nV p-p typical from 0.1 Hz to 10 Hz' amp:

http://www.analog.com/media/en/technical-documentation/data-sheets/ADA4522-1_4522-2_4522-4.pdf

A.

[guenthert]

Whoa, what happened here?   Where did the comparison table go which was here just a few minutes ago? 

[montemcguire]

I've been testing an ADA4522-2 in a DC servo circuit, and it seems to work very well for that use. This is basically an integrator with an RC lowpass pre-filter, but for that circuit and the high impedances around it, the ADA4522-2 has a very very low 'end of the day' residual offset, somewhere around the 20µV range. Another interesting property of this family of amplifiers is the fact that it can work with 55V total supply voltage - very unusual in this day of 5V op amps!

This part has been around for a while - not sure if the quad part is new or if they just decided to promote it a little more aggressively, but it's an interesting low cost amplifier that can be pretty useful in specific situations.

[d-smes]

What about the Maxim equivalent MAX44241/..43/..46 series?   https://datasheets.maximintegrated.com/en/ds/MAX44241-MAX44246.pdf    I like that CMRR, PSRR, Avol, and GBW appear superior to LTC2057 (Ib and Ios not so good).  Voltage noise is comparable to ADI part but current noise is not specified.  Bias/offset current is 4x higher, but quiescent current is 2x lower.  Can one infer some estimate of current noise from this?

[Kleinstein]

High bias current usually also means high noise. If the noise current is not specified, I would not expect especially good values. There is even a chance to get rather poor ones. So you might have to measure yourself.

Maxim has the similar MAX44520. Here they specify current noise. So for that one noise data are about on par with the ADA4522 / AD4528 at both voltage and current noise. So I won't expect the maxim technology to be any better.

Like with BJT based OPs, there is a trade-off between voltage noise and current noise, bias current.

[T3sl4co1l]

Gotcha for some amps: correlated input noise current.  More relevant to high Iq and low Iib bipolar types (e.g. LT1028), where current mirrors are used to cancel input bias current.  The same cancellation current flows through both inputs, so the noise spec is only met when the inputs have equal source impedances!

Not sure if this applies to any choppers, but it might.  Depends on CM vs. differential charge injection.

Tim

[Kleinstein]

With the AZ OPs the input bias current is usually opposite on both inputs. So no compensation of the voltage drop with equal resistors.

I would expect some correlation in the input current noise, but it could be other way around as with BJT OPs. So no simple compensation, but adding faster than normal. Another tricky thing with the current noise is, that it is in the form of high frequency spikes. Thus things like the input capacitance and even power supply decoupling might have an influence an low frequency noise.

[T3sl4co1l]

With the AZ OPs the input bias current is usually opposite on both inputs. So no compensation of the voltage drop with equal resistors.
...
Another tricky thing with the current noise is, that it is in the form of high frequency spikes.

Hmm, I wonder if there's a transformer trick you could pull, here.

Tim

[David Hess]

Gotcha for some amps: correlated input noise current.  More relevant to high Iq and low Iib bipolar types (e.g. LT1028), where current mirrors are used to cancel input bias current.  The same cancellation current flows through both inputs, so the noise spec is only met when the inputs have equal source impedances!

Not sure if this applies to any choppers, but it might.  Depends on CM vs. differential charge injection.

Tim

Bipolar amplifiers with bias current compensation also have mostly uncorrelated input bias current so there is little advantage to using matched source resistances for better precision which is interesting in light of how the noise is affected.  The noise curves in the LT1028 datasheet do actually show increased noise when the input resistance is unbalanced and the application notes specifically discuss why this is in the typical thorough manner of LT.  I do not see why this would not apply to the LT1007 or LT1124 but it is not discussed in their datasheets.

Chopper input noise is so screwed up by charge injection that I can almost understand not specifying it; the results depend too much on the application.  I think every application I have used them in involved low AC impedance at the inputs via the feedback capacitor and balancing non-inverting input capacitor so the charge injection just gets swamped; if the inputs were at a higher impedance, I think I would have noticed the noise from charge injection.

It is going to be a shame to lose the LT web site; it works so well compared to the AD web site.  I wonder how much their documentation will be rectified in the Orwellian sense.

[MiDi]

I found something interesting when I did research on chopping amplifiers that I thought is valueable to share here as it seems it contains quite some details of the ADA4522 internals.

Reducing Switching Artifacts in Chopper Amplifiers - Yoshinori Kusuda: The specs on page 168 for Chapter 6 should look familiar