A discussion on Precision OpAmps

[Kleinstein]

Using some kind of chopping under SW control with extra swtiched is a possible way, but this usually is rather slow chopping, less than 25/30 Hz to get at least 1 power line cycle for each configuration. With a SD ADC with extra settling time, or to reduce the number of switching spikes one may chose an even lower frequency. So one would still care about the noise a relatively low frequencies. So those CMOS OP with super low bias are not really the right choice. The switches anyway tend to have a leakage current of a few pA and thus no need for the amplifier to be much better. So good JFET OPs like the OPA140 or ADA4622 are the more suitable choice.

The very low bias CMOS OPs may be good for a kind of electrometer input with switching via low leakage relays, not so much for a more normal voltmeter input.

[Ole]

Fros the input I do need both the signal OP-Amps and the Bottstrapping OpAmps, the latter being either AD549L or OPA104. But I a m not completely sure about the Signal OpAmp.
The two contestants are the LMP7721 (2fA typ and 20fA max @25°C with 2µVp-p Noise and 50µV typical and 180µV max Vos)
and the ADA4530 (<1fA typ and 20fA max @25°C with 4µVp-p Noise and 9µV typical and 300µV max Vos)

[Kleinstein]

The LMP7721 and ADA4530 are very low bias OPs, suitable for a rather special electrometer type input. This is very low bias ( < 1 pA), but quite some noise and drift (e.g. > 2µV_pp,  > 2 µV/K).
These would normally not be combined with a feed forward driven supply directly from the input, but more with classical bootstrapping with direving the supply from the feedback path. For the feedback path there is no need to have very low bias / good precision like the old and expenside AD549, OPA104 - there a more moderate TL031 or TL072H should be sufficient.
Such a very high impedane input would have relay switching at the input. It also needs care with guard traces / cut outs.

For a more normal voltmeter input I would more look at something like OPA145, max4238, ICL7652 or AD8628. They have more bias current, but also much lower DC noise and drift. If switching at the input is done with JFETs, one can consider extra low bias auxiliary buffers (for the gate signal and maybe the protection). Here the AD549 is more like overkill and the TL031 may be good enough.
A good more normal voltmeter input is moderately low bias (e.g. < 50 pA), but low noise (e.g. < 1 µV_pp). Electronic input swithing may be used to compensate for input drift, but adds input bias. An zero drift OP could avoid the repeated input switching.

[Ole]

I think with an OP-Amp like the AD8655 (1pA typ. Ib, 50µV Vos typ, 0,4-2,3µV/dK dVos and 1,2µVp-p) as the Signal carrying Amplifier and the OPA828 (1pA typ. 50µV Vos typ, 0,3-1,5µV/dK dVos and 0,34µVp-p) as the bootstrapping Amplifier in a parralel structure and a low noise OpAmp to supply the AD8655 (those could be the OPAx140)

[Kleinstein]

I think with an OP-Amp like the AD8655 (1pA typ. Ib, 50µV Vos typ, 0,4-2,3µV/dK dVos and 1,2µVp-p) as the Signal carrying Amplifier and the OPA828 (1pA typ. 50µV Vos typ, 0,3-1,5µV/dK dVos and 0,34µVp-p) as the bootstrapping Amplifier in a parralel structure and a low noise OpAmp to supply the AD8655 (those could be the OPAx140)

This makes not sense for a normal voltmeter. These are nice fast OPs with good higher frequency performance and low noise there, but more mediocre in the low frequency range. 1.2 µV 0.1 - 10 Hz noise is still more on the poors side for DC precision, especially if most of the noise is 1/f noise.

One still has to decide if one want's auto zero switching before the amplifier (need resonable good speed and good noise of the 1-10 Hz range) or not (needs very low offset dirft and good reasonable noise for the 0.01 - 1 Hz range). The speed for the AZ switching also depends on the ADC used. Some are relatively slow and some loose quite some time on switching (higher order fitlers with SD ADCs).

[Ole]

I´ve continued looking for low noise, low Bias Current OpAmps.
What I have found is: ADA4625 (15pA Ib with 15µV Vos, 0,2-2,1µV/dK and 0,15µVp-p) and the OPAx140 (0,5-10pA Ib 30µV Vos and 0,25µVp-p)
These are both fitting in noise characteristics and in typical bias current.
The ADA4610 is in a similar realm in terms of noise (0,45µVp-p), but has a much higher offset (0,2-1,8mV) and a higher drift (0,5-12µV/dK) but still a respectable bias current of 5-25pA.
Other OpAmps that I am going to look into are the ADA4625, LTC6240 and ADA4637.