I mean Linear did the same thing for their filters in AN124f where they choose 1300uF-1.2k and 165uF-10k for 0.1Hz HPF.
Ok good point: that explains (partly) why for the LTC6655 (5V) I get different results against the data sheet.
Datasheet value is 0.25ppmpp = 1.25uVpp.
whereas I got 2.2uVpp for the MSOP and 2.7uVpp for the LS8-package.
Of course LT will not change its cirquit so they would have to update their datasheets with higher 1/f noise levels.
I don't think this is right. I simulated the AN124f filter and it's not far off: as you can see in the picture (normalized to 0dB gain) the -3dB points are at .192Hz and 9Hz which shouldn't reduce the total noise by much more than about 8% compared to .1 to 10Hz.
I think the confusion that has arisen is that zlymex was pointing out above that AN124 .1Hz HPF is incorrect and I'm guessing you assumed that AN124 LPF also had the same problem as zlymex's original LPF with a cutoff at 6.5Hz. It doesn't really matter of course how it arose but it means that there must be some other explanation of the discrepancies between the LTC6655 specs and the measurements.
zlymex measured the LTC6655 noise, in reply 2, as approx .33ppm compared to the .25ppm spec, (33% above spec) but I assume that was using the .16Hz - 6.5Hz filter (as calculated by Alex) - the schematic has been updated since but presumably the results are still from the original circuit. I estimate that the bandwidth reduction should reduce the measured noise by about 15% which means the .1 - 10Hz measured noise would be 33%/.85 = 39% above the spec. The .25ppm spec figure is as usual a typical figure so the part still meets the specs.
Andreas's measurements above, at 76% to 116% above the typical are much worse, and whilst technically within spec are very disappointing for a part which is expressly marketed as being low noise. I estimate that the .14Hz -3dB HPF error compared to .1Hz would only have an error of around 2%, so pretty negligible. So the question is, was Andreas particularly unlucky with the various parts he tested, or did the LT engineer/marketing bod who wrote the datasheet previously work at Vishay
or is Andreas's LTC6655 test setup / layout inducing excess noise for some reason?
To estimate the integrated 1/f noise with differing bandwidths, I used the formulae:
Vnoise rms = Vnw x sqrt(FC x ln (FH/FL))
where:
FC = 1/f cutoff frequency
Vnw = noise density well above FC
FH = HPF -3dB frequency
FL = LPF -3dB frequency
Vnw is used for frequencies above FC when FH > FC
The only voltage reference datasheets I could easily find showing 1/f noise spectrums were the LTZ1000 and the ADR4520 (bandgap), both of which showed FC to be around 2Hz and the AD584 with FC nearer 150Hz. I used 2Hz in the above estimates; increasing it to 150Hz increases the impact of the LPF errors but decreases the HPF errors. Eg. The AN124 bandwidth errors only change the integrated noise from 92% to 91% (of .1 to 10Hz) when FC is increased from 2 to 150Hz.