chopping amplifier: how does this remove 1/f noise

[gujaratibhai]

Hi folks,

I redesigned a 2-stage amplifier but am told to add chopping to it to fight the 1/f noise. I have read about chopping and it's advantages (Enz papers) but they seem too theoretical and have difficulties translating it to *me* using it. I have a snapshot of an existing design where chopping was applied and it is attached to this image.

If you have any insight or a tutorial/paper that can help describe the chopping and how it removes the flicker noise, I'd greatly appreciate if you shared. The box "X" represents chopping block consisting of cross switches that swap connections every clock cycle using ph1/ph2 (non-overlapped clocks).

Thanks,

Regards,
Gujaratibhai

 

[FvM]

To work as a chopper amplifier, there must be cross switches in front of the amplifier input. I don't see it in your schematic.

 

[gujaratibhai]

Hi FvM,

Unfortunately, there is none. The 3 cross switches is all i see. I have also seen results where with chopping enabled 1/f noise gets reduced quite significantly. I just don't understand the knobs on what dictates the improvements and by how much.

Thanks,

Gujaratibhai

 

[crutschow]

1/f increases as the frequency goes lower (hence the 1/f designation) and can be prominent in DC amplifiers since they have a low frequency cutoff of infinity.
To remove this noise a chopper amplifier is used which chops the DC to create a higher frequency AC signal at a frequency above the 1/f low frequency corner.
This gives an AC signal whose amplitude is proportional to the DC (or low frequency) signal level.
This AC signal is amplified and then detected (demodulated) to recover the DC signal.
The reduction in noise is determined by the ratio of the high frequency amplifier noise level compared to its 1/f noise level (which is normally higher for the same bandwidth).

 

[FvM]

I agree with the explanation. But the schematic in post#1 doesn't show a reasonable chopper amplifier topology. Main noise source is the amplifier input stage. A chopper amplifier would modulate the DC input voltage in front of the input stage and demodulate it at the amplifier output.

 

[dick_freebird]

They are chopping (flipping) the drains of the diff
pair, and the loads, does that not suffice to reverse
the front end mismatch? It would appear desirable
in the areas of input kickback noise, switch leakage
as an Iio/Iib term and so on.

 

[FvM]

I presume the gates of the differential pair are the amplifier input terminals, respectively input noise and offset of this amplifier stage can't be cancelled by the chopper.

 

[gujaratibhai]

Master thesis found at link below seems to be using the same architecture for a fully differential amplifier.

I will review and hope it had further insights. Thesis link:

https://digital.lib.washington.edu/.../Mandic_washington_0250O_10933.pdf?sequence=1

 

[FvM]

The missing circuit block is designated "passive mixer" in the thesis.