The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Simulators >> RF Simulators >> simulation of chopper amplifiers
https://designers-guide.org/forum/YaBB.pl?num=1203079866

Message started by rajath on Feb 15^th, 2008, 4:51am

Title: simulation of chopper amplifiers
Post by rajath on Feb 15^th, 2008, 4:51am

Hi all,

I'm trying to design a low noise amplifier and have decided on a chopper stabilized architecture. I'm using spectreRF simulation using pnoise to evaluate the output noise of a chopper stabilized amplifier, but I seem to get the same amount of noise at the output as a normal dc noise analysis (with all modulator switches closed). Im not sure yet if this is a simulator issue or a design issue. I have tried adding a sinusoidal signal at a low frequency after modulating the input signals and that is definitely attenuated out at the output. So I'm assuming (hoping) this is a simulator setting issue. Does anyone have experience with simulating chopper stabilized amplifiers using pss/pnoise?

thanks in advance,
rajath

Title: Re: simulation of chopper amplifiers
Post by vivkr on Feb 15^th, 2008, 6:57am

If the added sinusoidal noise is attenuated, then obviously you do not have a settling issue. How does the noise profile look and how are you
doing the PNOISE analysis? What options did you set there?

Vivek

Title: Re: simulation of chopper amplifiers
Post by thechopper on Feb 15^th, 2008, 9:53am

Hi Rajath

The chopping modulation process does NOT add any white noise power to the output as compared to the non-chopped case. Since this is a modulation process (no sampled involved) the noise power at the output
does not get increased and actually keeps the same as if the amplifier were non-chopped.
Additional note: As for flicker noise...if you already low pass filtered the amplifier's output signal at the proper frequency, you should not see significant contribution to the total noise power since
the flicker also gets modulated to the chopper frequency.

Hope this helps.
Tosei

Title: Re: simulation of chopper amplifiers
Post by rajath on Feb 15^th, 2008, 7:08pm

Thanks vikr and chopper!
I agree that the flicker noise should get attenuted (I do have an LPF at the output), but when I print out the noise summary of pnoise, the major contributor is still flicker noise, and the profile is similar to a dc noise analysis without chopping. My current pss beat frequency is the chopping frequency (500KKz), with 2 harmonics. My frequency of interest for noise is 22Hz to 22KHz. The pnoise is calculated from 0.1Hz to 1Mhz and no. of sidebands chosen is 10.

thanks,
rajath

Title: Re: simulation of chopper amplifiers
Post by Ken Kundert on Feb 15^th, 2008, 9:59pm

The settings look okay. It is a debugging exercise at this point. Try turning off the chopping clock and see what happens to the noise. If it doesn't change, then perhaps the chopping is not working at all. If it does change, then you may simply have a sneak path that allows the flicker noise escape to the output while remaining at baseband. Perhaps there is a large DC offset or an asymmetry in your chopping circuit. What happens to your DC offset when chopping; is it as low as you expect? You might want to increase your sweep frequency to overlap your chopping frequency (use a sufficiently small frequency step) to see if flicker noise is getting mixed up. No peak at the chopping frequency suggests that the chopping is not working.

-Ken

Title: Re: simulation of chopper amplifiers
Post by rajath on Feb 17^th, 2008, 10:20pm

Thanks a lot Ken for all the suggestions. I will check them out and try to get it working :)

cheers,
rajath

Title: Re: simulation of chopper amplifiers
Post by rajath on Feb 18^th, 2008, 4:49am

Hi,
After a bit of debugging and analyzing the spectrum, I found the solution for the problem even though I haven't fully understood the reason yet. The problem was with the modulating/demodulating switches: I had only one switch for each signal which meant that for half the period, the output of these switches were floating. When I changed this modulator such that it dragged the node to ground during the other half of the chopping period, the flicker noise reduced substantially. I can't figure out why the noise leaks out when the nodes are kept floating.
After this modification, the main source of LF noise is the inverter in the path of the chopping clock - another issue I need to figure out :)

regards,
rajath

Title: Re: simulation of chopper amplifiers
Post by thechopper on Feb 18^th, 2008, 4:58pm

Hi rajath,

I'm not sure I'm getting the picture correctly. How exactly were you leaving the switches floating. Usually mod/demodulation switches are cross-coupled pairs: on one phase two switches (assuming differential signals) let the signal go through with its sign unchanged (the other two remain open), while on the other phase the other two switches let now the signal go while changing its sign (multiplied by -1), while the switches used on the previous phase are now left open.

Also I do not get how the inverter driving the switches could be the major source of noise. This is not the usual case and would mean you have an almos noisless amplifier...

Could you please clarify?

Thanks
Tosei

Title: Re: simulation of chopper amplifiers
Post by rajath on Feb 21^st, 2008, 1:27am

Hi Tosei,
Sorry, did not realize there was a reply...
I do not have a differential signal and the current configuration is a simple opamp with unity feedback (i'm still in the process of evaluating this architecture). In my first attempt, I had a two single MOS switches for both the input and the feedback signal respectively, which turn on and off. Hence, when the switch is off, the input of the amp goes floating. I know this is a bad idea but haven't understood why it results in high 1/f noise.
Currently I have two switches for each line - one switch connects to the signal and the other to ground. The amp inputs are dragged to ground in the negative phase. I think the inverter noise appears because of the node being grounded in half the cycle. The noise decreases if I change this voltage from ground to a higher voltage like 200mV.
Is this the best way to implement the switches? can I have a cross coupled switch between the two inputs of the amp - doesn't make much sense to me since the voltages would be very close in the two lines if the circuit is working fine...
thanks in advance...

regards,
rajath