The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> Analog Design >> PSRR of Chopped Bandgap
https://designers-guide.org/forum/YaBB.pl?num=1295432049

Message started by Berti on Jan 19^th, 2011, 2:14am

Title: PSRR of Chopped Bandgap
Post by Berti on Jan 19^th, 2011, 2:14am

Dear all,

I have simulated the PSRR of a bandgap which uses a two-stage amplifier with the first stage chopped and a large off-chip compensation capacitor connected to the supply.

If the bandgap is unchopped, the PSRR looks ok. If the chopper is actived the low frequency PSRR is reduced and is peaking around a frequency which depends on the large off-chip capacitor and the chopping frequency.

I can imagine that the chopping function will affect (modulate) the supply disturbance and consequently influence the PSRR. But what the exact mechanism behind it? Can somebody explain? References?

Cheers

Title: Re: PSRR of Chopped Bandgap
Post by wave on Jan 19^th, 2011, 11:26am

Berti - I'm really curious about this OFF-chip capacitor for compensation.
I would think on chip is possible; not to mention the chance to pick up noises off chip.

Can you shed some insight - why? what size C? what bandwidth?

regards,
Wave
;D

Title: Re: PSRR of Chopped Bandgap
Post by thechopper on Jan 19^th, 2011, 5:51pm

Hi Berti,

You should be careful when considering PSRR of a chopped amplifier, since the chopped amp is not an LTI system, and as such the PSRR concept cannot be directly applied since - as you suggested - intermodulation products will arise.
However, if we consider PSRR of a chopped circuit as the transfer function of output components at a certain frequency generated by means of a disturbance on the supply voltage at the same or any frequency then the definition is consistent.
I assume that your lower PSRR at DC with the chopper ON is the result of a pac analysis. Following the definition for PSRR I described (which actually is what a pac analysis measures as I´m sure you know) then such result makes sense: assuming your circuit is unbalanced (otherwise at least for a differential circuit the PSRR is infinite) then you will have a finite PSRR and therefore the supply noise will be converted into differential signal by means of a certain transfer function that depends on the topology of your circuit. If such noise gets injected in between the input and output chopper switches, and if it happens to have a frequency content AT the chopper frequency or one if its harmonics, then the result at the output of your opamp will look like a DC shift. That means offset is created at the output due to a supply noise injected into the chopper. synchronized with the chopper frequency. The result is a reduced PSRR at DC.

For the same reason is that you see a high rejection at the chopper frequency: any supply noise injected into the chopper after having been converted into a differential signal will not create a noise component at the chopper frequency, unless such noise is a "DC noise" or simply offset.

Sorry for the long answer. Hope it helps
Best
Tosei

Title: Re: PSRR of Chopped Bandgap
Post by Berti on Jan 20^th, 2011, 5:53am

Wave, the external C is in the order of uF due to noise reasons. But I cannot give you more details. Moreover, the bandgap architecture is given and therefore out of discussion.

Tosei, thank you very much for the detailed answer.
I was thinking in the same direction... Indeed, I simulated the unchopped PSRR with XF and the chopped bandgap with PXF simulation.
It would however be nice to have a reference of a mathematical analysis to see what really happens. Especially the peaking in the transfer function I cannot explaint since it seems not to be directly related to the chopping frequency.

Cheers

Title: Re: PSRR of Chopped Bandgap
Post by Alexandar on Jan 21^st, 2011, 12:43am

The peak is probably a result of the fallof of the loopgain of your amplifier+feedback circuit. Does the peak shift when you change the loopgain?

Title: Re: PSRR of Chopped Bandgap
Post by Berti on Jan 21^st, 2011, 7:42am

Hi Alexandar

Indeed, the peak shifts when the loopgain is changed.
I've found that decreasing the loopgain has the same effect on the PSRR like the chopping. The attached figure shows a PSRR simulation. The blue curve is without chopping, the red curve is obtained with either chopping or reduced loop-gain.

It seems that the chopping basically decreases the gain of the first amplifier. Moreover, I still understand the peaking when the loop-gain is reduced.

Alexandar, can you please explain what you mean with "fall-off of loop-gain"?

Title: Re: PSRR of Chopped Bandgap
Post by Alexandar on Jan 24^th, 2011, 2:41am

By falloff I intended to say the high frequency behavior of the loopgain. The graph looks familiar, although it might be better to get rid of the spike u have in your PSRR. But again, I don't know the numbers or application, so maybe it is not a big deal.

The basic explanation of "chopping decreases the gain of your amp" is a bit too simplified, and imho incorrect.

Title: Re: PSRR of Chopped Bandgap
Post by Berti on Jan 24^th, 2011, 8:22am

Alexandar,

What I was thinking of with "chopping decreases the gain of your amp" is that since the amplifier bandwidth is low due to the external C, the phase of the amplifier might result in the two chopping stages being out of phase which might decrease the amplifier gain. But this seems not to be an issue since the PSRR doesn't increase at very low frequencies.

Regards