The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Simulators >> RF Simulators >> Double sampling
https://designers-guide.org/forum/YaBB.pl?num=1323887528

Message started by Sanjay Rajasekhar on Dec 14^th, 2011, 10:32am

Title: Double sampling
Post by Sanjay Rajasekhar on Dec 14^th, 2011, 10:32am

Hi guys,

I'm finding this a little confusing. Suppose, I have a double-sampled integrator (part of a sigma-delta ADC, being clocked at 10MHz, so sampling frequency is 20MHz). At the output of the integrator, I have a track and hold being clocked at 10Mhz, so sampling of the integrator output is at 10MHz. When I get input referred noise PSD from PNOISE sim, should I integrate it upto 5MHz (because output is sampled at 10MHz) or 10MHz (because input is sampled at 20MHz) to get the input referred noise power?

Thanks,
Sanjay

Title: Re: Double sampling
Post by Forum Administrator on Dec 14^th, 2011, 11:31am

Regardless of the input, the output updates at a 10MHz rate, and so the output spectrum will be periodic with a period of 10MHz. As such, you should integrate from 0 to 5MHz (because the signal is real, the spectrum from -5MHz to 0 will be the complex conjugate of the spectrum from 0 to 5MHz, so you only need to integrate one side).

-Ken

Title: Re: Double sampling
Post by Sanjay Rajasekhar on Dec 14^th, 2011, 9:36pm

Thanks Ken.

If I understand correctly, when output is being sampled at 10MHz, it is obvious that the power is within [0,5MHz] bandwidth, and when input referred noise is simulated using pnoise itself, the whole of the input noise power should still be within the same band. I got confused because of the input double sampling.

So here is a follow up question for the same circuit. Suppose I put a noise source of power kT/C at the input for my double sampled integrator. The bandwidth of this noise source is about 1GHz, so there is heavy aliasing when double sampled. The rest of the circuit is noiseless and infinite bandwidth (ideal components), so no inadvertent loss of noise power. When I ask pnoise to give me input referred noise, I compare the noise density it gives me with sqrt((kT/C)/(5MHz)). I expect these to be equal. However I always get input PSD of sqrt(0.5*(kT/C)/(5MHz)) from pnoise. This was why I thought maybe the input power is spread upto 10MHz. I'm looking for an explanation as to why this is happening. Any ideas?

Thanks and regards,
Sanjay