The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> Analog Design >> 2nd order SDM followed by sinc2 decimation filter?
https://designers-guide.org/forum/YaBB.pl?num=1501871794

Message started by neoflash on Aug 4^th, 2017, 11:36am

Title: 2nd order SDM followed by sinc2 decimation filter?
Post by neoflash on Aug 4^th, 2017, 11:36am

This attached diagram shows a sensing scheme that uses 2nd order sigma delta modulator followed by a second order sinc decimation filter.

In this configuration, the out of band quantization noise will be flat instead of rolling off in higher frequency.

In other words, the post filter noise spectrum will be approximately white and quite some noise will alias back.

I think it should be much easier to use 3rd order sinc filter. However, this paper made a different choice made me think that maybe this wide band noise is not a problem when OSR is high enough?

Regards,
Neo

Title: Re: 2nd order SDM followed by sinc2 decimation filter?
Post by Berti-2 on Aug 7^th, 2017, 4:51am

The frequency-response is sinc is sin(x)/x-like. For large OSR, the anti-aliasing filtering therefore becomes more efficient. Those filters are typically implemented as CIC filters.

I guess that's the reason why a second order filter is sufficient in this case.
(Your assumption that the quantization noise will be flat is consequently not true).

Cheers

Title: Re: 2nd order SDM followed by sinc2 decimation filter?
Post by neoflash on Aug 7^th, 2017, 3:28pm

The noise shaped of the 2nd order modulator will be 40dB/dec.

It then appears to me that a sinc-2 filter will not be sufficient to suppress the shaped noise.

Berti-2 wrote on Aug 7^th, 2017, 4:51am:

Title: Re: 2nd order SDM followed by sinc2 decimation filter?
Post by Berti-2 on Aug 8^th, 2017, 4:03am

If the OSR is large enough (>50), I would expect a sinc-2 filter being sufficient (because of the sin(x)/x characteristic, as I wrote in the previous post).

But you can write a simple Matlab script to check whether the decimation filtering is sufficient for a given OSR ... if you don't believe me. ;)

Title: Re: 2nd order SDM followed by sinc2 decimation filter?
Post by neoflash on Aug 11^th, 2017, 9:17pm

I believe you.

But that approach of high osr doesn't seem to be an economic choice.

A sinc3 with much lower OSR will do the job better, isn't it a lower cost way?

Berti-2 wrote on Aug 8^th, 2017, 4:03am:

Title: Re: 2nd order SDM followed by sinc2 decimation filter?
Post by nrk1 on Aug 15^th, 2017, 3:11am

Neoflash, Can you share the reference?

Title: Re: 2nd order SDM followed by sinc2 decimation filter?
Post by Berti-2 on Aug 17^th, 2017, 4:25am

Hi neon-flash,

You need to consider the combination of analog modulator and decimation filter.
The chosen OSR is often dictated by the analog part (modulator) and not the decimation filter.

Title: Re: 2nd order SDM followed by sinc2 decimation filter?
Post by neoflash on Aug 17^th, 2017, 4:19pm

nrk1 wrote on Aug 15^th, 2017, 3:11am:

Neoflash, Can you share the reference?

It is this paper. ISSCC-2017

9.7 A 6.9mW 120fps 28×50 capacitive touch sensor with 41.7dB SNR for 1mm stylus using current-driven ΔΣ ADCs
Hyunseok Hwang ; Hyeyeon Lee ; Hongchae Kim ; Youngcheol Chae
Publication Year: 2017, Page(s):170 - 171

Title: Re: 2nd order SDM followed by sinc2 decimation filter?
Post by Frank_Heart on Nov 18^th, 2017, 11:39pm

Hi, Neo,

Sinc3 would give better noise reduction, but more attenuation on input signal as well. People may care about signal loss more when noise reduction would not give more benefits.

-Frank