Designer’s Guide Community

Mohammed Omar

Junior Member

Offline

Posts: 17

AC simulation for integrate and dump circuit
Sep 06^th, 2009, 3:24pm

Hi All,

I was wondering how to do AC simulation, or get the Frequency response, of an integrate and dump circuit or generally for a linear time-variant circuit in Spectre.

An integrate and dump circuit operates in a periodic manner but to get the AC spectrum using PAC after PSS doesn't seem to give the famous sinc function response.

Regards,
Mohammed Omar

Back to top

IP Logged

Frank Wiedmann Community Fellow Offline Posts: 678 Munich, Germany	Re: AC simulation for integrate and dump circuit Reply #1 - Sep 7^th, 2009, 12:38am Take a look at http://www.designers-guide.org/Analysis/sc-filters.pdf.
Back to top	IP Logged

Mohammed Omar Junior Member Offline Posts: 17	Re: AC simulation for integrate and dump circuit Reply #2 - Sep 8^th, 2009, 4:53pm Thanks Frank. I guess this works for SC circuits and those with periodic operating points. I believe you are pointing me out to choose a correct phase of the PSS analysis. But will this analysis be valid for integrate and dump circuit? What about general time-varying circuits?
Back to top	IP Logged

Frank Wiedmann

Community Fellow

Offline

Posts: 678
Munich, Germany

Re: AC simulation for integrate and dump circuit
Reply #3 - Sep 8^th, 2009, 11:46pm

PSS analysis and the associated small-signal analyses like PAC and PXF only work for circuits with a periodic operating point. As the name "small-signal" implies, the input signals for the PAC and PXF analyses are assumed to be so small that they do not noticeably change the operating point. I am not familiar with integrate and dump circuits, sorry.

Back to top

IP Logged

Ken Kundert Global Moderator Offline Posts: 2386 Silicon Valley	Re: AC simulation for integrate and dump circuit Reply #4 - Sep 9^th, 2009, 1:20pm Integrate and dump are just a variation on switched capacitor circuits, and so PSS, PAC, PNoise, and the like will work on them just as they would conventional switched-capacitor circuits. -Ken
Back to top	IP Logged

Mohammed Omar

Junior Member

Offline

Posts: 17

Re: AC simulation for integrate and dump circuit
Reply #5 - Sep 11^th, 2009, 1:53pm

Thanks Frank and Ken for the prompt reply.

I simulated a simple integrate and dump circuit and got results inconsistent with those of the simple hand analysis.

Attached are the PSS and PAC results, I've been using a DC input for the integrate circuit; its schematic is attached too. I couldn't figure out how to attach them all in a single post, so I'll attach first the PSS results, followed by the PAC and then the schematic.

From what I understand is that this circuit should have a frequency response similar to the sinc function multiplied by a constant, but I can't figure out why it is not showing that shape.

Back to top

pss.png

IP Logged

Mohammed Omar Junior Member Offline Posts: 17	Re: AC simulation for integrate and dump circuit Reply #6 - Sep 11^th, 2009, 1:58pm Here are the PAC results
Back to top	pac.png IP Logged

Mohammed Omar Junior Member Offline Posts: 17	Re: AC simulation for integrate and dump circuit Reply #7 - Sep 11^th, 2009, 1:58pm Here is the schematic
Back to top	Screenshot_005.png IP Logged

Ken Kundert

Global Moderator

Offline

Posts: 2386
Silicon Valley

Re: AC simulation for integrate and dump circuit
Reply #8 - Sep 12^th, 2009, 12:26am

You might want to study the point behind Figure 6 in http://www.designers-guide.org/Analysis/sc-filters.pdf.

Also, I don't believe the shape should be a sinc function. Given the integrating nature of the circuit, I would expect it to have a sinc² shape. Plus there are other things that will affect the shape, such as the finite turn on and turn off time of the switch, the on resistance of the switch, simulator time steps, etc.

-Ken

Back to top

IP Logged

Mohammed Omar Junior Member Offline Posts: 17	Re: AC simulation for integrate and dump circuit Reply #9 - Sep 14^th, 2009, 2:20am Thanks Ken. I got your point. I'll check things out and let you know how things go. Thanks a lot for your help. Regards, Mohammed Omar
Back to top	IP Logged

Mohammed Omar Junior Member Offline Posts: 17	Re: AC simulation for integrate and dump circuit Reply #10 - Sep 25^th, 2009, 9:22am Hi Ken, I was wondering why you believe the shape should be a sinc² function not a sinc function; I believe the integrate and dump circuit should have the same frequency response of the integrating ADC. Thanks. Mohammed Omar
Back to top	IP Logged

Ken Kundert

Global Moderator

Offline

Posts: 2386
Silicon Valley

Re: AC simulation for integrate and dump circuit
Reply #11 - Sep 25^th, 2009, 12:28pm

The transfer function of an idealized sampler is flat. The transfer function of a sample and hold is the same as an idealized sampler except with the output passed through a zero-order hold filter. This filter has a pulse as an impulse response, which in the frequency domain has a transfer function of sin(x)/x, or sinc(x). An integrate and dump is the same as a sample and hold, except the input is first passed through a similar filter, one with a pulse as an impulse response. This filter performs the integration that the circuit is named after. As such, the overall transfer function of an integrate and dump should be sinc²(x) because the transfer function of each filter is sinc(x) and the signal passes through two of them. I would expect the same to be true for an integrating ADC.

-Ken

Back to top

IP Logged

Mohammed Omar

Junior Member

Offline

Posts: 17

Re: AC simulation for integrate and dump circuit
Reply #12 - Sep 25^th, 2009, 1:20pm

Thanks for the fast reply.

It seems I'm missing something, what is the difference between an idealized sampler and the zero order hold?

From what you I are saying, if the integrate and dump circuit has a pulse response, then it should have the sinc(x) response in frequency, but this isn't the case.

For the integrating ADC, I believe the resulting spectrum is that of sinc(x) not sinc²(x); Check Figure 3 in at this linkhttp://www.analog.com/static/imported-files/tutorials/MT-027.pdf.

Thanks a lot Ken.

Back to top

IP Logged

Ken Kundert

Global Moderator

Offline

Posts: 2386
Silicon Valley

Re: AC simulation for integrate and dump circuit
Reply #13 - Sep 26^th, 2009, 12:51am

An idealized sampler outputs impulses, so the output is zero everywhere except at the sample points. It has a flat transfer function.

A traditional sample and hold is the equivalent of an idealized sampler followed by a sinc() filter (a filter that has a unit pulse as an impulse response)

An integrate and dump is a traditional sample and hold preceded by a sinc() filter. Thus, there are two sinc() filters in succession, resulting in a sinc²() transfer function.

Don't trust me or the ap note you reference. Work it out for yourself.

-Ken

Back to top

IP Logged

Mohammed Omar

Junior Member

Offline

Posts: 17

Re: AC simulation for integrate and dump circuit
Reply #14 - Sep 27^th, 2009, 12:06pm

Thanks for the clarification.

I did the Integrate and Dump derivation and I got what you're trying to say.

The problem now is why doesn't the integrate and dump circuit show the desired response using PAC. It gives the same results I've posted before; there is no nulls in the PAC response and I don't believe anything would differ if I add an idealized sampler, i.e. The response should be the same before and after an idealized sampler, or am I missing something?

Back to top

IP Logged