Designer’s Guide Community

rfx

Junior Member

Offline

Posts: 11

PSSR verifictaion for oscillators.
May 27^th, 2005, 9:31am

Hi,

I have run the PSS and PXF to get PSRR for oscillator.
I am not sure the result is correct, so I wanted to
verify the result from PXF. I have tried following;

1) PSS with periodic input signal. I put 'vsin' source for vdd and added DC and amplitude 1uV and frequecny1 MHz. When I tried to simulate this after setting up PSS simulation,
Spectre gave me following error;
" Error found by spectre during periodic steady state analysis `pss'. `Vdd' is a periodic input signal, which is inconsistent with autonomous circuits. "

2) I have tried TRAN simulation with same setup. It runs fine. But it was difficult get frequency spectrum with reasonable resolution to see.

If anybody have experience with PSS solution for oscillator in the presence of other perisodic signal(s), please let me know.

Thanks,

rfx

Back to top

IP Logged

Andrew Beckett

Senior Fellow

Offline

Life, don't talk to
me about Life...

Posts: 1742
Bracknell, UK

Re: PSSR verifictaion for oscillators.
Reply #1 - May 31^st, 2005, 10:57am

You can't have both an autonomous circuit and a driven source - because you'd need to simulate over the beat period of both periods, which is chaotic for a start!

Secondly, you don't need the large signal periodic input if you're running PSS and PXF. PXF calculates the transfer function from all sources to the specified output - so provided you have a DC source for your supply (I guess you do), simply tell it the output of the circuit when setting up pxf, and then when plotting the outputs, select the relevant input source.

Note of course this is telling you the small signal transfer function - which is normally sufficient for PSRR.

Regards,

Andrew.

Back to top

IP Logged

rfx

Junior Member

Offline

Posts: 11

Re: PSSR verifictaion for oscillators.
Reply #2 - May 31^st, 2005, 1:34pm

Thanks for the quick answer.

I have done PSS and PXF to find PSSR of oscillator.I want to verify the result from PSS and PXF, i.e.;

I want to add an AC signal at Vsupply with 1MHz frequency and 1mV (or -50dBm) amplitude; I should see
sidebands around fosc +/- 1MHz -30dBc (PXF tells me I have 20dB gain at 1MHz).

I do believe there is a steady-state solution whenboth an autonomous circuit and a driven periodic source, so it should be feasible to simulate (at least, I was able to simulate this setup with eldoRF 5 years).

/rfx

Back to top

IP Logged

Andrew Beckett

Senior Fellow

Offline

Life, don't talk to
me about Life...

Posts: 1742
Bracknell, UK

Re: PSSR verifictaion for oscillators.
Reply #3 - Jun 1^st, 2005, 1:26am

In that case, set the the PAC magnitude on the supply source, and do a PAC analysis as well. Then you can look at the signal level at the output caused by this small signal input.

The methods used in SpectreRF cannot currently handle mixed driven/autonomous circuits (with large signal inputs,
that is). In PSS you need to simulate a complete beat period, which would be impractical, as I've outlined. With QPSS potentially it might be possible, but currently QPSS can't handle autonomous circuits (it's not that easy). There are other methods that could be explored, but nothing is available in SpectreRF right now for this situation, AFAIK.

However, in your case doing either PXF or PAC as I've outlined above is probably sufficient - and that will work fine, since all you have is the oscillator as a large signal variation to contend with.

Regards,

Andrew.

Back to top

IP Logged