The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Simulators >> RF Simulators >> PSRR in oscillator using PXF
https://designers-guide.org/forum/YaBB.pl?num=1048364433

Message started by oreke on Mar 22^nd, 2003, 12:20pm

Title: PSRR in oscillator using PXF
Post by oreke on Mar 22^nd, 2003, 12:20pm

Hello,

I wanted to look at power supply rejection in a VCO, and tried to give PXF a run. I got a little confused after trying to digest all the parameters explained in the documentation.

I have a fundamental freq of about 2GHz, and I would like to see how any power supply disturbance up to a few MHz looks around the fundamental frequency.

In an earlier post, Ken Kundert explained, that for "sweeptype=relative":

f_out = realharmnum*f_fund + f

where f is the swept frequency; and

f_in = sideband*f_fund + f_out

Using the explanations above, I suppose I could set

realharmnum = 1
sideband = [-1]
sweep f from say 1KHz to a few MHz

and I should have swept the power supply node from 1KHz to a few MHz (from DC), and can use "freqaxis = out" to look at how it looks around the fundamental freq. Is this correct?

Is there an alternative setup for the same results? If I choose "sweeptype = absolute", and use the same sweep range, does that sweep the output or the input or both?

Thanks,
oreke

Title: Re: PSRR in oscillator using PXF
Post by Jitter Man on Apr 3^rd, 2003, 9:15am

Oreke,
The sweeptype parameter specifies how the output frequency range relates to the sweep range. If sweeptype=absolute then
f_out = f
and if sweeptype=relative then
f_out = f + relharmnum*f_fund.
So if you were to set sweeptype=absolute and leave sideband=[-1] then you would be computing the transfer function from (1kHz->1MHz)-2GHz to (1kHz->1MHz). This is a down conversion term. It the the transfer function from the lower sideband of the fundamental to baseband.

With the setup you described (with sweeptype=relative) , you were computing the up-conversion term (from baseband to the upper sideband of the fundamental). To get this same transfer function with sweeptype=absolute you would have to change the sweep range. Rather than sweeping from 1kHz->1MHz, you would sweep from 1kHz+2GHz to 1MHz+2GHz. However, this is problematic for several reasons. First, you set the sweep parameters before the simulation occurs, and at that point you really don't know the oscillation frequency precisely. Second, you can no longer plot the results with a logrithmic x-axis. Finally, it is more difficult to resolve the offset frequency from the plot.

The freqaxis parameter does not affect the transfer functions that are computed by SpectreRF, it only affects how they are plotted. Remember that the transfer functions computed by SpectreRF include frequency conversion, so the input and output frequencies may be different. freqaxis allows you to specify whether you want the results plotted versus input frequency or output frequency.

[glb]The Jitter Man[/glb]

Title: Re: PSRR in oscillator using PXF
Post by Mark Gehring on Sep 10^th, 2003, 11:47am

This is a common problem with all SpectreRF set-up. I wish that someday the forms would show you explicitly what the input and output freqs will be, so that this confusion can be avoided. I always have to try this a few times before I get it right.