Designer’s Guide Community

adler

New Member

Offline

Posts: 3

confused with pnoise simulation setting
Dec 23^rd, 2015, 5:45pm

hello
while doing pnoise simulation,i have no idea of how to set the maxside band, and how to choose the option of "jitter mode" or "source mode"?
now, i have some views on the two problems above
1. if beat frequency is 5G, while i want to check 1M~15GHz Random jitter,
so the pnoise simualtion frequency integration range is 1M~2.5G, and the max sideband is (15G-2.5G)/2.5G=5
2. if doing driven circuit(buffer chain,divider,pfd...) pnoise simulation should choose "jitter mode"
if doing autonoumous circuit(vco...)pnoise simulation should choose"source mode"
3. BTW,what is the relationship of "jitter mode" and "source mode" with FM/PM noise option?
appriatiate for your kindly answer

Back to top

IP Logged

Ken Kundert

Global Moderator

Offline

Posts: 2384
Silicon Valley

Re: confused with pnoise simulation setting
Reply #1 - Dec 23^rd, 2015, 8:05pm

1. See Simulating switched-capacitor filters with SpectreRF for informaiton on how to choose maxsideband.
2. To determine what kind of analysis to perform, you must think about the circuit that follows the DUT in your system. For example, if the circuit that follows the DUT is a thresholding circuit, then you should do a jitter analysis (for example, the buffer drives a frequency divider and perhaps a mixer, both are threshoulding circuits and so you should use a jitter analysis. The PFD drives the loop filter and VCO, these are not thresholding circuits and so you should use a time-average noise analysis. Having said this, there are some important things you need to know:

a. The term 'source mode' is completely misleading. The correct name is 'time-average noise', meaning that it computes the average value of the noise (as an unaided spectrum analyzer would measure).
b. Oscillators are a special case. Because the noise produced by an oscillators is virtually all in the phase, the power in total noise and the power in the phase noise are the same. As such, you can use a simple noise analysis (source mode or time-average noise). You can use a specialized analysis (jitter or AM/PM), but it will be more expensive and should give the same result.
c. The calculator phase noise function does not actually compute the phase noise, it computes the normalized noise. It only represents the phase noise when applied to an oscillator.

-Ken

Back to top

IP Logged

adler

New Member

Offline

Posts: 3

Re: confused with pnoise simulation setting
Reply #2 - Dec 23^rd, 2015, 8:16pm

hello,Ken
thanks for your reply,
as you said, for threshhold circuit, we can choose jitter mode,
since pfd drive lpf and vco,we need to choose source mode or time domain mode,
but, what we should choose if we want to simulate the total pll phase noise, including driven circuit(divider..)and autonoumous circuit(vco...)
thanks again

Back to top

IP Logged

adler

New Member

Offline

Posts: 3

Re: confused with pnoise simulation setting
Reply #3 - Dec 23^rd, 2015, 8:29pm

Ken Kundert wrote on Dec 23^rd, 2015, 8:05pm:

1. See Simulating switched-capacitor filters with SpectreRF for informaiton on how to choose maxsideband.
2. To determine what kind of analysis to perform, you must think about the circuit that follows the DUT in your system. For example, if the circuit that follows the DUT is a thresholding circuit, then you should do a jitter analysis (for example, the buffer drives a frequency divider and perhaps a mixer, both are threshoulding circuits and so you should use a jitter analysis. The PFD drives the loop filter and VCO, these are not thresholding circuits and so you should use a time-average noise analysis. Having said this, there are some important things you need to know:

a. The term 'source mode' is completely misleading. The correct name is 'time-average noise', meaning that it computes the average value of the noise (as an unaided spectrum analyzer would measure).
b. Oscillators are a special case. Because the noise produced by an oscillators is virtually all in the phase, the power in total noise and the power in the phase noise are the same. As such, you can use a simple noise analysis (source mode or time-average noise). You can use a specialized analysis (jitter or AM/PM), but it will be more expensive and should give the same result.
c. The calculator phase noise function does not actually compute the phase noise, it computes the normalized noise. It only represents the phase noise when applied to an oscillator.

-Ken

hello,Ken
thanks for your reply,
as you said, for threshhold circuit, we can choose jitter mode,
since pfd drive lpf and vco,we need to choose source mode or time domain mode,
but, what we should choose if we want to simulate the total pll phase noise, including driven circuit(divider..)and autonoumous circuit(vco...)
thanks again

Back to top

IP Logged

Ken Kundert

Global Moderator

Offline

Posts: 2384
Silicon Valley

Re: confused with pnoise simulation setting
Reply #4 - Dec 23^rd, 2015, 10:02pm

Not time domain noise. What I said was time-averaged noise.

Concerning what type of noise analysis to use on the whole PLL, you need to think about what follows the PLL. If it is a mixer, use jitter or because a switching mixer is a thresholding circuit.

Quote:

2. To determine what kind of analysis to perform, you must think about the circuit that follows the DUT in your system. For example, if the circuit that follows the DUT is a thresholding circuit, then you should do a jitter analysis (for example, the buffer drives a frequency divider and perhaps a mixer, both are threshoulding circuits and so you should use a jitter analysis. The PFD drives the loop filter and VCO, these are not thresholding circuits and so you should use a time-average noise analysis.

-Ken

Back to top

IP Logged