The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> Analog Design >> How to simulate the switching kick-back
https://designers-guide.org/forum/YaBB.pl?num=1068464506

Message started by sulla on Nov 10^th, 2003, 3:41am

Title: How to simulate the switching kick-back
Post by sulla on Nov 10^th, 2003, 3:41am

Hi,

This is my first time to design a PLL circuit.
Since the charge pump current of my PLL is small, ~2uA, I try to find out how big the switching kick-back of the switches in the charge-pump.
Can you give me suggestions to simulate it?
Thanks,

ps. I use spectre to simulate my PLL.

Sulla

Title: Re: How to simulate the switching kick-back
Post by EVI on Dec 1^st, 2003, 6:44am

Hi Sulla,

The BSIM3 model has two modes - quasistatic and non-quasistatic. Quasistatic mode is the default mode and it assumes that the channel charge gets distributed equally between source and drain when the switches turn off, which is quite inaccurate model of charge injection effect. Non-quasistatic mode simulates the transistor channel as a distributed R-C system which presents much more accurate model of the channel :) but takes about 50 times longer to simulate :(. So in order to simulate the kick back effects accurately you have to switch your models to non-quasistatic mode. To do that go to the model file and change nqsmod parameter from 0 to 1.

Eugene.

Title: Re: How to simulate the switching kick-back
Post by rf-design on Dec 1^st, 2003, 2:11pm

I did not notice a 50x speed reduction if the nonstatic channel effect is turned on. Already if you test only the charge-pump the simulation time is very short.

I think you mention as a switch-back effect the charge injection of the charge pump. That is an additional constant charge which make an equivalent time offset for the input pulses. The best way is to simulate the charge-pump together with the phase-detector. If the reference and the divided frequency are equal and arrive at the same time you get an charge every period. That should be counteract by a time offset. So the charge result from the static and transient imbalance of the up/down current sources. Furtheron if you make an FFT of the charge you get the spur spectrum which modulate via the loop filter the VCO. With some calculation you also get the reference spur.