as the basics of mixed signal IC design , the VCO and it's varactor are the most important blocks in the PLL , speical most of the phase noise come form the VCO
the varactor must be sheilded very well , the substarte arround it is tied good
also tie arround the inductor
the VCO must be away from the switching " noisy " digital circuits "
like pfd
that one is tricky.
a) split as much as you can the dividers and "analog" (CP, VCO) power supply.
b) guradrings /trenches usually help
c) distance - allow at least 50um of substrate (no substrate tie etc) between digital and analog portions. (hi resistive substrate help more than anything else)
d) you can shield VCO output, Vctrl etc with ground (kinda coax)
but the best is distance.... between structures, metal lines.....
that one is tricky.
a) split as much as you can the dividers and "analog" (CP, VCO) power supply.
b) guradrings /trenches usually help
c) distance - allow at least 50um of substrate (no substrate tie etc) between digital and analog portions. (hi resistive substrate help more than anything else)
d) you can shield VCO output, Vctrl etc with ground (kinda coax)
but the best is distance.... between structures, metal lines.....
thanks alot ,
but i think the CP is a noisy block as it cause supply glitches and its contols are at refrence freq. so haw can i make it share the VCO which is the most sensetive block for noise ??
but i think the CP is a noisy block as it cause supply glitches and its contols are at refrence freq. so haw can i make it share the VCO which is the most sensetive block for noise ??
it's a bad idea to "share" CP with VCO.
put two noise blocks together will endup with more noise which u can not controll. I think Teddy's method is very good.
ok - if you surround the signal line with gnd all arround you can calculate you will get simillar effect as a coaxial cable - not exactly but close. There is a whole theory what is the best spacing and stuff but it does not matter.
For the CP and VCO spacing - yes charge pump is a major source of jitter at Fcomp. Spacing of cca 50um on chip and decent substrate tie and perhaps well isolation (or trench) will help a lot.
Another thing one should consider is that output stage will probably modulate the power lines and the field arroud those could affect your VCO noise - do not put it too close.
Honestly it is all black magic - hard to prove (especially if you do not create testchips for it) so you should do what makes you feel good . For example I believe 45deg angles are must - electrons don't bump their heads to the wall....