Designer’s Guide Community

‹ Previous topic | Next topic ›

Pages: 1

N.O. Clock Gen for Pipelined ADCs (Read 3252 times)

RobG

Community Fellow

Offline

Posts: 570
Bozeman, MT

N.O. Clock Gen for Pipelined ADCs
Apr 29^th, 2010, 9:46am

I think I have heard every option on the proper way to build a N.O. clock generator for a pipelined ADC:
1) A single generator with all the buffers inside the loop
2) A generator for each stage
3) A generator with buffers outside of the loop (a bit dangerous because if the delays don't match you could get overlap)
4) A single clock gen positioned at the end of the pipeline (so the first stage gets the clocks slightly later than the last stage
5) A single clock gen positioned at the beginning of the pipelined so the first stage clocks before the other stages.

I'm sure there are others. What are your opinions on the subject?

rg

IP Logged

Ken Kundert Global Moderator Offline Posts: 2386 Silicon Valley	Re: N.O. Clock Gen for Pipelined ADCs Reply #1 - Apr 30^th, 2010, 1:56am N. O. means non-overlapping?
Back to top	IP Logged

RobG Community Fellow Offline Posts: 570 Bozeman, MT	Re: N.O. Clock Gen for Pipelined ADCs Reply #2 - Apr 30^th, 2010, 7:41am Yes, non-overlapping. Sorry for the confusion.
Back to top	IP Logged

vivkr

Community Fellow

Offline

Posts: 780

Re: N.O. Clock Gen for Pipelined ADCs
Reply #3 - May 3^rd, 2010, 12:00am

I have also seen:

1. "digital" nonoverlap where the system clock is much faster than the ADC clock, and the N.O. is thus chosen as 1 clock phase or so; not as bad as you might think, especially since there are practically no variations in the N.O.

2. On a similar note, but for higher speed systems which had a DLL on-chip, one could similarly generate a N.O. Might look a little riskier if the N.O. phases are not routed symmetrically, but that's a risk you always have with every method.

Vivek

IP Logged

Pages: 1

‹ Previous topic | Next topic ›