i have a 8GHz differential clock to convert to single ended. Is taking one end of the clock enough? what are the risks of doing so?
3
-
2\$\begingroup\$ I would think that clock is differential for a reason, especially at 8GHz. I would do everything I could to keep it that way. You also probably can't take just one end and ignore the other since it's probably some exotic high-speed signalling level. Don't cut corners at 8GHz. \$\endgroup\$– DKNguyenCommented Oct 17, 2019 at 21:34
-
\$\begingroup\$ "I have a 8 GHz differential clock": In which shape? Freespace RF wave? Balanced coax? Coplanar microstrip line? Twinax? very short unmatched distance on PCB? On silicon wafer? \$\endgroup\$– Marcus MüllerCommented Oct 18, 2019 at 0:19
-
\$\begingroup\$ this is on silicon wafer ........... \$\endgroup\$– EmeraldCommented Oct 18, 2019 at 23:00
Add a comment
|
1 Answer
\$\begingroup\$
\$\endgroup\$
Unless you have experience with common-mode noise, modelling coax, stripline, microstrip, ground effects and understand the group delay distortion effects of unequal skew paths down to the millimeter and know what an eye pattern looks like and how to optimize, you can easily destroy the integrity of a data signal as you suggested.
NB7VQ14MMNG is a popular chip for selecting external Group Delay equalization for Differential In and Out, but it does show a single-ended output in a simple block diagram. These might be useful for SMA connected test points to a single-channel DSO
