[time-nuts] Why a 10MHz sinewave output?
Hal Murray
hmurray at megapathdsl.net
Tue Feb 7 04:58:15 UTC 2012
magnus at rubidium.dyndns.org said:
> Oh... nothing really beats "it's what customers traditionally asks for"
> Squarewave out provides high slew-rate which reduces the effect of
> additional noise.
Right. But if you have a single frequency you can easily filter out most of
the noise.
-----------
As clock speed has increased over the years, a new field has emerged. A good name is signal integrity. That covers clock distribution, data distribution, and power supply bypassing/regulation/decoupling. It's basically all the analog stuff needed to make digital logic work in the real world.
The technology for distributing more bits is also useful for reducing noise/jitter.
There are whole families of chips for clock distribution. Many include PLLs which can correct for trace length, make other frequencies, and/or do the spread spectrum thing to "reduce" EMI. The ones I'm familiar with are targeted at digital applications. Jitter within a small fraction of a bit cell is fine. The target market doesn't care about time-nut class super clean clocks.
There are other families of chips (or parts of big chips) for driving/receiving clocks and data between boards/boxes. Most of those are now differential so I assume twisted pair is cheaper than coax. SATA between your motherboard and hard disk is a good example. It runs at 1.5, 3, or 6 gigabits/second. Wiki says up to 1 meter.
If you want a good example of the technology in this area, check out gigabit ethernet over CAT5. It's 5 level encoding (2 bits/baud) at 125 megabaud/sec over 4 pair in both directions over each pair.
Optical stuff is still single-ended. :)
There are some very low cost optical links for distribution of audio. The key idea is to use plastic rather than glass. Bandwidth is limited but cost is low.
--
These are my opinions, not necessarily my employer's. I hate spam.
More information about the time-nuts
mailing list