Clock stability?

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

featherpillow

Well-known member
Joined
May 14, 2005
Messages
214
Location
USA
I don't know if anyone on our lovely board here has experience in the digital realm...

I've always thought that a PLL clock is more stable than a crystal clock. Is this true, or am I mistaken?
 
[quote author="featherpillow"]I don't know if anyone on our lovely board here has experience in the digital realm...

I've always thought that a PLL clock is more stable than a crystal clock. Is this true, or am I mistaken?[/quote]

A PLL (phase locked loop) is something that has to have a variable frequency oscillator somewhere, the thing that the phase detector drives to synchronize with the incoming signal. This oscillator can be a VCXO (voltage-controlled crystal osc). Because of its need to be "pulled" up or down from its nominal center frequency it tends to be, all other things equal, less stable in the absolute sense than a non-tunable crystal osc. It will also usually have a lot more phase noise. And to the extent that it is working, its stability is just that of the source it's locking to. However (big however): it can have lower phase noise/jitter than the source, if the source is noisy/jittery, and if the loop filter is slow and makes the system have a big "flywheel" effect. This is why many pieces of equipment "re-clock" and get benefits.

For a system master clock you just want a really stable low phase noise crystal oscillator. There are some still better oscillators based on atomic transitions but good crystals, especially if oven-stabilized, are plenty good enough.
 
OK, that makes sense. I'm having an argument with someone via email who says that re-clocking benefits are a myth because in order for the re-clocked device to see any benefit, the jittery/drifty internal clock has to feed its signal back to the master clock so that compensation can occur. He also says that a good crystal clock will have inherently less jitter than a PLL. It looks like he's right on the latter account.
 
As well as cost-sensitive uC apps, they are evidently getting popular in microwave stuff now, and are competitive with some fairly bulky approaches. Crystals don't go high enough and frequency multiplication exaggerates noise and burns power.

From a chart out of an article touting some new ceramic resonators it does show crytals in their frequency range being at least a few orders of magnitude higher Q than conventional ceramics.

For a comparison of the two see the table on pg. 11 of this ref.:

http://www.murata.com/catalog/p17e14.pdf
 
interesting read. I do see that cer.osc. spec a couple order of magnitudes worse than crystals but not as bad as RC networks.. I guess it really just depends on the application and the necessity of precise and accurate timing. I've never liked RC timing especially since I was put on a project that was already in the "done, get it out the door" stage where I found a 12pf cap timing cap varied it's capacitance as the temp in the device changed. this caused all kinds of headache especially when I told the bosses what I thought it was and they didn't believe me.. rather didn't want to believe me since that meant recalling a number of devices that were "upgraded" with this daughterboard.
 
:shock: no kidding..

I actually took my boss out of his office and into the lab and sat him down in front of the box.. i opened the lid and took the heatgun and barely blew on the board and watched it go haywire. I then took the soldering iron and placed it near each part on the board with no effect until I got near(not touching) the cap and the thing went crazy again.

Speechless bosses are priceless!
 
Back
Top