Can a frequency Divider cause Jitter?

[Lars Clausen]

Some Clock's use a frequency divider IC after the clock, to save money by only stocking 2-3 different frequencies.

The other 8 freqencies required to cover all CD / DVD models can then be derived from the 3 fundamental frq's, by diving, for example:

33.8688 MHz (Fundamental oscilator freq.)
div by 2 = 16.9344 MHz
div by 2 = 8.4672 MHz
div by 2 = 4.2336 MHz (used in Philips CD 100 a.o.)

etc. In this case one clock can supply 4 different frequencies.

Is this a good way to go, or will the frequency divider cause jitter on the otherwise clean Clock signal? What is your opinion?

[Jax]

It is the only way to go. Use a fast CMOS divider and decouple it with many ceramic capacitors so it does not disturb the oscillator. Use 22 to 33 ohm resistors in series with the outputs of the divider, SMD resistors as close to the output pins as possible are prefered.

Synchronous counters make the best dividers.

The divider contributes very little to the jitter, most is from the oscillator.

[TNT]

Lars !

I think You should state Your connection to LC-Audio in Your profile !

You are LC-Audio, aren't You ?

It is custom to do so if You have commercial connection to the bussines.


BR /

[Elso Kwak]

Originally posted by Lars Clausen
Some Clock's use a frequency divider IC after the clock, to save money by only stocking 2-3 different frequencies.

The other 8 freqencies required to cover all CD / DVD models can then be derived from the 3 fundamental frq's, by diving, for example:

33.8688 MHz (Fundamental oscilator freq.)
div by 2 = 16.9344 MHz
div by 2 = 8.4672 MHz
div by 2 = 4.2336 MHz (used in Philips CD 100 a.o.)

etc. In this case one clock can supply 4 different frequencies.

Is this a good way to go, or will the frequency divider cause jitter on the otherwise clean Clock signal? What is your opinion?

Hello Lars,
I am one day off from jester course, with permission of the course leaders.
I cannot believe this is happening!.
Am I dreaming?. Jocko please squeeze my arm! No, no, do not pull my leg!
Or am I confusing said course with this cosy forum?
Maybe Per-Anders can shed light on this? He is from Denmark too, you know, not from "Barcelona"!
Have a nice day.
Edit:
Oops I am wrong, Per-Anders is from Sweden. Sorry Per!

[Lars Clausen]

TNT: Some people are against others posting their commercial connections, and have made strong objections to me about this.

But in all honesty i will clearly state that i am of course LC from
LCAudio Technology.

However i do not use this forum for advertisement, only discussions about what interest me, audio technology. I hope no one finds that the fact that i am working in the audio industry should prevent me from being interested in audio technology, or using this forum?

Lastly i will also tell you that i agree totally with you TNT, it would be great if everyone openly stated their business relationships in their profile. It would make it easier to understand why some people criticise others for no apparant reason. Seen any of those posts lately ??

All the best to everyone from L C

[rbroer]

Elso,
You couldn't resist, could you

[Elso Kwak]

Originally posted by rbroer
Elso,
You couldn't resist, could you

Hi Rudolf,
Yes indeed

[rfbrw]

Originally posted by Lars Clausen
Some Clock's use a frequency divider IC after the clock, to save money by only stocking 2-3 different frequencies.

The other 8 freqencies required to cover all CD / DVD models can then be derived from the 3 fundamental frq's, by diving, for example:

33.8688 MHz (Fundamental oscilator freq.)
div by 2 = 16.9344 MHz
div by 2 = 8.4672 MHz
div by 2 = 4.2336 MHz (used in Philips CD 100 a.o.)

etc. In this case one clock can supply 4 different frequencies.

Is this a good way to go, or will the frequency divider cause jitter on the otherwise clean Clock signal? What is your opinion?

If you do not use dividers, how would you suggest maintaining synchronicity between e.g. the separate Mclk,Sclk and LRclk?

[Lars Clausen]

rfbrw: Of course - typically in the servo chip, dividers are used, for the purposes you mention.
But the servo chip is not all that sensitive to jitter, only the DAC is. That is why a refernce clock should always be connected as directly to the DAC as possible.

What i meant to discuss in this tread was, the use of dividers on the reference clock, to scale down from say 33.8688 MHz to 16.9344 MHz, BEFORE the clock signal is injected to the jitter sensitive DAC.. ? Any opinion on that?

[Elso Kwak]

Originally posted by Lars Clausen
rfbrw: Of course - typically in the servo chip, dividers are used, for the purposes you mention.
But the servo chip is not all that sensitive to jitter, only the DAC is. That is why a refernce clock should always be connected as directly to the DAC as possible.

What i meant to discuss in this tread was, the use of dividers on the reference clock, to scale down from say 33.8688 MHz to 16.9344 MHz, BEFORE the clock signal is injected to the jitter sensitive DAC.. ? Any opinion on that?

Hello Lars,
A NON-OS DAC does not use a masterclock. How about that?
Still this kind of DAC benefits from a low-jitter clock in the transport.