Jun 2, 2007 at 12:11 PM Thread Starter

#### Regus

I have been designing an oversampling DAC for the past year, year and a half (in my rare spare time) and I have been thinking about input and DAC clock variances...

I mean if the speed of the input clock is off even by 1% we are talking 441 error samples a second in a simple 44.1kHz scenario...

Now a TENT XO clock is specified for +/- 50ppm that would be 0.005% now if we asume both ends of the system uses something like this we are talking a maximum variance of 0.01% which is still 4 error samples per second... now while this might qualify as acceptable it is asuming both clocks are TENT XO clocks or equivalent...

Now I was wondering is it possible to design a low jitter clock the speed of which can be electrically addjusted to within say 500 ppm of the target frequency in 10ppm steps or better.

With such a clock it would be possible to compare input and DAC frequency and adjust the latter to match the input without having to fight with input jitter throughout the whole system...

Or is the some brilliantly simple solution I have overlooked?

Jun 2, 2007 at 2:28 PM
http://www.tentlabs.com/InfoSupport/...34/page34.html

PPM is a unit used to specify the clock's performance to the specified frequency, but has nothing to do with jitter. I think what you may be looking for is an ASRC? These don't seem to be viewed kindly in terms of jitter performance, however.

Jun 2, 2007 at 3:04 PM
Quote:

 Originally Posted by Regus /img/forum/go_quote.gif Or is the some brilliantly simple solution I have overlooked?

You can change an XOs frequency by changing the supply voltage. I don't know if the Tent XO is a VCXO (voltage compensated) or not - if yes, you must change the voltage quite a bit to get a change of a few ppm, if not, you need quite a stable regulator with very fine voltage control.

Jun 2, 2007 at 3:27 PM
Quote:

 Originally Posted by Pars /img/forum/go_quote.gif http://www.tentlabs.com/InfoSupport/...34/page34.html PPM is a unit used to specify the clock's performance to the specified frequency, but has nothing to do with jitter. I think what you may be looking for is an ASRC? These don't seem to be viewed kindly in terms of jitter performance, however.

Yes and you don't get error samples or loose samples due to jitter...

The problem with asyncrhous resampling is that you you either get error samples or loose samples depending on which clock is faster but if one were to tune the internal frequency to match the external frequency on average not on a clock by clock basis it you get ridt of the jitter from the outside...

And I can't seem to find ICs to take the core frequency from the source and multiply it so i can use it for the oversampling DAC - and it is either that or synchronizing two clocks...

Jun 2, 2007 at 3:29 PM
Quote:

 Originally Posted by balou /img/forum/go_quote.gif You can change an XOs frequency by changing the supply voltage. I don't know if the Tent XO is a VCXO (voltage compensated) or not - if yes, you must change the voltage quite a bit to get a change of a few ppm, if not, you need quite a stable regulator with very fine voltage control.

hmm are you saying that I could adjust the frequency of an regular Tent XO by changing the voltage?

So that theoretically I could adjust it using some kind of creative feedback circutry?

Jun 2, 2007 at 4:26 PM
He does offer a VCXO in addition to the regular XO modules. The XO modules probably do not vary with voltage (much). The VCXO is only variable by +/- 100ppm though. The VCXO is used in his XO DAC module, which retimes the incoming bitstream.

http://www.tentlabs.com/Components/VCXO/index.html

Jun 2, 2007 at 4:36 PM
Quote:

 Originally Posted by Pars /img/forum/go_quote.gif He does offer a VCXO in addition to the regular XO modules. The XO modules probably do not vary with voltage (much). The VCXO is only variable by +/- 100ppm though. The VCXO is used in his XO DAC module, which retimes the incoming bitstream. http://www.tentlabs.com/Components/VCXO/index.html

Yeah looks good except I worry 100ppm is too little I doubt one can be sure a CD player og soundcard comes within 100ppm, and I really need 33.8688 MHz which doesn't come in VCXO...

Jun 2, 2007 at 4:46 PM
Well, looks like the non-VC option is right for you, why not go for it?

-comes in the frequency you need
-you're able to change frequency by more than 100ppm

Jun 2, 2007 at 5:59 PM
Quote:

 Originally Posted by balou /img/forum/go_quote.gif Well, looks like the non-VC option is right for you, why not go for it? -comes in the frequency you need -you're able to change frequency by more than 100ppm

Well by how much can I change it how accuratly and exactly how do I go about doing it?

Jun 2, 2007 at 6:56 PM
You need a PLL + VCXO... you cannot pull a normal XO