Re-clocking a must?

[Peter K]

Hi,

On every thread regarding zero oversampling DAC's re-clocking is a major item.

I am intending to build the 2 x TDA1541A (S1- is I can find some). I planned to use the I2C from my CD-PRO2 and connect it directly into the two TDA1541A, like the way the "ZERO" of Allessandro Galavotti is designed.

However looking at the attention that re-clocking usually has in the discussions on this forum, I wonder if a Zero dac can be build without re-clocking.

I would appreciate your help.

Thanks!

Peter

[Peter Daniel]

Reclocking is not such a big deal and you can easily implement it. OTOH it is also not a requirement. I suggest you try both.

I'm doing the same. I built TDA1543 DAC without reclockin. After few days will try adding reclocker and will see if it's improvement or not

[Elso Kwak]

Originally posted by Peter K
Hi,

On every thread regarding zero oversampling DAC's re-clocking is a major item.

I am intending to build the 2 x TDA1541A (S1- is I can find some). I planned to use the I2C from my CD-PRO2 and connect it directly into the two TDA1541A, like the way the "ZERO" of Allessandro Galavotti is designed.

However looking at the attention that re-clocking usually has in the discussions on this forum, I wonder if a Zero dac can be build without re-clocking.

I would appreciate your help.

Thanks!

Peter

Hi Peter,
Reclocking is not a must but it was done by Kusunoki in his NON-OS TDA1543 DAC.
I found the sound became more analog and got more depth in the soundstage. Also instruments got more "separated". Most important is that the clock used for the reclocking is low jitter.

[Dr.H]

Hi Elso,

I tried the asynch reclocking using a 66MHz oscillator (similar in construction to guido's low jitter) and it seemes to work fairly well, certainly the reclocked bck line in the cd player seemed to have less overshoot when examined using a scope. I have not listened for an improvement yet and will be doing so shortly. I am curious to hear about your experiences in TDA1543. What about reclocking data? I noticed that the waveform for data looked pretty much the same with and without recloking BCK.

How does one determine if the clock is low jitter? The 66 MHz osc. I have I purchased from a general electronics shop.

One other thing: I built the KWAK 5 clock using AD8561 and measurement wise it seems very similar to the guido tent osc. in cleanness of the clock output (at 11.xxMHz) as measured on my 20MHz scope. For sound, they actually are virtually identical. Good sound from both.

Regards
Ryan

[Elso Kwak]

Hi Ryan,
Jitter improvement can not be seen on the scope.
Kusunoki reclocked only the Bitclock and the LATCH (Wordsync) but I found reclocking the DATA also is worthwhile. In fact not reclocking the latter gave a tremendous loss of definition of the sound. The above pictured sound-experiences in post #3 are for the TDA1543. Strangely enough asynchronous reclocking gave more improvement for the AD1865. More jitter sensitive chip???

If a ready made cantype oscillator is low jitter may be seen from the specs. Be careful, jitter is specified as peak-peak, r.m.s. or 3-Sigma...But consider specs a just a indication. Final test is listening.

You could try version 6 of the KWAK-CLOCK as this works well in a Philips player. This will give more bass slam. I suppose you have a Philips CDP as your frequency is 11.2896 MHz.

All clocks have different sonic signature. Lately I tried Guido's , Netaudio and mines in my Philips CD-650. If you hear no difference go for the cheapest solution, I would say.

[guido]

Peter,

You better connect your dac to the i2s bus and not to the i2c bus... Otherwise there will not be much sound.

Have a look at the 2*tda1541 post from a while back. If you want to go for two tda's it might be usefull. I myself have little time at the moment, but i am still progressing slowly.

Greetings,
GuidoB.

[jackinnj]

Here's a link to an EDN -- current issue -- on jitter.

http://www.e-insite.net/ednmag/index...e=5%2F1%2F2003

or the PDF version of the same:


http://a330.g.akamai.net/7/330/2540/...ges/293235.pdf

[jwb]

<b>Of course</b> jitter improvement can be seen on a scope. How else were you expecting to see it? The only ways I know of visualizing jitter are with an eye diagram on an oscilloscope, or by mixing the signal under test with a signal of known quality and observing the spectrum of the result. Either way, the proof is on the CRT, not in some magical phenomenon.

[Guido Tent]

Originally posted by jwb
&lt;b&gt;Of course&lt;/b&gt; jitter improvement can be seen on a scope. How else were you expecting to see it? The only ways I know of visualizing jitter are with an eye diagram on an oscilloscope, or by mixing the signal under test with a signal of known quality and observing the spectrum of the result. Either way, the proof is on the CRT, not in some magical phenomenon.

Hi,

Like you indicate, you need additional hardware to make jitter vissible. The eye diagram method relies on the analogue processing within the scope (deflection jitter counts here) and has limitted resolution.

However, your suggestion is based on frequency domain measurements, using a low noise reference (and in-phase) oscillator, and low noise mixer.

That will give you a fairly good indication, but the non plus ultra mathod is measuring time (as jitter related to timing).

Therefor I use a wavecrest time domain analyser. I still do not have an unambiguous correlation with TDA and spectral measurements, therefor I stick with TDA.

all the best,

[jwb]

As always, I am envious of anyone with access to time-domain instruments. However I will note that inexpensive techniques can also be productive. For example, you can measure random jitter by simply building two devices under test, mixing their output in-phase, and measuring the spectrum. No stable timebase is needed, although the quality of the PLL will be evident.

I am in awe that modern instruments have such stable timebases, and PLLs operating up to tens of gigahertz. This stuff isn't covered in Horowitz and Hill!