Asynchronous reclocking

[Bricolo]

Originally posted by Peter Daniel


I will try reclocking with TDA1541 as well. I just don't like it with the 1543 chip.

Maybe this will be part of your "correct implementation of the TDA1541"

[Oli]

Been a long while since I've contributed...

I tried reclocking 2x AD1861 using Elso's reclocker. Very impressive results. I wouldn't go back. Sound stage opened right up and the result was less harsh, with more audiable detail. This is ironic since asynchronous reclocking loses information. Obviously the audiable benefits outweigh the information losses.



Question-

Can anyone help with some advice on using silver-mica capacitors for my reclocker. currently I'm using NPO ceramic. My question is....

Will silver-mica produce better results, since these will obviously be more bulky and may have higher inductance, resulting in more impedance at high frequencies. These have to operate at 100Mhz where inductance becomes a serious issue!

Elso suggested the use of a polypropylene capacitor to couple the signal- is this appropriate at 100Mhz?

[Elso Kwak]

Hi Oli,
Your result confirms my idea that the ASR improves the sound with the Analog Devices DAC’s, R2R types. I did not want to miss it with the AD1865N-K. I don't have an explanation why the result with the TDA1543 is much less spectacular.
I don't think we loose information as long as we use two flip-flops in series. In this case there are no dropouts and scope pictures look AOK.
Silver mica caps work just as well as NPO ceramics but are much more expensive. I had more problems finding the right type of inductors. The "resistor" style inductor did not work well.
This a page picturing the type I am using:
http://217.34.228.137/Pages/MOLD/page84.htm
The 10nF polypropylene improved the sound a bit over a ceramic disc cap. I tried several types here. The frequency of my ASR is 60 -66 MHz, though a 100MHz crystal is used.

[borisov57]

Hi,
Elso: I don't have an explanation why the result with the TDA1543 is much less spectacular.

Maybe because with AD1865 you are reclocking Latch enable, which triggers output conversion, so if you have low jitter Latch enable, you have low jitter analog output.
In case of TDA1543 you are reclocking word clock and bit clock, where one off this signal triggers analog output through some internal logic with its own additional jitter.

Have a nice day,

Borisov57

[dddac]

The only true comparison is real A-B. Like I use in the dddac1543. Actually you can switch, whilst playing from normal clock to asynch reclocking mode. Everone, no exceptions, who I demonstrated this and also people who built the same design and did the test themselves were the same opinion. it works........ more space, more clear soundstage, no discussion needed, blind test, clearly detectable. No theorie, no datasheets, no different designs, no cable plugging, just click and there it is.....

BUT, it only works, if you use a XO clock which is 11.2896 Mhz. I tried 12, 18 and 24Mhz and it produces clearly noticable IM distortion......

regards,

doede

[Peter Daniel]

I have a Tent clock at that frequency and would be willing to try it out. Which circuit are you using?

I tried 24xxxMhz and you probably know my opinion

[dddac]

Hi Peter,

I used 12, 18 24 Mhz, VERY bad......

look at this pic..... shows a 1kHz sine wave spectrum reclocked with 12 Mhz. You can hear the xtra tones with female voices easily, brrrr

schematic is on my site at download.....

www.dddac.de

doede

[dddac]

to compare....

[Oli]

Cheers for the info Elso!

I can't argue that asynchronous reclocking will lose data if you have performed measurements and found the waveform is preserved. Regardless, asynchronous reclocking sounds best to me!

My thought would be that any data/fsync/ws transitions that occur whilst clock transistions occur would give rise to metastable states- i.e the flip flops would latch up for an unspecified period of time and data would be lost during this time. I thought the second flip-flop simply maintained synchronicity. Once the first flip-flop unlatched from lock-up the timing could be re-established with this second flip-flop. I.e Data can be lost, but timing is preserved.

My prototype reclocker is built on Veroboard. I am doing a proper PCB for the whole DAC. In your view will silver-mica capacitors provide any sonic benefits over NPO ceramics? I may as well use the best component I can- my time is more precious.

What if I replace the aforementioned polypropylene capacitor with a silver-mica capacior too? This capacitor does sit in the clock path.

[Pedja]

Originally posted by dddac
I used 12, 18 24 Mhz, VERY bad......

look at this pic..... shows a 1kHz sine wave spectrum reclocked with 12 Mhz. You can hear the xtra tones with female voices easily, brrrr

Doede,

Reclocking with non Fs multiple frequencies won’t in itself introduce that distortion.

Pedja