How much jitter...

[A 8]

Anyone measured/have an opinion how much jitter toslink adds?

[JohnW]

A very good question, in the past I have found that optical link introduces HF Phase Noise - above say 1MHz, but very much less at LF. Also due the the Inherent Ground isolation - earth loops are removed.

I will make a test jig tonight and post the results.

[JohnW]

The Jitter of the measurement set up looped-thru (without Optical transmission loop), so Clock source (10MHz or 4MHz), Output 50 Ohm buffer, Trigger splitter and measurement set-up (Agilent DCA 86100A + HP 83484A 50GHz Scope) is 2.5pS RMS Jitter.

Below is the jitter of the complete set-up WITHOUT optical transmission loop.

50pS DIV, 2.5pS RMS Jitter

[JohnW]

Complete SPDIF optical loop, using the Toshiba TX176 / RX176 devices, 10MHz Clock

1nS DIV, Jitter = 2578pS

[JohnW]

Complete SPDIF optical loop, using the Toshiba TX176 / RX176 devices, 4MHz Clock

Even though the results appear poor, it must be remembered that the Phase noise is HF and RANDOM in nature.

SPDIF receivers have good HF jitter attenuation, and Zero attenuation at LF, so it could be argued that the Ground & RF isolation offered by optical connection is worthwhile due to the reduction in LF Mains harmonic components. Random Jitter is always preferable to discrete components.

While Transformers might offer Good Ground isolation, they offer poor RF Isolation.

1nS DIV, Jitter = 750pS

[Jaka Racman]

Hi,

since I was also concerned about toslink jitter i made my own test setup. Results are for TORX173 receiver and Hercules Fortissimo II computer audio card (transmitter unknown). Clock is around 2.8MHz.

It is very interesting that pins 5and 6 (case) of the receiver must be grounded, otherwise jitter is terrible. I also tested Infineon SFH551V receiver, but the results are much worse.

Here are my results: 3.9ns peak to peak jitter accumulated over 10 min of measurement. I am interested what the results will be after DIR1703 receiver (75ps jitter claimed in datashet)

Best regards,

Jaka Racman

[A 8]

Great posts,

I do not have the means to measure but find paying close attention to the powersupply (and grounding) of the toslink units as having a more significant effect on "percieved" sonic performance ( less jitter).

Would you guys agree or is it similar in sensitivity to any other silicon used in digitalaudio?

[A 8]

I just realized this was supposed to be a reply in the thread about converting coax -> toslink.
Circuit to convert coax digital signal to optical?

Anyway, its still interesting on its own and seems to draw attention so perhaps it should stay as is?

[JonPike]

Sure! I'm very interested in these kinds of tests.. since only those who have access to the very expensive analyzers can really give accurate results.

If you guys are willing/have the time, I'd really like to see some results downstream of various recievers.. or with ASR, or sample rate adjusting chips, different clocks, transformer input, etc... How much spare time do you have??

Would be great knowlege for the DIY community..

Jon

[Jaka Racman]

Would you guys agree or is it similar in sensitivity to any other silicon used in digitalaudio?

I was surprised how much TORX173 is sensitive to decoupling. Usual 1uF ceramic cap I use in breadbording was not enough. I had to parallel it with 220uF OSCON. I did not use the recommended 47uH inductor.

Best regards,

Jaka Racman