One argument against optical SPDIF transmission is jitter. I want to show the impact of a TOSLINK transmission line on jitter. This is a purely visual approach showing the jitter. I do not do exact jitter measurements!
The setup is as follows: A WM8804 drives an Everlight PLT/133 optical transmitter. This is connected via a 2m optical cable to a Everlight PLR/135 receiver. The SPDIF interface chip used an external 27MHz crystal as clock source.
Have a look at these two oscilloscope pictures. The first shows an audio transmission at 48kHz sample rate, the second at 192kHz sample rate .
Interesting difference – isn’t it? At 48kHz sample rate, there is almost no jitter visible, but at 192kHz, there is a lot of jitter.
This shows only, that jitter is measurable. Is it audible? Many modern DAC chips should have no problems with this kind of jitter. But it cannot be excluded, that there might be audible differences, especially on high sample rates.
Apart from the jitter there is one major advantage of an optical digital connection: there is no electrical connection between transmitter and receiver. This means, there is no risk for ground loops, which are usually a bigger problem, than jitter.
What happens, if we use a high-quality sine wave from a function generator as the input? Look at this:
No visible jitter! Why does the signal look so much better? Is it only a better signal quality created on the input? I don’t know it yet. But I suppose, that the SPDIF interface chip also has an impact on the jitter. In our tests, the 48kHz sample used an internal master clock of 256xfs, while the 192kHz test used only 128xfs for the internal master clock. Therefore it is possible, that the internal clock configuration of the chip has a major impact on jitter, even with the same external clock source (a crystal oscillator in our case).
There are a lot more questions than answers, e.g.:
- What is the impact of jitter from the external clock source?
- Does the internal master clock configuration have an impact on jitter?
- How do other SPDIF sender/receiver perform?
We will look into some of these aspects in the future.
References:
Thanks for these photos! With all the talk about jitter, few people actually investigate actual digital signals except by going through digital receivers. I’d like to investigate such things myself. I currently run 24/96 digital from Mac Mini to Inday Toslink splitter, thence to Maudio CO2 converter, then through about 85 feet of 1694a RG-6, then into Tact digital preamp. Works and sounds great, but I wonder about all that conversion. I wish I could avoid the two splitter/converter boxes, but I need multiple Toslink outputs, and I haven’t seen a single box which takes one Toslink input and produces 2+ Toslink outputs and 1+ coax output. I am sure the Inday box first converts the optical to an internal voltage signal which then operates 4 Toslink LED’s. If only it used that voltage to also drive a SPDIF output as well, I’d be happier.
I personally detest USB and DSD, but that seems the way everyone else is going. Thus there aren’t as many SPDIF and Toslink products as there might have been. Also, as much as my setup processes the digital, almost anything else (such as conversion to HDBaseT) would be far worse, I suspect. At least all I have in my digital line are linear devices with very small rise time, rather than conglomerations of triggers, clocks, modulators and servos.
WRT your picture of “jitter” what you are seeing is the jittery oscilloscope triggering. An oscilloscope is not necessarily made for triggering on SPDIF signals, though it shouldn’t be too bad either. But an actual spdif receiver might be able to lock onto the actual SPDIF transitions more cleanly while rejecting everything else. Then, of course, it would apply some degree of buffering and servo lock. But servo lock is not perfect (and asynchronous resampling even worse IMO, but that’s the way most people go now) so it would be best to have the initial reception of the SPDIF transitions as clean as possible. And that means, as clean a “digital” signal as possible (at the point where the “digital” signal is really an analog voltage being switched).
I think it is far harder to hear differences among different digital connections than many audiophiles realize. So it’s useful to have some way of objectively testing them, because regardless of how hard it may be to hear, I want to get it as good as possible.
The jitter trace is not an oscilloscope triggering error – you can see the edge that it triggers on is clean.