Originally Posted by DonH50
Doesn't Amir's AP test rig provide the proper termination? I sort of figured it would but do not know (and am too lazy/busy to look it up).
Hi Don. I have been avoiding this point as to not confuse OP since it is unrelated to the topic at hand. But seeing how folks are reading into my lack of response, here we go
I think Arny confused the output that I showed with that of an oscilloscope where its default impedance is very high (1 megaohm or higher). As you correctly observed, such was not the case with my measurement. It was performed on an audio analyzer
and the input I used is its unbalanced connection that is made to measure S/PDIF. In that regard, it is normally terminated to 75 ohms. The AP does however have the option to remove the termination under its more advanced setting. Indeed, I had used that option as part of this larger test which was to determine the termination impedance of the source/cable. In the graph I post here however, termination was in place at 75 ohms. So the data was as presented.
Here is the same test except this time I have overlaid the unterminated input of the AP on the terminated version for both the generic high-bandwidth cable and the low-bandwidth one:
As expected, the usual effect is there. Without the termination resistor, the output voltage shoots up and there is slight amount of ringing together with some rolling off the high frequencies. You can see this in the green vs yellow (terminated vs. not respectively). Similar thing goes on with the low-bandwidth cable with its jagged response. It is just a different shape of bad
. With respect to OP's question, all four of these scenarios created a reliable transmission although the analyzer did raise a warning, lighting up its "confidence" flag for the worst case signal: unterminated low-bandwidth. So from data
transmission point of view, all was well.
It was a different story with respect to quality
of the timing for each sample though. Jitter which was already bad for the low-bandwidth cable at around 5,000 picoseconds, shot up to 40,000 picoseconds (no doubt the reason for above warning from AP). As a reminder, the high bandwidth generic cable was at 500 picoseconds and was unphased by impedance change.
As always when we look at digital audio transmission, we need to keep in mind that it must convey two things: (1) the digital audio samples and (2) "when" those samples need to be output. The former is very robust in consumer applications with short cables. The latter less so although in this case I could not harm a high-bandwidth cable even after screwing up the "transmission line" (within the limits of my measurement tool).
Finally, let me note that the low bandwidth cable is actually a high-end audiophile cable. It is not however designed for this application. It is an analog interconnect with a filter network. I used it here as an easy way to simulate what cable bandwidth does to the transmission and jitter without having to manually build such a circuit. So unless you have really, really long cables, you are not going to see the degraded response I am showing here. Think of it as a learning tool as you read these measurements. It is not meant at all to justify expensive S/PDIF cables. Indeed the data says you the cheapest thing worked great in this scenario (again within the limits of my measurements which is a few hundred picoseconds of jitter).