TOPIC:

What is the latency, if any, between the plate amps if they are daisy chained together via AES? I cannot find anything about this. I would assume there is some, but how much? Thanks.

Almost none. Just look further down the topic list for this forum.

So, based on what I read, the latency between the AES input and output has not been measured. Right? Just sayin it's a "zero delay buffer" isn't really enough unless you actually measured it. I don't have the means to measure it, so that's why I asked.

Do you know what a buffer is?

Yes, of course. Even buffers have a small delay.

It seems logical to me that we'd want the left and right channels to be as identical as possible, so that's where I'm coming from in asking this question. I can't find much of any information about latency in AES/EBU connections. So, I don't have any numbers to determine whether or not this is an issue. I don't know how to calculate the delay in the 110 ohm AES cable either. I mean, this is the reason we normally make speaker cables in ordinary analog systems the same length: we want each channel to be as identical as possible. Thus, I'm considering setting up the AES connections as parallel rather than series (aka "daisy chain").

Inspection of the board in the plate amp reveals an input transformer for the AES, and I assume the relay is to switch that input between AES and analog? Regardless, it is unlikely that the input impedance is exactly 110 ohms under real world conditions. See, for example, this paper on that subject:
www.scientificonversion.com/technical_paper_3.html

So, assuming that the input impedance is not perfect, then there will be reflections in the line, at least for the 6.1 MHz clock signal.

Which all means to me that a parallel connection for the AES/EBU wiring is superior. Doing that requires a $300 box though.

With speaker cables, different lengths will have different RLC parameters which could at least in theory cause measurable frequency response variations (however slight). The difference in delay along the cable however would equate to a few microns (at the speed of sound) so its hard to see how there can be any significance to it. Assuming we can take devteam's word for it regarding the circuit, I'd apply the same reasoning to the buffered link.

I'm not sure I follow the point about the impedance/reflections...?

Dirk,

You're overthinking a non-issue.
There is no reason to implement a parallel scheme with your digital feed. The buffered feed-through is included in the plate amps for precisely the usage you're thinking about.
Why attempt to fix a problem that doesn't even exist??

Dave.

john.reekie wrote:

With speaker cables, different lengths will have different RLC parameters which could at least in theory cause measurable frequency response variations (however slight). The difference in delay along the cable however would equate to a few microns (at the speed of sound) so its hard to see how there can be any significance to it. Assuming we can take devteam's word for it regarding the circuit, I'd apply the same reasoning to the buffered link.

I'm not sure I follow the point about the impedance/reflections...?

If the impedance is not correct, then at RF, reflections occur from the point of cable connection back down the cable. Since AES includes RF signals (the 6.1MHz clock for example), then reflections will cause intermodulation distortion in the signals in the cable. These distortion signals will be in the form of common mode and differential mode signals. They will be further disrupted by the transformer and buffer in the plate amps, causing further disruption to connected devices.

From the article I posted: "The circuitry of a high-resolution digital audio interface is especially sensitive to common-mode noise, crosstalk and electromagnetic interference. The design of interface circuits and printed circuit board layouts must consider this interference susceptibility to realize high quality, low jitter transmission of digital audio signals.

New product designs should be tested for interference susceptibility to detect and debug the effects of noise and interference on clock jitter and data errors."

You are aware that there is an ASRC on the digital input?

It seems to me that, in the absence of suitable measuring equipment to confirm your worries, a simple listening test would suffice. Just set up two speakers next to each other, connected by the digital link, and use the rear volume controls to play one or the other.

These plate amps are not FCC compliant, first of all. Second, the AES interface buffer is really inadequate for a high quality system. Third, we don't know that this is not a problem unless someone has actually measured the amount of error correction required at the tail end of a daisy chain of plate amps. For these reasons at least, this is not "overthinking it" nor "attempting to fix a problem that does not exist". So, if you understand the consequences of my three points above, you will see that there is indeed a problem, or, at the very least, a low quality system.
See:
tech.ebu.ch/docs/other/aes-ebu-eg.pdf

Just because "it works" and "we've never had a problem" does NOT mean that the system is optimum, or even good.

What on earth does FCC compliance have to do the AES/EBU buffer?

Dirk,

I have measured the latency. I wouldn't have commented if I hadn't. I only have two though....not ten.
I can describe how I measured it, but maybe you have a better scheme? Why not simply measure a couple of your units to satisfy yourself regarding this?

What amount of latency would invalidate the interface in your opinion?

Dave.

dreite wrote:

Dirk,

I have measured the latency. I wouldn't have commented if I hadn't. I only have two though....not ten.
I can describe how I measured it, but maybe you have a better scheme? Why not simply measure a couple of your units to satisfy yourself regarding this?

What amount of latency would invalidate the interface in your opinion?

Dave.

Well, thanks for measuring it. What was the result? How much cable did you have in the cascaded chain? You need at least 30 feet to see problems. I have not looked up what would be considered an acceptable latency in this application. I do not have a proper analyzer to actually measure it, nor should I be required to buy one. It's up to the manufacturer to provide that kind of testing.

Did you measure jitter in cascaded plate amps?

Using 5 or 6 of these cascaded in a home theatre would be a bad idea in my opinion. The document I cited above states that AES cables of 10 meters can have serious problems. So, if the total length of the cascaded chain is over 10 meters (30 or so feet), then that's when problems would show up. It's not hard to get 30 feet of AES cabling in a home hifi arrangement.

john.reekie wrote:

What on earth does FCC compliance have to do the AES/EBU buffer?

You apparently do not understand how RFI works then... and how cascaded interference can be a problem. Please see the documents I have provided for more information.