[Scrith]

I was browsing the PS Audio forums and came across an interesting post from Paul McGowan (of PS Audio), in response to a question about why their new DAC doesn't have an AES/EBU input. Here's the relevant part:

Quote:

In hindsight, we didn't include the AESEBU for two reasons: not enough rear panel space and, more importantly, degraded performance from most CD transports using the AESEBU. What? Yes. The problem is that very few CD and DVD players produce a true AESEBU signal that is indeed better than the single ended SPDIF out. On these players, I prefer to use single ended SPDIF or optical. Here's the deal. In most CD players and DVD players, the AESEBU is identical to the SPDIF output, except there's an extra inverter added to the signal. So, to make an AESEBU out, you simply take the SPDIF out and run it through an inverter chip (to make the opposite phase)and then put that signal on the XLR jack. That's it. Every less-than-$3K CD player does it this way and I am not a fan. I am not a fan because one side is different (slightly delayed) than the other side (because it had to go through an extra gate). The only other way to do this is with a high end output pulse transformer. To do that properly, you need a really good transformer like a Jensen silver (and even that rounds off the edges). Cost on these is really high - and most never bother to spend the extra dollars. I'd be willing to bet money your transport does not have the proper method of producing AESEBU. So 99% of all CD players and DVD players uses the wrong method - few use the right method. We believe that for a $1K DAC (heck, for any priced DAC) it's better to build it without compromises and with 99% of the transports in mind. I would rather use a perfect SPDIF output to the DAC than a compromised AESEBU input to make people feel better. And that's what most people have on their players.

So, I'm curious...any feedback on this? Are many (or most) of the AES/EBU outputs out there (especially off simpler devices, such as sound cards) a compromise?

By the way, Mr. McGowan is a friendly, knowledgable, responsible person who has taken the time over the years to post more than 5000 times at his company's forum. Please don't demean yourself by taking this as an opportunity to insult him or his company...I am just curious about the possibility that the much-heralded AES/EBU interface may, in fact, not be all its cracked up to be, in some cases.

 

[Garbz]

The problems of AES/EBU are well known. They suffer very similar issues to even S/PDIF, such as termination. The XLR plugs are not coaxial and cause reflections in signals just as the RCAs do for the 75ohm S/PDIF line.

He is very right about the inverter stage, however I do not think that this is a good reason for not including the stage, rather just an excuse to avoid the added expense. If the AES/EBU is formed carefully by inverting one side, and buffering the other to create an identical propegation delay on both sides, or by transformer coupling straight to AES/EBU (IMHO the way it's meant to be done) it can be much better, for one it is more immune to reflection, noise, and receivers can often cope much better recovering AES/EBU signals then S/PDIF. It is a common problem to see this implimented incorrectly. Even a $7000 marrantz unit simply used the +ve output from Q and the -ve from ~Q of a D type flipflop. After going to the trouble to reclock the output they completely neglected the differences in propegation in a D-type flipflop.

The reason that professional equipment can ignore most of these issues is the word sync accompanying this equipment. When a single external clock source or a master / slave approach is used, which this implimentation allows, synchronous reclocking at the DAC becomes trivial and the AES/EBU line really just needs to get the data across without worry for much else.

It is unfair to say one interface is worse then the other. Infact the AES/EBU interface is much better, but as always it breaks down when engineers can't be bothered or are pressured by cost reductions to poorly impliment either standard.

 

[Edwood]

Quote:

not enough rear panel space

This statement being 90% of the reason why it wasn't included.

-Ed

 

[Scrith]

Or is it?

 

[Sleestack]

Quote:

Originally Posted by Scrith Or is it?

Doesn't seem any smaller than this:

 

[Herandu]

Banter aside, I have frequently wondered about this precise issue. I have looked at many a service manual as part of my job, but couldn't quite see the audio benefits of some of the AESEBU circuits used in some units. The sound from them always seemed poorer from the AESEBU. I thought it was just me, until reading this topic.

 

[Garbz]

Nah it doesn't matter how good a standard is. It'll always break down on poor implimentation.

As an asside that looks like a 50 ohm BNC jack on the back of the Bel Canto. respect--;

 

[Iron_Dreamer]

Well, it sounds like we've gotten to the bottom of this. I'll keep using my properly implemented AES/EBU outputs, and sneeze at audiophiles and their obsession with $5000 SPDIF cables, which must be 1.5M +/- 0.01nM.

 

[Jon L]

Quote:

Originally Posted by Garbz As an asside that looks like a 50 ohm BNC jack on the back of the Bel Canto. respect--;

How the heck can you tell 50 Ohm vs. 75 Ohm BNC jack by looking at it?

 

[HumanMedia]

Quote:

Originally Posted by Jon L How the heck can you tell 50 Ohm vs. 75 Ohm BNC jack by looking at it?

The dialectric on either the male or female juts out further on the 50 ohm. On the 75 ohm it is more recessed.

 

[Solude]

Anyone hazard to guess whether the Stello CDT200 is a properly implemented AES output? Still up in the air as whether I should get a Zu Ash AES, Stereovox HDXV or something entirely different

 

[Garbz]

Quote:

Originally Posted by HumanMedia The dialectric on either the male or female juts out further on the 50 ohm. On the 75 ohm it is more recessed.

Very close, the 75ohm one doesn't actually have one at all. Not surrounding the pin anyway. I have seen some cheap imports at a store though which are neither 50 nor 75

 

[Edwood]

Quote:

Originally Posted by Scrith Or is it?

Look closer, PS Audio used a single PCB, all the connectors are in a single line, they are all PCB mounted, not panel mounted. There isn't enough room in PS Audio's implementation. They would either have to have some panel mounted components or have a riser board like Bel Canto does.

And I wouldn't call the craptastic vampire RCA SPDIF a "stellar" example of the optimum implementation of perfect 75 ohm digital output.

-Ed

 

[Jon L]

Quote:

Originally Posted by Garbz Very close, the 75ohm one doesn't actually have one at all. Not surrounding the pin anyway. I have seen some cheap imports at a store though which are neither 50 nor 75

Oh, Craapp! I've got some desoldering to do

*Edit*

In the spirit of "Glass-is-half-full," I will be able to report if changing 50 Ohm female BNC jacks x2 to 75 Ohm jacks will make an audible improvement

 

[Daroid]

I agree that just inverting the signal on one of the XLR pins is not the way to implement this, but it works, and with inverters/video rail-to-rail opamps with an extremely high slew rate the problem is minimized.
I don't think pulse transformers can be very expensive, a few dollars, but it requires more customization and compensation for inductance that manufacturers don't care about. The AES/EBU interface has become the reputation of being a defacto standard in the recording studios, hence, it must be good - and balanced is better than unbalanced, correct?

I "think" that sound cards sporting AES/EBU does it the correct way due to the simpleness and far shorter signal path, at least the signal is not inverted in any way but unfortunately not transformer coupled.

It's not an insult, but as Edwood said, why use an RCA for the SPDIF input? 99% of manufacturers don't get the SPDIF output right either, but it's understandable that he uses it since basically no-one use 75 Ohm BNC on their sources.