[art]

After you read Steve's white paper.............check out this picture of a typical example of how not to make an SPDIF pcb. I especially like the wiring to/from the RCA jacks.



Here is the link so that you can get a better look: http://www.diyaudio.com/forums/showthread.php?postid=626678#post626678

One of the worst examples of SPDIF outputs is in the venerable CD-80. Besides lots of mystery coax, it has probably the worst transfomer ever. Designed soley to pass an emmisions standard test, as it can not pass a square wave at the data rate.

Pat

[Uptown Audio]

Hi Pat (and all),
Your comments make a lot of sense and show that you know your cables and filters. I have been using separate DACs for some time now and have also used quite a few cable types at home. As you might imagine, I have extensive experience with digital connections through my business as there are loads of connections to be made in audio around here.
I have noticed differences in digital coax cables for years and found several cables that I really liked with most equipment and others that I did not. I have also used analog cables in a pinch, but never really even considered it a practice for the reasons that you have illustrated.
At the house, I am using what has become my favorite cable and I doubt that it is made any longer, but years ago Audio Alchemy made what they called a true 75 ohm digital coax cable. It seems a flimsy bit of coax, maybe RG59, with and RCA on one end and an F connector on the other I believe. I don't have it in front of me, but I believe that is correct. The F connector then screws onto a transformer, which is said to be best used at the input of the DAC. The transformer has an RCA connector on it's other side. If the construction is modular as I describe it, one could always use another section of coax to extend it a bit. I think it is one meter now. I have used other types with mixed results, but this one sounds dead neutral and has the highest resolving ability to my ear.
I have also enjoyed an Audioquest silver conductor, digital cable, which uses welded RCA terminations for what they say is a true 75 ohms. It sounds nice in many systems although I abhor most all silver, analog cables. I know about reflections in optical cables and about glass Vs plastic as well construction details that cause reflections, but I don't know as much about RF reflectivity. What wavelength are we talking about and what length of cable absorbs the lowest harmonic (or whatever we are talking about absorbing here!)?
-Bill

[Tweaker]

The GW Labs DSP has done the trick.The Bolder Cable coax is happy again. Even with a 50ft XLR microphone cable from the DSP to the Ultracurve the sound is sensational. I will order a digital XLR to replace that from Signalcable and call it good.
One thing I'm not too crazy about the DSP,though, is the upsampling feature. The soundstage widens, instrument expand in size but at the expense of some slight loss of focus. However, everything else the DSP does it does well.
Thanks for the help and education, everyone.
  -Jon

[art]

If you understand reflections in fiber, then you realise why using a single-mode fiber in less than 1 km lengths screws things up. Fiber was designed to carry high-speed data at spans where copper craps out.......1 km or so. Using it in shorter spans cause too much light to get reflected back to the laser, and they do not like that.

As for RF reflectivity..........I use a device called a Time Domain Reflectometer. It has a step output with a very fast rise time, perhaps 150 pSec. So, it can be considered a wideband device. Much wider than the range of frequencies one would encounter in SPDIF. However, it can easily be demonstrated how poor the average input or output is. I have found that the relection coefficient needs to be <0.01, or -20 dB return loss in order to make the input "sound right". (The stuff that I produce is typically 8 dB better than this.)

The cable itself does not absorb the reflection (unless it is especially lossy, which can be good..........unless it has rotten dispersion characteristics), but instead delays them until they arrive at a point in the waveform that can not affect the decision point.

FWIIW.......back when I had a "real job", I was the cable and connector whizz kid of the lab.

Not because I knew more than everyone else..........I was just the only guy that was not bored to death by it.

Pat

[ctviggen]

Art,

What's "fringing"?  I used to be an engineer (I have a BSEE and MSEE but am now a patent attorney) and have never heard that term used before.  Or do you mean "ringing"? I searched for "fringing" on Google but really can't see anything that relates to cables and the response thereof.

Also, there is no reason why an AES couldn't be implemented as well or better than SPDIF.  If Proceed spent the time to implement AES better than SPDIF, then perhaps they are correct that their implementation of AES is better than their implementation of SPDIF.  Without actually testing the bits that are decoded from both the AES and the SPDIF, we'll never know.  If I had a good digital scope, I'd do a test to see what the difference is.  

I will note that a previous listening test done between using an optical cable (a glass one) to my Proceed AVP and using the SPDIF over a Magic Digital One IC to an Ack Dack 1.2d, I thought the Ack Dack 1.2d was better than the DAC in the Proceed AVP.  Using the AES connection, though, from my Proceed PMDT to the Proceed AVP and comparing this with an Ack Dack 2.0, which certainly has better bass than the 1.2d, has made this a much closer race.  Wayne from Bolder Cable has made me a digital IC with BNC connectors on both ends, as the Proceed PMDT offers optical, RCA, BNC, and balanced digital outputs.  Since the Ack Dack 2.0 is BNC, I figured that would be the best connection -- BNC to BNC.  I'll retest once Wayne's cables get here.

[art]

Fringing is something us microwave types talked about. Yeah, probably not used in conjunction with cables that often. But basically..........

Even though it is a 110 ohm connector, the field has too many discontinuities with the mish mash of those big pins.........sometimes in dielectric........sometimes in air............just too many transistions. A BNC would not have the similar problem. Does that help to explain it any better?

Well, the AES/EBU does not work as well. Look at the amount of reflections even the best one has. And who said that Proceed got it right????????? I would doubt that they were even close. Companies like them and Wadia make some of the worst interfaces that I have seen.

You will not see any difference in the bits, unless you have a fancy 'scope that can measure jitter.

The original AES/EBU was even worse, and even though they will never admit it, having jerks like me kvetch about how bad it was forced them to change. I even went as far as to complain to an famous industry big wig on his cell phone one day, which seemed to annoy him. But at least he listened. Impedance misamtches was not his forte, but he got a brief lesson over the phone that day.

Pat