Are all spd/f inputs created equal?

[omtayslick]

Do some work better than others? Do some sound better than others? Can there be any variation in performance from one to another?

 

[Massive Master]

A S/PDIF should *pass* a signal - Not create it or change it.

It's the converter on either side of it that's going to affect the sound. Passing data from S/PDIF out to S/PDIF in should be bit-for-bit.

 

[IronFlippy]

S/PDIF is a digital signal transmitter, which means that there is not going to be any difference in quality between devices that utilize it.

 

[omtayslick]

Well, that's what I thought. Then I read these lines from this Martin Walker SOS review:
"However, unlike the Delta 1010, I couldn't get the Delta 66 S/PDIF I/O to produce bit-for-bit copies when I sent a recording to and from a DAT recorder, but the difference was only one bit's worth of dither noise, which shouldn't bother most people."

http://www.soundonsound.com/sos/jan01/articles/maudio.asp

Any insight on this?

 

[SonicAlbert]

There could be a number of problems that are not allowing him to get bit to bit copies. The reviewer could have set seomthing wrong somewhere, although that seems unlikely. The drivers for that particular interface could have a problem. Or the unit could just be malfunctioning.

The reason he mentioned the issue is because it is completely unacceptable. Digital I/O should pass data without changing it.

 

[Robert D]

Al briefly hit the point I was going to make. In all but the grossest of designs, there really isn't much the hardware designer can do to screw up a S/PDIF port. There is more chance to screw up the driver. I've used S/PDIF quite a lot over the years, and I've never encountered a S/PDIF implimentation that I thought sounded wrong, or that I had any driver type problems with. I can't say I've ever really checked bit for bit accuracy though.

 

[SonicAlbert]

I can't say I've ever really checked bit for bit accuracy though.

That's why I have to say that I really appreciated that the reviewer in that article did so. That was very thorough of him, and certainly very important information to include in the review.

How would one check for bit to bit accuracy, anyway?

 

[dgatwood]

Any error in S/PDIF transport qualifies as a bug.

If this were in Mac OS X, I would guess that it was caused by a bug in the documentation. The sample code in Apple's audio device driver doc was wrong up until a year ago or so. There was a conversion error that resulted in an off-by-as-much-as-one error when converting a positive float value to integer or vice-versa. In one direction, the example used 32767, while in the other direction, it used 32768.

Now the tester in S.O.S. was apparently testing on Windows, so that's probably not the problem, but it is a good example of how easy it is to get a sporadic off-by-one error in the least significant bit of an audio stream.

 

[Robert D]

That's why I have to say that I really appreciated that the reviewer in that article did so. That was very thorough of him, and certainly very important information to include in the review.

How would one check for bit to bit accuracy, anyway?

Yep, good someone's paying attention. You can use a scope, a logic analyzer, or a more dedicated piece of gear. There are software methods too.