Airport Express - is not bit perfect.

[same same but different]

I just discovered that.

 

Do simple thing: connect AE's optical output with the glass cable to a ADC that has optical input. Set ADC to follow its input (as a clock source). Play 16/44 file as usually via iTunes, sending the file to AE, record the file in the computer. And then compare the record with the source file by simple subtract operation. Make sure the files are sample to sample aligned.

 

If you open the difference file, you'll see the image like this:

http://dl.dropbox.com/u/2703292/pics/AE_difference.png

 

Then, do exactly the same procedure with MBP optical out. Set up MBP's output to be 16/44 too.

 

Compare the files, and you will see there almost black spectrum with a very few small mistakes. Overall, the record and the source files are almost identical.

 

Sad.

 

[same same but different]

P.S.

AE N

Volume in iTunes was 100%, equalizer and enhancer were off.

Cable - Lifatec 0,6m glass.

 

 

[souptin]

What tools do you use for the recording and subtractions? I'm interested in trying this out for myself - I have two AEXs, the original b/g and the newer n model, and I have found the b/g less prone to streaming dropouts despite it being snail like for file transfers.

 

[The Computer Audiophile]

I've tested the Airport express several times and found it but perfect every time.

 

[same same but different]

souptin:

 

I used Audition and Logic.

Tried few times with exactly same result.

 

All the time ADC was set exactly to follow optical flow as a clock source.

To make sure it's operate correctly, I tried MBP's optical out - and it was perfect.

 

The Computer Audiophile:

How you've checked it, please?

 

[mayhem13]

Hopefully the same method used by Stereophile nearly 8yrs ago where it tested bit perfect. Could be Apple's changed something in the newer versions.....I certainly hope not.

 

[same same but different]

Try it by yourself.

 

I'll check it with AE G tonight.

 

[same same but different]

Speaking of MBP optical output, more details. I've analized difference file, it is not completely black (silent). RMS is jumping around -90… -120 db take to take. If you increase the amplitude on 60 db twice, you'll see, probably, jitter affected mistakes:

http://dl.dropbox.com/u/2703292/pics/MBP_optical_out_plus120db.png

 

 

[mayhem13]

At those levels, you'll never be able to hear it. Hopefully that will help you sleep better at night......as long as the bit perfect part gets resolved. I've got three of the N version in use in my network and this is really bothering me. Apple has been known to do some pretty silly things for the good of the whole. I don't sweat the small details like low level jitter and noise.....but bits are not a small detail.

 

[same same but different]

mayhem13: Agree, those levels literally not that much important. I may only guess it's jitter trace. Maybe not. Not really so important.

 

I would be happy to see someone's substract result that prove AE is perfect.

 

Again, I'll try again with G version.

 

I can share the files then, if it helps.

 

[mayhem13]

Please do.

 

[phatoldsun]

The HDCD indicator lights up on my Marantz SR-6003 when my AE N streams a ripped HDCD encoded disc in AIFF or ALAC. When I move the volume control in iTunes to 99% or less, the HDCD indicator turns off. I read in previous threads on this forum that if the HDCD flag is retained to an HDCD decoder then this is a reliable means to test for bitperfectness.

I think The Computer Audiophile tested this using the HDCD decoder on the Berkeley DAC.

 

[goldsdad]

The iTunes volume control is a digital volume control; it works by changing the bits of the samples therefore the output cannot be bitperfect at less than 100%, resulting in the HDCD light correctly turning off.

 

Edit: Please ignore my post. I misunderstood your post as looking for an explanation for the light going off. Sorry!

 

 

[same same but different]

Tried AE-G.......

 

Absolutely bit to bit equal to the source file.

 

 

Back to N. Few attempts - nope, same picture, completely different from the source file.

 

Updated firmware (just noticed recently issued) - 7.5.2. Didn't work, all the same.

 

Can anybody check this please? I got no other AE N around to check with.

 

[VandyMan]

Your test method will fail if anything causes the samples to be misaligned. Dropouts or significant jitter (which the AE definitely has) could cause this.

 

[VandyMan]

hit enter too soon by accident

 

[same same but different]

Why then AE G seems work perfectly?

 

And why AE G is terrible sonically ?

 

[mayhem13]

Maybe Vandyman can quantify can quantify at what point jitter is significant and audible?

 

[The Computer Audiophile]

Hi mayhem - I don't think there is a single answer that applies to all listeners. The only data would have to be some sort of listening tests on jitter audibility, but the test results would of course be limited to the exact situation and people who took the test.

 

 

 

[The Computer Audiophile]

"same method used by Stereophile nearly 8yrs ago"

 

Hi mayhem - The AE has changed significantly in the last 8 years. In fact the S/PDIF receiver was changed and jitter increased greatly. I know at least one manufacturer had to increase the jitter window on one of its digital inputs in order to accept audio from the newer AEs.

 

I don't know of any tests done with a newer unit.

 

[same same but different]

"In fact the S/PDIF receiver was changed and jitter increased greatly."

 

You mean transmitter, don't you?

 

Are you saying that due to high jitter values in N version, it cause to such a wrong values transmitted so that files are not identical from the very beginning?

 

I'm getting another AE N unit to check. Will share tonight result.

 

 

[VandyMan]

My point was not that the jitter is or is not audible. You are testing for bit perfect data by aligning samples from two streams. If the samples fall out of alignment at the start or due to timing differences or dropouts (since it is wireless) within the stream, your test will not work.

 

[same same but different]

Version:

 

My sound card may have jitter minimizer inside, that completely clear the flow from low jitter values. For example, let say AE-G has 258 ps jitter (stereophile), and guess it's been taken off by this minimizer.

 

Then, MBP's optical out has jitter about 4 ns. It's per Realtek chip spdif specifications. Such values are harder for my card, there are some very small mistakers, but basically all goes well.

 

AE-N... What if it got so high jitter that can't be avoided by the card, therefore I've got absolutely different file recorded in this case?

 

[same same but different]

VandyMan:

 

I did aligned all flows to the very first sample before subtract procedure. No mistake.

 

[VandyMan]

Did you verify that the number of samples from the alignment point to the end is the same?

 

Even if you do, I'm not sure there is any good way to be certain that there was not a dropout at some point in the stream pushing all subsequent samples off by one frame or more without checking all of the data.