Review and Measurements of Empirical Audio Synchro-Mesh

[Empirical Audio]

With the PLL off; with it active, the highest “deviation” point is not above -110dB.

I assume your SM and the Gustard with the PLL active would sound identical, the difference in noise floor and “deviations” should high be inaudible.

This would be an interesting thing to test in a DBT, maybe this summer.

 

[Empirical Audio]

One question in my mind still is what difference the Synchro-Mesh would make if it were feeding the Gustard DAC with the PLL turned-on. Two levels of jitter reduction. If it's still on the bench, this would be interesting.

I noticed a difference feeding one iPurifier into a second one, so I use two in series now.

 

[gvl]

This would be an interesting thing to test in a DBT, maybe this summer.

Unsure if this was suggested in this or the other discussion but perhaps you can try the Blumlein's challenge for a warmup:

https://www.audiosciencereview.com/...-choose-the-8th-generation-digital-copy.6827/

 

[bequietjk]

I must hear the Synchro-Mesh at SOME point in my life.

 

[sergeauckland]

I must hear the Synchro-Mesh at SOME point in my life.

I thought the result of the measurements was that you won't hear it.

S

 

[Empirical Audio]

Unsure if this was suggested in this or the other discussion but perhaps you can try the Blumlein's challenge for a warmup:

https://www.audiosciencereview.com/...-choose-the-8th-generation-digital-copy.6827/

Got them downloaded. You want me to do this myself or invite some over for a DBT? It's too bad he has the spoiler available. Should have done that with PM or some other way. I'll not look at it though.

 

[bequietjk]

I thought the result of the measurements was that you won't hear it.

S

Jokes!

To be honest, because the item is called 'Synchro-Mesh' it is enough for me to want to listen. That is one of the coolest names I've heard. I know this is a measurement-based forum and that's what a great amount of us are here for. But I would be lying if I said that I only care about how this unit measures. In fact I would love to hear all of Amir's recommended and non-recommended measured devices. This is just the way of loving audio. This device just seems to be the hot topic.

 

[Killingbeans]

I guess that means I hate audio

Maybe I have some sort of mental disability, but I don't see how being "the hot topic" makes something worth spending time on?

 

[bequietjk]

It's just genuine curiosity. The discovery of great inventions.

Surely you don't hate audio.

 

[gvl]

Got them downloaded. You want me to do this myself or invite some over for a DBT? It's too bad he has the spoiler available. Should have done that with PM or some other way. I'll not look at it though.

Just do it for yourself. I thought I'd crack it but it was way more difficult than I expected. I also discovered recently I can no longer hear tones above 15kHz. The spoiler wasn't there originally.

 

[Empirical Audio]

Just do it for yourself. I thought I'd crack it but it was way more difficult than I expected. I also discovered recently I can no longer hear tones above 15kHz. The spoiler wasn't there originally.

Should I be picking which of A and B I think is more live sounding or which of A and B sounds most like the Reference?

 

[gvl]

Should I be picking which of A and B I think is more live sounding or which of A and B sounds most like the Reference?

The latter. Basically you need to identify which A or B is the 8th generation copy, or inversely, which A or B is the same as the reference. There are real measured differences between A and B, and while not apples to apples they are arguably more significant than the wider skirt at the base of the fundamental with the Gustard's PLL. I could pick one correctly, others just sounded the same to my ears and I gave up.

 

[Blumlein 88]

Just do it for yourself. I thought I'd crack it but it was way more difficult than I expected. I also discovered recently I can no longer hear tones above 15kHz. The spoiler wasn't there originally.

Yes, I added the spoiler after two weeks I think. But don't click on it and you won't see the answers.

Also you can use an ABX program like Foobar, and see if you can pick the one which is different. Won't matter if you've seen the spoiler.

 

[restorer-john]

I have to agree with Amir on this one. Whilst a range of manufacturers implemented separate clock mechanisms to feed back to the transport (my old Tag Mclaren kit did) and improved PLL mechanisms and ASRC have essentially solved the problem, it was still a fundamentally bad decision to use an embedded transport sourced clock for the DAC chip. Sample rates arent really relevant to this point.

There is absolutely nothing 'transport sourced' about the clock for D/A converters in CD players. They have their own dedicated clock. There's nothing fundamentally bad either- the designers knew other one way sources such as digital broadcasting were around the corner in the late 70s.

The digital processing (error correction, interpolation, RAM and RAM control, EFM demodulation etc) usually has its own separate clock or it may use the D/A clock in certain implementations.

The waveform after the PLL(s) is delivered to the EFM demodulator and clocked into a (typically) 16k bit RAM where it is corrected/interpolated and then clocked out as brand new data. Jitter is a complete non-issue in CD players.

The EFM demodulator controls the spindle motor servos to ensure the RAM doesn't under or over flow.

Sample rates are quite relevant to my post as PLLs are designed with lock ranges which are typically centered around a single frequency. That was relatively simple in the CD era with a single 44.1KHz. Then along came 32KHz and 48Khz which necessitated twin PLLs in many implementations. With sample rates up to obscene numbers now, ASRC is the only option.

 

[March Audio]

There is absolutely nothing 'transport sourced' about the clock for D/A converters in CD players. They have their own dedicated clock. There's nothing fundamentally bad either- the designers knew other one way sources such as digital broadcasting were around the corner in the late 70s.

The digital processing (error correction, interpolation, RAM and RAM control, EFM demodulation etc) usually has its own separate clock or it may use the D/A clock in certain implementations.

The waveform after the PLL(s) is delivered to the EFM demodulator and clocked into a (typically) 16k bit RAM where it is corrected/interpolated and then clocked out as brand new data. Jitter is a complete non-issue in CD players.

The EFM demodulator controls the spindle motor servos to ensure the RAM doesn't under or over flow.

Sample rates are quite relevant to my post as PLLs are designed with lock ranges which are typically centered around a single frequency. That was relatively simple in the CD era with a single 44.1KHz. Then along came 32KHz and 48Khz which necessitated twin PLLs in many implementations. With sample rates up to obscene numbers now, ASRC is the only option.

Sorry thats not correct. We are not talking about an integrated CD player, we are talking about two boxes connected via SPDIF. In a classic two box player/dac config the DAC word clock is generated in the transport and embedded in the SPDIF stream. It is then extracted at the DAC end typically via PLL which has to follow the vagaries of the source, including data induced jitter, and is not perfect in itself. The clock should be at the dac chip. Its one of the benefits of USB connection.

Sample rates have nothing to do with this fundamental flaw.

http://audioworkshop.org/downloads/AES_EBU_SPDIF_DIGITAL_INTERFACEaes93.pdf

Yes this is essentially a problem solved now, but that still doesnt make it a good design decision.

 

[Frank Dernie]

It should be mentioned and corrected in this context, that Sony, from their very first separate transport/ D/A converter, the CDP-1/DAS-R1, recognized the issue with the possible future use of external D/A converters and designed a synchronized (clock sent to DAC and slaved) connection to the D/A converter. It's certainly not a 'poor design decision'.

That was not their very first separate converter, though the first was paired with a complete CD player with a digital output as the transport.
I bought one and used it for years. The transport doesn't work any more but the DAC does.

 

[mansr]

In a classic two box player/dac config the DAC word clock is generated in the transport and embedded in the SPDIF stream. It is then extracted at the DAC end typically via PLL which has to follow the vagaries of the source, including data induced jitter, and is not perfect in itself.

Most S/PDIF receivers these days (not sure since when) use only the preamble of each subframe for PLL tracking in order to avoid data dependent jitter.

The clock should be at the dac chip. Its one of the benefits of USB connection.

Yes, it's better. It also requires some kind of feedback mechanism, which makes both ends more complicated. Adding an S/PDIF output to an otherwise stand-alone CD player is trivial. Supporting an external clock or flow control is much more involved. Since most buyers were probably not going to use the digital output, it made sense to keep the cost down while still providing the feature.

If we only ever did things the best possible way, we'd get very little done at very great cost.

 

[March Audio]

Most S/PDIF receivers these days (not sure since when) use only the preamble of each subframe for PLL tracking in order to avoid data dependent jitter.


Yes, it's better. It also requires some kind of feedback mechanism, which makes both ends more complicated. Adding an S/PDIF output to an otherwise stand-alone CD player is trivial. Supporting an external clock or flow control is much more involved. Since most buyers were probably not going to use the digital output, it made sense to keep the cost down while still providing the feature.

If we only ever did things the best possible way, we'd get very little done at very great cost.

Valid points and yes it's become a non issue as techniques have developed.

I don't think however it was beyond the wit of man to have used a more technically effective and yet still cost effective solution.

 

[Empirical Audio]

The latter. Basically you need to identify which A or B is the 8th generation copy, or inversely, which A or B is the same as the reference. There are real measured differences between A and B, and while not apples to apples they are arguably more significant than the wider skirt at the base of the fundamental with the Gustard's PLL. I could pick one correctly, others just sounded the same to my ears and I gave up.

Okay, My wife and I both did this, at different times. She is always busy...

Result: I picked 4 out of 4 correctly.

My wife picked 3 out of 4 correctly.

The thing that we both noticed is that in at least 2 of the tracks, the 8X version sounded more focused. This made it difficult, because the altered track actually sounded better than the original, but didn't match the reference. Usually, we listen for the better sounding track, not trying to match two tracks, so we had to resist this tendency.

I have a theory about this improvement in focus in the 8X tracks:

As I believe I mentioned in the other thread, some of the early re-clockers used several PLL receivers in series to reduce jitter because one was not good enough. They went from receiver to transmitter to receiver etc.. S/PDIF to I2S to S/PDIF to I2S and on and on. In this case, we are going from S/PDIF or I2S to analog to S/PDIF or I2S to analog and on and on. I think the effect of the series PLL's might be the same, reducing the jitter in the end result and thereby improving focus. Because the altered tracks actually sounded better than the originals indicates perhaps that the compression and harmonic distortion added by the multiple A/D and D/A is less obvious than eliminating some jitter. This reinforces my desire to see the relevance of these different distortions studied with DBT tests. It's not unlike amplitude sensitivity. If you cannot hear it, it's not important.

Steve N.

 

[Empirical Audio]

Most S/PDIF receivers these days (not sure since when) use only the preamble of each subframe for PLL tracking in order to avoid data dependent jitter

If this is true, might it explain why different offsets change the sound?

 

[SIY]

If this is true, might it explain why different offsets change the sound?

Assumes facts not in evidence.