mytek chosen by David Chesky of audiophile label Chesky Records

[lucey]

Originally Posted by David Spearritt

These converter qualities are independent of my music or my chain or my anything. This is cold hard accuracy at sampling an analog voltage waveform and producing it again from the numbers. Nothing to do with music, I'm afraid. Remember, I am only talking A/D and D/A.

In simple terms there is the analog I/O path, the converter chip itself and the clock... these are the parts of "the converter" that make the sound. Clearly tiny amounts of measured difference between products equate to clear sonic differences.

Do you disagree with these ideas as a premise?


My point is that if you take products from Weiss, Lavry, Prism, Pacific Micro, DAD, Mytek, etc, etc ... the specs are all very good ... these are all "quality" converters. But they do sound different. If your chain is refined, that difference might matter to you. And listening is the only way to get there.

Do you disagree with this?

There is no perfect anything ... it's all about the sound.

[Roland]

I can't see what specification wise you are looking for, particularly with AD or DA's the specs on the top brands are all pretty similar, give or take usual manufacturers "license". You talked about jitter in earlier posts, but as I mentioned above this isn't a problem for any gear that is reasonably designed. RME's converters spec fantastically well, however, I would not expect them to be considered top of the tree, they are built to a reasonably low price point, I suspect that even Mr Carstens (one of the most brilliant electronics minds in the business) would probably concede that. I feel far more important is the filter design and implementation as this is IMHO where the differences in the sound of converters comes from. Many earlier digital recorders were accused of lacking in bass weight and slam, almost certainly caused by the analogue front end. What makes this all the more likely is that many converters share the same digital circuitry yet sound different. The clocking side is fairly academic and has been well understood for some time now, the only real differences lie in the analogue implementation, all the more reason to listen.

That being said, the differences between different converters is very small these days, a well respected designer I was talking to today told me that the top quality components for a good AD were only a matter of a few pounds.

Regards


Roland

[jslevin]

Originally Posted by lucey

My point is that if you take products from Weiss, Lavry, Prism, Pacific Micro, DAD, Mytek, etc, etc ... the specs are all very good ... these are all "quality" converters. But they do sound different. If your chain is refined, that difference might matter to you. And listening is the only way to get there.

Strongly agree with this.

Great converters do sound different, and the differences do not correlate to the specs.

Seems like part of what David is missing here is that ADC and DAC are not purely digital processes — they don't have perfectly predictable results where you know the exact sequence of 1's and 0's that are the "correct" answer — you can't get two phenomenal converters to null out totally — and they do have analog components.

They have specs, but so do purely analog devices — no more so and no less so — and all devices exist within their specs, plus or minus a margin of acceptable accuracy, not on the exact point of each spec. Why would anyone think that analog/digital devices can be evaluated purely on spec? It doesn't really make sense, on top of which it doesn't fit the practical experience of any great engineer that I know of.

JSL

[David Spearritt]

I am aware that A/D and D/A have analog stages. But as Mike Story from dCS says the digital conversion uses two bits of information, the signal was this big at this time.

Now if the clock is accurate that sorts the time out and making sure the exact analog waveform fed to the device appears at the sampling point then that's the second condition met.

If there is another analog stage before the sampling circuit, "sweetening" the signal to perhaps "disguise" the effects of lousy clocking then so be it. An analog modification before the A/D or after the D/A is colouration and could be argued to be undesirable. Conversion is a cold accurate science, its not an infinitely variable fuzzy thing like analog audio.

If you can hear differences between high end converters that have similar clock specs, then surely those differences must be due to analog colouration. Also, I wonder how many listening tests to converters have isolated the A/D from the D/A. That is, were volume accurate samples taken from all and then played back through one consistent D/A?

I am trying to assemble some common specs for some popular converters It seems not many manufacturers are publishing specs, I wonder why. Recalling the old salesman rule ... "where there's mystery there's margin".

Here's what I have so far.

Converter TC SK48
Clock Stability A/D +/-30 ppm
Clock Jitter
A/D D/A Conversion Jitter < 42 ps
Jitter Rejection <-3 dB @ 10 Hz, < -100 dB @ 600 Hz
DIO Interface Jitter < 1000 ps
Source StudioKonnekt 48

Converter TC System 6000
Clock Stability A/D +/-30 ppm
Clock Jitter
A/D D/A Conversion Jitter
Jitter Rejection < -3 dB @ 50 Hz < -65 dB @ 500 Hz < -100 dB @ 1.4 kHz
DIO Interface Jitter < 1000 ps
Source Tech Specs

Converter RME Fireface 800
Clock Stability A/D
Clock Jitter
A/D D/A Conversion Jitter
Jitter Rejection
DIO Interface Jitter < 700 - 2000 ps for very low to very high jitter signals
Source RME: Support TechInfo

Converter Prism Orpheus
Clock Stability A/D +/-10 ppm
Clock Jitter
A/D D/A Conversion Jitter
Jitter Rejection >60dB/dec above 100Hz
DIO Interface Jitter
Source Orpheus technical Specifications

Converter Focusrite Saffire
Clock Stability A/D
Clock Jitter
A/D D/A Conversion Jitter < 250 ps
Jitter Rejection
DIO Interface Jitter
Source Focusrite Audio Engineering | US Site | Products | Saffire | Saffire

Converter MOTU 828 MK3
Clock Stability A/D
Clock Jitter
A/D D/A Conversion Jitter
Jitter Rejection
DIO Interface Jitter
Source MOTU.com - 828mk3 Clock and sync

Converter Metric Halo ULN2
Clock Stability A/D
Clock Jitter
A/D D/A Conversion Jitter
Jitter Rejection
DIO Interface Jitter
Source ULN-2 Expanded

Converter SSL Duende
Clock Stability A/D
Clock Jitter
A/D D/A Conversion Jitter
Jitter Rejection
DIO Interface Jitter
Source Solid State Logic | Music

Converter Mytek
Clock Stability A/D
Clock Jitter < 10 ps
A/D D/A Conversion Jitter
Jitter Rejection
DIO Interface Jitter
Source MYTEK DIGITAL USA

Converter dCS 995
Clock Stability A/D +/-1 ppm
Clock Jitter
A/D D/A Conversion Jitter
Jitter Rejection
DIO Interface Jitter
Source dCS Professional Products

Converter Lynx Aurora
Clock Stability A/D
Clock Jitter
A/D D/A Conversion Jitter
Jitter Rejection 3000:1
DIO Interface Jitter
Source Lynx Forum

[Roland]

Hi David, for an totally non-scientific experiment today I listened to a file from my pyramix system clocked to the Madi Wordclock, and the same file, but this time clocked only to the Madi digital signal.

Madi (I believe) exhibits slightly more jitter than say Adat and some other forms of digital transmission, however as the data is transferred it is buffered into the DA circuit.

If the circuit is a fixed rate converter, only if the data buffer emptied would this be a problem, if it's the sample rate converting type as I understand it, it is never a problem as the conversion is asynchronous. An electronics engineer I know is at present working on a Madi circuit which is in basic terms a multiplexed AES digital signal. In order for the circuit to work the data must be sent in correct order so in my mind that rules out the "overtaking" theory that I've seen expounded in some forums.

The result of my rough test? Not a jot of difference that I could discern. IMHO the differences people hear in almost all modern converters are down to the implementation of the analogue circuitry and filtering (the sweetening circuit as you put it is a necessary evil) which can significantly alter the sound.

Regards


Roland