Thread: Mytek Stereo 192

Originally Posted by earflappin

Barrows, how did this improvement in SQ manifest? Does the ASRC and DPLL smooth out in the sound in pursuit of its goal to reduce jitter? Thanks.

Not in my experience. With the ASRC and DPLL active, my DAC sounds a smidgen more "digital", with a slight electronic haze or hash. With synchronous clocking, providing the masterclock directly from my async USB interface, the DAC sounds more natural, not smoothed over (ie no loss of detail), but instruments sound more like musical instruments and less like electronic facsimiles of instruments. Timbres seem more real, especially the difficult (for digital) timbres of violins and horns (think Miles' muted trumpet) which are more like the real thing. I hear somewhat similar differences when comparing the PS Audio PerfectWave DAC with the ASRC on or off.

Originally Posted by barrows

Miska: just so we are clear, I have no problem with those who need DSD 128 for whatever reason, my only point was that it is my view that a majority of audiophiles do not.

I would say majority doesn't care about DSD in first place. Majority is not even interested in computer audio. If someone doesn't need DSD128, it is still IMO bad idea to ruin that feature. Same as the SDIF-3 inputs on the mastering version of Mytek I have.

In any case, I would consider it better to replace the USB 1.1 chip with XMOS chip (even roughly same physical size), rather than removing the USBPAL block.

In my own DAC, I achieved a nice improvement in sound quality, by running the ESS chip synchronously, which deactivates the ASRC and DPLL. Of course, I went to great lengths to try and make sure the incoming data was as jitter free as possible, because running in this mode does pass any jitter present on through to the output.

I think it was covered couple of times already on this thread, that the Sabre ASRC would not be active. And in any case, the ASRC is not active with DSD, which is the subject here.

Could you please provide the background data why do you think USBPAL is not async? At least it is less sensitive to computer timing variations than FW.

For the sake of it, I re-measured jitter again with DSD, using USBPAL and XMOS feeding the AES input (MuFi V-Link192).

Both results look practically the same...

Any possible PLL problems are clearly visible in jtest, since PLL doesn't change data at all, it just affects timing. ASRCs can be more tricky to study.

Originally Posted by barrows

FireWire sounds better on the Mytek because they use the Rigisystems USB interface, which is not asynchronous (as far as I can tell). While Mytek may still call their USB input asynchronous, I think they are referring to the asynchronous re-clocking of the ESS chip, and not the USB receiver itself. I do not see any mention of the Rigigsystems inteface being asynchronous at their site. In any case, async or not, it seems clear from reports that the Rigisystems USB interface is not a top performer. Users of the Audio Research DAC have also reported that it sounds better with a really good FireWire or USB to SPDIF converter. Guess what, Audio Research also uses the Rigisystems USB interface.
Too bad these DACs have not chosen to go with an XMOS based, asynchronous USB interface...

Don't think so. The product release sheet specifically refers to the "ultra low 10ps jitter clock for asynch USB2 and FW operation" and points to a specific spot on on the DAC board for the location of the clock. The same document refers to the ESS DAC and chipset and points to a totally different spot on the board in the same picture.

firedog:

"Don't think so. The product release sheet specifically refers to the "ultra low 10ps jitter clock for asynch USB2 and FW operation" and points to a specific spot on on the DAC board for the location of the clock. The same document refers to the ESS DAC and chipset and points to a totally different spot on the board in the same picture.'

Could you point me to that sheet, I would like to see it, or is it easy to find at the Mytek site. I believe that the clock it refers to is the masterclock oscillator used for the ESS chip. An async USB receiver requires (unless it is generating clocks through DDS, which would not be 10 pS) 3 separate frequency oscillators to work: one for the USB receiver, and two for the two families of audio data, 44.1 base and 48 base.

Miska, I am pretty sure the ESS will be clocked from its local oscillator even with DSD streams in, but it can be configured mant different ways, meaning the ASRC would still be active, but I will double check. Do you have the real data sheet on the ESS, I have requested it a couple of times with no luck yet (since they "released" it publically without the DND)?

As mentioned, I do not "know" if the USBpal is async or not, what I know is that users have reported compromised sound using DACs with the USBpal, and, nowhere on the Rigisystems site does it claim that the USBpal operates asynchronously-these two things make me suspect that it is not. You might have the ability to confirm this though? Since you appear to be quite intimate with computer software, hardware, and programming, you might be able to figure it out for us?

An async USB receiver requires (unless it is generating clocks through DDS, which would not be 10 pS) 3 separate frequency oscillators to work: one for the USB receiver, and two for the two families of audio data, 44.1 base and 48 base.

There are at least 22.579 MHz (for 44.1 base) and 24.576 MHz (for 48 base) clocks on the board, next to the FPGA to which the ESS is connected to, not far from the ESS.

Miska, I am pretty sure the ESS will be clocked from its local oscillator even with DSD streams in, but it can be configured mant different ways, meaning the ASRC would still be active, but I will double check. Do you have the real data sheet on the ESS, I have requested it a couple of times with no luck yet (since they "released" it publically without the DND)?

I have my doubts if ESS has implemented ASRC that operates in DSD domain. Otherwise it would need to convert DSD to PCM for the ASRC.

No, I don't have the ESS datasheet, I didn't even ask.

As mentioned, I do not "know" if the USBpal is async or not, what I know is that users have reported compromised sound using DACs with the USBpal, and, nowhere on the Rigisystems site does it claim that the USBpal operates asynchronously-these two things make me suspect that it is not. You might have the ability to confirm this though? Since you appear to be quite intimate with computer software, hardware, and programming, you might be able to figure it out for us?

I suspect that it is async. For example if you look at the evaluation board photo there, you can see three clocks on the board
USB Audio - Rigisystems
one small close the the USB processor and two bigger ones closer to the FPGA.

"I have my doubts if ESS has implemented ASRC that operates in DSD domain. Otherwise it would need to convert DSD to PCM for the ASRC.

No, I don't have the ESS datasheet, I didn't even ask."

Yeah, without the data sheet it is impossible to know, except, when I get a DSD interface running with my ESS DAC, then I will likely be able to figure it out by trial and error... For instance, if the ESS volume control continues to work with DSD...the volume control runs at 48 bits...-I can also go back to async clocking and see what happens-but having the data sheet would be better.

Yes-I have seen the pictures of the USBpal, and noted the clocks. I cannot DL the Mytek pdf here, but will in the next couple of days. I would like to really know how it is working without the marketing language. Things are a little hard to figure out just reading literature, especially when considering how many configuaragle options there are with the ESS chips (async or not, DPLL bandwidth settings, modulator settings, word length settings, one can even turn off the OSF and run it as, apparently NOS, really do not understand that on a SDM DAC?)

Originally Posted by barrows

For instance, if the ESS volume control continues to work with DSD...the volume control runs at 48 bits...

It works with DSD too (if you enable it), at least in Mytek. But it still doesn't tell anything about ASRC.

Volume could be done similar way as CS4398, with a DSD processor capable of performing volume control without any down-conversion. This is completely different path than the PCM oversampling and volume control. In addition there's an option to have "Direct DSD" that bypasses all processing goes straight to conversion stage.

Or it could be similar to WM8741/2 where you can choose between DSD-to-PCM conversion with volume control (this is followed by the normal PCM OS) or "Direct DSD" that bypasses all processing and goes straight to conversion stage. I think the PCM-to-DSD works at 352.8 kHz there.

Or it could be something completely different...

I've understood that it would be similar to the CS4398's "DSD Processor" that performs pre-filtering and volume control at DSD64/DSD128. However, I have very strong doubts about it performing ASRC at max 6.1 MHz rates.

one can even turn off the OSF and run it as, apparently NOS, really do not understand that on a SDM DAC?)

For example WM8741/2 support that too, with input at 352.8/384 kHz. Another example is AK4399 with support up to 768 kHz at 32-bit.

Third example is PCM1792/4A with support up to 768 kHz. At 768 kHz options for delta-sigma rate multiplier are 4x and 8x (-> max 6.1 MHz modulator rate).

These are still not "NOS" in similar sense, because there's still the delta-sigma modulator in play. Only way to have these modern DACs in true NOS mode is to input already delta-sigma modulated data and have the DAC perform in "Direct DSD" or similar mode.

Originally Posted by pacwin

You are overstating the case for USB versus Firewire. Its a preference not as you claim "compromised sound". You obviously don't have one handy.

My basis for this conclusion is both the reports from owners of the Mytek, and owners of the Audio Research DAC 8. Both products use the Rigisystems USBpal interface, and a large majority of owners of both DACs have stated "preferences" for the sound produced by inputs other than USB. To me, this is a very good indication that the USBpal is problematic sonically. As an example of another USB interface; one does not hear of (any?) users of DACs with XMOS based interfaces (Meitner, EMM, Weiss, etc) who prefer inputs other than USB.

When you talk about the USBPAL and the version of that implemented on the Mytek mainboard, you are talking about the Cypress EZ-USB® FX2LP USB microcontroller
CY7C68013A - 100AXC - Cypress Semiconductor

So? None of this negates the sonic results expressed by the majority of users of DACs with the USBpal.

Given the amount of processing power in the Mytek DAC (8051 controller in the USB, ARM RISC in the Firewire, 2 x FPGA (Xilinx and Cyclone III(probably involved with the clocking)) buffering and so on its more or less far fetched to think that its not an asynchronous design.

I never suggested the Mytek DAC does not operate, in its entirety, asynchronously, even the ESS chip alone is enough to call a DAC asynchronous. My suspicion was that the USBpal interface on its own, MAY not be asynchronous. Even if the USBpal is asynchronous, it is clear to me, from the reports of those using DACs with it, that it does not offer SOTA USB audio performance.

Question for Mytek - I have read in this thread that the latest versions have -6dB attenuation jumpers on the board. Are these currently in the distribution network? Can dealers tell which versions they have by serial number? Thanks.

Originally Posted by mytek

I'm trying to make sense of this...I'd rather offer facts than hypothesis....The USB2.0 solution used in our DAC is asynchronous, and uses the clock based on ultra stable clock generator built into the DAC. So in the sense of "audio performance" it's as good as it gets - clock jitter is a non issue. Same with Firewire. Unlike with XMOS, no Class 2 drivers can be used, it's a different approach in this sense- but there are also advantages to this for example faster throughput and lower latency. Can you elaborate what is "SOTA audio" performance" ?

Best Regards , Michal at Mytek New York

Michal: "SOTA": state of the art.

You may nor be aware that another popular DAC amongst forum readers here is the Audio Research DAC 8. This DAC also uses the USBpal, and, as with many Mytek owners, people using this DAC have found the USB input lacking in sonic performance, especially when coupled with OSX, and of course then there is linux... Owners of the Audio Research DAC 8 have even resorted to adding an external USB-SPDIF (despite the inherent limitations of SPDIF) in order to achieve better sonic performance with a computer source, just as owners of the Mytek have often reported the FireWire option outperforming the USB. On the other hand, we do not hear reports like this from owners of DACs which use XMOS based interfaces. So subjectively, there are users who appear to find that the USBpal is lacking in comparison to other options. I do not know if Mytek evaluated any other USB options before choosing Rigisystems. In any case, after hearing the reports of those who felt the USBpal was lacking, I looked at the Rigisystems site, and noted that I could not find any mention of asynchronous operation, hence I speculated, that maybe, the USBpal was not operating asynchronously, and I stated my speculation as such.
While I have heard other USB interface developers suggest that writing a dedicated driver has advantages over using the USB Class 2 Audio spec, in practice, it does not seem that these interfaces are outperforming the Class 2 Audio compliant options, at least for audio playback purposes, where latency is not really any concern. I understand that in the recording studio, or for video use, latency is important, but for audio playback it is a non issue.

I have no explanation for why some find the USBpal to be a poor performer, it does seem that this interface offers better performance in a Windows environment, perhaps the OSX drivers are just not quite adequate for the task?

And: thanks for your response to this thread, I am sure everyone at Computer Audiophile appreciates your presence here.

Originally Posted by tailspn

Excuse me barrows, but what dog do you have in this hunt? To the best of my knowledge, you appeared for the first time July 4th, and have done nothing but speculate negatively on how the Mytek USB port actually works, and therefore, how the unit sounds. You further compound that by comparing it to another manufactures completely differently designed DAC using a like port, and what a bad reputation that DAC has for sound. You don't own the Mytek product, and forgive me if I'm wrong, but it appears you have no intention of purchasing one. You have a Twisted Pear Buffalo II, and are anxiously waiting for whatever incarnation of USB input module Russ is going to produce. Why are you dumping on this product?

Just curious….

Sorry tailspin, but you might need to re-read my posts. I have made no assertions about how the Mytek DAC might, or might not sound from any input, I would not do so, having never heard it myself. Additionally, I never stated that the Audio Research DAC 8 has "a bad reputation for sound", only that owners of that DAC have had a challenging time getting decent sound from the USB input, especially with OSX. If you have read this thread, you will see many users who have a preference for the sound of the FireWire input on the Mytek. To me, the reports of the USBpal, as used in the Mytek and Audio Research DAC 8 indicate that this interface seems to have some shortcomings. Indeed, some reports have come in that the (XMOS based) USB interface in the Weiss 202U actually outperforms Weiss' FireWire interface, conclude from this what you may, or not...
The only "dog" I have in this race is an interest in different USB interfaces, and what qualities and properties might make one interface superior to another. I have pointed out an apparent contrast between the subjective reports of the sound quality of the USBpal versus driverless solutions using XMOS based USB interfaces. I have not seen any reports of DACs which use (XMOS based) Class 2 Audio compliant interfaces having reduced sound quality via USB vs. other inputs-my interest lies in which USB interface(s) types, and designs, might offer the best performance.
No I do not own a Mytek, and yes, I have no intention of purchasing one, but that does not mean I am not interested in its design and approach to audio reproduction. I am sorry if you interpret my posts as "dumping", for the record it is clear to me that Mytek has offered an excellent DAC at a very competitive price point. My only questions are about the USBpal interface, which is not, actually, designed or produced by Mytek.

For reference: I currently use an SOtM asynchronous USB receiver with a linux based (voyage/mpd) server. I find the linux approach to offer distinct sound quality advantages over OSX. But I am always looking for the "best" possible interface.

Originally Posted by mytek

I believe it's true most dealers would have this new hardware (rev 5). We certainly ship this version when ordered in our online stores. Best Regards, Michal at Mytek New York

Michal-

Okay, but when are you going to post instructions on how to find the jumper and change it's position? It has been out quite a while now from what I understand, but no change in the manual to reflect this. I wrote the Mytek help desk asking this same question a while back and never got an answer.

Thanks

Barrows,

By the way, XMOS is not "driver-less" for Windows. And luckily it has also it's own native ASIO driver. They could implement a true native ASIO DSD support into their firmware and ASIO driver if they wanted to...

Originally Posted by Miska

Barrows,

By the way, XMOS is not "driver-less" for Windows. And luckily it has also it's own native ASIO driver. They could implement a true native ASIO DSD support into their firmware and ASIO driver if they wanted to...

Indeed, XMOS based interfaces are not driver less for Windows, but this is not XMOS' problem, a driver is required for Windows because Wnidows itself is not compliant with the USB 2 spec which allows for high resolution playback-this is a fault in Windows (and I have heard rumours it may be fixed in Windows 8, anyone?) not with XMOS' software. I use and prefer linux, but, the XMOS interface approach is highly programmable for those who understand such things, you might even be able to program it for native DSD with ASIO if you want that function.