[EliasGwinn]

Scrith,

Regarding the -30 dB attenuation setting, the sound should not change if the final output level is the same. The attenuators are purely passive, just before the output. The only change is the output impedance (lower output impedance is better), but the -30 dB is the lowest output impedance setting (43 ohms) of them all. So, I don't know why you are hearing something going missing when you go from -20 to -30 dB.

Quote:

Originally Posted by Scrith /img/forum/go_quote.gif Well-known high-end audio manufacturer Ayre will soon be releasing a DAC supporting USB input that supports an asynchronous USB mode (via USB firmware licensed from Wavelength Audio, I believe). This asynchronous USB mode allows clock(s) on the DAC to determine the transfer rate of data from the computer, rather than relying on the computer's (inaccurate and jitter-prone) clock(s) to determine the rate at which data is received (and thereby necessitating resampling on the DAC side to overcome this poor timing). This new firmware is particularly interesting because it works with the existing Windows and Mac USB drivers.

The problem with this implementation is that it requires the operating system to re-sample the audio to the clock of the converter. This is a problem because the re-sampling algorithm that Windows and Mac uses is absolute garbage.

Even if the master clock from the outboard DAC tells the computer to play at the same sample rate (44.1k for a CD, for example), the computer still must re-sample the data to be synchronized to the DAC's master clock.

So, it is true that the audio is transferred using the DAC's clock, but it is at a very costly tradeoff. For outboard DAC's that struggle with jitter, it may be a worthwhile tradeoff.

Quote:

Originally Posted by Scrith /img/forum/go_quote.gif I mentioned this a few months ago and haven't heard much about it since then, so I'm wondering if there is any update. Does Benchmark have any plans for a new DAC (or an update to the firmware for existing DAC(s)) that supports this new variation on USB audio data transfer? This evolutionary step seems very important, because it finally eliminates all computer clock related jitter from the source (without providing a band-aid solution like resampling or intermediate buffering).

This USB mode (asynchronous) is not new. When we designed our USB interface, we made the decision to not use this mode. This is because, as I mentioned, it requires the computer to re-sample the data. Therefore, with our method, the computer can simply pass the data to the DAC1 bit-transparently.

Is there a lot of jitter on the data coming from the computer using our solution? There sure is! But, luckily, the DAC1 can handle any amount of jitter with no loss in sonic quality. In fact, the performance of the DAC1 using the USB from a computer is equivalent to that of the DAC1 using a high-end transport with a high-quality 75-ohm coaxial cable.

The DAC1's USB solution is the best of both worlds...it eliminates the horrible OS re-sampling and it is immune to the computer's jitter!

Thanks,
Elias

 

[EliasGwinn]

Quote:

Originally Posted by BitPerfect /img/forum/go_quote.gif Cool, I'll give that a try then. I see Hosa and others have inline balanced attenuators that are switchable to 20, 30, and 40 dB. Means I'd need to buy two of them. That feeds perfectly into my finely tuned neurotic-audio-obsessive personality. I can lose at least two nights sleep wondering if they're perfectly matched. No mention of output impedance on most of them, but the amp has 28 kOhm nominal input impedance, and the cables are short, so I should be alright. I'll probably only lose one night's sleep over that one. Thanks!

If you have that dream about the interconnects tying your feet and hands together, and the optical cable starts shining its light at you, DON'T LOOK INTO THE LIGHT!!

 

[emmodad]

Quote:

Originally Posted by BitPerfect /img/forum/go_quote.gif Cool, I'll give that a try then. I see Hosa and others have inline balanced attenuators that are switchable to 20, 30, and 40 dB. Means I'd need to buy two of them. That feeds perfectly into my finely tuned neurotic-audio-obsessive personality. I can lose at least two nights sleep wondering if they're perfectly matched. No mention of output impedance on most of them, but the amp has 28 kOhm nominal input impedance, and the cables are short, so I should be alright. I'll probably only lose one night's sleep over that one. Thanks!

small caution: most all fixed or adjustible XLR-style "in-line attenuators" you find (such as the Hosa, IIRC) are designed for mic lines ie they are designed to be used in low-impedance signal chains. Only fixed ones I've seen that are designed for audio line-level are from rothwell in the UK (google on "rothwell attenuator"). Elias may know of something else.

Other options would be to follow the DAC-1 by a passive stereo attenuator of reasonable quality and good tracking, so you can set it and forget it; or a pair of high-quality stepped mono adjustibles.

For the passive stereo, it gets expensive in the "audiophile" realm (Placette, Goldpoint level controls, other similar designs from Europe); or a Presonus Central Station is a good bang-for-the-buck solution for simple, clean high-bandwidth passive level control, using the TRS main inputs (the RCA inputs go through some active circuitry to do -10 to +4 level shifting). You can find them in good seasonal sales from folk like Musician's Friend for say USD 300-400 or so. Of course, YMMV concerning your connections single-ended, balanced, etc. Another inexpensive solution from the pro space is an a-designs ATTY passive stereo attenuator, IIRC USD 100 or so.

With any of those, you can either use DAC-1 in fixed output mode and then control volume with the inserted atteniuator; or use the DAC-1 vol control with the attenuator following in set&forget. Elias may have comments about gain staging, depening on the gain of your particular amps....

tweakaudio Ultimate Attenuators: a small audiophile and recording business in Boulder Creek, CA builds amongst other things these excellent (and not cheap, $350?/pair, but very clean) mono attenuators that are essentially high-quality resistors and stepped pots soldered directly together with RCA or XLR male and female connectors. they're tiny, beautifully made, and install directly at the input jacks of your amps. A pair of these would quite possibly appeal to a finely tuned neurotic-audio-obsessive personality...

 

[BitPerfect]

Quote:

Originally Posted by EliasGwinn /img/forum/go_quote.gif If you have that dream about the interconnects tying your feet and hands together, and the optical cable starts shining its light at you, DON'T LOOK INTO THE LIGHT!!

Dude! You have that dream too?

 

[BitPerfect]

Quote:

Originally Posted by emmodad /img/forum/go_quote.gif small caution: most all fixed or adjustible XLR-style "in-line attenuators" you find (such as the Hosa, IIRC) are designed for mic lines ie they are designed to be used in low-impedance signal chains. Only fixed ones I've seen that are designed for audio line-level are from rothwell in the UK (google on "rothwell attenuator"). Elias may know of something else. Other options would be to follow the DAC-1 by a passive stereo attenuator of reasonable quality and good tracking, so you can set it and forget it; or a pair of high-quality stepped mono adjustibles. ...

Thank you for your time and generosity in detailing this. You've undoubtedly saved me from making a mistake.

Your mention of the the tweakaudio nude ultimate attenuator reminded me that I had sent something about that to an Audiogoner friend a couple of years ago. I had completely forgotten about it. An email search shows he subsequently bought an NHT PVC Pro for his small office system, which at around $90 might be another candidate for balanced VC. They also have an unbalanced unit with Jensen transformers.

I will do some research on all your suggestions. Again thanks!

 

[G-U-E-S-T]

Regarding passive attenuators that attach to your amp, please consider Scott Endler's Shotgun Attenuators. They are absolutely the best passive solution I've heard to date. I strongly recommend avoiding the rothwell attenuators - they are the worst I've ever heard and they sound really bad in comparison, although I can't explain why.

 

[BitPerfect]

Quote:

Originally Posted by G-U-E-S-T /img/forum/go_quote.gif Regarding passive attenuators that attach to your amp, please consider Scott Endler's Shotgun Attenuators. They are absolutely the best passive solution I've heard to date. I strongly recommend avoiding the rothwell attenuators - they are the worst I've ever heard and they sound really bad in comparison, although I can't explain why.

Helpful! I had not seen these. I like that there is an XLR version. I'm in favor of first trying the simplest thing that could possibly work. This is pretty darn simple (as are the rothwells for that matter). Appreciate the feedback.

 

[emmodad]

you're welcome. funny, I had thought of the NHT but forgot to mention it, also seems to get good comments.

do google on the Ultimate Attenuators, they may not be the correct solution for you but the pictures are quite something to see....

glad to be of assistance to someone from the home playground of MythBusters

and those shotguns look potentially cost-effective too..

 

[Mazz]

Quote:

Originally Posted by EliasGwinn /img/forum/go_quote.gif The problem with this implementation is that it requires the operating system to re-sample the audio to the clock of the converter. This is a problem because the re-sampling algorithm that Windows and Mac uses is absolute garbage. Even if the master clock from the outboard DAC tells the computer to play at the same sample rate (44.1k for a CD, for example), the computer still must re-sample the data to be synchronized to the DAC's master clock.

Elias, could you clarify this surprising statement? Are you implying that it's limitation of the USB audio protocol spec, or of the default USB audio drivers in Windows and the Mac that forces resampling by the OS, or something else entirely?

I had a (quick) look at the USB 2.0 Audio spec (although didn't look at the 1.x version) and didn't see any obvious indications that you can't have the computer specify a sample rate that should be used by the DAC's clock in async mode. If that's correct, that suggests you should be able to have the computer stream a bit-perfect stream to the DAC provided the DAC can provide a master clock that matches the sampling rate for the audio stream. (Building the hardware implementation is another matter - I remember reading no standard USB chipsets handled it in the past...)

So if it's true that the USB spec can handle it, that suggests that the default USB audio drivers don't do it. And that means you can only do this with custom drivers (or proper resampling code in your audio player), both of which are certainly feasible, although at the cost of complicating the development - and use of - the product.

 

[omegared]

Quote:

Originally Posted by omegared /img/forum/go_quote.gif Unfortunately no. Let me try what Quaddy suggested first to ensure ALL usb root hub do not have power save enabled. Will update you again.

Hi Elias. My sychronisation problem is resolved! I left the system playing over several nite and allowing it to idle during this period. I'm still able to play music thru DAC1 without any problem! Apparently, like Quaddy adviced, you need to unplug the usb device before going to system devices to disable usb power down when idle. If not, the settings will reset itself upon system shutdown / reboot.

 

[Scrith]

Regarding asynchronous USB support, here is a quote from the Wavelength Audio page:

"The Firmware that runs the USB controller (TAS1020B) of the product is located on the DAC module and therefore changes the way the DAC and the computer talk to each other. Settings like bit size (i.e. 16 or 24 bits) and sample rate (i.e. 44.1K, 48K, 88.2K and 96K) will be communicated to the computer. All the DAC's on this page use Asynchronous USB mode."

This seems to indicate that the DAC is telling the computer which format to use for the data it sends the DAC (16/24 bits, 44/48/88/96K), as Elias said such a system might.

I am a bit confused by Elias' statement that the computer will always resample to the requested rate, even if the data is already at that rate. Are you sure about this, Elias? If this is not the case (and it certainly seems unlikely that a computer would resample in this case), then the asynchronous method would appear to have a major advantage (the DAC is controlling the rate at which it receives data).

I appreciate that the DAC1 is "the best of both worlds" but I think the best option for a DAC would be to have no resampling whatsoever, with a DAC that is controlling the rate at which it is receiving data from the source and using some internal buffering to overcome any latency in the arrival of the requested data between the source and the DAC.

 

[EliasGwinn]

Quote:

Originally Posted by Mazz /img/forum/go_quote.gif Elias, could you clarify this surprising statement? Are you implying that it's limitation of the USB audio protocol spec, or of the default USB audio drivers in Windows and the Mac that forces resampling by the OS, or something else entirely? I had a (quick) look at the USB 2.0 Audio spec (although didn't look at the 1.x version) and didn't see any obvious indications that you can't have the computer specify a sample rate that should be used by the DAC's clock in async mode. If that's correct, that suggests you should be able to have the computer stream a bit-perfect stream to the DAC provided the DAC can provide a master clock that matches the sampling rate for the audio stream. (Building the hardware implementation is another matter - I remember reading no standard USB chipsets handled it in the past...) So if it's true that the USB spec can handle it, that suggests that the default USB audio drivers don't do it. And that means you can only do this with custom drivers (or proper resampling code in your audio player), both of which are certainly feasible, although at the cost of complicating the development - and use of - the product.

The problem stems from the fact that no two clocks are ever equal. Even if they are both said to be at 44.1 kHz, one may actually be 44100.01 Hz and the other may be at 44099.99 Hz.

This is why recording studios use a master clock. Even if they set all their A-to-D's to 44.1 kHz, the samples would not be truly synchronized. This results in pops and clicks.

In a USB playback system, unless the driver and the media software are directly linked, the software operates using an internal clock. This data stream is then sent to the audio stack in the operating system.

At this point, the OS determines how to stream the data to the peripheral device (the USB audio device). If the OS is receiving instructions to stream it at a specific sample rate (e.g., the DAC's master clock), then the OS must re-sample it to match that rate. On the other hand, if the OS receives instructions to stream it using the original clock, then no re-samplign occurs.

Thanks,
Elias

 

[EliasGwinn]

Quote:

Originally Posted by omegared /img/forum/go_quote.gif Hi Elias. My sychronisation problem is resolved! I left the system playing over several nite and allowing it to idle during this period. I'm still able to play music thru DAC1 without any problem! Apparently, like Quaddy adviced, you need to unplug the usb device before going to system devices to disable usb power down when idle. If not, the settings will reset itself upon system shutdown / reboot.

Great to hear! I'm sure this will also prove valuable to other people who's computers are cutting off the USB connection.

Thanks,
Elias

 

[EliasGwinn]

Quote:

Originally Posted by Scrith /img/forum/go_quote.gif I am a bit confused by Elias' statement that the computer will always resample to the requested rate, even if the data is already at that rate. Are you sure about this, Elias?

Please read the post I wrote a few minutes ago concerning this. Basically, its because the sample rates will never be the same, and there must be a master clock, to which all others are sampled to.

Quote:

Originally Posted by Scrith /img/forum/go_quote.gif If this is not the case (and it certainly seems unlikely that a computer would resample in this case), then the asynchronous method would appear to have a major advantage (the DAC is controlling the rate at which it receives data).

This is questionable. I can understand why one might think that it would be advantageous to have the DAC control the stream rate. But, keep in mind, the computer is still the device that is pushing the data out. In other words, the DAC isn't "pulling" the data out by itself. It is merely specifying the rate. The computer still has to push the data (at the DAC's rate), and so we don't really bypass the computer's imperfections, let alone the USB cable, and the EMI environment that surrounds a computer.

So, even if the computer did not re-sample, and the device was receiving bit-transparent (or otherwise high-precision) audio, the asynchronous method still may not have less jitter.

Quote:

Originally Posted by Scrith /img/forum/go_quote.gif I appreciate that the DAC1 is "the best of both worlds" but I think the best option for a DAC would be to have no resampling whatsoever, with a DAC that is controlling the rate at which it is receiving data from the source and using some internal buffering to overcome any latency in the arrival of the requested data between the source and the DAC.

Ideally, a digital stream would never need to be resampled before conversion. However, in reality, jitter will always be present when data is being transmitted between devices, and so there is a need to address jitter. Benchmark's jitter immunity is a result of the data being re-sampled asynchronously (not based on the incoming clock). This re-sampling is done on-board using a high-resolution ASRC chip (AD1896) millimeters away from the D/A chip.

Our philosophy is to avoid any processing in the computer, do the hard work in our box where we can control the quality of the processing.

Thanks,
Elias

 

[G-U-E-S-T]

Hi Elias,

I have a question please about the analog inputs on my DAC1 PRE. Do they have any provision at all for interrupting any potential ground loop current or other EMI/RFI flowing through the interconnect cable shield?

I like very much how the digital inputs are isolated, but I worry about the possible lack of decoupling/isolation on the analog rca inputs. As you know, this can be a problem when interfacing some single-ended equipment - and I would guess (as a layman) that any possible inlet for EMI/RFI (even the analog ins) could possibly also affect the internal DA circuitry as well.

Thanks in advance for your reply!