Absolute Sound Review Of Alpha USB on-line

1audio

Well-Known Member
Sep 17, 2010
31
2
353
SF Bay Area
Thunderbolt

Thunderbolt will never replace USB for slow peripherals. Its an enhancement over Displayport with high throughput speeds and the ability to connect PCI express like interfaces to the host computer, but only a few. The power and overhead makes no sense for a mouse. The chip will be more expensive than a commodity mouse. But it may be useful for audio, Apogee is supposed to be working on it.

The other PC manufacturers are not jumping on and have been waiting for a while to see if it has legs (It was around for a while as Light Peak). This is a classic technology to solve a problem that no one really has.

The best strategy in the computer biz is to work back from the desired output finding the most appropriate solution and get the hardware and software that makes it work. And expect to replace it in a few years. This has not changed since the days of the IBM 1120. Audio consumers are more oriented to really long tech cycles and don't want to buy stuff that is obsolete in a short while. Even though most high end audio components fall from grace in a matter of weeks from introduction.
 

Phelonious Ponk

New Member
Jun 30, 2010
8,677
23
0
Those features help when you don't have an asynchronous connection and the PC is generating timing. To the extent the Alpha is doing that now, it should not matter.

I use a usb converter that re-clocks and galvantically isolates. Maybe that's the only reason why I can't hear a difference between Amarra, Pure Music and iTunes.

Tim
 

amirm

Banned
Apr 2, 2010
15,813
37
0
Seattle, WA
Almost every DAC today re-clocks. That by itself though doesn't fix the timing jitter because some of that still gets through. An asynchronous device doesn't "re-" clock. It is a master and has its clock that is independent of input completely. Hence the name. With Re-clocking, there may still be differences due to source acting differently.
 

1audio

Well-Known Member
Sep 17, 2010
31
2
353
SF Bay Area
Jitter

Here is a bewildering number:

To achieve 24 bit accuracy on a 100 KHz sine wave you need less than 100 FemtoSeconds Jitter. Any higher and you won't get the signal to noise that 24 bits offers.

However there are no audio frequency 24 bit accurate ADC's also at lower frequencies you can tolerate higher jitter, for example 24 bits at 10 KHz ONLY requires 1 PicoSecond jitter performance.

Here is a useful calculator: http://www.maxim-ic.com/tools/calculators/index.cfm/calc_id/jitter

These levels are so low that anything can degrade them from noise leakage around the system to subtle vibration at the oscillator. Asynchronous interfaces help however execution is essential. The Alpha interface is large for what it is to reduce coupling from input to output. That is also the rationale for not building it into the DAC.
 

WDW

Well-Known Member
May 24, 2011
94
49
925
BC, Canada
1audio

Thanks for your comments...USB will live on precisely because it is so inappropriate for audio devices that these recent fixes, ALPHA USB, et.al, will have some life into the future. Funny, but heartening.
At the risk of further raising the ire of our moderator, can you explain the Auriliti server...is it intended to completely bypass a computer.....WDW







Here is a bewildering number:

To achieve 24 bit accuracy on a 100 KHz sine wave you need less than 100 FemtoSeconds Jitter. Any higher and you won't get the signal to noise that 24 bits offers.

However there are no audio frequency 24 bit accurate ADC's also at lower frequencies you can tolerate higher jitter, for example 24 bits at 10 KHz ONLY requires 1 PicoSecond jitter performance.

Here is a useful calculator: http://www.maxim-ic.com/tools/calculators/index.cfm/calc_id/jitter

These levels are so low that anything can degrade them from noise leakage around the system to subtle vibration at the oscillator. Asynchronous interfaces help however execution is essential. The Alpha interface is large for what it is to reduce coupling from input to output. That is also the rationale for not building it into the DAC.
 

DonH50

Member Sponsor & WBF Technical Expert
Jun 22, 2010
3,947
306
1,670
Monument, CO
Here is a bewildering number:

To achieve 24 bit accuracy on a 100 KHz sine wave you need less than 100 FemtoSeconds Jitter. Any higher and you won't get the signal to noise that 24 bits offers.

However there are no audio frequency 24 bit accurate ADC's also at lower frequencies you can tolerate higher jitter, for example 24 bits at 10 KHz ONLY requires 1 PicoSecond jitter performance.

Here is a useful calculator: http://www.maxim-ic.com/tools/calculators/index.cfm/calc_id/jitter

These levels are so low that anything can degrade them from noise leakage around the system to subtle vibration at the oscillator. Asynchronous interfaces help however execution is essential. The Alpha interface is large for what it is to reduce coupling from input to output. That is also the rationale for not building it into the DAC.

Graph here: http://www.whatsbestforum.com/showthread.php?1322-Jitter-101
 

Nicholas Bedworth

WBF Founding Member
May 7, 2010
312
0
0
Maui, where else?
What Demian presents certainly corresponds to my listening experience. 100 ps is just OK, 2 ps is really good, and fractions of a ps are divine. Some house clocks such as the Grimm CC-1 ($2495) have 200-300 femtoseconds RMS measured across broad ranges of the phase noise plot. So does the Audiophilleo, possibly others. Getting 100 femtosecond RMS jitter is probably going to be expensive, but 200 or 300 fs isn't really that hard, provided one is willing to use an external clock, or has a USB-S/PDIF device with excellent internal clocks.

Any form of digital interface, FireWire, Thunderbolt, S/PDIF, AES has jitter of one form or another. Per Amir's tag line, eliminating jitter is the occupation of a saint, or, perhaps, a person who doesn't particularly value his sanity.

Up in the $20-30K range, DACs such as the MSB don't reclock, and also they essentially double-up their multiple DAC modules to get 25 bits, yielding 24 bit reliably. For this level of performance, one must pay. :)

Current USB, managed using asynchronous protocols, is perfectly adequate for the incredibly slow data rates that obtain in the audio world, relative to HD video and networks.

Historically-minded enthusiasts may recall that a major factor in FireWire's short, medium and long-term failure in the computer market place was that Apple demanded extremely high licensing fees way back when, and the world moved on. It's a great protocol, that's for sure, but USB turned out to be good enough, and it's simpler to implement, and much cheaper. That's why it's universal.
 

1audio

Well-Known Member
Sep 17, 2010
31
2
353
SF Bay Area
Getting 100 femtosecond RMS jitter is probably going to be expensive, but 200 or 300 fs isn't really that hard, provided one is willing to use an external clock, or has a USB-S/PDIF device with excellent internal clocks.

You can get whatever level of jitter you want if you spec it right. Most jitter specs for crystal oscillators are based on SONET specs, with a bandwidth of 12 KHz to 5 MHz. This can be misleading for audio. I did a quick calculation based on one of the best oscillators made for close in phase noise (a BVA) and depending on the range included in the calculations the numbers are quite different. For a 10KHz to 5 MHz it works out to 50 fS. The same oscillator using the AES numbers I found somewhere starting with 100 Hz and ending at 10 KHz gives a less pretty number of 150 fS and extending to 10 Hz to 50 MHz gets a realistic 2.8 pS. The other issue is what range is relevant? technically nothing above the sample rate will get down converted into the audio spectrum so you stop at FS (192 KHz worst case). And then the jitter has much less effect on low frequencies so where do you stop? And is it audible? More controversy and heat than clear light.

Up in the $20-30K range, DACs such as the MSB don't reclock, and also they essentially double-up their multiple DAC modules to get 25 bits, yielding 24 bit reliably. For this level of performance, one must pay. :)
24 bit accuracy in a DAC is incredibly hard to get. Analog devices lists one 20 bit industrial DAC, TI lists one at 20 bits, LTC stops at 18 bits. I have two examples, both are Kelvin-Varley dividers with 24 bit accuracy, and take about 20 seconds to set to a voltage. So then we relax what we mean by 24 bits. Its not absolute accuracy anymore (code in = volts out +/- less than 1 lsb) but something else, usually a more course accuracy that it related to SNR etc. And then the bits are a little less meaningful.

In a sense any spdif receiver reclocks, as in regenerates the clock from the data stream and some do a very good job of it. Its possible to further "reclock" if you can generate a cleaner clock that is still in sync with the audio stream. There are a number of techniques that work, Lavry has one of the more sophisticated ones as does this device: http://www.dagogo.com/View-Article.asp?hArticle=30 with a fifo and independent clock domains. the techniques are not that difficult or expensive. Execution is the key as it always is.
 

1audio

Well-Known Member
Sep 17, 2010
31
2
353
SF Bay Area
At the risk of further raising the ire of our moderator, can you explain the Auriliti server...is it intended to completely bypass a computer.....WDW

Since you started this thread, sure. The moderator can move it if desired. Its perhaps easiest to think of the box as an audio computer. It does what is necessary to take the content of a USB stick or USB connected drive and play it through the S/PDIF or USB output. You can find out more here: http://auraliti.com/uploads/Auraliti_Simplified.pdf We should start a specific thread if you want more details.
 

Nicholas Bedworth

WBF Founding Member
May 7, 2010
312
0
0
Maui, where else?
Hand-built DACs

The MSB DACs are apparently built by hand, from scratch, using precision resistors. They're definitely not orderd from a catalog, hence the $30K price. :) My recollection is that they do 25 bits and in pairs, to get close to real 24 bit outputs. It's quite a stunt! They use several additional and quite interesting tricks to get such accuracy. http://www.msbtech.com/products/dac4diamond.php?Page=platinumHome

We're using laboratory phase noise measurement equipment, either the Symmetricom 5120a or the new Time Pod, and if you look at the areas of significance to audio, say, 1 to 100 kHz, the RMS jitter using internal clocks in the best USB-S/PDIF devices will be from 2.5 ps or lower, depending on the clock frequency. With the external clocks, one can see a factor of 10 less (again, this is from memory, but let me check the data when time allows). In any event, the very low phase noise offsets, from say 0.1 Hz to 100 Hz, seem to be critical for realism and presence.

The MSB high-end DACs don't regenerate the clock from S/PDIF. They discard all incoming clock information, dump the digital music data into a big buffer, and clock it out using OXCO internal clocks. This approach may be similar to that of Dan's. There's no attempt made to sync the DAC with the incoming data stream. This of course is also my favorite architecture. We used this design on the input side A/D side starting about 25+ years ago with a 5 MHz dual-ported memory in which the input clocking and output clocking are uncoupled. Works just as well on the D/A side as well.
 

microstrip

VIP/Donor
May 30, 2010
20,806
4,698
2,790
Portugal
(...) The review is of course very positive although I was puzzled by him saying he found a difference between a PC and Mac source. The point of an asynchronous and isolated bridge is to eliminate such differences. I will have to ponder what remains to explain the differences Robert Harley perceived.

I read the review a few days ago in TAS and only now saw this interesting thread.

It is always nice to have preliminary appreciations of new products, but the excessive insistence of Robert Hartley review on the importance of the Macintosh contribution to sound quality is unsettling. A review with another platform such as Windows7 seems to be urgently needed. I would also like to hear the opinion of the manufacturer on this subject - I can not believe they are not aware of this problem, and I have no doubts they presently can advise on it better than anyone else.
 

amirm

Banned
Apr 2, 2010
15,813
37
0
Seattle, WA
I just checked with Berkeley folks. They say they have tested Windows 7 ASIO drivers against the Mac and there is no sonic difference. Michael says that Robert used an XP machine with ASIO drivers and he (Michael) believes that doesn't sound good. I have no explanation for that OS either if bit-exact rendering is used. Regardless, I hope everyone runs Windows 7 anyway and in that sense, this point in the review is moot.
 

DonH50

Member Sponsor & WBF Technical Expert
Jun 22, 2010
3,947
306
1,670
Monument, CO
Aside: Most clock sources spec jitter well away from the carrier (center, clock frequency), and usually include a curve to show close-in phase noise (that can be converted to jitter). Jitter rises very rapidly near the center frequency, heading toward infinity right at the clock frequency (there are practical limits, natch). Because of that, the wideband jitter number typically provided for an oscillator (clock) is fairly useless for audio circuits, where most of the signal bandwidth is "close-in" to the clock.

Random jitter is only one part of the picture anyway, with deterministic jitter, sinusoidal jitter, etc. also contributing. I have long suspected that deterministic jitter is the true culprit; jitter caused when one sample's value is related to the previous sample. Asynchronous clocking also help to eliminate that error source; other schemes simply isolate the clock from the data stream so the two do not interact. Hopefully...
 

1audio

Well-Known Member
Sep 17, 2010
31
2
353
SF Bay Area
I think we are on the same page with jitter issues. They are quite complex once you get to actually making something.

the new Time Pod, .

What is a Time Pod? I'm pretty familiar with available solutions for measuring phase noise and this is new to me.
 

Orb

New Member
Sep 8, 2010
3,010
2
0
Aside: Most clock sources spec jitter well away from the carrier (center, clock frequency), and usually include a curve to show close-in phase noise (that can be converted to jitter). Jitter rises very rapidly near the center frequency, heading toward infinity right at the clock frequency (there are practical limits, natch). Because of that, the wideband jitter number typically provided for an oscillator (clock) is fairly useless for audio circuits, where most of the signal bandwidth is "close-in" to the clock.

Random jitter is only one part of the picture anyway, with deterministic jitter, sinusoidal jitter, etc. also contributing. I have long suspected that deterministic jitter is the true culprit; jitter caused when one sample's value is related to the previous sample. Asynchronous clocking also help to eliminate that error source; other schemes simply isolate the clock from the data stream so the two do not interact. Hopefully...
Good points Don,
just to add Paul Miller has measured and shows an anecdotal trend between phase noise related jitter and its effect on sound-image, most recent example was the review of the Yamaha music server where phase noise looks to have contributed to affecting the image/sound and was picked up by 2 of his reviewers, which correlated with his own views and measurement for its jitter.
Its correlated jitter has exceptionally low.

But one area that seems to only come to light recently in audio publishing is that of analysing what is happening from DC to 5hz, where an analysis was done after subjective preference occurred between an isochronous DacMagic USB product and then used with aysnch Halide Bridge (which of course limits this problem and was the preferred product).
Paul Miller followed the following article up using his own software and he calculated Type1 jitter to be equal to 1680ns, surmisation was that the suppression of the buffering-resampling reaches it limits at around 2-5hz in the isochronous DacMagic, and involved demodulated signals - June 2011 Opinions Hi-finews.
Link to initial article by Jim Lesurf.
http://www.audiomisc.co.uk/Linux/Sound3/TimeForChange.html

Cheers
Orb
 

microstrip

VIP/Donor
May 30, 2010
20,806
4,698
2,790
Portugal
Amir,

Do you know when you will receive the alpha USB? It seems as it is not CE certified it can not be sold in Europe, unless we buy it in Switzerland.
 

amirm

Banned
Apr 2, 2010
15,813
37
0
Seattle, WA
Unfortunately there was yet another parts delay. To be on the safe side, I say we are looking at end of July.
 

1rsw

Well-Known Member
Jul 21, 2010
134
1
365
Unfortunately there was yet another parts delay. To be on the safe side, I say we are looking at end of July.

I wish! This is getting to be a little frustrating. After the wait for the DAC I should have expected it but damn am I getting impatient. From what I read, there could be yet another delay. Having said all that, I do really appreciate their attention to detail and it seems some of this is related to wanting to be 100% sure the devices being shipped are built/operating to their standard. There are many companies that would just try and get em shipped and deal with issues later.
 

ack

VIP/Donor & WBF Founding Member
May 6, 2010
6,774
1,198
580
Boston, MA
If it helps, due to the Japan earthquake, we are seeing significant automotive part delays all the way to CPUs (yes, I've been waiting for some heavy duty Lenovo hardware to show up for months now and the missing piece is one CPU).
 

amirm

Banned
Apr 2, 2010
15,813
37
0
Seattle, WA
Indeed. As product delays go, this is one of the worst examples I have seen. Latest status report says they are a couple of weeks away and it is still due to part issue. Let's see if that does it.
 

About us

  • What’s Best Forum is THE forum for high end audio, product reviews, advice and sharing experiences on the best of everything else. This is THE place where audiophiles and audio companies discuss vintage, contemporary and new audio products, music servers, music streamers, computer audio, digital-to-analog converters, turntables, phono stages, cartridges, reel-to-reel tape machines, speakers, headphones and tube and solid-state amplification. Founded in 2010 What’s Best Forum invites intelligent and courteous people of all interests and backgrounds to describe and discuss the best of everything. From beginners to life-long hobbyists to industry professionals, we enjoy learning about new things and meeting new people, and participating in spirited debates.

Quick Navigation

User Menu

Steve Williams
Site Founder | Site Owner | Administrator
Ron Resnick
Site Co-Owner | Administrator
Julian (The Fixer)
Website Build | Marketing Managersing