[grokit]

 
Quote:

Anetode:- I think you've pretty much got it! Although I wouldn't underestimate the number of "inept boutique products" out there.

The reason I've recommended the Lavry da924 (or the more consumer friendly version, the da2002) is simply because it has been a standard used in many world class mastering studios for many years. Amazing really that in this fast moving area of technology that a roughly 10 year old design is still at the top. However, I'm sure during the last 10 years that technology has allowed for the same or almost the same level of quality for considerably less money. A worrying trend I see though is that many DAC manufacturers are not aiming at the best quality but at the biggest numbers. Presumably driven by marketing departments who obviously feel it's easier to sell bigger numbers than it is to sell quality. Even to the point that bigger numbers mean worse quality! The marketing seems to be working, I wonder how many audiophiles are under the impression that 24/192 (or even worse 32/384) is higher quality than 24/96 or 16/44.1? Might be interesting to find out!

G

 
For 16/44 music? I would bet more than half of us.

 

[gregorio]

For 16/44 music? I would bet more than half of us.

From what I've read on head-fi, substantially more than half! I'm quite willing to explain, if anyone is interested?, why 24/192 (and higher) are compromised formats relative to say 16/44. Not sure if that is a discussion for the science forum?

G

 

[grokit]

Hmm... I thought it was basically a discussion about various methods of up-conversion vs. bit-perfect.
 
Please go on, I'm game 

 

[gregorio]

Hmm... I thought it was basically a discussion about various methods of up-conversion vs. bit-perfect.
 
Please go on, I'm game 

OK, thought it would be more appropriate to post it on the Science Forum, Here it is.

I've lit the touchpaper now just sit back and see what happens!

G

 

[DaveBSC]

Quote:

That makes sense.  I'm receiving a Berkeley Alpha DAC in the next few weeks, and I've been struggling with the best way to do a USB to S/PDIF conversion.  I haven't been able to find good information about the actual engineering design that goes into building a USB-S/PDIF converter to determine whether there's just a lot of markup and black magic going on, or if there's a good reason to get a high-end converter like the Alpha USB or Wavelink.  Or, if there is a real difference between one of the $1k+ converters and the $200-400 ones, if I might be better served going with a pro solution like the DCD-8 or Big Ben.

There is a significant difference between the adaptive mode converters like the Bel Canto and the Centrance stuff, and quality asynchronous converters. This is especially true of the PCM270x based, 16/48 limited USB inputs that must go through an internal S/Pdif conversion process before going on to I2S. If you can't tell the difference between one of those inputs and straight S/Pdif or AES/EBU, this isn't the hobby for you.
 
That being said, asynchronous is not a magic bullet that solves everything. The Hiface is the prime example of this. It's 44 and 48 based clocks are of reasonably good quality, but the input power is compromised, the power to the clocks is compromised, and the output is way out of the S/Pdif spec. What makes the JK MK3 version so compelling is that all of these problems are solved, and because John doesn't run a high-end company that sells through a dealer network (markup on top of markup) what you end up with can be considered a really good value. The Audiophilleo2 is similar money, but it's completely bus powered like the standard Hiface. Audiophilleo talks about how great their regulation is, but I don't believe that regulated USB can be as good as squeaky clean battery power.
 
I'm not really familiar with the DCD-8 or Big Ben, but the RME Fireface products are sometimes talked about around here as a "professional" alternative to USB converters, and their performance in terms of jitter is not impressive in the slightest.
 
http://www.rme-audio.de/en_support_techinfo.php?page=content/support/en_support_techinfo_steadyclock
 
 
 
 
 

 

[gregorio]

There is a significant difference between the adaptive mode converters like the Bel Canto and the Centrance stuff, and quality asynchronous converters. This is especially true of the PCM270x based, 16/48 limited USB inputs that must go through an internal S/Pdif conversion process before going on to I2S. If you can't tell the difference between one of those inputs and straight S/Pdif or AES/EBU, this isn't the hobby for you.

I've never had to use one of these format converters, so I don't know much about what they do or how they do it. Is the difference in price/performance of these units mostly based on their level of output jitter or something else? If it's based on their output jitter, why buy an expensive unit if the jitter is reduced below audibility when the signal reaches the DAC?

G

 

[DaveBSC]

Quote:

I've never had to use one of these format converters, so I don't know much about what they do or how they do it. Is the difference in price/performance of these units mostly based on their level of output jitter or something else? If it's based on their output jitter, why buy an expensive unit if the jitter is reduced below audibility when the signal reaches the DAC?

G

 
I don't think its based on output jitter. Most of the good asynch converters should be less than 50ps, which shouldn't be an issue. I think the performance differences are mostly down to good power, and a quality S/Pdif output stage. The USB spec was never intended to transmit audiophile grade power - it's made to power thumb drives and charge cellphones and whatnot. Running hyper sensitive clocks and the S/Pdif output on this supply is simply a bad idea, which is why the Hiface and the Halide Bridge are bottom of the barrel in terms of sound quality. You may be able to clean the USB supply to an extent, which is what the Audiophilleo and Sonicweld converters do, but I think the best way to go is to avoid USB power altogether, either by way of a linear AC fed power supply, or batteries.
 
The Alpha USB and the Wavelength Wavelink use a mix of AC or battery power and USB power for their input sections, while the JK MK3 Hiface is at least near 100% battery, if not completely battery powered. I'm not entirely sure how the Off-ramp works. At least most of its power comes from external DC, but I'm not sure whether its input section is USB powered or not.
 
Another positive about completely avoiding USB powered is that it can then be cut at the source via something like the SoTM USB card. The really expensive USB cables try to separate their power and signal sections as much as possible to avoid interference. If there is no USB power being transmitted, there's no possibility for interference, and a Belkin cable should at least theoretically sound just as good as a Locus Cynosure.
 
 

 

[gregorio]

Thanks for the explanation, but I still don't fully understand. Why do these format converters need good power supplies for hyper sensitive clocks and spdif outputs?

Even ten times the amount of jitter you mentioned coming out of the format converter to the input of the DAC should be reduced way below audibility.

G

 

[DaveBSC]

Quote:

Thanks for the explanation, but I still don't fully understand. Why do these format converters need good power supplies for hyper sensitive clocks and spdif outputs?

Even ten times the amount of jitter you mentioned coming out of the format converter to the input of the DAC should be reduced way below audibility.

G

That I don't believe is true. <100ps yes, 500+ps, should start to noticeably degrade the sound.
 

 

[Elysian]

I think I read on one of the pro forums that the jitter on a well-implemented S/PDIF connection is around 50ps, and a quality USB implementation is closer to 5, but I could be remembering wrong.
 
Berkeley's handling the power for their converter this way:
 
http://www.berkeleyaudiodesign.com/products4.html
 
Quote:

Great care was taken in the design of the Alpha USB to isolate the noisy computer/USB environment from the digital audio output. The USB receiver and processing are powered by the computer, while the output master clocks and line drivers are powered by a separate isolated linear power supply.

Two key factors combine to achieve the excellent audio performance of the Alpha USB - the unprecedented degree of electrical isolation between the USB input connection and the audio output and the very low noise/low jitter performance of the individually tested audio output master clocks.
...
• Mains power: 100/120/240VAC, 50/60Hz, IEC power input connector
• Power consumption: 3W, designed for continuous operation

 
Anyway, I've been reading a lot of pro people say these converters should be a relatively straightforward affair, while the audiophiles disagree vehemently.  I can see the way the power is implemented and the clocking to account for the difference between higher and lower-end converters, and the price on the expensive offerings probably partially factors in the sheer amount of R&D time that went into it.  It's a toss-up whether all that extra time makes an audible difference or not.  The INT202 seems to be an overrated piece of gear, and several DIYers have told me the Wavelink and Audiophilleo2 are not only cheaper, but do a better job.

 

[DaveBSC]

Quote:

The INT202 seems to be an overrated piece of gear, and several DIYers have told me the Wavelink and Audiophilleo2 are not only cheaper, but do a better job.

 
I think when done properly, USB is the best way to get audio from a computer. Using a NAS drive via ethernet to a DAC is a different matter, that's a music server, not a computer. The first step is making sure the port isn't big shared with wireless mice, keyboards, and other peripherals. The SoTM card is definitely better than just a standard motherboard port. Cables are up to the individual, but with a good asynch converter, a basic shielded cable should do the job. Still, it wouldn't hurt to experiment with a Wireworld - provided the dealer has a return policy. It does make sense to use the shortest possible USB cable, and a long S/Pdif cable rather than the other way around.
 

 
 

 

[grokit]

 
Quote:

There is a significant difference between the adaptive mode converters like the Bel Canto and the Centrance stuff, and quality asynchronous converters. This is especially true of the PCM270x based, 16/48 limited USB inputs that must go through an internal S/Pdif conversion process before going on to I2S. If you can't tell the difference between one of those inputs and straight S/Pdif or AES/EBU, this isn't the hobby for you.
 
That being said, asynchronous is not a magic bullet that solves everything. The Hiface is the prime example of this. It's 44 and 48 based clocks are of reasonably good quality, but the input power is compromised, the power to the clocks is compromised, and the output is way out of the S/Pdif spec. What makes the JK MK3 version so compelling is that all of these problems are solved, and because John doesn't run a high-end company that sells through a dealer network (markup on top of markup) what you end up with can be considered a really good value. The Audiophilleo2 is similar money, but it's completely bus powered like the standard Hiface. Audiophilleo talks about how great their regulation is, but I don't believe that regulated USB can be as good as squeaky clean battery power.
 
I'm not really familiar with the DCD-8 or Big Ben, but the RME Fireface products are sometimes talked about around here as a "professional" alternative to USB converters, and their performance in terms of jitter is not impressive in the slightest.
 
http://www.rme-audio.de/en_support_techinfo.php?page=content/support/en_support_techinfo_steadyclock

 
Well said, these statements totally agree with my experience, with one glaring exception: Which Bel Canto converters are you referring to? I have an older DAC2 that is offline ATM because it is single-ended, but I have resisted selling it because I like the way it sounds. Anyways you're just talking USB when you say adaptive mode, correct? Because when I look at BC's specs for my unit (which does not feature a USB input), they are pretty clear that it's asynchronous:
 
"The new Asynchronous Sample Rate Converter (ASRC) used in the DAC2 provides a minimum of 139 dB of dynamic range. Result: Any filter residue is well below the noise floor of the original recording, providing better imaging and high frequency coherence."
 

 

[gregorio]

That I don't believe is true. <100ps yes, 500+ps, should start to noticeably degrade the sound.

My problem in understanding, I think is resolved. If I've understood correctly, there is no benefit to these expensive units, they are selling on the marketing hype of accurate clocks and low jitter. Which is pointless because any decent DAC will remove the jitter. Actually DaveBSC 500ps of jitter would start to degrade the sound, if it wasn't removed by the DAC's jitter reduction circuitry. 500ps of jitter would be pretty trivial to cope with for any decent DAC. Even a decade ago professional DACs were available which could easily reduce to a signal with many nanoseconds of jitter (>4,000ps) down to just around 5ps or so of jitter. Feed them any crappy clock or signal into the input and you get beautiful pristine clock to the DA chip.

I think I see what's going on now. I see now Elysian has posted "lot of pro people say these converters should be a relatively straightforward affair". I'm with that camp, definitely, all it needs to do is convert the format, nothing too complex there. What's the point of spending money on an expensive clock source in a format converter when the clock signal is going to be regenerated in the DAC anyway? It's like owning a top of the range power conditioner and then buying another power conditioner so your top of the range power conditioner only runs off conditioned power. Nonsensical if you ask me!

Thanks for the info and helping me understand what's going on.

G

 

[DaveBSC]

Quote:

Well said, these statements totally agree with my experience, with one glaring exception: Which Bel Canto converters are you referring to? I have an older DAC2 that is offline ATM because it is single-ended, but I have resisted selling it because I like the way it sounds. Anyways you're just talking USB when you say adaptive mode, correct? Because when I look at BC's specs for my unit (which does not feature a USB input), they are pretty clear that it's asynchronous:
 
"The new Asynchronous Sample Rate Converter (ASRC) used in the DAC2 provides a minimum of 139 dB of dynamic range. Result: Any filter residue is well below the noise floor of the original recording, providing better imaging and high frequency coherence."
 
 

 

 
You're making a mistake here in terminology. ASRC and asynchronous USB conversion are two entirely different things. ASRC has nothing to do with USB. It takes an incoming digital signal 16/44, 24/88, whatever, and upsamples it, usually either to 24/96 or 24/192. 96 and 192Khz are not direct multiples of 44.1, which is where you get the asynchronous part. A DAC with synchronous sample rate conversion would upsample 16/44 to 24/176.
 
Asynchronous USB slaves the computers clock to the S/Pdif converter, rather than adapting to the computer's clock. The Bel Canto USB link uses Centrance adaptive mode tech. Output jitter is 600+ps, and it is audibly worse sounding, even when connected to a good DAC, than asynchronous converters which are 1/10 of that.
 
 

 

[grokit]

Ah, thanks for that Dave. That USB Link doesn't seem like a very good deal at all for $500, which is more than I paid for the DAC2.
 
The DAC2 has pretty good jitter control on its own, as I have hooked up USB reclocker/converters to it that make a difference with other DACs and The Song Remains the Same as with a straight converter on the DAC2. The anti-jitter measures don't make a difference on that one at all.