A conversation with Charles Hansen, Gordon Rankin, and Steve Silberman

[chriss71]

The USB driver does not 'ensure reliability', although typically you don't expect many errors as the document says.

 

I will make it bold for you, OK!

 

Reliability - Bit Error of 10 to the Power of -10 or better

 

I have tested it many times, and I always came up with a perfect stream...

[Richard Dale]

I will make it bold for you, OK!

 

Reliability - Bit Error of 10 to the Power of -10 or better

 

I have tested it many times, and I always came up with a perfect stream...

 

If you are going to shout at me, I don't think there is much point in continuing this conversation.

 

They are talking about levels of unreliability. In the context of a network protocol 'reliable' has a very specific meaning. See this wikepedia link:

 

Reliability (computer networking) - Wikipedia, the free encyclopedia

 

"In computer networking, a reliable protocol is one that provides reliability properties with respect to the delivery of data to the intended recipient(s), as opposed to an unreliable protocol, which does not provide notifications to the sender as to the delivery of transmitted data."

 

Isochronous USB in an 'unreliable protocol' by this definition.

[chriss71]

If you are going to shout at me, I don't think there is much point in continuing this conversation.

 

That was not in my intention, sorry for that, Richard...

 

Yeah, it is a question about definition. I only can say, what I have observate long time. I always ended up in a bit perfect stream on the opposite flip side...

 

But I think our two points are clear, so let us agree to disagree....

[The Computer Audiophile]

Hi Guys - I just asked an educated, certified, experienced computer/audio engineer to enter this discussion and clarify several things, but it's not going to happen. This person said there's so much misinformation in this thread He or she wouldn't know where to start.

 

I think the armchair engineering is getting way out of hand and doing a disservice to those who read the thread.

[Superdad]

Hi Guys - I just asked an educated, certified, experienced computer/audio engineer to enter this discussion and clarify several things, but it's not going to happen. This person said there's so much misinformation in this thread He or she wouldn't know where to start.

 

I think the armchair engineering is getting way out of hand and doing a disservice to those who read the thread.

 

Yeah Chris, I have begged my engineering partner (he designs/routes/tests giant computer chips for a fabless house for 30 years; plus he is a brilliant digital/analog audio designer) to come join CA many times. He has mostly sworn off forums for the same reason as your guy.

[Jud]

Hi Guys - I just asked an educated, certified, experienced computer/audio engineer to enter this discussion and clarify several things, but it's not going to happen. This person said there's so much misinformation in this thread He or she wouldn't know where to start.

 

I think the armchair engineering is getting way out of hand and doing a disservice to those who read the thread.

 

Hi Chris. What you and Superdad have said makes me think of perhaps a different way to do something that's often been discussed as a possibility here at CA.

 

Rather than having to clear up misinformation, perhaps your (and/or Superdad's) engineer friend might author a tutorial or series of tutorials on the general subject of getting data from a computer to a DAC and potential problems an audio engineer must deal with in that regard. And rather than having to deal with the problems of forums, comments could be disabled.

 

Because I'm sure I've managed to give out misinformation in this and/or other threads in the course of speculating about what might cause the differences I believe I hear, and I would hate to be "sentenced" to doing so in perpetuity when there's a way to avoid it. And I'd be able to learn more about our hobby, which would be a really wonderful thing.

[rayooo]

Hi Guys - I just asked an educated, certified, experienced computer/audio engineer to enter this discussion and clarify several things, but it's not going to happen. This person said there's so much misinformation in this thread He or she wouldn't know where to start.

 

I think the armchair engineering is getting way out of hand and doing a disservice to those who read the thread.

 

+1 !!

[realhifi]

Thanks for the heads up to that discussion between those three. I believe Michael sums it up well in his response to some of the "armchair engineers" so well named by Chris. If you don't mind my posting his reply: "I'd recommend anything in the "Industry Voice" section of AudioStream for starters. But I think one issue we're dealing with is point of view. It should be obvious that the people designing the gear face "problems" we as listeners would never encounter. For example, the issues with adaptive versus asynchronous USB would certainly never cross a listener's mind but Gordon Rankin (and now many others) saw asynchronous USB as a solution to problems inherent in adaptive USB (mainly by allowing the DAC to control the timing of data transfer).

 

As far as your earlier question CG, why all the animosity and rancor, I wish I had a good answer but I only have guesses. One guess is some people seem to think that manufacturers and reviewers are in league to confuse people into buying stuff they don't need. And these people are obviously too smart to fall for this trap so they feel obligated to warn the less able. Of course this is utter nonsense and I say this with complete confidence speaking as I do from a position of authority at least in terms of knowing why I do things much more than someone who has never met me.

 

One of the main reasons for the "Industry Voice" section of AS is to give people access to the issues and ideas that concern the people involved in the making of the gear they may buy."

 

In particular the sentence about the industry and reviewers being in league to confuse folks to buy more is so true. On this and many sites it feels that many are feeling that (not to mention those sneaky wallet grabbing dealers!) are only doing just that and that all is snake oil.

 

Anyway, thanks for the link, I missed that on that site.

[Miska]

For example, the issues with adaptive versus asynchronous USB would certainly never cross a listener's mind but Gordon Rankin (and now many others) saw asynchronous USB as a solution to problems inherent in adaptive USB (mainly by allowing the DAC to control the timing of data transfer).

 

Asynchronous transfer mode was there in the standard for quite a while, so it was seen as necessary by the people who wrote the USB Audio specification. It just wasn't interesting since USB wasn't being used for hifi or serious recording purposes, mostly for USB headsets for VoIP calls and such where adaptive mode really wasn't such an issue.

 

Pro-audio equipment if it was using USB was anyway mostly using vendor specific proprietary protocols to do asynchronous transfer, much like RME, M2Tech and RigiSystems still are.

 

The entire hifi industry is sort of late wake-up to computer based audio, while it has been around for ages for gaming and pro-audio. Gaming sound cards just went for having complex on-boards DSPs for headphone and multi-channel speaker 3D-audio, and pro-audio was mostly using PCI or Firewire.

 

For example, these are not bad products in their segment either:

Sound Blaster ZxR - Sound Blaster - Creative Labs (Pan Euro)

Sound Blaster Recon3D Fatal1ty Champion Sound Card - Creative Labs (Pan Euro)

Apollo 16 Audio Interface with Realtime UAD Processing and Thunderbolt

 

I designed and built my first PC audio interface card in 1989, in the same days of AdLib and first SoundBlaster. But then I went mostly to passive sonar and audio analysis systems... (apart from pile of DIY DAC's for personal use)

 

Kind of funny thing is that the audiophile and pro-audio industry segments are still largely disconnected and the late wakeup audiophile industry is having a bit NIH-syndrome... There are of course number of companies who are now more or less players in both grounds, like Mytek and Benchmark, with strong pro-audio background.

[Audiophile Neuroscience]

I think the armchair engineering is getting way out of hand and doing a disservice to those who read the thread.

Rather than having to clear up misinformation, perhaps your (and/or Superdad's) engineer friend might author a tutorial or series of tutorials on the general subject of getting data from a computer to a DAC and potential problems an audio engineer must deal with in that regard. And rather than having to deal with the problems of forums, comments could be disabled.

1+ for tutorials by recognized professionals. Q and A sessions afterwards would be great rather than "comments" if this was preferable to the author.Of course there are always issues of potential bias or affiliations but once declared you take that on board. Notwithstanding possible agendas, I guess it is fair to say there is not a lot of experts willing to donate their time ?

[One and a half]

Q and A sessions afterwards would be great rather than "comments" if this was preferable to the author.Of course there are always issues of potential bias or affiliations but once declared you take that on board. Notwithstanding possible agendas, I guess it is fair to say there is not a lot of experts willing to donate their time ?

 

On the Audiostream site, CG used to post here at CA moons ago, Gordon Rankin now only rarely, both are well respected for audio and digital engineering. Whether they will be kind enough to give the basics without revealing too much of their IP, let's give it a shot.

 

The difficulty is that IP information might slip out and cause some problems, but usually you can get conflicting opinions form engineers on the same issue. There's always a few ways around things, to find the most efficient is often staring at you in the face and often a solution can't be correct because *I* didn't mention it....ego gets in the way!

 

The other link in the chain of course is the player software, something tells me it's not a simple script although that's how it can start. EG, from a terminal window in windows, you can type c:\music\01_beethoven_symp.wav and enter, and something will play. It's the process when the read is made from the HDD in between the file and the driver for the sound hardware or USB stream that needs some explanations.

 

Perhaps wait for part two of the article, it's not out in the open just yet.

[Audio_ELF]

The problem with any such "tutorial" is that each engineer will talk and speak to an absolute truth (as they see it) where as ask 10 engineers and you will get 11 different replies.

[Richard Dale]

1+ for tutorials by recognized professionals. Q and A sessions afterwards would be great rather than "comments" if this was preferable to the author.Of course there are always issues of potential bias or affiliations but once declared you take that on board. Notwithstanding possible agendas, I guess it is fair to say there is not a lot of experts willing to donate their time ?

 

I don't see the problems with this thread being anything to do with a lack of tutorials. There is a mass of information on this site about computer audio, and I personally have learned a great deal from it all. That has not brought me up to the level of knowledge of the likes of Gordon Rankin or Charles Hansen who have a mass of practical experience in designing DACs, engineering degrees etc, etc.

 

To me the problem with this thread and others is that people with no practical experience or deep understanding of the issues have taken it upon themselves to say that 'Gordon Rankin is wrong' or whatever. No amount of extra tutorials will help with that. I like to think that I know what I know, and I know the limits of what I know, and I know when I should defer to someone who has more knowledge and experience than myself. I am not sure it is possible to teach that kind of self awareness of your own limitations to everyone out there.

[mav52]

 

To me the problem with this thread and others is that people with no practical experience or deep understanding of the issues have taken it upon themselves to say that 'Gordon Rankin is wrong' or whatever. No amount of extra tutorials will help with that. I like to think that I know what I know, and I know the limits of what I know, and I know when I should defer to someone who has more knowledge and experience than myself. I am not sure it is possible to teach that kind of self awareness of your own limitations to everyone out there.

 

 

So true, it happens in a lot of these threads. Want-to-be's designers, engineers and manufactures challenge actual designers and engineers on products these actual designers/engineers actually produce and sell. I'm just amazes me that these want-to-be's ( know more than a actual manufacturer) aren't designing/engineering and selling their own products, after all they talk a good game, but sadly it's pretty much just talk, the proof has always been and still is in the end result and that is a product being sold to consumers.

[Sam Lord]

I think a general response to the oversimplification of audio fidelity from digital sources could be to conduct a thought experiment: a null test, not of data but of waveforms, which pits the exact voltage and time of a signal entering and/or leaving a DAC versus a perfect signal. The data "one" has a threshold voltage that varies slightly with conditions, an average that does the same, a variation with time that does the same... et cetera, et cetera. All of the parameters of transforming digital data into analog signal vary with the physical environment. Just ask your network "bits is bits" fellow to achieve a voltage vs. time null of any given pair of signals to below -130dBFS, and he'll back off a little.

 

I've found Hanson and Rankin to be extremely strong designers. In particular I recommend Hanson's thousands of posts on DIYAudio forums for proof. I wish Bruno Putzeys had been aboard that discussion, that guy has turned the amplifier world on its head. His converter work is also world-class, see the recent design papers from Mola-Mola Audio.

[Superdad]

I think a general response to the oversimplification of audio fidelity from digital sources could be to conduct a thought experiment: a null test, not of data but of waveforms, which pits the exact voltage and time of a signal entering and/or leaving a DAC versus a perfect signal. The data "one" has a threshold voltage that varies slightly with conditions, an average that does the same, a variation with time that does the same... et cetera, et cetera. All of the parameters of transforming digital data into analog signal vary with the physical environment. Just ask your network "bits is bits" fellow to achieve a voltage vs. time null of any given pair of signals to below -130dBFS, and he'll back off a little.

 

I've found Hanson and Rankin to be extremely strong designers. In particular I recommend Hanson's thousands of posts on DIYAudio forums for proof. I wish Bruno Putzeys had been aboard that discussion, that guy has turned the amplifier world on its head. His converter work is also world-class, see the recent design papers from Mola-Mola Audio.

 

Great points. And thank you for calling my attention Mr. Putters and his new firm Mola-Mola! Just read their whole site and his development blog. Wow, a VERY sharp mind.

I'd love to hear their innovative DAC/preamp.

[Sam Lord]

Thanks Superdad. Heh, Mr. Putters...he'd like that. Like many of the best designers, Bruno is very generous with his time; I've spoken with him at great length at AES conventions.

 

Just for reference, folks might want a taste of what this designer offers. Note, I have no affiliation with Bruno or his companies, don't even own any gear (yet). His DAC maintains a 140dB SNR and unmeasurable THD/IMD, yikes. And this is with a non-gain-ranging design.

[Superdad]

Thanks Superdad. Heh, Mr. Putters...he'd like that. Like many of the best designers, Bruno is very generous with his time; I've spoken with him at great length at AES conventions.

 

Just for reference, folks might want a taste of what this designer offers. Note, I have no affiliation with Bruno or his companies, don't even own any gear (yet). His DAC maintains a 140dB SNR and unmeasurable THD/IMD, yikes. And this is with a non-gain-ranging design.

 

Oops, "Putters" was my iPad's auto-correct!

Sorry Sam, but what is a "non-gain-ranging design."

[Miska]

Great points. And thank you for calling my attention Mr. Putters and his new firm Mola-Mola!

 

OK, more 1-bit DACs and something to chew for those who claim that 1-bit can't be good. Running at 100 MHz.

DAC

(IOW, it's an up-sampling "DSD-DAC" with 100 MHz sampling rate)

 

Although he uses three DSP processors, the first stage still seems to be resource constrained looking at how it has been implemented. Probably not that bad performance regardless.

 

I would love to play with this one!

[Superdad]

OK, more 1-bit DACs and something to chew for those who claim that 1-bit can't be good. Running at 100 MHz.

DAC

(IOW, it's an up-sampling "DSD-DAC" with 100 MHz sampling rate)

 

Although he uses three DSP processors, the first stage still seems to be resource constrained looking at how it has been implemented. Probably not that bad performance regardless.

 

I would love to play with this one!

 

But they say it is 3.125mHz/32 bits. Would like to hear his line stage circuit as well. Does not look like the Mola-Mola Makua is in production yet.

[Sam Lord]

...Sorry Sam, but what is a "non-gain-ranging design."

I can't draw it for you, and someone like Miska should answer this because he won't screw it up like me, but anyway... a gain-ranging DAC design switches circuits, I suspect modulators, as the input data crosses certain thresholds. Every circuit has an ideal DC operating point and an AC range that is acceptably linear. But when you have signals like 24bit PCM specifying a dynamic range of 144dB, you can switch circuits to output a signal that either has less amplitude and noise or more of those. So, the lowest noise occurs in the operation of the lowest-output circuit. The idea sounds great until you face the monumental task of making the switching process silent and smooth in its effect on the modulators. I could not make a good DAC or ADC if my life depended on it, but I couldn't make a well-performing gain-staging device if the universe(!) depended on it. I recall two ADCs that use gain ranging; 1) the Prism AD2 Dream; and 2) the Salzbrenner Stagetec ADC, see the latter:

"Truematch Reference Microphone Converter"

They have claimed S/N ratios of 130dB and 153dB respectively, IIRC. There might be others out there.

OK, more 1-bit DACs and something to chew for those who claim that 1-bit can't be good. Running at 100 MHz. DAC

(IOW, it's an up-sampling "DSD-DAC" with 100 MHz sampling rate)

Although he uses three DSP processors, the first stage still seems to be resource constrained looking at how it has been implemented. Probably not that bad performance regardless. I would love to play with this one!

Miska, I always learn a lot from your posts. What do you mean by "resource-constrained" in this case? Just curious. Thanks!

[Superdad]

I can't draw it for you, and someone like Miska should answer this because he won't screw it up like me, but anyway... a gain-ranging DAC design switches circuits, I suspect modulators, as the input data crosses certain thresholds. Every circuit has an ideal DC operating point and an AC range that is acceptably linear. But when you have signals like 24bit PCM specifying a dynamic range of 144dB, you can switch circuits to output a signal that either has less amplitude and noise or more of those. So, the lowest noise occurs in the operation of the lowest-output circuit. The idea sounds great until you face the monumental task of making the switching process silent and smooth in its effect on the modulators. I could not make a good DAC or ADC if my life depended on it, but I couldn't make a well-performing gain-staging device if the universe(!) depended on it. I recall two ADCs that use gain ranging; 1) the Prism AD2 Dream; and 2) the Salzbrenner Stagetec ADC, see the latter:

"Truematch Reference Microphone Converter"

They have claimed S/N ratios of 130dB and 153dB respectively, IIRC. There might be others out there.

 

Wow, thanks for the conceptual explanation of that Sam. Fascinating. Now I get to got and ask my engineering partner to explain that in more detail--which will mostly sail over my head!

 

Can you cite some examples of DACs that DO use "gain-ranging?"

[Miska]

What do you mean by "resource-constrained" in this case? Just curious. Thanks!

 

Because of the light-weight 6th order polynomial interpolator with low 3.125 MHz output rate. Since it has the 0.001 Hz word clock pull, it's practically a polynomial ASRC to reduce jitter at the same time.

[Jud]

it's practically a polynomial ASRC to reduce jitter at the same time.

 

I thought I remembered reading something about its role in reducing jitter at Mola-Mola's site.

[mav52]

I thought I remembered reading something about its role in reducing jitter at Mola-Mola's site.

 

I saw something about Polynomial in a patient description

Patent US20120066280 - Asynchronous Sample Rate Conversion Using A Polynomial Interpolator With ... - Google Patents