or Connect
AVS › AVS Forum › Display Devices › Ultra Hi-End HT Gear ($20,000+) › How to replace your home theater pre-pro with a HTPC!
New Posts  All Forums:Forum Nav:

How to replace your home theater pre-pro with a HTPC! - Page 3

post #61 of 166
What is the advantage of that card over an asynchronous external USB dac ?
post #62 of 166
Quote:
Originally Posted by HFGuy View Post

What is the advantage of that card over an asynchronous external USB dac ?

I looked at the card and it feeds I2S straight into the DEQX processors, and powered by the downstream DEQX gear. Great idea.

I shot an email to MSB to ask if they can develop something similar to use with their DACs and powerbase.
post #63 of 166
Quote:
Originally Posted by xianrenppsg View Post

you are right,AES I/O card is really not much different than a USB or old RS232 serial card function wise.
32.gif

USB audio has a significant difference from AES. AES & SPDIF is a realtime stream with the clock loosely embedded. USB audio is packetized. As this must be memory buffered it affords the opportunity for very good jitter reduction. The same thing can be done with AES or SPDIF but it's optional. With USB audio it's mandatory.

This is why USB audio can outperform AES/SPDIF. Jitter reduction is built into the topology while with consumer SPDIF, it's most often not implemented for cost reasons.

And so called "re-clocking" or "clock cleaners" really don't work well at all. In order to reduce jitter you really need to decode to parallel bits and re-serialize. PLL tracking re-clockers trade jitter for reduce rise time which has problems as well.
post #64 of 166
Quote:
Originally Posted by Glimmie View Post

And so called "re-clocking" or "clock cleaners" really don't work well at all. In order to reduce jitter you really need to decode to parallel bits and re-serialize. PLL tracking re-clockers trade jitter for reduce rise time which has problems as well.

Reclocking worked very will in the PS audio architecture (which is asynchronous and reclocks AES/EBU and S/PDIF), and the MSB DAC, both of which sound a lot better in asynchronous mode than synchronous.
post #65 of 166
Thread Starter 
Quote:
Originally Posted by edorr View Post

I looked at the card and it feeds I2S straight into the DEQX processors, and powered by the downstream DEQX gear. Great idea.

I shot an email to MSB to ask if they can develop something similar to use with their DACs and powerbase.

EXACTLY, it's not going USB to AES or SPDIF and then into the DAC which requires a clock recovery. It's async USB with I2S out direct into the DSP. It sounds awesome, really catapults the DEQX processors into a higher league for that USB input.
post #66 of 166
Quote:
Originally Posted by Nyal Mellor View Post

EXACTLY, it's not going USB to AES or SPDIF and then into the DAC which requires a clock recovery. It's async USB with I2S out direct into the DSP. It sounds awesome, really catapults the DEQX processors into a higher league for that USB input.

I don't understand the USB input though. If this is a PC / mac card, it would just be an I2S output card for DEQX. If it has an USB input and I2S output, it really is a USB to I2S converter on a computer card. If the latter, why would they not stick this in a box like any other USB converter. Also note that USB to I2S converters exists (Empirical Audio Offramp).
post #67 of 166
Quote:
Originally Posted by Glimmie View Post

. PLL tracking re-clockers trade jitter for reduce rise time which has problems as well.

Reduced rise time ?
post #68 of 166
Quote:
Originally Posted by HFGuy View Post

Reduced rise time ?

Bad choice of words. The re-generated clock will have a reduced response time to frequency changes. And as clock recovery depends on tracking the data transitions, you have now introduced a slight variable timing skew. Yes, jitter will measure lower but bit change sample time is still not accurate.

Not much of an issue for audio but a far larger problem with SDI and HDSDI. Re-clockers in digital video are often worse than none. They should only be applied when needed. Too many re-clockers in a chain will actually increase long term jitter to the point of link failure. This was an early problem in SDI routing switchers. Some re-clocked on both input and output so after several passes through the matrix, the signal failed whereas if just left un-reclocked, it would be well within the cliff.
Edited by Glimmie - 11/5/13 at 2:58pm
post #69 of 166
Quote:
Originally Posted by edorr View Post

Reclocking worked very will in the PS audio architecture (which is asynchronous and reclocks AES/EBU and S/PDIF), and the MSB DAC, both of which sound a lot better in asynchronous mode than synchronous.

"Asynchronous" sounds to me like they are sample rate converting. Many products now do that as it's so cheap to implement these days. But you WILL get a framing error sooner or later. You probably won't ever hear it but if two clocks are even a sub fraction off frequency, sooner or later a buffer is going to under or over flow. Fact of physics.

But hey, I too will trade a subtle single click or pop every few minutes versus excessive jitter.
post #70 of 166
I am not at all that familiar with the strange naming convention used in audio DACs but I am an analog designer at a company that designs serdes, I design PLLs and CDRs for 28Gbps. So I find it confusing how a CDR based DAC cannot be vastly superior. I do agree that the recovered clock is very poor but there is NO reason you would want to use the recovered clock rather than a new clock source. With a CDR you would want the highest possible closed loop bandwidth to maximize your jitter tolerance, but that is the opposite of what you want from the clock used to retime the recovered bits. A CDR to recover the data and they a simple elastic buffer to a low bandwidth PLL using a external reference clock is the best method and frankly at this datarates is so simple I cannot imagine it not being the standard.
post #71 of 166
Quote:
Originally Posted by Glimmie View Post

"Asynchronous" sounds to me like they are sample rate converting. Many products now do that as it's so cheap to implement these days. But you WILL get a framing error sooner or later. You probably won't ever hear it but if two clocks are even a sub fraction off frequency, sooner or later a buffer is going to under or over flow. Fact of physics.

But hey, I too will trade a subtle single click or pop every few minutes versus excessive jitter.

There is a way around that too. Divide the recovered clock several times, then use it as a reference to an ultra-low bandwidth PLL, this will prevent the elastic buffer from running empty.
post #72 of 166
Quote:
Originally Posted by Glimmie View Post

"Asynchronous" sounds to me like they are sample rate converting. Many products now do that as it's so cheap to implement these days. But you WILL get a framing error sooner or later. You probably won't ever hear it but if two clocks are even a sub fraction off frequency, sooner or later a buffer is going to under or over flow. Fact of physics.

But hey, I too will trade a subtle single click or pop every few minutes versus excessive jitter.

No, they both run at native sample rate. The MSB has a 0.5s delay because of all the processing in the reclocking circuitry and uses a $5k optional femto clock, so I doubt they are cutting any corners.
post #73 of 166
Quote:
Originally Posted by edorr View Post

and uses a $5k optional femto clock.

Only in high end audio could you charge that much for something so unexceptional smile.gif
post #74 of 166
Quote:
Originally Posted by edorr View Post

No, they both run at native sample rate. The MSB has a 0.5s delay because of all the processing in the reclocking circuitry and uses a $5k optional femto clock, so I doubt they are cutting any corners.

No two devices can run at EXACTLY the same frequency independently. Impossible in this known universe.

To avoid any bit loss the two clocks must be synchronized. Could be via a slow rate PLL, but they must be locked. No two oscillators are identical. Even two Rubidium standards will drift apart. Case in point: Before there were video frame synchronizers available in the late 1970s, the TV networks often would have two Rubidium standards calibrated together, one in the truck calibrated against the unit at the NOC. By the end of a typical live sporting event they would be out of phase enough to cause color shifts at cuts. That's about 2 degrees at 3.58mhz. And gen-locking the network to a remote truck is out off the question for reliability reasons.

So clocks must be at least loosely locked or you will have a buffer under / overflow. No audiophile company can get around this. It's basic laws of physics and time.

And I'm not impressed with the $5K Femto clock. A good used Rubidium standard was about $10K in the 1990s and there's no way the "Femto clock" is even close to half that accuracy. I have worked with Rubidium standards over the years and even these will not magically sync a CD transport to a DAC for very long. Remember that 48/96khz audio is roughly the same bandwidth as NTSC analog video.

IME, the name of the game today in most entertainment grade digital audio systems, that includes pro broadcast stuff too, is sample rate conversion. It's much easier thanks to modern silicon fab than dealing with clock lock problems and the side effects are negligible.

EDIT: I read the blurb at MSB. Seems they are buffering the audio in a 500ms memory. But what they don't tell you is that memory will under or overflow at some point resulting in a soft click which they may in fact mute out as they can "see it coming". But it will always be audible but probably not noticed.

This is basic (and crude) sample rate conversion - deep buffer memory.
Edited by Glimmie - 11/5/13 at 4:54pm
post #75 of 166
Thread Starter 
Quote:
Originally Posted by edorr View Post

I don't understand the USB input though. If this is a PC / mac card, it would just be an I2S output card for DEQX. If it has an USB input and I2S output, it really is a USB to I2S converter on a computer card. If the latter, why would they not stick this in a box like any other USB converter. Also note that USB to I2S converters exists (Empirical Audio Offramp).

It's a USB card that plugs into a slot on the 'motherboard' of the DEQX processor.
post #76 of 166
Quote:
Originally Posted by Nyal Mellor View Post

It's a USB card that plugs into a slot on the 'motherboard' of the DEQX processor.

Got it. This was not clear from the description on the web page.
post #77 of 166
Quote:
Originally Posted by Glimmie View Post

No two devices can run at EXACTLY the same frequency independently. Impossible in this known universe.

By "both" I meant the PS audio DAC and the MSB Dac.

Quote:
Originally Posted by Glimmie View Post

To avoid any bit loss the two clocks must be synchronized. Could be via a slow rate PLL, but they must be locked. No two oscillators are identical. Even two Rubidium standards will drift apart. Case in point: Before there were video frame synchronizers available in the late 1970s, the TV networks often would have two Rubidium standards calibrated together, one in the truck calibrated against the unit at the NOC. By the end of a typical live sporting event they would be out of phase enough to cause color shifts at cuts. That's about 2 degrees at 3.58mhz. And gen-locking the network to a remote truck is out off the question for reliability reasons.

So clocks must be at least loosely locked or you will have a buffer under / overflow. No audiophile company can get around this. It's basic laws of physics and time.

And I'm not impressed with the $5K Femto clock. A good used Rubidium standard was about $10K in the 1990s and there's no way the "Femto clock" is even close to half that accuracy. I have worked with Rubidium standards over the years and even these will not magically sync a CD transport to a DAC for very long. Remember that 48/96khz audio is roughly the same bandwidth as NTSC analog video.

IME, the name of the game today in most entertainment grade digital audio systems, that includes pro broadcast stuff too, is sample rate conversion. It's much easier thanks to modern silicon fab than dealing with clock lock problems and the side effects are negligible.

EDIT: I read the blurb at MSB. Seems they are buffering the audio in a 500ms memory. But what they don't tell you is that memory will under or overflow at some point resulting in a soft click which they may in fact mute out as they can "see it coming". But it will always be audible but probably not noticed.

This is basic (and crude) sample rate conversion - deep buffer memory.

At moments like this I thank my lucky stars for being blissfully ignorant of digital engineering theory, and just being able to recognize good sounding gear when I hear it - much like I suspect a career in gynecology would be highly detrimental to my ability to enjoy having sex smile.gif
post #78 of 166
Do they mention the integration range on that "femto" measurement ? Or do they just give a meaningless number ?
post #79 of 166
Quote:
Originally Posted by HFGuy View Post

Do they mention the integration range on that "femto" measurement ? Or do they just give a meaningless number ?

To be honest, the issue did not come up in my discussions with MSB. I'm in the business of writing checks, listening to music, and swapping gear when I have the upgrade itch. I leave the technical detail to the engineers.
post #80 of 166
Quote:
Originally Posted by edorr View Post

To be honest, the issue did not come up in my discussions with MSB. I'm in the business of writing checks, listening to music, and swapping gear when I have the upgrade itch. I leave the technical detail to the engineers.

I've enjoyed reading your entire process of replacing the pre/pro, I still have alot to learn with this A/V stuff. On the odd occasion I can speak with any authority here I'd like to contribute.
post #81 of 166
Quote:
Originally Posted by HFGuy View Post

I've enjoyed reading your entire process of replacing the pre/pro, I still have alot to learn with this A/V stuff. On the odd occasion I can speak with any authority here I'd like to contribute.

All contributions are appreciated. We all bring our biases to this hobby and that is OK and keeps things interesting.

I have not fully arrived yet. I envisage running Dirac on my server, with lynx MCH straight into 3 x 2 channel DACs, but have not gotten it to work yet, so I am using the Oppo / Vanity into Trinnov as a (hopefully) interim solution. While Trinnov may be better than Dirac, I suspect the benefits of shortening the signal path and running my MCH from a server instead of spinning disc on the Oppo will outweigh the DRC benefits of Trinnov. Besides, I'd lose 2 boxes (Rackspace is at a premium for me), and save a bundle. Most importantly, I would have my MCH accessible on my iPad trough Jriver which outweighs all other considerations.
post #82 of 166
Quote:
Originally Posted by edorr View Post

To be honest, the issue did not come up in my discussions with MSB. I'm in the business of writing checks, listening to music, and swapping gear when I have the upgrade itch. I leave the technical detail to the engineers.

Well some of us are just trying to help you write smaller checks and still get every bit as good performance. But hey, the stuff you buy is very good stuff. I just don't always agree with the marketing claims

My wife is a purse fanatic. I've been to them all many times, Louis Vitton, Prada, Coach, Channel, you name it. But I never heard one of their sales people say these purses offer any more functionality than a Wall Mart $12.99 pleather special. Functionally they do the exact same thing, carry keys, makeup, and cell phone around.

If these ultra high end audio companies would just say basically the same thing I would have no problem. But this junk science and legitimate engineering theory taken way out of context they sometimes use to justify their prices is ridiculous - and quite often borders on fraudulent.
post #83 of 166
Quote:
Originally Posted by Glimmie View Post

Well some of us are just trying to help you write smaller checks and still get every bit as good performance. But hey, the stuff you buy is very good stuff. I just don't always agree with the marketing claims

My wife is a purse fanatic. I've been to them all many times, Louis Vitton, Prada, Coach, Channel, you name it. But I never heard one of their sales people say these purses offer any more functionality than a Wall Mart $12.99 pleather special. Functionally they do the exact same thing, carry keys, makeup, and cell phone around.

If these ultra high end audio companies would just say basically the same thing I would have no problem. But this junk science and legitimate engineering theory taken way out of context they sometimes use to justify their prices is ridiculous - and quite often borders on fraudulent.

The diminishing returns in high end audio are well known. However, unless reviewers are in on the ploy and buyers of the stuff are tonedeaf, outright snobs or both, I still operate under the assumption that spending megabucks on a very highly acclaimed electronics buys you some incremental performance improvement.

The only marketing claim behind the femto 140 clock is that it's accuracy is 140 femtoseconds. There is actually detailed jitter measurement of the MSB galaxy clock posted on their website. Whether any of this translates into audible differences is in the realm of subjective observations, but the engineering detail and measurements are there.

http://www.msbtech.com/products/galaxy.php?Page=platinumHome
post #84 of 166
That link is horribly inaccurate and very misleading !!! I am actually enraged after reading it. They throw out some good engineering terms and explain them completely wrong or twist it to their benefit. Sorry but anyone who doesn't present the units of phase noise in dBc/Hz doesn't get an opinion. Without the units in dBc/Hz i can improve their measured results with 2 resistors.

I also want to point out they used an FFT to measure phase noise. And I am not sure how they came to the conclusion on that -170dB number, my math shows -144dB.
Edited by HFGuy - 11/6/13 at 1:08pm
post #85 of 166
Quote:
Originally Posted by HFGuy View Post

That link is horribly inaccurate and very misleading !!! I am actually enraged after reading it. They throw out some good engineering terms and explain them completely wrong or twist it to their benefit. Sorry but anyone who doesn't present the units of phase noise in dBc/Hz doesn't get an opinion. Without the units in dBc/Hz i can improve their measured results with 2 resistors.

I also want to point out they used an FFT to measure phase noise. And I am not sure how they came to the conclusion on that -170dB number, my math shows -144dB.

If true, I would not be entirely shocked. However, if I asked them to reply they would vigorously defend their charts / measurement and sound just as credible to me. I personally would have no clue what either one of you are talking about. Therefore, I stick with my check writing and listening routine. I got the clock on trial, with no questions asked option to return it if I don't like it. to be perfectly honest, I was not overwhelmed by the much touted improvement with the clock upgrade, but If I ever wanted to resell the DAC, I won't be able to without the clock, so I kept it. I don't lose sleep over this stuff - life's too short and there is too much good music to be listened to.
post #86 of 166
I am by no means saying they make a poor product, or that their goal of designing the lowest jitter clock isn't a worth while goal ( I won't get into the whole debate of ENOB and SNR), but I object to using poor engineering to market. From reading that link they clearly didn't design the clocking circuit, nor would I expect a small audio company to have the resources to do it properly. I wouldn't have said anything if they would just say "hey look here, we are providing a really clean clock", but the second you try to browbeat the uneducated public you had BETTER be 100% accurate.

I am also really not impressed with whole cover up the circuit board in a sexy looking case, let your customers see what you actually did. Reminds me of the companies that have this massive CNC milled aluminum chassis the size of a small dog, and inside it's just an LM3886 or something similar.
post #87 of 166
Quote:
Originally Posted by HFGuy View Post

I am by no means saying they make a poor product, or that their goal of designing the lowest jitter clock isn't a worth while goal ( I won't get into the whole debate of ENOB and SNR), but I object to using poor engineering to market. From reading that link they clearly didn't design the clocking circuit, nor would I expect a small audio company to have the resources to do it properly. I wouldn't have said anything if they would just say "hey look here, we are providing a really clean clock", but the second you try to browbeat the uneducated public you had BETTER be 100% accurate.

I am also really not impressed with whole cover up the circuit board in a sexy looking case, let your customers see what you actually did. Reminds me of the companies that have this massive CNC milled aluminum chassis the size of a small dog, and inside it's just an LM3886 or something similar.

When you pay $10K (or $5K) for a piece of electronics the size of a matchbox you know you are getting shafted...... The optional RS232 interface on their Analog DAC is $1,000. This means the connectivity is all there, and they sell you $50 worth of electronics for a grand to plug into it. The USB interface is over $2K - again, $100 worth of stuff. I think this is sheer brilliance! As long as it sounds good and I can get a reasonable percentage of my money out upon resale I don't care.
post #88 of 166
Audio isn't that bad, a good phase noise analyzer is about $250k, a good BERT can easily run well over that ... and we will only get 1-2 years of use out of it.

As a comparison point we have designs that are 50fsec (16MHZ GPLL) but at 28GHz .... and will end up in products that are under $100.
post #89 of 166
Quote:
Originally Posted by HFGuy View Post

Audio isn't that bad, a good phase noise analyzer is about $250k, a good BERT can easily run well over that ... and we will only get 1-2 years of use out of it.

As a comparison point we have designs that are 50fsec (16MHZ GPLL) but at 28GHz .... and will end up in products that are under $100.

I know that the basic hardware to build very high precision clocks can be had for peanuts. What the $5K - $10K pricepoint is based on I have no clue. Thenagain, dCS, Grimm, Antelope rubicom clocks all cost you an arm and a leg.
post #90 of 166
Quote:
Originally Posted by Kwikas View Post

I'm very interested to know your thoughts on the JRiver / Dirac combo vs the Datasat.

Do you have any thoughts or comments thus far?

I sure do. It is better in every way. I have now had them all for months and there is no comparison. I have now completely removed my Meridian 861 (it's collecting dust). My Theta Casablanca is no hooked up to a video game computer and I no longer have a Datasat (returned it). It was the coolest of all the units, but not worth it's money to me. For many people, this will be the best they can get if you're not prepared (or knowledgeable enough to build your own system and troubleshoot every issue you will have).

There is no doubt it is superior in every way to all the processors listed above. I had them all hooked up in the same room for many months (not minutes or days). The exasound is so smooth and non-harsh compared to all of them it's night and day. I can literally turn up the volume to the point that my ears cannot take any more volume from just pure decibels of sound.

Even my wife said the exasound is better than all of them.

I am using JRiver 19 with Dirac and the Exasound e28 with upgraded clock. There are only a few issues that I have/had with this setup:

1. I am screwed/limited to only 8 channels and I doubt the manufacturer cares enough about 9+ channels to actually setup a driver to use more than one (I have asked many times and I don't get a promising answer, more of an avoiding answer) which makes me believe it will never happen. That gave me serious heartburn for months but finally had to move on and live with it the way it is (for now).

2. He claimed that this DAC worked with Dirac and it was seriously misleading and not quite truthful. In fact it took him months to come up with a solution for me. The DAC would playback through Dirac as a playback device but you couldn't use it as an output device during the recording playback/tone generation process of the room calibration. So this meant getting a totally separate sound card system just for the calibration process and then applying the results to a completely different playback system (the exasound DAC). This pissed me off for months and I pushed and complained and fought until a solution was forced to be found on his part. It's a clunky solution and still misleading but it supposedly works (I still haven't tried it from frustration of how much time I have spent doing his work for him). Therefore I am applying a Dirac filter that was created using an RME Hammerfall DSP Multiface 2 from my recording studio machine. I really don't care if the results are the same or just as good recording/calibrating from one sound card and then playing out through the DAC. I wasn't told I would have to go through all of this to get finished results and it amounted to months of frustration, troubleshooting and waiting.

Computer wise I never had any issues that I wasn't able to troubleshoot very quickly. JRiver took a bit of time to setup and comprehend all the amazing features. It's literally a Datasat in software format. Only once mastered, seemed more powerful and much better of a solution beyond a doubt.

3. 3D movies are an issue for several reasons. First off JRiver doesn't play 3D movies. So I purchased TotalMedia Theater 6 and integrated the 2 of them (so JRiver opened TMT6 when accessing a 3D movie). Problem here is, TMT6 doesn't work at all with ASIO drivers. So that is a fail also.

All in all, I am ok with the few issues I have had at this point. I have learned an incredible amount of information and also able to prove in every way it's a more superior solution to all the above listed. I have owned a meridian for 9 years, Theta for 7 and the Datasat was setup by the manufacturer. And yet the exasound is the clear winner in sound. But in total ease of use, it is the hardest of all solutions as the others require almost no knowledge in comparison. Simply do some speaker calibrations (or better yet pay someone to do it for you) and then pop in a Blu Ray and all is done.

If I had to do this all over again, I would do it all over again and choose the exasound. Now I just need to nag and nag and nag the creator of the exasound to get me a driver that works with two of them at once. The biggest issue he gave me was that being asynchronous and so powerful at maximum resolution there is really no more bandwidth available for a second one on the USB bus. I said well USB 3 has much more to spare (this is a USB 2 interface). I also mentioned you could have one use the USB 2 interface on the board/card and then USB 3 for the other. since they use separate busses.

Hope that helps.

Maestro2be
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Ultra Hi-End HT Gear ($20,000+)
AVS › AVS Forum › Display Devices › Ultra Hi-End HT Gear ($20,000+) › How to replace your home theater pre-pro with a HTPC!