High Resolution Multi-Channel Digital Interface - Page 2 - diyAudio
Go Back   Home > Forums > Source & Line > Digital Source

Digital Source Digital Players and Recorders: CD , SACD , Tape, Memory Card, etc.

Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Reply
 
Thread Tools Search this Thread
Old 15th November 2003, 06:43 PM   #11
Thunau is offline Thunau  United States
diyAudio Member
 
Thunau's Avatar
 
Join Date: Mar 2001
Location: PA USA
All I can say, wow! You are doing some really cool work there. Sending multiple channels of PCM data directly to the digital amp is something very desireable. I work in pro audio world. We are using DSP processors that can network with eachother using a single CAT5 cable. They can usually carry 8 channels of 24 bit audio at 96kHZ or 16 channels at 48kHz. My guess is, they are using the same chips that you are utilizing for your TX/RX duties.

What would be awesome is the ability to play back a CD or DVD in a computer, do a digital crossover in the same computer, come out of Ethernet card into a digital amp with a CAT5 connector and your receiver board. Some folks on this very board are already writing applications for digital crossovers inside a PC. It seems like a perfect logical step. The question is how to make an Ethernet card output raw PCM. I guess a special driver or a translating application would be needed.

BTW, how does the Equibit amplifier do volume adjustments?
__________________
"Most people just say what they know, the wise ones know just what to say."
  Reply With Quote
Old 17th November 2003, 02:01 PM   #12
diyAudio Member
 
Join Date: Aug 2002
Location: Wisconsin
Quote:
Originally posted by Thunau
I work in pro audio world. We are using DSP processors that can network with eachother using a single CAT5 cable. They can usually carry 8 channels of 24 bit audio at 96kHZ or 16 channels at 48kHz. My guess is, they are using the same chips that you are utilizing for your TX/RX duties.
They probably are doing something different. Having a proprietary interface doesn't require as much flexibility.

Quote:
What would be awesome is the ability to play back a CD or DVD in a computer, do a digital crossover in the same computer, come out of Ethernet card into a digital amp with a CAT5 connector and your receiver board. Some folks on this very board are already writing applications for digital crossovers inside a PC. It seems like a perfect logical step. The question is how to make an Ethernet card output raw PCM. I guess a special driver or a translating application would be needed.
Ethernet would be completely different from what I am doing, even though it can use a CAT5 cable.

The approach I described would easily work by tapping into the I2S lines of a computer sound card. The biggest problem with using a soundcard for a digital crossover is that most non-professional units are limited to eight channels. That's enough for stereo, but not multichannel.

I'm planning on focusing on a non-PC approach. I think that FPGAs (and their ability to accommodate large parallel data paths) are the ideal platform to implement massive time-domain FIR filters. How's this for a thought: one second of 192KHz fs requires 192000 taps per channel (to correct down to 1Hz). For 24bit data, each tap should be 72bit to avoid rounding errors and provide headroom for normalization. Six two-way speakers plus two subs will require fourteen channels (20channels for three-way). That's 2688000 72bit taps and 2688000 24x24 multiplies per second! This is the benchmark goal I have for a true brute force FIR without the use of FFTs.

More near term, I'm taking a less ambitious intermediate step using TAS3103 digital audio processor chips to implement an all-digital solution, including crossovers. This will limit me to IIR filters and digital delays for driver time-alignment. I think this should still be quite an improvement over what I have now. I need to get some practice with MLS measurement and correction techniques. Once I have that in place, along with a full complement of digital amps, it should be much easier to transition to the FIR approach.

Quote:
BTW, how does the Equibit amplifier do volume adjustments?
They can use either digital attenuation, variable power supply voltage, or a combination of both.

Regards,
Brian.
  Reply With Quote
Old 18th November 2003, 05:10 PM   #13
diyAudio Member
 
Tieftoener's Avatar
 
Join Date: Nov 2002
Location: Coastal AL
I just have to let you know that I'm REALLY impressed! And to let you know that we appreciate you sharing your knowledge and experience with this new concept that makes a lot of old cronies (and even a youngster like me) who are attached to a "pure" analog signal take a step back. There's a lot of talk with this type of approach, but not a lot of action. Kudos to you on a fanciful amount of labor of love. Even if the system sounded like crap, you should still be proud that you managed to integrate the whole thing on such a universal level. Outstanding work.... simply outstanding!
__________________
Tieftoener -- You're ears can sense a movement in air that causes your eardrum to move less than 1/10th of the diameter of a Hydrogen atom! Don't abuse the one of the most amazing organs your Creator gave you!
  Reply With Quote
Old 3rd December 2003, 06:58 PM   #14
diyAudio Member
 
Join Date: Feb 2002
Outstanding stuff, Brian. Very elegant.

Coupling this with an XR-25 or 45 for 6 full digital channels has some very attractive possibilities.

Do you have any interest/intent of doing any small production runs of the boards? It looks like you've provided enough artwork to get the pcb's done at any of the online places, but assembly by us duffers would still be a challenge.

Time for me to try to reverse-engineer my Delta 1010 expansion connector. If I can reliably isolate the I2S lines, I might have to try this.
  Reply With Quote
Old 4th December 2003, 07:06 PM   #15
diyAudio Member
 
LineSource's Avatar
 
Join Date: Sep 2003
Location: SiliconValley
Default Digital Room Equalization at EACH speaker

Digital domain room equalization is impressive when properly executed. Several topologies are possible based upon how you answer the questions:

Where do the digital amps go?

Where does the digital crossover go?

Where does the digital room equalization go?

Can digital room equalization optimize several speakers simultaneously for an optimized summation? or just one speaker at a time against the room? This last question seems the most challenging. Do algorithms exist for optimizing seveal speakers simultaneously to get the best summation at the listening position?


My current thinking is to put the digital amp + crossover + room equalization at each speaker, most likely serially connecting a PC to each speaker to calculate and download room equalization until this algorithm can be built into each speaker's DSP. Separately equalize each speaker against the room.

Has anyone tackled digital room equalization?
  Reply With Quote
Old 4th December 2003, 07:42 PM   #16
Henckel is offline Henckel  Denmark
diyAudio Member
 
Join Date: Sep 2002
Location: Copenhagen, Denmark
Default Re: Digital Room Equalization at EACH speaker

Quote:
Originally posted by LineSource
.

Has anyone tackled digital room equalization?

Sure - look here for a guide:

www.mooneyass.com/DRC/DRC%20Guide%20v1.0.pdf

And the program called DRC can be found at;

http://freshmeat.net/projects/drc/?topic_id=114

Well worth the effort to get it working

Br

Morten
  Reply With Quote
Old 4th December 2003, 11:58 PM   #17
diyAudio Member
 
Join Date: Aug 2002
Location: Wisconsin
Quote:
Originally posted by dwk123
Coupling this with an XR-25 or 45 for 6 full digital channels has some very attractive possibilities.

Do you have any interest/intent of doing any small production runs of the boards? It looks like you've provided enough artwork to get the pcb's done at any of the online places, but assembly by us duffers would still be a challenge.
I have thought about it. I don't want to make a regular business out of making this type of thing, but it would be nice to sell enough to recover some of my development cost.

In this particular case, I feel that patching these boards into existing commercial products is a bigger challenge than soldering the parts onto the board. It didn't seem to me that anyone who could figure out where the I2S signals were located and could tap into them would care about having the board preassembled. I think that if I tried to sell these, the amount of support people would probably want would be overwhelming. This is a hobby and I want to keep it fun.

In the future, I've thought about designing boards that would just require plugging in. Perhaps a receiver board for the Panasonic SA-XRxx units, and a transmitter board that could plug into the I2S lines of a six or eight channel sound card. These types of devices could be made so that they are fairly straightforward to use for people that don't want to make and troubleshoot their own circuitry.

Unfortunately, I think that interfacing to any CD, DVD-A, or SACD player will always require tapping into PC board vias, traces, pads, and/or IC pins. Every player would need to be individually figured out. Careful layout and routing of the interface wires is necessary to prevent signal ringing and reflections.

Right now, I want to focus my time on new circuits. This interface was just the first step of my new all-digital system goals. I'm presently brewing up a multi-channel digital processor based on the TAS3103.

In the meantime, here's a few more details for anybody that might be interested in trying this on their own:

******************************

All of the parts for the PCM MuxIt Transmitter and PCM MuxIt Receiver are available from Digikey (prices shown are US$ at time of posting).

PCM MuxIt Transmitter BOM:
(1) 296-9737-5-ND.................SN65LVDS150PW MuxIt PLL.............$06.160
(1) 296-9738-5-ND.................SN65LVDS151DA MuxIt Transmitter...$05.020
(6) P49.9HCT-ND...................49.9ohm 0603 resistor.....................$00.090
(2) P100HCT-ND....................100ohm 0603 resistor......................$00.090
(1) P10.0KHCT-ND..................10Kohm 0603 resistor......................$00.090
(6) PCC1762CT-ND.................0.1uF 16V 0603 cap........................$00.103
(6) PCC2250CT-ND.................4.7uF 16V 1206 cap........................$00.594

TOTAL: $16.72


PCM MuxIt Receiver BOM:
(1) 296-9737-5-ND.................SN65LVDS150 MuxIt PLL...................$06.160
(1) 296-9739-5-ND.................SN65LVDS152DA MuxIt Receiver........$05.020
(3) 296-15234-5-ND...............SRC4192IDB 192KHz ASRC.................$15.740
(1) ECS-P53-B-ND (30.0 MHZ)..30.0MHz 3.3V oscillator....................$08.400
(1) ZXCM209TFCT-ND.............ZXCM209TFTA 3.08V Reset generator.$00.710
(5) P49.9HCT-ND...................49.9ohm 0603 resistor.......................$00.090
(3) P100HCT-ND....................100ohm 0603 resistor........................$00.090
(6) P10.0KHCT-ND.................10Kohm 0603 resistor.........................$00.090
(12) PCC1762CT-ND...............0.1uF 16V 0603 cap..........................$00.103
(11) PCC2250CT-ND...............4.7uF 16V 1206 cap..........................$00.594

TOTAL: $76.54

*********************************

I had my bare boards made at Advanced Circuits with their Bare Bones PCB service:
http://www.barebonespcb.com

I've had excellent results with their service. The boards don't have soldermask or silkscreen, but the price is great, especially for low volume boards. This service has one day turn time as standard! Now that this is available, I rarely do any hand-wired protos.
They charge a $10 handling fee, plus $5 or more for shipping (depending on the delivery option you choose).

The Gerber zip files I posted earlier can be directly loaded into the Bare Bones website.

Use the following info:

PCM MuxIt Transmitter:
X dimension: 0.864"
Y dimension: 1.272"
.zip file name: MX2.zip
Part name: MX
revision: 2
The price for one board is $25.55. Additional boards are then $0.55 each.

PCM MuxIt Receiver
X dimension: 2.034"
Y dimension: 1.472"
.zip file name: MR2.zip
Part name: MR
revision: 2
The price for one board is $26.50. Additional boards are then $1.50 each.

*********************************

I use 30AWG wire-wrap wire to tap into the I2S signals.

Generally it's best to connect as close as possible to the IC that generates the signals. Reflections are usually less of a problem with this approach. This is especially true for the MCLK signal.

I tap a ground wire as close as possible to each point that I tap a signal from the board. I then twist the wires together to minimize the loop-area impedance of the signal and its return ground path. This is very important to preserve the signal quality and to minimize ringing. If two or more signals being tapped are close to the same ground wire tap point, they can share and be twisted together with the same ground wire. Also keep all of the wires as short as possible.

Some examples of this can be seen in the pictures I posted, although much of the wiring is hidden under the boards (to keep the length short).

*********************************

These boards both use a 3.3V supply. Many times you can use a 3.3V supply from the unit you are tapping into. I don't have the measured current draw for these boards handy (I'll try to post it later). Some units are sensitive to having additional power drawn from their 3.3V supply.

For example, if the 3.3V supply is generated off a 5V supply by a small surface mount linear regulator, it may not have enough capacity.

Another case to watch out for is if the 3.3V supply is the regulated leg of a multitap switching supply. In this situation, extra current draw on the 3.3V supply will be accommodated, but the other supply voltages will increase beyond their spec.

Some units I've tried seem to have plenty of reserve (I've had good luck with Panasonic). Others seem to have their supplies closely engineered to accommodate the original circuitry and nothing more (this was the case on a Sony I did).

If necessary, you can put in a separate 3.3V supply.

*********************************

I always recommend getting the schematic and other service documentation for the unit you are modifying.

*********************************

For the transmitter, it's probably best to tap into the I2S signals that feed the disk player's DAC. It isn't necessary to disable the DAC. If a disk player sends the I2S signals through a separate base management / surround processing chip, then you have the option of tapping into them before they get this extra processing (my preference).

For the receiver, there can be a number of different places to tap into. In the case of the SA-XR10 that I am using, there are two choices:
a.) The connector going directly to the digital amplifier board.
b.) Tapping into the signals that originally came from the ADC (this approach requires that the ADC be disabled).

Initially, I'm using the second approach. This allows me to use the receiver's volume control. Once I have my digital processor working, I plan to feed the amp directly.


*********************************

I have a couple hundred hours of listening to this interface now. I have to say that I'm very pleased with it.

I just wish I could get new stuff done faster.

Regards,
Brian.
  Reply With Quote
Old 6th December 2003, 01:39 AM   #18
diyAudio Member
 
Join Date: May 2002
Location: Melbourne, Australia
Send a message via MSN to Mark Hathaway
Default AMAZING

Howdy Brian,

well done, that is some very fine work. I've been using the ad1896 myself in a variety of projects, I love it. I strongly suggest you loose the simple oscillator circuit for the ad1896's and adopt something like the Kwak clock, the resultant jitter from the ad1896 is only governed by the jitter on your master clock source. The sonic differences are quite amazing. Like everything else the ad1896 likes very clean separate supplies.


I'm currently designing a board to slip into my dvd-a player to give me 3 aes/ebu streams, ad1896 and tx chips, planning to use the ad1853 for analog outs as well. I found a dvd-a player with a Zoran Vaddis 5 chip for less then 200 USD, getting any info from Zoran is impossible, but I managed to get a datasheet from AVS Tech, the manufacturer of the 6 channel dac chip. Standard form its nothing exciting, switch mode psu and all, but a good bunch of power supplies, a decent clock ( uses 27 Mhz, mpeg clock freq, makes it easy) and a digital out and it'll be a rocket. Does dvd-a MLP. I've not looked inside many other players, so I'm not sure what all-in-one chips they use, but I'm guessing Zoran has a big market share.

How good is the Equibit amp? Have you compared it to anything? Tripath? a really good analog amp? e.g current feedback, no odd order distortion, 900khz bandwidth. I've a schematic if your interested. I only say this as the topology I have is as good if not better then Tact Audio's Millennium amp (equibit implementation, back in the toccota (sp?) days) which is considered a very good digital amp.


As you have said, IEEE-1394 is not a nice monster. From what I've heard and read, it sounds like crap compared to aes/ebu

Best bet is doing away with a tx/rx setup completely, or using the aes/ebu or your MixIt methods. Have you considered merging source and amp into one box and doing away with any transmitter/receiver setup? Players decoder into ad1896(s) into Equibit chip(s)


Once I've got this going I'll tackle a sacd player, again starting with an inexpensive implementation and adding to it. I haven't looked into sacd players for a few years now, but from what I'd seen, Sony was using the same chips everywhere and changing psu and component quality to differentiate models.


I'm following the DAX Groups project with close interest, a dvd-a and/or sacd player straight into a digital amp would be great. So far I think great analog amps are still a little in front of digital amps, but I'd love to be proven wrong.


Mark Hathaway
  Reply With Quote
Old 6th December 2003, 07:05 PM   #19
diyAudio Member
 
Join Date: Aug 2002
Location: Wisconsin
Default Re: AMAZING

Thanks, Mark.
Quote:
Originally posted by Mark Hathaway
I strongly suggest you loose the simple oscillator circuit for the ad1896's and adopt something like the Kwak clock, the resultant jitter from the ad1896 is only governed by the jitter on your master clock source. The sonic differences are quite amazing. Like everything else the ad1896 likes very clean separate supplies.
I'm using the ASRC with both the input and output ports in slave mode, so the oscillator is only running the internal ratio estimation logic. This allows me to keep the critical low-jitter MCLK for the DAC (in my case a digital amp) completely off of the interface board.

If the ASRC is an output master, then the quality of the oscillator is much more critical (as you are describing).
Quote:

I'm currently designing a board to slip into my dvd-a player to give me 3 aes/ebu streams, ad1896
The problem with this approach is that there will be three different recovered clocks from the separate AES/EBU streams. They will be asynchronous to each other. There also will be three different group delays. Once this happens, there isn't any practical way to get the three streams phase-aligned with each other again.
Quote:
How good is the Equibit amp? Have you compared it to anything? Tripath? a really good analog amp? e.g current feedback, no odd order distortion, 900khz bandwidth. I've a schematic if your interested. I only say this as the topology I have is as good if not better then Tact Audio's Millennium amp (equibit implementation, back in the toccota (sp?) days) which is considered a very good digital amp.
I haven't heard the Millennium yet, but it seems to have a very good reputation.

For a couple hundred bucks, I think the Panasonic SA-XRxx receivers are surprisingly good. These are the only Equibit units that I have heard so far. Since they don't represent an optimal Equibit implementation (especially the power supply), I haven't been able to experience the full potential of TI's Equibit chip sets. However, my expectations are very high.
Quote:
As you have said, IEEE-1394 is not a nice monster. From what I've heard and read, it sounds like crap compared to aes/ebu.
My main concerns with IEEE-1394 is that it is very complicated to implement, and it's also likely that high resolution audio will be encrypted.

IEEE-1394 fully buffers the data at the receiving end. The master clock can then bring the data directly out of the buffer. The quality should be as good as the clock and the rest of the circuitry that is downstream from this point.

AES/EBU uses a recovered clock (and it doesn't support multichannel unless some type of compression is used). For this reason I think that IEEE-1394 offers the potential for much higher quality.

Actual implementations will vary, of course.
Quote:
Have you considered merging source and amp into one box and doing away with any transmitter/receiver setup?
That certainly would be one very good and practical way to implement a final design.

At this point, I'm not even close to achieving a full system design. One of the reasons I was compelled to make this interface was that I wanted the flexibility to work on one section at a time.
Quote:
I'm following the DAX Groups project with close interest, a dvd-a and/or sacd player straight into a digital amp would be great. So far I think great analog amps are still a little in front of digital amps, but I'd love to be proven wrong.
I feel that digital amps are already competitive with more traditional DAC / analog amp designs (for a given price). Right now there seems to be a big gap between consumer products (like Panasonic) and the high-priced stuff (like TacT). I suspect that most of the people here will be concerned with how this gap fills out.

Regards,
Brian.
  Reply With Quote
Old 6th December 2003, 07:30 PM   #20
diyAudio Member
 
Join Date: Aug 2002
Location: Wisconsin
Exclamation Transmitter Board Correction

The dimensions I posted for the PCM MuxIt Transmitter board were incorrect.

Use this information instead of what I previously posted:

PCM MuxIt Transmitter:
X dimension: 0.952"
Y dimension: 1.360"
.zip file name: MX2.zip
Part name: MX
revision: 2
The price for one board is $25.65. Additional boards are then $0.65 each.

I'm sorry about the error.

Brian.
  Reply With Quote

Reply


Hide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
High resolution image of Levinson digital output board Dr.H Digital Source 11 31st October 2010 03:08 AM
Multi-channel PC audio interface - options? error401 Digital Source 6 5th March 2008 10:32 AM
Modify an A/V receiver to provide multi-channel digital pre-out mhakman Digital Source 3 10th April 2007 11:12 PM


New To Site? Need Help?

All times are GMT. The time now is 04:13 AM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright 1999-2014 diyAudio

Content Relevant URLs by vBSEO 3.3.2