DigiSpeakers - The project Prototype IV - diyAudio
Go Back   Home > Forums > Amplifiers > Class D

Class D Switching Power Amplifiers and Power D/A conversion

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 25th August 2008, 09:33 PM   #1
lineup is offline lineup  Sweden
diyAudio Member
 
lineup's Avatar
 
Join Date: Dec 2005
Location: the north
Arrow DigiSpeakers - The project Prototype IV

Click the image to open in full size.
This is the block diagram. Output: BTL, PBTL speakers
Quote:
The major difference between this prototype and previous versions is the amplifier.
We are still using the TAS5504 as our PWM processor
and the TAS5142 (or TAS5342 if available) as our power stage.

Both are from Texas Instruments.
However, we now configure the output stage as Bridge-Tied-Load (BTL)
and drive the speakers differently via software depending on the product (more on the products in a minute).

At 96KHz and below, the TAS5504 has 4 independent processing channels that can be configured to accept either channel (left or right) from any of the I2S interfaces.
Above 96KHz, the TAS5504 runs out of processing power and can only process 3 channels (see page 11 in this document). The processing pipeline for each channel is impressive.

Each channel has seven bi-quad filters, base, treble, volume, loudness and dynamic range control.
Also, each channel has an output mixer with gain control to mix any of the processing channels at output.
With this flexibility, it is possible, via software, to create several products from the proposed design.
LOGO:
Click the image to open in full size.

Member of DigiSpeaker crew & this forum:
tylerjbrooks

Project internet homepage:
DigiSpeaker - The end of analog music

Prototype IV Description (Block Diagram above):
Prototype IV


_________________________________________

Lineup - watching the end of analog music
Quote:
Bad news on the doorstep...
I couldn't take one more step.
I can't remember if I cried
When I read about his widowed bride

Holly's recent bride was pregnant when the crash took
place; she had a miscarriage shortly afterward.

But something touched me deep inside,
The day the music died.
Recorded and released on the American Pie album in 1971, the single was a number-one U.S. hit for four weeks in 1972.
The song is an abstract story of his life that starts with the death of Buddy Holly ......
__________________
lineup
  Reply With Quote
Old 25th August 2008, 11:56 PM   #2
diyAudio Member
 
Join Date: Aug 2008
Location: Seattle, WA
Default Fun Project

Hey LU. You might have noticed in my Introduction post that I said I had been 'kicking a TAS5142 around my office' for the past year or so. This project is what I meant. My friend and I just recently decided to get a little more serious about it.

I just completed the BOM for the design. It worked out to be a more expensive than I had hoped (I estimate $103 in parts at 1K prices) so I have my work cut out for me.

I made all my footprints. I think I am just about ready to push the autoroute button. Wish me luck!
  Reply With Quote
Old 29th August 2008, 04:35 AM   #3
lineup is offline lineup  Sweden
diyAudio Member
 
lineup's Avatar
 
Join Date: Dec 2005
Location: the north
Default Prototype IV, bill of material, BOM

.
tylerjbrooks has published the current BOM, bill of materials
for Prototype IV
at http://www.digispeaker.com/

I attach PDF of the list of material.
As what I can see all can be bought at DigiKey.
For some discount = lower price.
Quote:
Prototype IV: BOM
Submitted by tylerjbrooks on Tue, 08/26/2008 - 17:04.

The bill of materials for Prototype IV has been complete. Here it is as an OpenOffice spreadsheet.

This BOM needs some work. Right now, the parts come out to be about $103 assuming we get some additional discount from a distributor for doing most of our buying through them. The $103 price is a 12% discount (too much?) from the 1000 part pricing. Mostly from DigiKey.
Attached Files
File Type: pdf prototypeiv_bom.pdf (71.1 KB, 63 views)
__________________
lineup
  Reply With Quote
Old 29th August 2008, 04:34 PM   #4
diyAudio Member
 
Join Date: Aug 2008
Location: Seattle, WA
The schematics for prototype IV can be found in this post:
http://www.digispeaker.com/content/p...%3A-schematics

I am doing the layout for it now.
  Reply With Quote
Old 29th August 2008, 04:42 PM   #5
lineup is offline lineup  Sweden
diyAudio Member
 
lineup's Avatar
 
Join Date: Dec 2005
Location: the north
thanks tyler

i really fancy this project
it is certainly the future of 'perfect audio reproduction'

digital - digital - speaker

and like analog active systems with amplifier in box
this idea has got some real advantages

keep it up, please, my friend
i am watching you & your helper(s)

sorry about the quality of my BOM, parts list PDF,
it is not showing all columns
... but I also provided the link to your website
for more accurate info

regards
Lineup Analog Amplifiers - forum for members only
( already 25 nice members from all over the world! )
__________________
lineup
  Reply With Quote
Old 29th August 2008, 08:19 PM   #6
diyAudio Member
 
Join Date: Aug 2008
Location: Seattle, WA
Thanx LU. My friend and I hope to get a nice sound out of DigiSpeaker. We like the class D amplifiers because they are small and powerful.

Notice that DigiSpeaker is really a general computing platform that happens to have a nice amplifier as a peripheral. Each node has a pretty powerful CPU (Freescale MPC5200), memory, mass storage (SD cards in the Gigabyte range), network (via powerline modem) and home automation interface (via Insteon) in addition to an amplifier.

Our measurements indicate the music decode and playback doesn't really impact the CPU significantly. We hope to use all that extra power for other things. Home automation. Security. Home web site. Media center. ...etc. Jump over to our website and poke around. You will find a bunch of ideas about what to do when the walls of your home are turned into a computing resource.
  Reply With Quote
Old 1st September 2008, 07:07 PM   #7
diyAudio Member
 
Join Date: Oct 2007
Copied from the STA323W thread...

-----------------------

I posted some info on how the STA32X fits into my plans at the following:

Plateamps for active speakers

As you can tell, the STA32X is one of many options I'd like to make available for building active speakers. I've got a number of these amps working, but there's still a lot of work to be done to make them a "product". However, the software was a long pole, and I've made a lot of progress on that recently

------------------------

Neil, I checked out your slide presentation. You are building something similar to our (Tyler and me) Digispeaker, http://www.digispeaker.com, project. Tyler has the STA326 eval board because it is an amp we evaluated and decided not to use.

Digispeaker takes your concept up another level. A Digispeaker needs digital input. It has a powerline modem and can receive this input over the AC wiring in your house. The digital source can be Internet radio, your PC, an A/D input board, or another Digispeaker (they have internal SD cards to hold the music).

The current Digispeaker design supports two models - a low end one which can biamp a pair of speakers and takes 44.1/48Khz input. Or a high end mode where you use two triamped Digispeakers (L/R) and they accept 192/24 input. The amp is a TAS5504 with a TAS5342 output stage.

Digispeakers are networked and work as a synchronized multiroom audio system. In the high end configuration you put two in each room and they are synchronized on left/right channels.

If you poke around on the web site you see that we support Insteon. Insteon lets you replace an existing light switch with a Keypadlinc, http://www.smarthome.com/2486dwh8.html. Buttons on this control the audio source, volume, etc. and your light too. The system is also capable of being controlled by an iPhone/iPod touch or any other IP device.

The full design is on the website. Take a look and let us know what we can do to improve it.

----------------------------------------------------

Originally posted by jonsmirl
Neil, I checked out your slide presentation. You are building something similar to our (Tyler and me) Digispeaker, http://www.digispeaker.com, project.

I looked at your web site, and what you are doing is interesting, but my focus is quite different. I'm more interested in active speakers with built-in digital crossovers, and that is why I need multiple amps and why I'm taking advantage of the biquads. Biquads are a great way to make great crossovers. The presentation is actually part one of two: the possibilities get more interesting with the 6-8 channel amps like the TAS5518 or STA309. Right now I'm working on a very interesting speaker that uses two of the Spherex Xbox amps. The Spherex amps use the DDX-8001 controller--basically the same as the 8-channel STA308.

There is a much older briefing that outlines where I was heading with these amps several years ago and provides some background. Actually, I'm still following the same basic plan: http://home.comcast.net/~neilandbar...akers/N-way.ppt

I agree with the benefits of an all-digital signal path, but I'm not commited to it. The digital amps require a well-designed power supply, because they are basically a power DAC. Also, you can still get lower distortion with a Class-D amp that has a well designed feedback loop. So in addition to the all-digital amps I'm also looking at digital amps that have analog inputs such as the TDA8290 and the AD1994.

Another difference we have is that I'd rather keep the audio networking "open" to allow different solutions. I agree that the powerline audio is a good approach, but so is wireless and wired ethernet. A lot of vendors are coming out with digital media players that you will be able to buy cheaper than you or I could build them, so why not have this be an off-the-shelf module that can be used with standard media server software? For this multi-amp speaker I'm working on I was leaning toward the Zyxel DMA1100P media player, but I haven't gotten that far yet.

BTW, the powerline audio concept is not new: the first wave came and went without commercial success. STT came out with the homeplug audio systems several years ago (see http://www.stt.com.tw/product/spk85.html), although they didn't try to address the high-end market with a full digital path. I've got 3 or 4 of the receiver boards that I scavanged from the Radio Shack Accurian HomePlug speakers, which were a commercial flop. I've also got a transmitter--you are welcome to all of them. The receivers and transmitter are actually the same board with different firmware, and they use the older Intellon 5100 chip (along with MP3 compression to reduce the data rates). It works OK in my house, but I couldn't get it to work at my mom's apartment complex because there were too many dropouts. The 5500 chip should be more robust and the 200M AV chips should be even better, but it's worth noting that powerline audio may not work unless you have an interference-free line.

But even though we have different objectives, we should probably continue to compare notes and look for opportunities to collaborate. About a year and a half ago I set up the Audio Developers web site, but I haven't done anything with it. The concept was to have a single site where multiple developers could sell or share complementary capabilities, such as electronics, software, cabinets, crossover design, measurement services, etc. A key idea was to get members of this "collective" to talk to each other to look for ways they could help--and this could be shared software, PCB design services, or other intellectual property that could help members be successful. If this is interesting to you maybe I need to get this effort organized.

My plan two years ago was to retire and start doing this fun audio stuff for work, but I'm married, and my wife had different ideas. The up side is that we have a nice house in the country...the down side is that I'm still working at a 9-5 job to pay for it.
  Reply With Quote
Old 1st September 2008, 07:09 PM   #8
diyAudio Member
 
Join Date: Oct 2007
More from STW323 thread

quote:
The presentation is actually part one of two: the possibilities get more interesting with the 6-8 channel amps like the TAS5518 or STA309.

Why do you need more channels? The TAS5508 only has three fully capable channels. So does the TAS5518. Those extra five channels don't work at 192Khz. I see from your older project you were trying to drive a dozen speakers.

We are using the three full channels to tri-amp in 192K mode with two Digispeakers (one for L/R). For lower cost use a single one and it can bi-amp at 44.1/48. It can also drive two full range speakers at 192K. The TAS5504 has the biquads needed to do this.

Note that you could use 10 Digispeakers in one room each capable of driving three speakers. You can then program the biquads to drive each speaker anyway you want to. Send 1Khz bands to each speaker, etc.

Digispeakers can take a 7.1 AC3 stream in and pick off individual channels. The mpc5200 CPU can do 1GFLOP and mp3 decoding takes less than 10% of the CPU. ac3 decode takes about 25%.

We're getting ready to start looking at speakers. We need to locate ones that are matched to our amps capabilities.
quote:
The digital amps require a well-designed power supply, because they are basically a power DAC.

Power supply is a 90W, L6599 based half-bridge LLC adaptor with very stable output voltage and low EMI. It has a power factor correction front end (L6563 based) so it complies with the most stringent regulations. It's 92% efficient at full load. Power supply volume control is also implemented. This is a complex power supply to design.

What would be an alternative amplifier with a feedback loop?
quote:
Another difference we have is that I'd rather keep the audio networking "open" to allow different solutions.

The boards have Ethernet on them too. It can be let unpopulated to reduce costs ($8).
There is also a USB jack which can be populated ($5). Plug an 802.11 wireless stick into it.
quote:
A lot of vendors are coming out with digital media players that you will be able to buy cheaper than you or I could build them, so why not have this be an off-the-shelf module that can be used with standard media server software?

The boards have a SD card socket. They understand SDHC and can take 8GB+ plus cards. To your PC they just look like remote disk drives. Copy your tunes into the SD cards and now you don't need a media server. The Digispeakers know about each other and can play tunes located on other nodes. Of course we can source from a media server or stream too.

If you want to play vinyl, digitize it first. Use a 192/24 DAC. Won't wear out your vinyl this way.
quote:
Powerline works OK in my house, but I couldn't get it to work at my mom's apartment complex because there were too many dropouts.

The 85Mb is more noise resistant that the older 14Mb.

To get around drop outs, the nodes use the 85Mb net to stay ahead of the stream. For example they may copy the next 4-5 songs on the play list into RAM on all of the nodes before synchronously starting to play them. Same for a net stream, it get buffered and then synchronously played. You can turn this off for live events.

In general the music is not streaming around the net in real-time. It moves around faster than real-time and highly synchronized clocked are used to make everything play at the same time.

If it is really a problem switch the node to Ethernet or 802.11.

----------------------------------------------------------

quote:
Originally posted by jonsmirl


Why do you need more channels? The TAS5508 only has three fully capable channels. So does the TAS5518. Those extra five channels don't work at 192Khz.

I'm just using SPDIF, so I don't care about 192KHZ audio. At 44.1KHz the TAS5518 provides 8 channels of processing. I'm using the other amps for flexibility or reinforcement (multiple amps per frequency band). I've got a GUI that lets you pick different configurations plus multiple amps per channel (for example, two amps for a bass channel to drive two bass speakers). The current pick list is:

2-way
2-way stereo
2-way stereo plus sub
3-way
3-way stereo
3-way stereo plus sub
4-way
4-way plus sub
5-way
6-way

The speaker I'm working on right now is a five-way with 2 amps for the sub (it actually uses 4 8" subs, two drivers per amp). So I'm using two 2 6-channel amps; one per speaker. The ST308 actually has 10 biquads per channel, so that's a lot of processing available for crossovers and EQ. I'll post some pictures when I get the nerve--it is pretty weird
You can play with an early version of the GUI--the code is at:
http://www.audiodevelopers.com/plat...ultichannel.zip
quote:

Power supply is a 90W, L6599 based half-bridge LLC adaptor with very stable output voltage and low EMI. It has a power factor correction front end (L6563 based) so it complies with the most stringent regulations. It's 92% efficient at full load. Power supply volume control is also implemented. This is a complex power supply to design.

90W may not be adequate. My approach was to use an unregulated main supply and then use multiple local switchers like the LT1074 for each amp. The LT1074 is very simple and it easily puts out 100W. The picture shows an early prototype that used a TI evaluation board with the LT1074 board next to it. There is a toroid on the other side of the mounting plate
quote:

What would be an alternative amplifier with a feedback loop?
There are quite a few digital amps with analog input that use feedback to lower distortion and provide good PSRR. Philips, MPS, Sanyo, Tripath (R.I.P.), UCD, etc. The only amps that don't use analog feedback are the ST DDX, TI Purepath and the Cirrus stuff.
quote:

The boards have Ethernet on them too.
The point I was trying to make is that the digital media player market is maturing, and there are many benefits to leveraging what comes out of this technology. So I think it makes a lot of sense right now--at least for my application--to keep this a separate module rather than integrating a specific implementation into the design. Your design shows the powerline network chip on the same board as the amplifier, and I was trying to get you to think about that . I was planning on using an off-the-shelf audio server to send linear PCM across the network to a media player, which would provide SPDIF to the digital amps. It's an all-digital path with no compression, and I can buy this capability from multiple vendors.

But it sounds like your objectives are substantially different, in that you are trying to send 192KHz audio and you have a networking concept that isn't supported by the current generation of audio servers and media players. That's a level of complexity that I'm not up to right now, but if you can make it work it should be a unique and promising capability. There may even be a good market for high-end networked home audio installations that need this capability, and you might just be at the right place at the right time to benefit.

Anyway, it sounds like we've got some complementary work going on here. I've got a head start on using the biquads to implement crossovers that could probably help you out. And I've got some smaller digital amps that could fill out a product line with some low-cost amp alternatives. But I don't have anything comparable to your networking and AC3/MP3 decoding (other than off-the-shelf media players), and I don't have the programming skills to deal with the huge amount of open source audio processing and networking code. So it might make sense to get together and think through a strategy for working together (?).
  Reply With Quote
Old 1st September 2008, 07:10 PM   #9
diyAudio Member
 
Join Date: Oct 2007
Last piece from STW323 thread

quote:
90W may not be adequate.

Note that is 90W of DC power, not 90W of audio power. We can change the power stage to the 310W chip and build a new power supply if needed. We are designing for whole house audio applications, not 7.1 movies.
quote:
There are quite a few digital amps with analog input that use feedback to lower distortion and provide good PSRR. Philips, MPS, Sanyo, Tripath (R.I.P.), UCD, etc.

Do any of these have digital input? I'm the software guy, not the EE. Why do the TI designs not use feedback? Is it because of the support for power supply volume control (PSVC)? We've done some testing and PSVC makes a huge difference at low volumes.
quote:
The point I was trying to make is that the digital media player market is maturing,

You can plug an iPod into the USB port. The integrated media server feature comes almost for free. The nodes are running Linux. It costs $1 to add the SD card socket. Later we'll write software to transparently sync the contents of your iPod onto the SD cards.
quote:
It's an all-digital path with no compression, and I can buy this capability from multiple vendors.

Each channel of uncompressed audio in the system utilizes about 2Mb of bandwidth. If you are running 6 channels that is 12Mb. That's why 14Mb powerline doesn't work get a little noise and you can hear the drop outs. That is same same reason Sonos has switched from 802.11G ($20) to 802.11N ($70). Powerline is about $25.

For your application get a NSLU2. Find a USB 1.0 hub (not a 2.0 one, of if it is 2.0 make sure it is multi-TT). Buy several USB audio adpaters with SPIF out like this one. NSLU2 can support 4-5 of these. Load Linux on to the NSLU2. Add on a USB 2.0 external drive, get SATA or IDE. Put a 1TB disk in it. For under $400 you have a 1TB audio server capable of serving four zones. It can also funcition as a media server and server videos/stills to something like a DSM-520 I'm currently using this setup in my house.
quote:
Your design shows the powerline network chip on the same board as the amplifier, and I was trying to get you to think about that

After reading some TI app notes, they are suggesting it is better to put the amp on the power supply board. That might make sense, you could then have 100W, 300W, etc boards.

Powerline support can be left unpopulated. It will be the subject of some intensive testing from us. The main reason for powerline support is to dramatically lower the cost of retrofiting wholehouse audio into existing homes. Instead of running audio wires to a central location, just look around in the ceiling until you can find some AC. In my last house it was $8,000 for the audio wiring not including any electronics. You are looking at the prototype board designs, we haven't decided on a final form.
quote:
But I don't have anything comparable to your networking and AC3/MP3 decoding

I have been doing software over 20 years. I work on the Linux kernel and used to work in the Microsoft networking group. The mpc5200 with floating point was picked to make hacking easy. We could have used an ARM chip without FP and saved $8. But then you would need to write audio codecs in fixed point assembly. Fixed point assembly codecs are available but not many people have the skills to work on them.
  Reply With Quote
Old 1st September 2008, 07:48 PM   #10
diyAudio Member
 
Join Date: Oct 2007
From st thread

quote:
Originally posted by jonsmirl

Note that is 90W of DC power, not 90W of audio power.
So with an 80% efficient amp you'll get about 70W of audio
quote:
Do any of these have digital input? I'm the software guy, not the EE. Why do the TI designs not use feedback?
The class-D amps with feedback are usually analog input. The one exception that I know about is the TAS5706, which uses a closed loop power stage to increase PSRR. It looks like TI is starting to integrate this approach into its power stage product line.

The article that created a backlash against the all-digital amps is the one by Putzeys. It's a bit dogmatic but it has some well-taken points. My response is to provide a wide range of class D products, to appeal to all audio religions

quote:
You can plug an iPod into the USB port. The integrated media server feature comes almost for free. The nodes are running Linux. It costs $1 to add the SD card socket. Later we'll write software to transparently sync the contents of your iPod onto the SD cards.

Cool stuff, but I don't know whether people will buy it. Or whether someone will come out with a really cheap version that will make the DIY stuff OBE.
quote:

Each channel of uncompressed audio in the system utilizes about 2Mb of bandwidth. If you are running 6 channels that is 12Mb. That's why 14Mb powerline doesn't work get a little noise and you can hear the drop outs.
The Accurian speakers that I was playing are strictly 2-channel, MP3 compressed, so very low bitrate. There were dropouts because the original homeplug technology wasn't refined. I've seen some reports on the newer powerline networking products and the newer standards perform much better. But all of these standards need to be applied with care, using suitable buffers and some attention to the operating environment. For example, Bluetooth audio works great until you turn on the microwave.
quote:
For your application get a NSLU2...I'm currently using this setup in my house.

Maybe someday. Windows Media Player recognizes the digital media players as devices and that's OK for me in the short term.
quote:

After reading some TI app notes, they are suggesting it is better to put the amp on the power supply board. That might make sense, you could then have 100W, 300W, etc boards.
Agreed. One of the TI app notes shows the distortion graphs for using a relatively short length of wire between the power supply and the amp--it's surprizing how much benefit you get from having the power supply close by. Also, because of the PSVC, the amp and power supply become more tightly coupled. So it makes sense to put these on one board. But I wouldn't get carried away with PFC. I don't think anyone in the audio community is too concerned about that right now.
quote:

I have been doing software over 20 years. I work on the Linux kernel and used to work in the Microsoft networking group.

My background is hardware and systems but I've got enough programming skills to be dangerous (mostly assembly code). My approach is to let the biquads do the DSP, so I can get by with a small controller on the I2C bus to control the chips. All of the number crunching is done in VB.net, and I just sent the results to the controller for storage and loading into the amps. I'm able to hack away at this stuff because the tools are good, but it's not something I enjoy or something that I am good at. What you are doing with the mpc5200 is out of my comfort zone by a lot.

There are always a lot of tradeoffs between cost, development time and capabilities, but if you can get a diverse group to work collaboratively you don't have to make as many compromises in the design. That's why I keep suggesting some additional dialogue to look for ways to help each other. I'm getting ready to disappear for a while, working long hours on a proposal in Seattle, but maybe after things slow down...
  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
Digital X-over Prototype blu_line Digital Source 5 8th May 2010 03:44 PM
my new prototype zilog Power Supplies 5 25th September 2008 06:03 PM
Guitar Amp prototype SS TDA2050 project brucevicknair Chip Amps 0 29th May 2007 08:27 PM
prototype PCB's in the US mikelm Solid State 4 6th September 2006 02:20 AM


New To Site? Need Help?

All times are GMT. The time now is 08:22 PM.


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