YAY!
Finally after many years I have fully active ZDT 3.5 speakers with custom DSP box. Just "finished" today.
Some details, the DSP box is my PCM4202 DAC, with the ADAU1452 DSP, and the ADAU1966 DAC (despite it's problems), as on my website/gitgub (although needing more documentation).
There is a passive attenuator box so I don't blow up my speakers or amps and can run the low level side with the highest voltage level possible, because why not!
So far my tuning has been to approximate the ZDT 3.5 crossovers/transfer functions in the ADAU1452 using whatever I can find in SigmaStudio to do the job. So it needs a little fine tuning. But it still sounds awesome.
Everything after my Yamaha amp, which is acting as a preamp, has been built or fettled by me. Both the Cyrus One and Rotel (6 channel amp) have been completely gutted and refurbished (including every solder joint being reflowed as most on the Rotel were dry).
Finally after many years I have fully active ZDT 3.5 speakers with custom DSP box. Just "finished" today.
Some details, the DSP box is my PCM4202 DAC, with the ADAU1452 DSP, and the ADAU1966 DAC (despite it's problems), as on my website/gitgub (although needing more documentation).
There is a passive attenuator box so I don't blow up my speakers or amps and can run the low level side with the highest voltage level possible, because why not!
So far my tuning has been to approximate the ZDT 3.5 crossovers/transfer functions in the ADAU1452 using whatever I can find in SigmaStudio to do the job. So it needs a little fine tuning. But it still sounds awesome.
Everything after my Yamaha amp, which is acting as a preamp, has been built or fettled by me. Both the Cyrus One and Rotel (6 channel amp) have been completely gutted and refurbished (including every solder joint being reflowed as most on the Rotel were dry).
Attachments
Wow, so I replaced the (questionable and suddenly obsolete) ADAU1966 DAC board with an AK4468 board, in my DSP box in post 41.
Specifically it is my AK4458 board fitted with an AK4468 chip as they are pin compatible. This allows me to run the PCM4202 and AK4468 chips in 192kHz/24bit mode.
This thing sounds awesome.
First song I listened to gave me goosebumps all over (and more...). Maybe a blind listening test would show no difference, but in any case this setup rocks.
I am still using earlier revision boards for testing. Need to do some more design changes, and update to latest designs. Would make fixing some reset issues I have easier.
But this setup is really good, beats my MiniDSP boxes. Sounded like it was raining in the room with the MiniDSP 2x4 boxes.
Specifically it is my AK4458 board fitted with an AK4468 chip as they are pin compatible. This allows me to run the PCM4202 and AK4468 chips in 192kHz/24bit mode.
This thing sounds awesome.
First song I listened to gave me goosebumps all over (and more...). Maybe a blind listening test would show no difference, but in any case this setup rocks.
I am still using earlier revision boards for testing. Need to do some more design changes, and update to latest designs. Would make fixing some reset issues I have easier.
But this setup is really good, beats my MiniDSP boxes. Sounded like it was raining in the room with the MiniDSP 2x4 boxes.
I'm looking for ways to build a DSP-driven active speaker, and I ended up finding your github repository (wow!), plus all the work you did.
I'm tempted to build your AK4458 board, but it looks as if the AK-family of chips, at least in the USA, is backordered and at least 18 weeks out.
My fall back is to consider at cheap AD1938 boards on Aliexpress/eBay or even your AD1939 board. But if I can find a pre-built DAC board, that would be my preference: as much as I can assemble an SMD board, I don't particularly enjoy the process (I only have a soldering iron/hot air station, no reflow oven)
Do you have any other suggestion for DACs worth looking into? Is the AK4468 so much batter than an AD1938/9?
I'm tempted to build your AK4458 board, but it looks as if the AK-family of chips, at least in the USA, is backordered and at least 18 weeks out.
My fall back is to consider at cheap AD1938 boards on Aliexpress/eBay or even your AD1939 board. But if I can find a pre-built DAC board, that would be my preference: as much as I can assemble an SMD board, I don't particularly enjoy the process (I only have a soldering iron/hot air station, no reflow oven)
Do you have any other suggestion for DACs worth looking into? Is the AK4468 so much batter than an AD1938/9?
My view is that a lot comes down to the way the DAC is implemented and supporting circuitry, especially power supplies and clocks. A lower spec DAC implemented properly would be better than a high end chip implemented poorly.
My designs use good quality regulators on the analogue supply locally on each board and other supply filtering. With clock and signal traces I have impedance matched (roughly) the PCB traces and ribbon cables (26AWG wire size), and also tuned the termination resistors. You can feed in any high quality clocks you want, there are clock headers on ADC/DAC boards for this, as long as it meets what the IC needs.
Biggest change using the AK4468, instead of the ADAU1966, has been increasing sample rate on my crossover box to 192kHz (from 48kHz). So that might be why it sounds "better”, although I do have more background noise with AK4468.
I’ve not used the AD1939 devices properly so cannot comment yet, they power up and do things, but not tried them in my decent setup.
My ADAU1966 board uses a 5V for AVDD (VAA rail on the schematic). It was designed for the non-A part which is not available. The ADAU1966A part will fit, but requires 3.3V for AVDD supply. R309, R310, R311 would need to change to give 3.3V output. I would need to work it out.
I have built all my boards with a standard iron (no hot air), it’s a slow process, but works. If you want something working easily I would probably buy a cheap AD1938/39 off ebay or aliexpress to play with. You can always build something else later on like an AK4468 board and then compare them.
Better to get something up and running and enjoy it. Most of the boards I am using from my designs are Rev 1 boards with some issues. I have new designs, but not built them yet. Prefer just using it and enjoying it!
My designs use good quality regulators on the analogue supply locally on each board and other supply filtering. With clock and signal traces I have impedance matched (roughly) the PCB traces and ribbon cables (26AWG wire size), and also tuned the termination resistors. You can feed in any high quality clocks you want, there are clock headers on ADC/DAC boards for this, as long as it meets what the IC needs.
Biggest change using the AK4468, instead of the ADAU1966, has been increasing sample rate on my crossover box to 192kHz (from 48kHz). So that might be why it sounds "better”, although I do have more background noise with AK4468.
I’ve not used the AD1939 devices properly so cannot comment yet, they power up and do things, but not tried them in my decent setup.
My ADAU1966 board uses a 5V for AVDD (VAA rail on the schematic). It was designed for the non-A part which is not available. The ADAU1966A part will fit, but requires 3.3V for AVDD supply. R309, R310, R311 would need to change to give 3.3V output. I would need to work it out.
I have built all my boards with a standard iron (no hot air), it’s a slow process, but works. If you want something working easily I would probably buy a cheap AD1938/39 off ebay or aliexpress to play with. You can always build something else later on like an AK4468 board and then compare them.
Better to get something up and running and enjoy it. Most of the boards I am using from my designs are Rev 1 boards with some issues. I have new designs, but not built them yet. Prefer just using it and enjoying it!
Last edited:
Yes, I can tell that your boards are well thought out. I think you shared pretty amazing work, and I'm surprised by the lack of interest around it.
I was looking at your uCodec project, based on the AD1939. And places like PCB Prototype & PCB Fabrication Manufacturer - JLCPCB could assemble the board for not a lot, if needed. But you are saying that you have not tested (enough) that design, yet? Just making sure I understand your post correctly
And, yes, my goal is to experiment a bit. But there are few DAC boards available, and chips are for the most part backordered (e.g. I just checked, and the AD1938 and AD1939 are backordered at both Mouser and Digikey in the USA)
So it looks as if the AD1938 board usually sold as codec for the ADAU1452 is my best bet for now
I was looking at your uCodec project, based on the AD1939. And places like PCB Prototype & PCB Fabrication Manufacturer - JLCPCB could assemble the board for not a lot, if needed. But you are saying that you have not tested (enough) that design, yet? Just making sure I understand your post correctly
And, yes, my goal is to experiment a bit. But there are few DAC boards available, and chips are for the most part backordered (e.g. I just checked, and the AD1938 and AD1939 are backordered at both Mouser and Digikey in the USA)
So it looks as if the AD1938 board usually sold as codec for the ADAU1452 is my best bet for now
@robca: AK4468 and most of their audio catalogue are out of stock, because they are currently all out of production 
. the factory burned down last year and they were unable to find somewhere with suitable fab to produce them, given it was their own fab, with a specialised process.
at this stage, it looks like next year, if at all
. the factory burned down last year and they were unable to find somewhere with suitable fab to produce them, given it was their own fab, with a specialised process.at this stage, it looks like next year, if at all
Fire was not good for AKM parts, although even before the fire it took ages for me to find stock of both the AK4458 and AK4468 devices (pin compatible).
Some components are still hard to find stock of, even the DSP chips - either the ADAU145x or the newer ADAU146x - seem to have limited availability at times. Luckily for the uDSP board I had some samples from Analog I found in a box.
With the uCODEC and uDSP boards, I can't remember where I got to. I think I had both powered up, and was running my crossover code on the uDSP board, although that wouldn't work with the uCODEC board. I was doing it last year, but had to pack up for an electrical re-wire on my rental property.
Not even sure where the boards have gone, must be in a box somewhere...
Some components are still hard to find stock of, even the DSP chips - either the ADAU145x or the newer ADAU146x - seem to have limited availability at times. Luckily for the uDSP board I had some samples from Analog I found in a box.
With the uCODEC and uDSP boards, I can't remember where I got to. I think I had both powered up, and was running my crossover code on the uDSP board, although that wouldn't work with the uCODEC board. I was doing it last year, but had to pack up for an electrical re-wire on my rental property.
Not even sure where the boards have gone, must be in a box somewhere...
Hi Paul,
May I ask you what was the conclusion of your findings about the voltage bias here?
Open Hardware DSP Platform Blog - Open Hardware DSP Platform
Thanks a bunch!
May I ask you what was the conclusion of your findings about the voltage bias here?
Open Hardware DSP Platform Blog - Open Hardware DSP Platform
Thanks a bunch!
My conclusion is a possible silicon bug. Nothing wrong with the design from what I could see, but the CM voltage just doesn't change as it should for a 5V supply.
The ADAU1966 (not the A part) became unavailable and then obsolete/NRND very quickly. It's why I designed the AK4458 board, although that also has problems with supply as well now!
I've not tried the ADAU1966A. Should work on the same PCB if the AVDD is changed to 3.3V
The ADAU1966 (not the A part) became unavailable and then obsolete/NRND very quickly. It's why I designed the AK4458 board, although that also has problems with supply as well now!
I've not tried the ADAU1966A. Should work on the same PCB if the AVDD is changed to 3.3V
@ohdsp
Thank you for your contributions to the community!
An XMOS AVB system would be an excellent match for a DAC that supports TDM16 input and standalone operation, and your github was the only open source project I could find using the ADAU1966. I was excited to see you were active on this thread here, even though the project is aging on github.
I built a XEF216-based platform paired with the ADAU1966A, but it was an expensive dumpster fire and I was forced to abandon the effort.
Do you think there is a chance we could work on your revised ADAU1966A together?
I am not an excellent engineer in most ways, but I would love to help in any way I can to make the project a reality.
Thank you for your contributions to the community!
An XMOS AVB system would be an excellent match for a DAC that supports TDM16 input and standalone operation, and your github was the only open source project I could find using the ADAU1966. I was excited to see you were active on this thread here, even though the project is aging on github.
I built a XEF216-based platform paired with the ADAU1966A, but it was an expensive dumpster fire and I was forced to abandon the effort.
Do you think there is a chance we could work on your revised ADAU1966A together?
I am not an excellent engineer in most ways, but I would love to help in any way I can to make the project a reality.
Biggest issue now it part availability. Just looking back at the DSP board and the TLV1117LV33DCYR is out of stock on 260+ day lead time. ADAU1966 board uses the same regulator. The ADM811ZART seems not available, and alternatives I have been using are not available either.
Problem is I could design all of this out somehow, and then run into the same problem again.
The component market just doesn't work so well for small projects at the moment!
Problem is I could design all of this out somehow, and then run into the same problem again.
The component market just doesn't work so well for small projects at the moment!