|
About 8 months have passed since I finished the 16-chip DDDAC build with headphone output and even though I don't listen to headphones at all, the DAC has been working like a charm. Why then a new build? There are two reasons: first, I wanted to do a few more radical things with the DAC power supplies, and second, I was hoping to integrate a computer interface directly in the DAC.
I began the project shortly after the original 16-chip DDDACs were built, since I suddenly had a complete PCB left over (I gutted the original 8-chip DDDAC to recover all the parts needed for a new build that could be salvaged). I also had extra heat sink material and TDA1543 chips left over from building the newer DACs that were looking for a new use. Since I had something to listen to for a while, I was taking my time with the build, looking for possible improvement throughout the design.
The most visible change to the original DAC design is that I disabled all the original on-board voltage regulators except for the 8.5V analog supply for the DAC tower. Those supplies were moved off-board and changed to much higher quality designs, one supply for each chip, not just for the clock like first implemented in the 16 chip DDDAC in 2005. For the CS8412 chip, I also implemented a "Wildmonkeysects" filter circuit (google it and you shall find). Almost all capacitors in the unit are now Blackgates and all the resistors on the DAC board are Holcos. Every wire inside the DAC has been deep-cryo treated, just like all the switches and higher grade connector jacks at the back. I also included a toggle to allow the use of a WBT NextGen RCA jack as an alternate to the BNC jack for the SPDIF input. Up to this level, the DAC was pretty much an improved version of the last 16 chip build. However, there was that USB thing yet to do....
For a few weeks I was messing with an M-Auido Transit board, replacing parts on it with higher grade components, and trying to replace the USB power with a battery-fed custom supply. I even figured out where the SPDIF can be tapped an was in the process of implementing some sort of transformer-based connection to the CS8412 on the DDDAC board when I got word from Doede Douma that he was working on a new DDDAC design that would also include a USB - I2S interface that completely eliminates the need for SPDIF, transformers and other band aids. So for a few months, the new DAC was put into a box while I focused on building a new preamp.
In January I received a "beta" USB board for review from Doede. Much better quality board than his original DAC, and already fully built, saving me the time to solder the bulk of the parts. But since I had a few Blackgates and Panasonic FCs laying in the parts bin, I replaced some of the key capacitors right away. Installing it as an alternate input to the DDDAC tower only took the addition of a 3PDT switch to toggle the Data, FS and BCK outputs from the USB board with the same lines coming from the CS8412. That way, a direct A/B comparison between SPDIF (CD transport) source and USB (computer) source is possible with this DAC, using the very same power source and DAC tower, wiring, etc, just toggling the inputs.
Initial listening tests showed that I had a ground loop problem with the computer source that only went away when using a battery-powered laptop as source. Under those conditions, and going through a preamp, I was virtually unable to distinguish the PC source from the modded Toshiba CD transport on the SPDIF input. However, bit sounded quite a lot better than what the older 16-chip DDDAC put out, proving that the investment in the better power supply and other tweaks wasn't just for show. However, I was still hoping to get more from the USB board, and it turned out that it was all a matter of how to configure the computer to get the most from that interface. I had to install a special driver (ASIO4all) to be able to use the computer volume control without Windows kmixer messing with the signal. Once I had that working, I solved the ground loop issue with an expensive USB optical interface, originally designed to move USB 1.1 devices far from the source computer, but it also provides complete galvanic isolation between computer and USB device, while maintaining enough bandwidth to handle even 96kHz/24 bit upsampled audio. With those fixes in place, I tried hooking up my power amps directly to the DAC and the result was one of those moments when you realize that a lot of things you have been working on in the past have suddenly become totally obsolete. In this case, I instantly realized that dropping the preamp completely from the signal path and using the PC for volume control and transport duty was a dramatic improvement over the sound I was able to get from the same DAC feeding it via SPDIF from the transport and using a preamp. That new preamp I am 75% done with (and have about $600 in parts invested in) moved very quickly into a parts box. I am not sure when I'll finish it, but it has quickly dropped to low priority. My audio rack now holds nothing but a DAC, the computer sits remote in a basement room, while the screen and mouse are on a KVM extender.
Room correction will be next, and eventually, I'll be exploring multi-channel sound cards in the computer that will allow for a digital crossover. At that point in time, I'll have to build a new 4 channel DAC. Maybe in 12 months I'll be able to post about that build on this web site as well.
Meanwhile, there are a few tweakable areas on this DAC that may be addressed in the near future. The switches on the front panel are not quite up to my liking, and the mute circuit I had to add to cut back on the power-up pop from the DAC isn't really 100% silent, plus I have to build a new battery box for this DAC, since the headphone DAC will be for sale with the big battery box.
The battery box will need to be pretty big to include a second box - I recently finished a custom 5V battery supply for the optical USB cable receiver end. Description of that build can be found here. This supply also fixes the problems I had with Windows XP not recognizing the DAC after I powered it down for recharging.
The next project will be the build of a big file server to store and play all my music - it's already in use, but the full report will get posted when I am done with the ripping and software config.
Here's a link to my current Foobar driver and output setup: DDDAC with ASIO4ALL.
The build pictures below skip some of the contruction details that are already covered on the DDDAC16 pages from last year.
|
This picture was taken during early testing, but the circuit is pretty much all completed. The big red toggle switch in the back is the USB/SPDIF input selector. At the time this image was taken, none of the power supply circuits necessary for the SPDIF operation were plugged in (white plugs hanging in the air on the brown PCB at the bottom). The improvised power connector hanging off the side on the bottom cost me a few fuses, as it shorted to ground a few times during these open case tests. The preamp is still visible on the image at the top. By now that Bottlehead Foreplay has been retired until further notice. |
|
A better view at the green USB board. Note the 12mHz Tent clock on the board, driving the PCM2707 USB chip that
is located on the bottom of the PCB.
At the bottom of the image you can see a gold colored capacitor. It is part of a complete 3.3V circuit that was supposed to power the M-Audio transit board, but as it turns out, Doede's USB board has a low noise voltage regulator of similar design built into the PCB, so now I am using my hand-built super clean supply to power a pair of power LEDs... On the left, you can see the new heat sink I rigged up for the voltage regulator that drives the DAC. Since I wanted to completely eliminate the need for fans in this DAC, I needed a better solution to move heat from the regulator to the chassis. A few minutes on the bench grinder in the garage and some heat sink epoxy created this new dual heat sink part that touches the outside walls of the chassis flush when the PCB is screwed into the DAC. This also allows easy removal of the entire board. |
|
Heat sink grease is applied to the contact surfaces of that heat sink before final installation into the chassis. With a load of 16 DAC chips, this regulator is the hottest part in the entire DAC and needs the best cooling. The photo also shows small slots in the base of the rear face plate, visible under the read and white NextGen plugs. The idea is that warm air will flow out the front mesh of the DAC and cold air will be pulled in from the low openings in the back. Early testing has shown that the left heat sink of the chassis gets much warmer than in the older DAC design, keeping internal temps at bay without the use of any forced air. |
|
The rear face panel is a rather complex custom part made by frontpanelexpress.com (about $30), but it fits perfectly and all the holes are machined perfectly to hold the jacks without a risk of them twisting in oversized holes. The wiring of the SPDIF jacks to a toggle switch isn't really "purist" but since a friend wants to do some testing of fancy wires on this DAC, while I believe the BNC beats them all, I decided to put that switch into this DAC. I don't plan on using the SPDIF interface very much anyway. All these connectors and switches were deep-cryo treated in a multi-day bath in liquid nitrogen with a slow warmup of yet another 3 days. Looking at this image I just realized that I forgot the 75 ohm resistors on the SPDIF inputs (shows you how much I use those...). |
|
One of the final assembly steps was to epoxy the tower to the chassis using heat sink epoxy material. For a while I was contemplating a mechanical solution, but since I have never needed to pull one of these towers out of any DACs, I decided to go the easy route and just glue it to the walls again. There's a cryo-treated DIP8 socket on the base of it, so it can be disconnected from the DAC board easily. An extra 100uF Blackgate N on top of the DAC supplements the 100uF next to the 8.5V supply regulator on the PCB. |
|
These larger Auricaps (5.0uF) were initially intended to be used in my new tube preamp, but since that project is now stalled, I decided to use them in the DAC instead of the 2.0uF caps previously used. The idea is that with the 10k input impedance of the power amp, the larger caps will give me better bass response. Mathematically, 10uF coupling caps in the DAC would be idea, but they wouldn't fit into the chassis, nor do I think I will be using my amps for much longer. These caps are the only passive component separating the TDA1543 chip output pins from the input of the power amps in my current setup, so they are rather important. In the future, I may try some Teflon caps once the budget allows for those. |
|
The rear panel from the outside with all jacks installed. I elected to keep the panel without engravings, because the cost for the labeling is rather substantial, plus there's not that much that needs to be done back there. One switch (the one on top) toggles between USB and SPDIF inputs, the other switch at the bottom picks the connector for the SPDIF input. |
|
Here I am adding another heat sink to the inside edge of the DAC tower, again with arctic silver heat sink epoxy. There's also heat sink grease on every single chip inside the DAC tower to increase the heat transfer to the aluminum plates between each chip. |
|
While the heat sink is clamped to the tower, I installed the battery checker board also salvaged from the original 8-chip DDDAC build. The LEDs are mounted right on the board since they can be seen through the mesh of the DAC front face plate very easily. This DAC has no attenuator on the front center, so it was a lot easier to position the LEDs in an area visible from my listening chair. Too bad they don't work right at this point. The red LED is always on - another bug to address before this project is fully completed. That gray stuff on top of the IC on the board is ERS paper - I use this or a similar adhesive emission blocking material on all ICs of the DAC. |
|
Here a look at the front face plates of the headphone DAC (now with black Neutrik jack) and the new USB DAC. I don't really like the look of these simple toggle switches, but for now they will do the job. Much easier to install, too. Both DACs are resting on half-sphere "Norsorex" feet, created by cutting some "unhappy bouncing balls" in half. These rubber balls are great absorbers and cost a lot less than some big name products or even sorbothane. Most surfaces of the DAC chassis are covered with a thick asphalt-based vibration damping sheet meant for automotive use. Again, low cost but very effective. I left some surfaces uncovered for better heat transfer. the rear panel with the connector plugs will need to get some damping treatment as well. |
|
Another look at the back side showing that I didn't measure right. the top edge of the face plate is about 0.7mm too high. Nothing a few minutes with the file can't fix, though. |
|
The inside wiring completed. Battery input and charger are the green and yellow wires inside the mesh that snake through the DAC to the switch on the front faceplate. The red switch on the front panel handles the mute switch, something only needed if you never power down your power amps, since there's quite a power-on pop coming from the USB board when it powers up. The rest of the wiring is mostly the supply lines to the PCB from the voltage regulators on the board with the big red Blackgate caps (1000uF/16V each). The messy green/blue/white wire bundles over the top of everything in the back are the I2S lines. The next USB DAC won't have any of those, since I will be implementing it as a pure USB DAC, in a layout that will also minimize the length of these lines.. |
|
And that power switch up front really needs to be changed. It is just a little too tight there for this setup. Perhaps I'll change the PCB in front of it by cutting off the unnecessary 3.3 V part in back, which will add about 1 inch of clearance at the front. Meanwhile, though, I am listening to this DAC and hope to keep the interruptions due to tweaking and fixes to a minimum. |
|
|
Here's a view in the dark, now with the finished battery checker I custom built for this DAC. IT is a 10 LED bar
display using a really inexpensive LM3914 IC and a few resistors and diodes to give me constant feedback about the
remaining battery level. So it does suck 100ma at full charge, but as the display drops LEDs towards the lower
charge levels, the power consumption of the checker also drops. It looks really nice at night, shinging through
the mesh of the front of the DAC.
I'm still looking for some nice on/off and mute switches to use, but for now the toggles that are in there are working just fine. It's just for looks if I should change that. With the battery checker done, and a battery power supply built and in use on my USB optical cable receiver, I am pretty much done with this project. Even though the DAC can do SPDIF like my old DDDACs, I don't use it for that any longer. All those ICs inside the DAC that are needed for SPDIF operation are actually unplugged from the power and fully disabled at this time. It's a pure USB DAC right now. |
|
And this is what happens to your audio rack when you move from a transport - DAC - preamp setup to a remote
computer feeding a USB DAC, doing the volume control on the computer. That huge isolation transformer now only
feeds the battery chargers and the AC supply for the USB receiver, which I mostly run on batteries anyway.
This minimalist system (DAC, power amps, speakers) goes right along with the less-is-better approach of the DDDAC design. No preamp sounds better than the best preamp, just like no opamps or tubes behind the DAC chips sounds better than the best solution with such parts in the signal part. When will I build the next DAC? Maybe in fall or winter. I am considering a 36 chip build based on the new DDDAC 1543 mkII PCBs, but that DAC looks so well thought out, there's hardly an option left for tweaking :) Once I figure out how to make that DAC even better short of stacking 120 DAC chips, I will get started on that project. Meanwhile, I need to finish the music server, room treatment, active crossover, speaker rewiring, build 4 power amps... |
|
Here's a closeup of my "quick and dirty" 5 volt SLA battery supply for the optical USB receiver end (it's a powered hub essentially)
I modified the wire so that I can switch from AC to battery power with a quick connect. The supply is nothing more than a
5V regulator with some 120uF Panasonic FC caps and a 0.22uF Wima. I'll be changing that to some Blackgates and XR7 ceramic
bypass once I rebuilt it in a small box with heavy duty wiring and an on/off switch that'll toggle a battery charger to
the battery when the unit is off. There's quite some current draw on this little thing, which is why there's the fat heat sink
on the regulator. My DAC is self powered, so it could be even worse if you run a DAC that's usb powered off this unit.
The on/off switch on this USB "hub" is essential to make sure WinXP doesn't drop the USB audio device when you power down the DAC: if you first power down the USB, then the DAC, then power up in the reverse order, the PC will not lose track of the USB audio device. Any other order and you need to reboot the PC, so the little power supply on/off feature on the USB receiver is actually a very useful feature to be able to keep the PC running while the DAC is off, avoiding the reboot that would be necessary to get the device recognized again if you just powered down the DAC. The final version of this supply can be seen on this page. |