 |
Up front a big thanks to Doede who was able to resolve any issue with some quick email replies. He also sourced some of the hard to find parts for me, well before he started offering part kits for his DAC. This was my first DAC project but it wasn't my first DIY audio project. I'd rate this build more difficult than the Foreplay preamp, because there are no step-by-step instructions. One must be able to read the schematics and interpret the PCB to build this DAC, as well as figure out all the wiring of the battery, switches, etc. So be warned if you want to build this DAC - it's not as easy as it may appear.
A first estimate of the parts cost was about $300 per unit, which isn't "cheap" for a DAC, but we're talking about a DAC that performs better than some DACs well over $1000, plus the estimate included only the highest grade components available. Auriga, Blacktops, etc - these things aren't cheap. Some of the chips were over $30 each, and by now have become very hard to find. If you embark on building this DAC, make sure you can source all the chips needed, especially the CS8412 receiver chip, as the source I used for it has closed shop. Once I had both DACs working well, we decidsplurgeplerge and order some very nice custom faceplates for the enclosures that came with plasticplasic ends. Those parts cost us about $65 per DAC, but we felt that it was worth it after all the other expense. If somebody is interested in the CAD files for these panels to build this DAC in the same enclosure, please email me. A list of the parts and costs for what I put into these two DACs can be downloaded in this Zip file - it is an Excel spreadsheet. The list shows sources and part numbers for most parts. Much of these parts are now available from Doede as a kit, although the higher grade parts I used are not available in his kit.
My build of these DACs varies from the original design in a few areas. For example, I placed a whole bank of capacitors between the battery and the DAC inputs, decoupling it from the regulators on the PCB. There I used parts I had on hand, but again, those were pretty high quality caps. I also used some LEDs to show that the unit is powered up, that the fan is powered (although I don't use the fan very much anymore - the copper heat sink seems to suffice to keep the DAC cool). I bypassed all the Blackgates that feed the IC power legs with some small Wima or Vishay MKP caps as well. The circuit around the clock chip is modified to match the instructions on Guido Tent's web site, and it includes ferrite beads as well as a 47ohm resistor on the output leg. The installation of these "tweaks" is shown in the photos below. The circuit for the toslink receiver is also changed to match what the datasheet of the receiver I used required. There areadditionalitonal capacitors and chokes in my DAC that aren't in the schematics, although I don't know if they make a difference.
The most interesting part of this project was the construction of the DAC tower, as I was hoping to get enough cooling from a proper heat sink design to be able to run the DAC without fan cooling. My approach works well, but the DAC still gets rather hot (although I think it stays below the rated max operating temperature). One day I ran it for 10 hours without fan and it still sounded good a that point. For longer life of the components I decided to cool it with a very slow running computer fan, which works really well. With a larger enclosure and larger heat fins on a custom PCB that allows for a larger cooling area around the DAC tower, once could possibly build this DAC with cooling good enough to not require a fan.
So how does it sound, now that it's all done? In my system it has completely replaced the DAC in my Cambridge Audio D500, which was sounding quite good to me after recent player mods, but now there's simply no comparison. The soundstage of the DDDAC1543 is so much better defined that there is now "space" between instruments where there wasn't any with the old DAC. Detail is simply amazing, and vocals are to die for - not a hint of harshness and extremely good pace. Bass is well defined, but intially I felt that there's a little lack of punch. This went away when I added some larger Auricaps to the output stage (going from just 0.47uF to 2.0uF) - the original DAC design was meant for a preamp with much lower input impedance than my Foreplay.
In summary - this DAC has convinced a lot of people who heard it that CDs can sound smooth, clean and musical. For the foreseeable future I don't need to worry about this part of my system. I'll be focusing my attention on a new transport (Toshiba SD3950 mod), a better rack , room treatment and a big upgrade of my amps once I can afford the parts required for that... the list never ends.
|
Here an overview of some of the parts I had to source to build the two DACs. This photo shows about 75% of the materials I needed to complete the job. Many small packages from various suppliers had to be ordered as I ran into problems or realized that I had overlooked a few items on the schematics. I changed my mind about the fans, about the number of batteries per DAC, about the chassis, the switches, about power connectors, and other items as I went along. The project began in December 2003 when I ordered the PCB, but I didn't really begin building the DAC before the end of May. Both DACs were completed and tested by the end of June. |
|
Closeup of the 7ah battery that powers the DAC for about 15 hours before recharging is needed. People who are opposed to battery powered DACs probably don't realize that it takes me 20 seconds to swap batteries ($7.25 each on ebay) and I have another 15 hours before I run out of power. It is extremely clean power at low cost. The expense to build an AC power supply that is as clean as battery power would have made this DAc considerably more expensive. Since this picture was taken, I have changed the connectors on the DC line to wire connectors that fit exactly on the shape of the terminals and have a snug fit. The wire I used is left over "OTA" speaker wire - 18awg should be plenty for the 0.5 amps the DAC draws. |
|
Construction of the DAC began with a rather curious process of stacking DACS with hand-cut copper fins, and gluing the whole thing together with computer heat sink "thermal silver epoxy." I did this first in pairs, then made stacks of 4, and in the final round I put 8 DACs on top of each other to form the core of the DDDAC1543. The copper in between each DAC chip carries the heat away from this oven (these are run at about 8.50 volts, not 5.0 volts as in most applications and they generate some serious heat). |
|
Before I continued with the heat sink design, I first had to solder all the pins of this tower. Do this carefully, and test every single lead from top to bottom. I had one bad stop in one tower and was wondering why one channel was running considerably lower output voltage than the other... Soldering these pins before the tower goes on the PCB is much easier, so doing it right the first time is highly recommended. Ignore the shape of the copper fins - at that stage of the build I still thought I could mount a large heat sink against the end of these bent plates. Space considerations in the small chassis I had available had me change that later. |
|
Once the DAC tower was built and at least a plan for cooling it was developed, I started with the PCB. Populating the board with ICs, capacitors and resistors was pretty easy. I spent more time looking for the proper parts in my large pile of small plastic bags, than actually soldering. That big red/orange capacitor on the board is a Blackgate N 47uF that feeds the DAC tower's power leads (at $10 the most expensive capacitor in this DAC). All the Blackgates in my DAC are slightly larger than the original design, as it had turned out in other people's tests that the 20uF was clearly not large enough for at least the DAC power leads. I put 33uF Blackgate NX caps on all the other power leads. The image also shows the battery checker circuit at the bottom of the PCB - I didn't worry much about that one at that stage, but should later spend a lot of time on it to figure out how to make it work as intended... |
|
On the second day I installed just a few more parts. Most of the time was spent grinding some fins off the large heat sink for the LM1085 voltage regulator (it needs one that large, trust me). The problem is that my heat sink was so large that a trimmer and a smaller Blackgate were in the way. Instead of mounting these below the PCB (thought of that too late plus that would have probably rendered my chassis too low for the whole thing), I used a large power grinder to remove some material from the heat sink. It's not pretty, but very functional. |
|
Although the schematics of the DDDAC1543 don't include a "power supply" section other than leads to the
battery, I felt it wouldn't hurt to put some capacitance between battery and DAC, nice and close to the
board. I had a couple of Panasonic FC caps left over from a recent upgrade of my preamp, as well as a
handful of small Vishay film caps. I paralleled the three 270uF Panasonic electrolytic caps with three
Vishay 0.47pf caps and three Wima MKP10 1000pF caps (not visible here yet). Later I added another two 2200uF
caps I salvaged from an old Denon CD player. There was still room for those and I figured it can't hurt ;-) The three pars of power leads going away from this breadboard "power supply" go to the individual circuits on the DAC board. I tested this with three different batteries and couldn't tell a difference from the single battery version, so even if you're a purist, save yourself the hassle with three chargers and three batteries - it sounds the same with just one. |
|
This is the "input board" - a small assembly that contains the toslink receiver and the coax input jack, as well as a few capacitors, resistors and a choke. The Toslink receiver is powered by a 5 volt link to the PCB which I connected just before it feeds the clock. This photo still shows a large switch - to make these parts actually fit on the small 2x4" rear faceplate, I had to later change the selector switch to the smallest micro switch I could locate. The DAC schematics don't show much of this little circuit board but if you look at the TORX179 toslink receiver datasheet, you'll see exactly what I have done here. At this stage I didn't use shielded wires yet, but in the final build there's a nice shielded wire running from the switch to the PCB. In the near future, I plan to change the RCA jack of the coax SPDIF input to a BNC connector. Those are simply better for 75 ohm connections. |
|
Before connecting it to my CD player, I had to test the DAC for electrical "soundness" first. My 0-20V 12 amp Hewlett Packard bench DC power supply came in handy (usually it powers the slot car track ;-)). Not a single capacitor vented, the supply showed a steady 0.5amp load, and after a few minutes, the DAC tower was getting warm. At that stage I was pretty much done with the DAC assembly. The battery tester didn't work yet, but that was because I had not been able to figure out that the power had to be fed to two inputs on the small board, not just one. Adjusting the battery tester was very easy with the bench power supply, where I was able to adjust the voltage in fine increments, allowing me to set the trimmers to the exact point where I wanted the LEDs to change. |
|
Here a view of the first DAC on the "bench" with the bench power supply on the kitchen floor :-) Even though it's a small DAC, you still need a large table to get the thing built. The picture also shows most of the tools needed. |
|
Here is my serial #0001 DAC working for the first time, battery charger all hooked up, coax wire from input to board. You can see a big blue trimmer sticking out from next to the now very copper-finned DAC tower. That trimmer was necessary because the original part I had put in there was a single turn trimmer that made it virtually impossible to set the vref voltage to the desired 3.85 volts. The blue trimmer is a $5 part with a 25 turn spindle, and even with this part it takes a fine touch to get the voltage just right. Note that there is a slight drift in that voltage, but to make the DAC sound decent, you should get as close as possible to this value (at least for the 8 DAC tower). The yellow caps are 0.47uF 450V Auricaps for coupling. These are directly between DAC output and the RCA connectors to the preamp. No opamps, no tubes, no transformers, just two nice caps in the signal path. At the very end of this project, I added a second pair of Auricaps (2.0uF/200V), which improved the bass response quite a bit. This change is highly recommended. If you are sourcing parts for the DAC, consider a pair of 3.0uF caps in this position. |
|
The initial "assembled" DAC was cooled with a small old CPU fan. It really makes a huge difference in terms of chip temperature and is why I chose to run a fan on the DAC permanently. However, one day I forgot to power up the fan and ran the DAC for 12 hours with just the cooling fins doing their job, and it sounded just fine when I realized my mistake. I'm sure the DAC will last longer with the cooling, so I generally run the fan. Having a fan that is quiet enough to not be heard from my listening position was important. The CPU fan had to go and a larger slower fan had to replace it on the final DAC chassis. |
|
This was the first time the DAC went on my rack to go between CD player and preamp. The cardboard mounting was actually very good sounding. No interference, just a bit "ugly" and a potential point of failure in this shape. At that time I had no idea how difficult it would be to get all this into a small project box. If you are building this DAC, just go for a large box, because it will make the final assembly considerably easier. |
|
Here's a closeup of the DAC tower with the extra fins. I cut many 0.5x1.5" pieces from the copper sheet I originally bought, then glued them to the ends of the copper strips that stuck out from the DAC tower. Bending them upward should improve air flow in case the fan is not running. This picture still shows the old vref trimmer and there is no grounding yet. The entire tower copper "structure" is now grounded. Later I did the same for all other chips except for the clock, which is in a metal canister already. |
|
Top view of the copper structure. Nothing pretty, but it gets the job done, and the material is $5.00 at any crafts shop. A pair of tins snips and the heat sink epoxy is all that's needed to build this. |
|
This photo of the tower shows the grounding wire I soldered to each layer of the tower. There's also the blue vref trimmer installed already. This is very close to the final assembly into the project box. The resistor on the top left corner of the PCB in this image also got a copper heat sink, glued to the body of it. The other ICs are all copper-clad and grounded. |
|
Below the PCB, I installed bypass capacitors for all the Blackgate caps on the power leads of each IC. Note the little brown piece of CAT-5 on the battery charger PCB on the right - that's what I had to do to make the board actually work. |
|
Here's a closeup of the "tweaked" clock circuit. According to Guido Tent's data sheet, the clock output
should go into a 47 ohm resistor, but there was no room on the PCB for that resistor. So I drilled some holes
right on the output lead of the clock and added the resistor between the chips. The clock itself has already been
treated with a nice coat of "rope weatherstripping" - my preferred material for damping. The two beads on the left
of the clock are also not part of the DDDAC1543 schematics, but should go there according to the clock schematics.
I didn't hear a difference, but it can't hurt putting these into that location. The clock chips arrived with these
in the packaging, so I felt I ought to use them :-) There's also a 0.1uF capacitor between clock chip and the neighboring chip. That one is somewhat hard to locate on the schematics, but should really go there. The smaller blue chip is one of the Blackgates that belong there on the regular PCB. |
|
Before ordering fancy faceplates for the cases, I had to see if everything actually fit into the project boxes I had bought. With plastic end caps, duct tape and some foam here and there, I managed to wedge all the parts into the boxes. Drilling holes for all the LEDs and connectors took quite a long time, and in the end things still didn't fit right. I learned a lot from this test assembly and was able to create exact drawings for the face plates needed in the final assembly. |
|
The back panel is extremely crowded. The power connectors here are total overkill (Neutrik PowerCon), but in the bigger picture of the DAC costs, spending a few dollars more for connectors that won't fail didn't seem out of place. The other DAC I built actually uses XLR connectors for the power lines. Both DACs have a much cheaper and less expensive connector for the fan power supply, which really would have worked fine for all the power connections. Maybe for the next DAC... |
|
Up and running in the project box with fan and plastic face plates for the first time. The fan is a simple 12V low-noise computer fan with blue LEDs indicating power. To make it run almost without noise, I bought an adjustable power supply from RadioShack. Now the fan runs at 6V and can barely be heard right next to the DAC. Drilling the holes into the box as well as cutting the large hole for the fan on the top plate were rather difficult without the proper tools. If you have a fancy case go pay somebody to do these things... Oh, and the whole thing is bonded with rope caulk to a 1/2" 7075 polished aluminum base plate that rests on a set of DIY roller bearings. |
|
Front view of the finished DAC. The face plate was custom made by frontpanelexpress.com - expensive, but absolutely perfect once you get the drawing right (free software available on their site). Front and back panel combined cost about 20% of the entire DAC parts cost... |
|
The rear panel is much more crowded with the fan power lead coming in above the charger and battery connectors. The small input selector switch was needed to fit between the coax and toslink inputs. |
|
Another view |
|
Side view with a view through the vent holes at the crowded interior. |
|
Here a view of the back of my CD player with the custom coax cable I built to connect the DAC. The cable construction is based on a foamed teflon core salvaged from a Beldon 89259, which I sliced open. I replaced the center conductor with some 30awg bare silver I had on hand, wrapped the foam core in a layer of teflon plumber's tape, then put the tin foil shield back on, added another layer of tin foil, wrapped that in a layer of teflon tape, then put the Beldon copper braid back on and wrapped the whole thing several times in Teflon tape. On the CD player end I installed a Vampire BNC connector (solder) and on the DAC side I used a Canare RCA crimp connector. The Canare connectors are very difficult to install unless you have the right tool, but I managed to get it to work. In the future, I'll move to BNC connectors on both ends of the connection. This cable sounds quite a bit better than the toslink ($70) glass connector I first used. |