DAC Update; The Bit Goes On by Sheldon D. Stokes In volume 5, issue 1 I published two DAC designs. In the months that have passed after my submission of the two designs, I have made some improvements in the larger, more expensive DAC. The larger DAC started it's life as a half a dozen boards mounted on a piece of plywood. I etched a board for each portion of the DAC so each piece could be changed and worked on more easily. My DAC is still on that piece of plywood (yes, I'm too cheap to buy a case right now). But now it is a single board, and looking much more professional. The areas that have been changed in the newest DAC version are in the power supplies, digital filter, and output stage. The power supply regulation stages have increased in numbers and gotten simpler. The regulators have been changed to the adjustable type for the digital and analog supplies. Adjustable regulators are electrically less noisy. They also seem to isolate the digital power from the analog power better than fixed regulators. I specify LM317 and LM337 regulators for analog and digital. If you want to really go off the deep end you should try Linear Technology LT1085/LT1033 adjustable regulators for the analog supply lines. They make a small but noticeable improvement. Each DAC chip has it's own regulators in this version (this brings the regulator count to eight). The filament supply also uses an adjustable regulator so the filament voltage can be changed easily if a different tube type is to be substituted. The high voltage line regulation has been simplified. It now uses a zener referenced pass transistor that feeds a pair of RC filters (each channel has it's own RC filter). The digital filter chip has been changed. I am using a fairly new release from NPC. This chip has de-emphasis provisions, using a FIR filter. Doing de-emphasis in the digital domain is much better than putting a filter network in the analog section. This new filter has a choice of two different response characteristics. The DACs sonic signature changes slightly with the different responses (which is a little puzzling). The analog section has also been changed. I started messing with my home-brew DAC using 6DJ8's in a standard resistor loaded configuration. But I couldn't leave well enough alone, I started trying other tube types. I increased the B+ and then used the 12AU7 for a while, then shifted to the 12AT7, which I feel is a much better sounding tube than the 12AU7. Then I raised the B+ some more and then tried a SRPP (shunt regulated push pull) configuration. This really had some promise, so I decided to try the 6DJ8 again in that configuration. I really like the 6DJ8, it is more finicky about the bias point used, but when dialed in, it sounds very good. After the bias is dialed in, it has a sonic signature that I find very pleasing (try using 4.25mA of bias current). But if you like the 12a_7 family, I will give you the information to use 12AT7's in this new DAC version. The board layout has some new features. First all the regulators have extruded heat sinks from Digi-key. But only the filament and b+ regulation devices need heat sinks. For all those of you who aren't used to working with tubes; the pass transistor in the B+ power supply is sitting at about 400 volts, which will give you quite a wake up call if you touch it. So I recommend a transistor mounting kit to isolate the transistor from the heat sink. My hands have a unique ability to find the B+ in any circuit. So this is my standard warning: BE CAREFUL, you can shock yesterdays meals out of you, or worse so be careful. There are provisions for three LEDs on the board. The first (PWR) is a power indicator. The second (LED1) is an indicator that the data is valid. It will flicker when searching or paused, or in time of transport woes. The third (LED2) is a de-emphasis indicator, it will light up when de-emphasis is engaged. As before the switch (S1) is for digital phase inversion. There is a bank of four DIP switches near the digital filter (U8). The first two switches (next to pins 17 and 18) set the sampling frequency coefficients for the de-emphasis filter. I have listened to the de-emphasis filter set at both 44.1 and 48 KHz while listening to a CD. The difference is very subtle, so if you frequently shift between 48 and 44.1 sampling rates you shouldn't really loose any sleep over the switch setting. That's why I didn't add all the glue logic needed to automatically set the sampling rate (it would be another 4 or 5 chips). Dip switch 3 (in front of pin 19) should be connected to pin 27 of the digital filter using a small piece of wire wrap wire (sorry I couldn't get it routed). After you have connected the pin on switch three (right in front of pin 19 on the digital filter) to pin 27, dip switch three is now used to set the filter response characteristic. Switch four (in front of pin 20) sets the digital filter timing mode. I included this switch, even though I can't really hear a difference between settings. The chart shown on the schematic describes the functions for the four dip switches. You will also need to run four wires from the analog supplies to the analog DAC chips. The pads are labeled near the DAC chips where the wires need to go. Putting a ferrite bead on these lines near the DAC chip would also be a good idea. The analog stage in the parts list is set up for 6DJ8's as I stated earlier, but it can be easily adapted to 12AT7's. To increase the filament voltage, change R5 from 499 to 250W , then to change the bias current, change R32, R33, R82, R83 from 680 to 1000W. You can also raise the B+ by substituting higher voltage zeners in the power supply. IF you were to replace the 62 volt zeners with 75 volt ones, you would have a 300 volt B+. Also included is an output mute board schematic and layout. This board shorts the analog output to ground for about 30 seconds at power-up. It also mutes the output as soon as the power begins to droop. I am including this board for several reasons. First, the nature of SRPP and cathode follower designs cause a DC offset on the output as the tube is warming up and moving to it's operating point. I measured 25 volts of offset for the first 10 seconds or so, by 20 seconds the offset is in the millivolt range. Any properly designed component should have protection from DC offset on it's inputs. But passing the buck is not my style, so I made a board that eliminates the problem. 99.9% of people will not need to use this board, but I don't want to hear that my design blew up a reader's components. I have used my DAC on my Counterpoint pre-amp, an NAD receiver, a Klyne pre- amp, and more. Any tube input device is surely safe, and any semiconductor device that is well designed should also be safe. I have never had a failure and I am not using the board. Secondly, it's a nice board to have around for power amps that warm up less gracefully than my DAC. The board works as follows: The board power is wired to the raw (before the regulator) power for the filaments of the DAC tubes. The outputs of the board are wired to the center pins of the output jacks of the DAC. Finally the ground of the board is wired to the ground plane junction of the DAC board. The heard of the circuit is a 555 timer chip that closes a relay when the RC circuit has charged up. The delay time can be changed by changing that values of the RC circuit. The recent changes make a small but noticeable improvement in my big DAC. If you are already in the throws of the older one, don't think that you now need to build this one. They will both sound very good if they are built with care and patience. And don't forget, my component choices are just suggestions. You can change many things in the design in order to "season to taste" If you like 10$ resistors, use them. If you think I used too much capacitance, use smaller caps, etc. The Torroid Corp. transformers are gorgeous, but expensive. If you are trying to save money, use standard EI core transformers. If there is enough interest, I will call a manufacturer and have boards made. So if you are interested drop me a line or call me. I can be reached at the following address and numbers (or through this magazine): Sheldon D. Stokes 102 Windy Cove Apt A. Hampton VA 23666 home: (804) 766-0392 lab: (804) 864-5520 E-mail: stokessd@sun.soe.clarkson.edu sheldon_stokes@qmgate.larc.nasa.gov