High-Voltage Entertainment in the Home

CDs and Their Discontents

(Low voltages only)

 

1.     Single Board DAC with a Tube

I started out with a Rotel 955 and this little SPDIF DAC kit from Scott Nixon, which is based on the very obsolete Cirrus CS8412 receiver (although, that's a CS8414 in there now) and Philips TDA1543 DAC, a chipset that was very popular in the DIY community around the turn of the century.  The 6DJ8 in the picture serves as a cathode follower buffer to give the TDA1543 enough power to drive the cabling; it essentially replaces the unity gain op-amp that would otherwise sit after the DAC.  By the way, that's a diamond-base Telefunken 6DJ8, pulled from an ancient HP O-scope extracted from a dumpster at the Randall Physics Lab, and no, you may not have it (the scope was late-50s vintage and the tube still looks brand new and tests at 100%).   I supplied the toroid and the RF filter; the board has a cunning power supply that pulls multiple voltages, up to 24 VDC for the tube, off the 12VAC trasnformer.  It compares well with the significantly modified Sony SACD player below, quieter and maybe just a little crisper on the percussion. Interestingly, the TDA1543, a long-obsolete chip with significant sample-to-sample variation in accuracy, has acquired something of a cult following, based on the idea of using masses of them in parallel to improve conversion accuracy by appealing to the Law of Large Numbers (this only applies if the conversion errors between chips are indpendent, which does in fact seem to be the case here.  It is also the case that, unlike the CS8412, the TDA1543 can be had in quantities for less than $2 apiece).  However, given its modest heritage, this is a robust little unit (unless you, ahem, use the wrong transformer out of your junk box and overheat the whole thing).  It now sits in my office with the Rotel. 

 

 

2. Sony SACD Player

Removed the muting circuit;

Removed the low-pass filtering in the feedback loop around the final op-amp;

Replaced the bypass and coupling capacitors in the audio stage with Panasonic low-impedance FC electrolytics, available from Digi-Key, which are astonishingly good (I actually preferred them to some polypropylenes coupling caps I tested);

Removed the headphone driver (it loads down the analog stage);

Replaced crucial late-stage coupling caps with copper foil-in-oil types;

Replaced the on-board 11.3MHz clock with one of Guido Tent's excellent X02 clocks;

Bypassed the jitter-inducing clock buffer (thanks to Pat Di Giacomo at Analog Research, who distributed Guido's clock);

Replaced the output stage driver with the high-quality Burr-Brown OPA-627.

I would like to figure out how to improve the performance of the 33.9MHz clock that is produced by a 3x multiplier off the 11.3MHz clock, so if you have any inspirations, please let me know. The original modification involved removing the second-order low-pass filter that keeps 11.3MHz out of the analog path. This was how I found out that my pre-amp's frequency range is flat out to 11.3MHz; at first, I thought my oscilloscope was broken. I couldn't believe that propagating RF energy throughout the system was a reasonable idea (sorry, Larry). I have since replaced the second order filter with a single cap to form a first-order filter that performs the same function but has no effect below 80kHz.

On normal CDs, the result is excellent, but not transcendent. The high end is smooth, not excessively grainy or strident, and percussion is pretty crisp and sharp, but still somewhat digital. High-end transients, like those generated by Shelly Manne's sock cymbals--and Shelly is at my house a lot--are where digital lets me down.  Bass notes have well-defined pitch. Location of instruments is excellent, and the background is fairly black, but not a higher volumes. I only have a few SACDs, and I can't really tell any difference between a good CD and SACD, so my opinion on SACD versus DVD-A is a big shrug, both these formats are DOA, anyway. Larry now tells me DVD-A is the way to go, but I won't purchase any new format disks until they force me, the same way they're trying to make me buy a digital TV, so I can see really crisp, clear corporate crud.

3.    DDDAC1543

(No picture--the DDDAC is visiting Rick Francis in Toledo).

Again failing to achieve audio nirvana--the CD player still gets beat pretty much every time by the turntable--I acquired a DIY DAC kit, the DDDAC1543 MK2, from Doeda Douma with PCM2707 USB and CS8414 SPDIF receivers that talk to the I²S bus and 48xTDA1543 I²S-to-analog DACs built on four 12-chip towers. The TDA1543, running at 8 volts, can produce 2V P-P output with a simple I/V resistor, but is not very linear.  Running 48 DACs in parallel results in about an additional 2.5 bits of linearity over and above a single TDA1543 (12-13 bits resolution), which puts us close to the resolution of a 16-bit CD (before noise shaping).   Many DACs also supply much current, so cabling becomes irrelevant (I think this is a real issue in pulling the analog directly off the PCM2707).  The setup allows for some interesting comparisons, as the above DAC has a CS8414 plus a single TDA1543, so we can compare the USB receiver versus the SPDIF receiver, and the single TDA1543 versus 48 of them.  

Since at least one intrepid builder indicated they have had problems assembling the towers, let me give you some hints.  Take the heat sinks apart like Doeda recommends, and file down the rough edges.  Assemble the boards, but don't put the parts in that would obstruct soldering the TDA1543s.  Start the tower as indicated.  Make sure every time you put in a TDA1543, you check, double-check and triple-check that it's pointed in the right direction, because unsoldering these babies is much less fun than soldering them.  Before you put a TDA1543 on the tower, bend the pins in ever so slightly with pliers so that they will touch the pins on the DIP package below before soldering.  Check the orientation and alignment, then put another heat sink part on top. Check the alignment again before soldering, making sure the pins are touching on both sides.  Then, tack solder exactly one pin on each side of the DIP package and re-check the alignment. If the package is level and the pins are all touching, tack solder the remaining six pins in any order.  Contrary to Doeda's advice, I did use heat sink grease on the end pieces.  Apply the grease and tie the end pieces tight to the tower with wire prior to soldering the heat sink pins.  I left the wire on.  Don't forget to put the rest of the pieces on the board.

The power supply has an RF filter, an Avel-Lindberg toroid good for 5A, Fairchild "stealth" diodes and a capacitance multiplier power supply with separate LM317 voltage regulators for each board.  The unit is dead quiet at any volume, and talks to the SPDIF output on the Sony, above, and the USB port on the Mac Mini without problem.  The 48 TDA1543s are sufficient to drive any reasonable cabling.  So far, my observation is that I am hard-pressed to tell the difference between the SPDIF and USB interfaces on this unit, but that 48 TDA1543s are much, much better than one.  Many DAC chips audibly enhances the dynamic range over a single chip (however, as Chris Beard has pointed out, the naive paralleling scheme may cause the chips to interact, and may also lead to cascading failure if one chip decides to short out.  So, this may be another transitional technology).   The sound is  rich, buttery, if you will, with better resolution on those pesky transients.  But, it still doesn't beat the vinyl, which leads us to the next step.

4. Apogee Mini-DAC

More details here on the Apogee Mini-DAC.  Like the DDDAC1543, this cute little unit from Apogee Digital accepts both SPDIF and USB data (and several other formats), and reclocks the input before the actual D/A conversion.  This unit, however, reads the input stream using a 'loosely-coupled' clock, puts the data into a FIFO buffer, and then writes the data out of the buffer using a 'stiff clock'.  I have thought for a long time that this is the path to immunity to transport independence, but this is the first unit I have seen to actually do it for less than a stratospheric price tag.  Plus, it is tiny, it has the USB interface and has balanced XLR outputs, how cool is that? 

Both this unit and the DDDAC1543, connected via USB, were immediately recognized by a variety of laptops and picked up the Sony's SPDIF stream. They both sound rich and balanced, with pleasing percussion, soundstage, and a rich bass.  Neither beats good vinyl.  Rick Francis, Michael Gartz and I compared the three DACs on Mike's Quad+Celestion subwoofer rig (featuring spiffy Larry Moore 300B amps), using a Mac Mini and a vanilla DVD player as transports.  The tubed DAC was immediately appealing with a punchy midrange, but faded fast against the competition.  After several hours of listening, we decided that the USB source was less veiled than the SPDIF source, and that the difference between the two was larger on the DDDAC1543 than on the Apogee.  However the difference did not go away entirely on the Apogee, which I find puzzling--the buffering is supposed to make the Apogee essentially source-independent.  We all preferred the Apogee over the DDDAC1543, but only by a small margin, agreeing that we could live very well with either, but that we preferred vinyl (and, Mike's got a nice vinyl rig, too).  I think we're running up against the limitations of the 16 bit technology.   We also preferred listening to the music off the hard drive, rather than directly from the CD.

So, my new CD drive is a Mac Mini with two mirrored Iomega 500GB  hard drives (that's right, four drives in extruded aluminum cases), managed by iTunes.  I use Apple Lossless Encoding with error correction, so an average CD occupies 35MB.  One drive is entirely full, the other only about 10%, and I estimate I can get my entire CD collection on the two drives.  I use the simplest disk mirroring possible; once a month, I turn on the mirror drives and type /bin/cp -R, because hipsters prefer the command line.

After all this, I am done messing around with CD for awhile.  I just found a quartet of 813s in the basement.  Amazing what turns up in Spring cleaning.

Return to the rig