Anyone tweaked a California Audio Labs Alpha DAC?

[beau2317]

I just picked up a good deal on a used California Audio Labs Alpha DAC. It has a tube output stage.

I have not looked in detail yet, but read somewhere on the net that it has a pair of 12AX7s, one per channel. Both triodes are wired in parallel, as a cathode follower, with a solid state current source.

B+ is said to be 140V

I was wondering if anyone has tried tweaking this? Or have any suggestions.

I've not been impressed normally with the 12AX7, but the DAC makes a pretty good sound out of the box.

What other tubes could I try?

[Jeb-D.]

That seems pretty weird to use 12ax7's as output tubes even if it is small signal stages.

Anyhow there is a lot of 12ax7 options try switching the tubes brands and variations first. It may not make a big sonic difference though since the output is as a buffer and is probably only swinging a couple of volts.

Switching to different tube types would probably require tweaking resistor values.

[beau2317]

Yes I thought it quite weird. But I get the feeling that this was put in more as a marketing exercise than for any real reason (there are a couple of SS gain stages before this).

Also this sort of thing often happens when you get SS engineers asked to design tube bit. They look for a standard circuit and then use the most common tube..

[nuvistor]

I have a 1993 version of this DAC with the potted power transformer bearing the Alpha logo. I agree that the tubes are a marketing feature (gimmick?) but they do influence the sound, unfortunately the goal apparently was to add veiling and reduce clarity, probably to reduce the stridency of the typical CD of that time.

The 12AX7 CCS-loaded cathode follower is the output stage of an op-amp made of discrete components, using resistive-loaded small-signal NMOSFETs with NPN CCS for differential input pair, followed by PNP transistor VAS with Miller compensation loaded with NPN CCS, which drives the 12AX7s. The opamp DC output voltage is kept close to 0V with a servo loop using a 741 opamp. I replace the noisy MOSFETs with a matched JFET pair (U406) and after awhile replaced the 12AX7 cathode follower with a 5965 and then a JFET-NPN follower powered from the +15V rail.

You noticed that one 12AX7 is used per channel, the sections are wired in parallel.

If you want to stick with tubes for the output stage, I suggest something with higher gm that will take the 140V B+ and change the output CCS current set resistor to a lower value for more current to suit the new tube, I liked 5965 with CCS at about 8mA, but the higher filament current required reducing the value of a resistor in series with the filaments. You probably could use 12AU7 with somewhat less current. The higher plate current increases temperature of the plate voltage regulator, and the power transformer runs warmer, so you might want to remove the cover in use.

I thought the JFET-BJT follower sounded pretty close to the 5965 without running warmer.

Although I used matched JFETs, discrete JFETS like J201 or PN4302 would probably work OK since the servo compensates for the Vgs mismatch, the original NMOSFET pairs were not matched.

Another possibility is bypassing the op amp and connect the CS4328 output directly to the output jacks, it's specd to drive 600 ohms, but I haven't tried this.

[beau2317]

Thanks for that info. I tried all the tubes in my parts bin that would theoretically work, 12AU7, 12AT7, 6072A and found that the 12AU7 sounded best.

I am considering taking the output directly from the DAC as well as some other options.

Cheers

[beau2317]

Wow I took the output directly from the DAC chip, with a 51R/10n filter and it sounds great. So much better than via the regular output.

Seems such a waste to put all those transistors and tubes in to make the sound so much worse.

Thanks for the tip.