Working with Current O/P DAC Chips

[kevinkr]

Used 10 ohm resistors with 5751 and AD 18 bit current output dacs about 17 yrs ago in a Sony CDP - this does work.. Friend now has the modified Sony which is still working.

Have used both grounded grid and common cathode voltage amplifier topologies.

Last dac with tube output was an Assemblage Dac 1 with 6SL7 and 6189 in a cascade for about 50dB gain.

32 - 50dB is the required range depending on the specific dac and the value if the I/V resistor chosen which in some cases should be chosen to assure that the audio voltage on the output of the dac is not high enough to cause linearity problems in the dac. Some older AD and burr brown dacs were specified with a max voltage on the current output before linearity began to suffer.

I started with conventional 1% mf and went to exotic Vishay bulk foil types, my impression was that if there was an improvement it was very subtle and probably not worth the money spent.

Sowter makes a transformer as well for this chore, and sounds far better than the active circuitry it was compared to. Buffering was by 6922 (basically novel floating cf outputs) - see www.triodeguy.com for the details of his adcom mods. (Note that the Adcom information is not available, however a very similar circuit is used in the SACD player mod.)

Any of these approaches are well worth the effort imho.

[Brian Beck]

Thanks for the tips Kevin. Of course, the link to the SACD design shows a transformer/floating CF output stage intended for a voltage output DAC. Filtering and buffering a voltage output DAC is a bit easier than doing the I/V conversion for a current output DAC.

FWIW, I posted this design for a voltage output DAC about 18 months ago. It works very well, although I'm not sure how it would compare to the transformer-coupled buffer. (It also makes a good line stage, less the LC filter).

Output stage amplifier and filter for voltage output DAC

[kevinkr]

Hi Brian,
Unfortunately I didn't look too closely at the SACD version, the original version in the ADCOM IIRC actually had a current output dac and I think the difference was the addition of a resistor across the secondary of the transformer, the reflected (to primary) terminating resistance being equivalent to something like 10 ohms or so..

I have run across a few instances of transformers (Sowter or Llundahl) being used for I/V conversion without buffers into relatively high impedance line stage inputs with exceptionally good results.

Your circuit is interesting, I particularly like the idea of the LRC based low pass filter. I was planning something similar for the DSD decoder in my sacd player using some gates to buffer the stream, which would then drive an LPF/integrator and buffer. It may never get built. I'm pretty disenchanted with sacd, particularly since I rediscovered vinyl recently..

[oshifis]

I just came across this thread. Check out my transformer/tube I/V stage that I finished recently:

Laszlo's Valve Output Stage with Lundahl transformer

[oshifis]

Originally posted by kevinkr
32 - 50dB is the required range depending on the specific dac and the value if the I/V resistor chosen which in some cases should be chosen to assure that the audio voltage on the output of the dac is not high enough to cause linearity problems in the dac.

With my arrangement the I/V resistance seen by the DAC is 2.8 ohms, the voltage on it is 4 mV RMS, 11.3 mV peak-to-peak, well in the safe region. The transformer voltage gain is 24 dB, the tube has 30 dB, that makes 54 dB (500x).

[Brian Beck]

Oshifis,

You've got a clever circuit which I've never seen before. It's nice to see something new instead of rehashed old designs.

I'm still trying to wrap my brain around the behavior of the two secondaries in the circuit, both as regards their effect on the SRPP behavior and also on each other, wired as they are. And I'm trying to understand the secondary load impedance that each half secondary "sees", given the mutual coupling through the tube. A narrative circuit description would be appreciated if you can talk us through it. A simulation would be valuable, but I prefer to understand the circuit first.

I looked up the data sheet of the LL1678, and I note that the DCR for each primary is 4.5 ohms. How do you figure 2.8 ohms is possible given that DCR? I would be impressed by anything lower than 10 ohms.

[Brian Beck]

Originally posted by kevinkr
I'm pretty disenchanted with sacd, particularly since I rediscovered vinyl recently..

Kevin,

I can sing along with that tune!

But at least most SACDs sound better than most CDs.

BTW, our local Best Buy's music department has maintained a "high-definition" section for DVD-A and SACD discs. Each time I pass by, the dedicated shelf space is squeezed smaller and smaller. I haven't checked in months, but it may have completely disappeared by now. I figured that this shelf-space metric was a telling indicator of the larger global problem.

[oshifis]

Originally posted by Brian Beck
I looked up the data sheet of the LL1678, and I note that the DCR for each primary is 4.5 ohms. How do you figure 2.8 ohms is possible given that DCR? I would be impressed by anything lower than 10 ohms.

You are right, I did not involve the DC resistance in the calculation. Nevertheless I thought on it, and 2 pairs of the primaries are connected in parallel, it can be seen on the diagram as the pin numbering (termination alternative C, but with separated secondaries). This way the DC resistance of the half primary is 2.25 ohms, plus the secondary 375 ohms transformed to the primary (1.5 ohms) which adds to the 2.8 ohms represented by the cathode impedance. This all makes 6.55 ohms. At 4 mApp this is 26 mVpp, around the permitted voltage compliance of the TDA1541A current output.
A simulation would be great (anybody?), I have no tools for this. I can measure THD, IMD, S/N, spectrum, frequency response, transient response, etc. I am driving only the half primary from a single DAC for now, but I am very pleased with its sound.

[Brian Beck]

Originally posted by oshifis
You are right, I did not involve the DC resistance in the calculation. Nevertheless I thought on it, and 2 pairs of the primaries are connected in parallel, it can be seen on the diagram as the pin numbering (termination alternative C, but with separated secondaries). This way the DC resistance of the half primary is 2.25 ohms, plus the secondary 375 ohms transformed to the primary (1.5 ohms) which adds to the 2.8 ohms represented by the cathode impedance.

I see; yes it would be 4.5/2 with paralleled primaries. But the primary impedance is not just the secondary DCR (375 ohms) reflected back to the primary; it would also involve reflected impedances from the rest of the SRPP circuit environment. The interaction of the two secondary windings within the SRPP makes it a bit more complicated.

Originally posted by oshifis
I can measure THD, IMD, S/N, spectrum, frequency response, transient response, etc. I am driving only the half primary from a single DAC for now, but I am very pleased with its sound.

Well, the proof is "in the pudding", but I'd like to understand the circuit behavior better. I'll have to "cogitate" on it some more.

BTW, you could drive 2mA DC though the unused primary via a simple CCS to balance the DC current, until you can implement the dual DAC scheme.

[oshifis]

Originally posted by Brian Beck
The interaction of the two secondary windings within the SRPP makes it a bit more complicated.

It can be imagined as the output transformer of a push-pull tube power amplifier, if it helps creating the simulation model. SRPP is push-pull.