I/U stage D1 clone with 2sk389 finetuned...

[Klaus]

Hi,

Friends of Pass´isch I/V conversion! I have finally finished fine-tuning of my modified Pass D1 output stage. Fore interested guys here are the results…

D:\Witte privat\Schaltpläne\D1 with CCS_r.JPG
To remember, that is the schematic. Nothing new, just additional CCS (imported from the Aleph P1.7) and the 2SK389 in stead of the IRF610. The only open issue was: which current gives the best sound for the 2sk389.

Well, it´s really amazing how strong the sound changes with the bias current through this FET. To make it short, I see the truth between 4,5 and 5mA. 4,5mA sounds a bit like OPA AD797 = warm, not to full, no aggression at all, extreme depth, good focus and dynamic, but not the best in class. At 5mA the story grooves more like the OPA627 or so = high focus and dynamic, more frontal than deep, harder, more precision – but close to a bit of "technical" sound. BUT: it really sounds so much better than any OPA! Exactly the difference like solid state versus Pass. Well, I´ve chosen the middle, 4,75mA. My lower BC550B has 0,61V U_be, so R2 became 128Ohms. R1 to be calculated accordingly.

BTW, my shown stage is fired by 4 x PCM1704. BB SRC4192 + DSP based four times oversampling + filtering with maximum impulse correct filter settings. Parallel voltage regulators fed from massive PSU, and so on… The sound is just incredible, no more words! Mr. Pass (and Heyink/Germany for the digital DAC stuff) fulfilled my last dream, after Amp and Pre just an appropriate DAC-output stage was missing. Done…

Regards

Klaus

[stefanobilliani]

That's fine Klaus.
Ah...I am a fan of the D1 output stage too , and have sucessfully implemented with Sony and Philips dacs. Fiew 2SK389 are lying around on the desk, and without a doubt I will try your circuit.
It does it fine also without the buffers btw and it is even more simple in the construction but it is another story...

Thanks

[stefanobilliani]

...it is yet more simple the use of a jfet CCS , for those interested , like the one described in the http://www.passdiy.com/pdf/diyopamp.pdf article with 2SK170 .The 2SK170 can handle 10mA at 30volts easily.

[Klaus]

Originally posted by stefanobilliani
...it is yet more simple the use of a jfet CCS , for those interested , like the one described in the http://www.passdiy.com/pdf/diyopamp.pdf article with 2SK170 .The 2SK170 can handle 10mA at 30volts easily.

Hi,

I tested the simple CCS with just one 2SK170. Due to strong variations of the unmatched FETs I had to set the R to the right value to get the required current. Also somehow the sound was worse, not so powerfull. Maybe the CCS combination of the bipolar transistor with its high gain and the bigger IRF610 gives the "stronger" CCS, working more ideal? I don´t know, but I do not care about some transis more or less...

I like the P1.7 CCS because the basis-emitter refernce voltage of the bipolare transi is a very good reference, no matching required.

Regards

Klaus

[CheffDeGaar]

Hi Klaus,

Nice work ! As Stefano, I'm a fan of the D1 I/V stage. I already use a variation of it in my balanced DAC (PCM63PK), and I'm very tempted to try your design But I've got some questions lying around...

- Have you tried replacing the upper CCS by a mere resistor (RL) to do the I/V conversion ? Better sound with the upper CCS ?

- In my version (may be just a bit overkill...), I use a servo to drive the Fet gate, in order to keep the source at 0V. Is your arrangement with a simple voltage divider (also used by Nelson) stable ? No temp. drift ?

- Gate stopper resistors (221R) at every transistor, except at the buffer... intentional ?

- Would a referencing of the buffer's CCS to the negative rail instead of the ground improve the PSRR ?

Thanks again for sharing.

[till]

No temp. drift ?

i made the D1 stage 1:1 like manual, and yes there is temp drift, at least with my not matched Fets. But it is stable after some time for warm up, so the solution is not to turn off the DAC. I use it on balanced AD1865N-K.

[CheffDeGaar]

Thanks Till.

I've tried it before servoing, but I was unable to get a stable 0V at the input, even after a full day warmup. Well, wasn't the exact D1 stage (was not published at this time) but same spirit... But the quiescent current was higher (15 mA), so thermal stability was harder to reach.

[Klaus]

Hi,

- Have you tried replacing the upper CCS by a mere resistor (RL) to do the I/V conversion ? Better sound with the upper CCS ?
--- I have never tried the "simple" variant with RL. I think that a CCS must be creating an advantage. Too many guys described that e.g. the sound of the P1.7 with CCS is harder, more stabile and silent than the sound of simpler stages, like P1.0, L and so on. Maybe it is the stabilization effect of the CCS – acting as second regulator and decoupling the voltage supply from the gain stage, because the supply fed current is constant. So I prefer this concept: simple voltage pre-regulator w/o feedback (the simple P1.7 PSU) and the CCS as "fine" regulator. The load resistor for the I/U conversion is somehow totally decoupled from PSU influences.

- In my version (may be just a bit overkill...), I use a servo to drive the Fet gate, in order to keep the source at 0V. Is your arrangement with a simple voltage divider (also used by Nelson) stable ? No temp. drift ?
--- when I tested the bias currents I frequently not adjusted the 0V level exactly. When I did it later I heard no difference. So I think that it is important for the DAC to feed into a constant potential, but it does not matter to much if it´s exactly 0V. Just not to bring the internal voltage regulators of the switched CCS´s in the DAC into the pain to follow the signal run – like at passive I/U-conversion.
I adjusted after warm up and get +/-5mV. I do not care about it any further…

- Gate stopper resistors (221R) at every transistor, except at the buffer... intentional ?
--- just like Pass is doing at P1.7 and D1…


- Would a referencing of the buffer's CCS to the negative rail instead of the ground improve the PSRR ?
--- maybe. I decided to use the ground as CCS reference to limit heat losses.


Thanks again for sharing.
--- isn´t that the sense of this nice forum?

Regards

Klaus

[LBHajdu]

What would be a good replacement for the 2sk389? Can another IRF610 be used as in the original D1? The 2sk389 has to be order from Europe. I don’t know why it’s not more popular in the states. I try to restrict myself to dose semiconductors that come from http://www.mouser.com/ or http://www.digikey.com/.

[tiroth]

You can order them from MCM, slectronics, electronix in the US. If you only need a few I can also supply them.