Last night I took out my meter and made some measurements in my SSMH 12AU7 build. I had some problems measuring the gate to ground voltage as it is a very high impedance circuit. My DMM would report only 13.71 and 13.62 volts on the gates with 11.82 and 11.71 volts at the tube heaters. The difference is less than two volts!
I had to bring out my old analog FET VOM with 10Mohm input impedance to get a more reliable measurement of ~16.1 and ~16.2 volts (the "~" is because it's an analog meter). Gate to source voltage was thus ~4.2 volts. Much better.
Shamazo, your measurements are definitely strange since they're not lower but much higher Might it be that you reversed R2 and R4?
I also measured the voltages at Q3 and Q4 in the constant current source plate loads. Q3 being my main concern since I knew it would bare the highest voltage drop of the two After the amp has run for a while I saw a voltage of 1.1 across Q4 while voltages of 23 and 24 volts showed up across each Q3. I must have gotten really mixed up a few days ago when reviewing the
datasheets for the BC327 since now that I recheck them I see Vceo listed as 45 V and Vces = 50 V. Since the only way the transistor would see more than 45 volts in regular operation would be that the tube became a short circuit (highly improbable, though not impossible), I deem that the BC327 is
just adequate for the application. Definitely go with the 2N5087 or BC556 if you want increased peace of mind.
By the way, measuring both the plate to ground voltate and the voltage across Q3 were also affected by my DMM's input impedance. I could only get reliable measurements with my FET VOM.
Finally, I noticed that the heater voltage was lowish so I measured the voltage drop across R13 and noticed it was too high. This was due to the higher current set by the CSS going through it. Since R13 sits at the top of the gate bias voltage divider, it was lowering the gate bias.
So I removed the CCS from that junction and connected them directly to the 48 V line from the power source. Since the CSS do a great job of decoupling the tube circuits from each other and from the output MOSFETs I think this configuration won't be a problem. Naturally, I'd definitely like to hear the opinion of someone with more experience building amplifiers with CCS plate loads. Another option would be to wire independent 'R13a' and 'R13b' for each CCS.
After moving the CCS directly to the 48V line the voltage at the MOSFET gates was now ~17 and the heater voltage went up to ~13.2. A little hot but quite alright. I did some listening tests and the amp worked quite alright. No hum and only by pushing the volume up real high was I able to hear the amp's noise floor through the Grado SR-80i
This showed something that I observed when I originally worked with the 12AU7 and 17EW8: I don't see a way to increase the MOSFET bias, that is to increase the voltage from gate to source in this amp. The voltage of the source seems to be "clamped" some 4.2 volts below the voltage set at the gate as when I increased the gate voltage the source voltage moved up correspondingly. Might this be because some sort of CCS-like behavior from the tube heaters? That by holding the drain-source current at ~150mA the source voltage is "clamped" below the gate voltage?
Shamazo: If you're seeing 19 volts at the heaters, what's the voltage at the MOSFET gates in your build?
Next I'll setup the amp and CCS loads to work with the 17EW8 and see how that measures.
cheers!