Below is an LTSpice model of a prospective PP amp with 1625 using UNSET topology. Of course the real amp will have adjustable bias and grid and screen resistors etc but first I just wanted to rough this out. Input will be from a balanced source. This model is biased for the outputs to remain in class A and develops about 20W at clipping. Spice says it seems to want 10-11K primary impedance transformers at this plate voltage of around 400V. Any major flaws? Other things to consider?
20W Class A with 6L6GC is possible (and also good, IMHO). No experience with this circuit but it is certainly possible with standard pentode output stage. By memory 350V plate voltage, 250V g2, bias for 70mA on each tube and 8K plate-to-plate. That's pure Class A with low THD around 2-3% wthout any feedback, With some feedback THD can be really low.
You have an IXCP10M45S as the CCS in the cathodes of your LTP. With the grids at ground potential the CCS will have about a volt across it which will not work. My testing with a 10M45 is that it wants at least 20 volts across it for best results. Anything more than that will not hurt it (up to 450 volts) but will not help either. I typically tie it to the negative bias supply in my Universal Driver board design which is often -150 volts. This can put too much voltage across the HK of the 12AT7 on start up because there is no current flow. Some would use a zener from the cathodes to ground to fix this, but I prefer a NE-2 neon bulb. The zener is a voltage varying capacitance, while the neon bulb is a fixed 1 pF capacitor below it's breakover voltage at which point it becomes a tiny VR tube clamping the voltage.
I have never actually tested the UNSET circuit with a 6L6 type tube but it should work fine.
I have never actually tested the UNSET circuit with a 6L6 type tube but it should work fine.
Ah, I wondered about that. The datasheet suggested it would work but maybe a read it wrong, but real world testing trumps all. Thanks for that.
So if I tie the bottom of the CCS to a safe HK voltage that would be best and simplest?
So if I tie the bottom of the CCS to a safe HK voltage that would be best and simplest?
Last edited:
I am updated my design and hope to begin prototyping one channel soon.
I bought one of Pete Millets universal PP driver boards and will use that for the balanced input and the driver circuit. I also made an UNSET PCB for the 1625, pic below. I modeled my circuit with using the 12GN7A in triode made for the driver and the LTSpice simulations are excellent. The planned transformer is 5.2K PP which Edcor has and I biased for mostly class A. Power output is about 25W at clipping and 20W in class A.
FFT at 1W
FFT at 10W
UNSET PCB
I bought one of Pete Millets universal PP driver boards and will use that for the balanced input and the driver circuit. I also made an UNSET PCB for the 1625, pic below. I modeled my circuit with using the 12GN7A in triode made for the driver and the LTSpice simulations are excellent. The planned transformer is 5.2K PP which Edcor has and I biased for mostly class A. Power output is about 25W at clipping and 20W in class A.
FFT at 1W
FFT at 10W
UNSET PCB
Not yet, these build are winter to spring projects. Only built it in spice.
Yes it is directly inspired by George. The front end is Peter Millet and the output is George.
You still have a problem with voltage for the current source in the diff amp tail.
Swap the + and - connections for V5 voltage source to make it a negative voltage supply.
Cheers,
Ian
P.S. interested as I have 12 NIB Aperex 807, the ones with ceramic bases intended for high altitude use in B52 bombers.
Swap the + and - connections for V5 voltage source to make it a negative voltage supply.
Cheers,
Ian
P.S. interested as I have 12 NIB Aperex 807, the ones with ceramic bases intended for high altitude use in B52 bombers.