Oh man, its been ~20 years. I don't know if I still have those notes, that was pre-internet era tinkering for me.
I do remember I ran the screen voltage fairly low, ran a negative voltage rail off of a backwards filament transformer, and used a zener to stabilize the screen. I had ended up with a crate of assorted tubes and scrapped a bunch of old Magnavox consoles and a few heathkit amps not long before, and had a lot of 7 and 9 pin tubes. I CCS loaded a 6J6 differential with a 6AU6 tail, ran a triode connected 6AU6 to feed the differential, and then ran push pull 6AQ5 into old Magnavox console pull output transformers. I remember ~10 watts or so at clip, and it worked really well. I eventually added a 6EU7 up front and a tone stack, and adapted it to guitar, and then proceeded to gift it to my college Eltec 208 professor.
I do remember I ran the screen voltage fairly low, ran a negative voltage rail off of a backwards filament transformer, and used a zener to stabilize the screen. I had ended up with a crate of assorted tubes and scrapped a bunch of old Magnavox consoles and a few heathkit amps not long before, and had a lot of 7 and 9 pin tubes. I CCS loaded a 6J6 differential with a 6AU6 tail, ran a triode connected 6AU6 to feed the differential, and then ran push pull 6AQ5 into old Magnavox console pull output transformers. I remember ~10 watts or so at clip, and it worked really well. I eventually added a 6EU7 up front and a tone stack, and adapted it to guitar, and then proceeded to gift it to my college Eltec 208 professor.
So kind of like Valve Wizard's (Toward the bottom here https://www.valvewizard.co.uk/ccs.html) but with negative supply?
LTP:
I have used the LM334 as CCS (constant current Sink) for the cathodes of a phase splitter.
The main issue is the minimum burden voltage of the LM334, versus the sum of the quiescent bias voltage, and signal's maximum minus voltage swing on the input grid.
The cathodes signal swing is 1/2 of the input grid signal swing on the driven grid, because of the equal impedance of the connected cathodes.
You have to pick a tube type so that the bias voltage minus the maximum negative signal swing, does not drive the cathodes' voltage below the LM334 minimum burden voltage.
If you need a lower burden voltage, then you either have to use a different CCS; use a negative voltage below the CCS, or use series batteries in the grid circuits to raise the cathode voltages.
Balanced Amp
I use a common resistor for the common cathodes.
If you use a CCS there, the maximum plate voltage swing is about only half of the plate voltage swing when you use a common cathode resistor.
I have used the LM334 as CCS (constant current Sink) for the cathodes of a phase splitter.
The main issue is the minimum burden voltage of the LM334, versus the sum of the quiescent bias voltage, and signal's maximum minus voltage swing on the input grid.
The cathodes signal swing is 1/2 of the input grid signal swing on the driven grid, because of the equal impedance of the connected cathodes.
You have to pick a tube type so that the bias voltage minus the maximum negative signal swing, does not drive the cathodes' voltage below the LM334 minimum burden voltage.
If you need a lower burden voltage, then you either have to use a different CCS; use a negative voltage below the CCS, or use series batteries in the grid circuits to raise the cathode voltages.
Balanced Amp
I use a common resistor for the common cathodes.
If you use a CCS there, the maximum plate voltage swing is about only half of the plate voltage swing when you use a common cathode resistor.
I have never heard a noise coming from my amplifier designs and builds.
Last year was the first time I connected any of my amplifiers to 100dB+ loudspeakers.
(I admit, that was my new balanced amplifier, there is no LM334 in it).
Food for thought:
In an LTP phase inverter, the noise of an LM334 CCS is effectively Cancelled.
That noise is one phase from one plate, and the opposite phase from the other plate (and the amplitudes are equal too).
Send that on to the push pull output stage, and it effectively cancels there in the output transformer, at the secondary output.
(another vote for using an output transformer). Yay!
I consider that circuit to be one of the very few times when Gaussian Noise is effectively cancelled.
Good luck with doing that with other circuits.
Hint: try and create two independent Gaussian Noise sources that are in anti-phase and equal amplitudes through time.
(Do not waste your time trying that. If it cancels, it is either not independent noise sources; is not Gaussian, or Both).
Last year was the first time I connected any of my amplifiers to 100dB+ loudspeakers.
(I admit, that was my new balanced amplifier, there is no LM334 in it).
Food for thought:
In an LTP phase inverter, the noise of an LM334 CCS is effectively Cancelled.
That noise is one phase from one plate, and the opposite phase from the other plate (and the amplitudes are equal too).
Send that on to the push pull output stage, and it effectively cancels there in the output transformer, at the secondary output.
(another vote for using an output transformer). Yay!
I consider that circuit to be one of the very few times when Gaussian Noise is effectively cancelled.
Good luck with doing that with other circuits.
Hint: try and create two independent Gaussian Noise sources that are in anti-phase and equal amplitudes through time.
(Do not waste your time trying that. If it cancels, it is either not independent noise sources; is not Gaussian, or Both).
Last edited:
A constant current sink in the cathode circuit of the tube will provide and extremely high effective resistance in that setting. The sink needs to operate at low burden voltage, or else a negative supply would be needed anyway.
With or without current source, you can bias the grids well above ground so you don't need a negative supply.
Yes, a floating differential self biased above a CCS as a tail will work well. A decent CCS wont need more than a few volts.