Such large cathode capacitor is overkill.
Even if you use CCS in the cathode, the AC blocking capacitor value not orders of magnitude larger, than in cathode bias case.
Sample:
Larger C1 gives lower -3dB point, but can cause anomalies (in this case due to the Schade feedback).
It's value must "fine-tuning", depending of CCS parameters (such stabilizer, simple or cascode BJT, FET etc.).
I wouldn't really call this a hybrid bias at all. It's strictly a CCS controlled bias, as long as enough voltage overhead for valve variation is accommodated, and allowable G1 leak resistors' value can be chosen by that. Seems like a great idea to me.
All good fortune,
Chris
All good fortune,
Chris
The same thing can be done with just a small sampling resistor and a servo to the G1 bias voltage. No voltage overhead required, no bypass cap time constant (instead, there must be one in the servo), different layout for similar result, but not fashionable.
All good fortune,
Chris
All good fortune,
Chris
Chris,
If you reacted to my sample, it's only demonstration of cathode blocking capacitor value, not to hybrid bias problem.
Can be combined with fixed bias, but I'm not sure that worth it.
No, I'd only meant that yours and post #1 are functionally the same CCS bias. This is a good choice for purely class A amplifiers, although even they will have some small variation of cathode averaged (DC) current with signal level. This variation in cathode current with level causes some envelope variation (envelope time constant set by bypass capacitance x reciprocal of Gm // CCS stiffness) of valve bias, a "breathing" effect. Only purely "fixed/but adjustable" G1 bias doesn't have this effect.
An ideal, perfect class A amplifier would have none of this effect, so if you're close enough to perfect, you could still get into Heaven. Depends on the criteria.
All good fortune,
Chris
An ideal, perfect class A amplifier would have none of this effect, so if you're close enough to perfect, you could still get into Heaven. Depends on the criteria.
All good fortune,
Chris
I don't see that this is hybrid bias. Agreed, you have a fixed voltage on the grid, but the valve's operating current is determined solely by the CCS and is unchanged by the voltage source. It is always CCS bias. What the voltage source does do is effectively vary Vak (for a fixed anode voltage). There is a disadvantage to this scheme compared to resistive cathode bias and that is in the phase response. A practical capacitor bypassing a small resistance (a few hundred ohms) is unable to contribute much of a phase deviation in the low frequency audio band. But when that capacitor is bypassed by a CCS, it can contribute a large phase deviation at low frequencies. The upshot is that CCS bias makes low frequency harder to achieve when global negative feedback is applied.
I could see myself using this bias method. If one was using an LM317 like device (max 37V) as the constant current element it's operating voltage could be kept in a comfortable linear operating region with the rest of the bias voltage supplied as a neg grid voltage.
Remembering that the time constant is set by the CCS in parallel with 1/Gm, which helps to moderate the effect.
All good fortune,
Chris
That is an interesting circuit. I don't recall seeing it before. Imagine one valve draws slightly too much current. Its cathode voltage rises slightly and that lifts the voltage on which the potential divider to the other valve's grid sits to go more positive, turning the other valve on more to balance the system. Those potential dividers (22k, 330k) must need careful choice of values to prevent the system driving itself into runaway. But if the circuit is from 1966, resistor tolerance will be pretty poor. Anything as tight as +/-5% was usually marked. I wonder if the extra complication was really justified.
And the (pure pentode) circuit has regulated g2 supplies. That must be straining the heater-cathode insulation of that ECC88. And if the HT was 400V, that 100k, 100k potential divider puts Vak for the ECC88 at 200V. ECC88 is only rated for 130V. Not a good design choice.
And the (pure pentode) circuit has regulated g2 supplies. That must be straining the heater-cathode insulation of that ECC88. And if the HT was 400V, that 100k, 100k potential divider puts Vak for the ECC88 at 200V. ECC88 is only rated for 130V. Not a good design choice.
It cannot runnaway. Reducing the 22k resistors simply forces the two valve to settle closer to the average of the two currents they would have conducted without the cross coupling.
Apparently it was 150V at the cathode of the ECC88.
Merlinb,
Thanks!
An interesting example of cross bias.
That kind of bias has been discussed in one or more threads of Tubes / Valves.
By any other name, that cross coupled bias has been called: "Garter Bias."
I leave the origin and beginnings of that kind of bias up to the reader/researcher.
A search of diyAudio should find lots of examples.
If one tube gets excess gas causing higher current . . . the other tube will "follow".
I never liked Garter Bias.
Thanks!
An interesting example of cross bias.
That kind of bias has been discussed in one or more threads of Tubes / Valves.
By any other name, that cross coupled bias has been called: "Garter Bias."
I leave the origin and beginnings of that kind of bias up to the reader/researcher.
A search of diyAudio should find lots of examples.
If one tube gets excess gas causing higher current . . . the other tube will "follow".
I never liked Garter Bias.
I use hybrid bias in a Citation 5 amplifier, which uses fixed bias. All I did is take out the 10 ohm resistors in the cathode that were used for measuring the bias and replaced them with 100 ohm resistors. I did it because it was suggested in the past by someone who has responded to this thread that it would help keep the tubes balanced. My system is sensitive, so I don't miss any watts lost in the cathode resistors.
Yes, it is very good to have very closely matched plate currents in the output transformer primary.
By the way, there was a very good push pull 300B amplifier in "Sound Practices" magazine that used those Citation output transformers.
Back to your modified cathode resistance . . .
The 100 Ohm cathode resistors, are they bypassed with a capacitor?
I am guessing they are not.
That will increase the plate impedance, rp.
If you use the KT88 / 6550 tubes in Beam Power mode, I bet you are using a medium amount of Global Negative Feedback.
That mostly takes care of the increased plate impedance, rp.
A Triode Wired KT88 has a plate impedance, rp, of about 800 Ohms.
When you use an un-bypassed 100 Ohm cathode resistor, that plate impedance will increase by an Additional 800 Ohms, so the total plate impedance, rp, is now 1600 Ohms.
(it turned out that both numbers were 800 Ohms, the increase in rp = u x cathode resistor; 8 x 100 Ohms = 800 Ohms.
Ultra linear gives the plate impedance, rp, something in-between the Beam Power mode, and Triode Wired mode.
Have Fun!
By the way, there was a very good push pull 300B amplifier in "Sound Practices" magazine that used those Citation output transformers.
Back to your modified cathode resistance . . .
The 100 Ohm cathode resistors, are they bypassed with a capacitor?
I am guessing they are not.
That will increase the plate impedance, rp.
If you use the KT88 / 6550 tubes in Beam Power mode, I bet you are using a medium amount of Global Negative Feedback.
That mostly takes care of the increased plate impedance, rp.
A Triode Wired KT88 has a plate impedance, rp, of about 800 Ohms.
When you use an un-bypassed 100 Ohm cathode resistor, that plate impedance will increase by an Additional 800 Ohms, so the total plate impedance, rp, is now 1600 Ohms.
(it turned out that both numbers were 800 Ohms, the increase in rp = u x cathode resistor; 8 x 100 Ohms = 800 Ohms.
Ultra linear gives the plate impedance, rp, something in-between the Beam Power mode, and Triode Wired mode.
Have Fun!
Last edited:
Decker,
The original cathode resistors were 10 Ohms. 10 Ohms at DC, 10 Ohms at 20Hz, and 10 Ohms at 20kHz.
Now, you have 100 Ohms, from DC to 20kHz.
A 1000uF bypass capacitor is 8 Ohms capacitive reactance at 20Hz.
A 100 Ohm cathode resistor and 1000uF bypass capacitor, is similar to . . .
An amplifier with a 1000 Ohm self bias resistor, and a 100uF bypass cap.
The original cathode resistors were 10 Ohms. 10 Ohms at DC, 10 Ohms at 20Hz, and 10 Ohms at 20kHz.
Now, you have 100 Ohms, from DC to 20kHz.
A 1000uF bypass capacitor is 8 Ohms capacitive reactance at 20Hz.
A 100 Ohm cathode resistor and 1000uF bypass capacitor, is similar to . . .
An amplifier with a 1000 Ohm self bias resistor, and a 100uF bypass cap.
jcalverez,
You are correct!
The longer recovery time of self bias after an overload is the higher impedance of the CCS (versus a self bias resistor).
But the total recovery time is according to 2 parallel impedances, which the bypass capacitor "sees":
1. The CCS impedance is in parallel with the cathode impedance, and that is what "drives" the capacitor to recover. A relatively long recovery time.
2. The Self Bias Resistor impedance is in parallel with the cathode impedance, and that is what "drives" the capacitor to recover. A relatively short recovery time.
You are correct!
The longer recovery time of self bias after an overload is the higher impedance of the CCS (versus a self bias resistor).
But the total recovery time is according to 2 parallel impedances, which the bypass capacitor "sees":
1. The CCS impedance is in parallel with the cathode impedance, and that is what "drives" the capacitor to recover. A relatively long recovery time.
2. The Self Bias Resistor impedance is in parallel with the cathode impedance, and that is what "drives" the capacitor to recover. A relatively short recovery time.