Hello guys, I am prototyping an amplifier with the output tube biased at 34mA (#45). I've decided to remove the cathode resistor (especially the cathode resistor capacitor!) having better bass plus more details and control the bias with the grid voltage (around 50V). For a peace of mind I would like to monitor the bias current. I've got a microammeters with the "0" in the middle, so if there is no current flowing the meter pointer remains in the middle.
On the 0R47 (relatively small resistance that should not affect the output impedance) current sensing resistor there is about 16mV voltage drop, opamp amplifies it to 1.6V. On the opposite there is 1.6V reference voltage after the opamp buffer. At 34mA I've got 0V between them and the current is not flowing. If the bias current will change I got the meter pointer moving to direction that depends on the voltage difference.
Is it an optimal solution? Is there a better idea?
More: Anyone know how to automatically slowly control the grid bias voltage V5 (50V) to control the bias current at the constant 34mA? 😎
PS. I am not checking the average voltage drop on the OPT as the copper has quite large temperature coefficient of resistance and they warm up easily.
On the 0R47 (relatively small resistance that should not affect the output impedance) current sensing resistor there is about 16mV voltage drop, opamp amplifies it to 1.6V. On the opposite there is 1.6V reference voltage after the opamp buffer. At 34mA I've got 0V between them and the current is not flowing. If the bias current will change I got the meter pointer moving to direction that depends on the voltage difference.
Is it an optimal solution? Is there a better idea?
More: Anyone know how to automatically slowly control the grid bias voltage V5 (50V) to control the bias current at the constant 34mA? 😎
PS. I am not checking the average voltage drop on the OPT as the copper has quite large temperature coefficient of resistance and they warm up easily.
Last edited:
You circuit looks cute. But without completely analyzing it, I think that the poor 50uA
meter movement will be hard pegged at one end, until the 45 tube warms up.
Where is the meter protection?
And your circuit does not have any integration of the cathode current, which will vary wildly during large signal conditions, the meter will be shaken to death.
Fix the meter issues, and at least you can see that the current is where you desire it, regardless of what you do for a fixed/variable/feedback - grid bias system.
As to automatic bias control using - grid bias, the requirements are a cathode current sense resistor, (in the 45's case 'center of filament' & with DC filament power), operational amp(s), negative bias supply voltage that runs the op amp(s), and a very long time constant integrator loop for the op amp(s).
I believe that the thermionicaudio has part or all of that.
meter movement will be hard pegged at one end, until the 45 tube warms up.
Where is the meter protection?
And your circuit does not have any integration of the cathode current, which will vary wildly during large signal conditions, the meter will be shaken to death.
Fix the meter issues, and at least you can see that the current is where you desire it, regardless of what you do for a fixed/variable/feedback - grid bias system.
As to automatic bias control using - grid bias, the requirements are a cathode current sense resistor, (in the 45's case 'center of filament' & with DC filament power), operational amp(s), negative bias supply voltage that runs the op amp(s), and a very long time constant integrator loop for the op amp(s).
I believe that the thermionicaudio has part or all of that.
Last edited:
It won't monitor the bias on a meter (but you could rig one up) but it will match bias on all output tubes so you don't have to bias it anymore.
I don't know, if is it better, but simpler. 🙂
1R "cathode resistor" paralleled with 50-100mV DC meter.
Panel Meter 100 mV DC, round - Tube-Town GmbH
IMHO amount of current is more important than the degree of change.
I would not say so. To be optimal, there are much components required to carry out a simple task, i.e. to monitor cathode current.
I would say optimal current monitor is simply 1...2.2 ohms cathode resistor and suitable current meter (some 100 uA) across it.
This size cathode resistor (< 10 ohms) has negligible effect to output impedance.
Oh, no. This is not my product, I merely recommend it.
It's true it uses opamps and a 10R sense resistor, but I find it to be transparent, and saving the hassle of rebiasing is nice (I use toroidal outputs and they like a balanced DC flow).
There's more info on it in this thread --> Automatic bias board.
You will find out quick enough if the needle bounces on loud bass passages.
Depends on the mass, resonance, and damping of the d'Arsonval movement.
Also depends on the 2nd harmonic distortion caused by the 45 and decreasing XL at low frequencies.
Depends on the mass, resonance, and damping of the d'Arsonval movement.
Also depends on the 2nd harmonic distortion caused by the 45 and decreasing XL at low frequencies.
PRR, awesome, thank you for that, I will try it, if it start bouncing at low frequencies as 6A3sUMMER said I will try to implement this (it's over complicated now but I've added simple meter protection and an integrator):
Inspired by John Broskie article I have created a simple auto-bias circuit for my needs.
I am sure Pavel's auto-bias module is great and universal in many different situations but for something as simple as SE something like this should be sufficient. Also I would need to use two modules and use only 50% of them in monoblocks.
I have used LT components as I have already the models in LTSpice, shunt regulator with accuracy of 0.05% and low temperature drift 10ppm/°C max. and high voltage op amp. The integrator works very slow, even 10Hz signal won't make a big difference. Resistors R5, R6 and especially R4 should have lower possible tolerance and small temp drift.
It works great, at least in the simulation 🙂
I am sure Pavel's auto-bias module is great and universal in many different situations but for something as simple as SE something like this should be sufficient. Also I would need to use two modules and use only 50% of them in monoblocks.
I have used LT components as I have already the models in LTSpice, shunt regulator with accuracy of 0.05% and low temperature drift 10ppm/°C max. and high voltage op amp. The integrator works very slow, even 10Hz signal won't make a big difference. Resistors R5, R6 and especially R4 should have lower possible tolerance and small temp drift.
It works great, at least in the simulation 🙂
Last edited:
Nice work. Next thing I'd like to figure out how to do is automatic bias with a microcontroller like an Arduino. I found this: https://scholar.uwindsor.ca/cgi/vie...gle.ca/&httpsredir=1&article=1118&context=etd
- Status
- Not open for further replies.