Was reading Merlin's HiFi Preamp book, he talks about the golden ratio and 1/3 Imax with 2/3 of B+ (or HT).
Since ra is roughly 3k in a 6922.
(My B+ is 250V):
Ra should be 2xra.. so 6k. Isnt it too low?
Imax then would be 41.6mA.
And center bias would be.. 2/3 of B+ = 167V ; Imax/3 = 13.8 mA.
Is this correct?
How should I choose proper plate resistor?
Thanks.
Since ra is roughly 3k in a 6922.
(My B+ is 250V):
Ra should be 2xra.. so 6k. Isnt it too low?
Imax then would be 41.6mA.
And center bias would be.. 2/3 of B+ = 167V ; Imax/3 = 13.8 mA.
Is this correct?
How should I choose proper plate resistor?
Thanks.
Use at least 3x Ra, or more if your B+ is on the high side.
With triodes, the higher the plate resistor, the lower the distortion.
Then adjust the cathode resistor for the desired plate current and dissipation (1.6W max).
You don't need the maximum possible peak to peak output voltage unless it's a driver stage.
This datasheet has some example values. For your B+, they specify 47k and 1.2k.
https://www.tubeampdoctor.com/media/pdf/24/19/71/tad-datenblatt_rt-6922_190912_version_1_0.pdf
With triodes, the higher the plate resistor, the lower the distortion.
Then adjust the cathode resistor for the desired plate current and dissipation (1.6W max).
You don't need the maximum possible peak to peak output voltage unless it's a driver stage.
This datasheet has some example values. For your B+, they specify 47k and 1.2k.
https://www.tubeampdoctor.com/media/pdf/24/19/71/tad-datenblatt_rt-6922_190912_version_1_0.pdf
Last edited:
Yes, notice that with the larger plate resistor, the distortion is about 10dB less.
The mechanism is simple. The anode resistance is nonlinear, but not the plate resistor.
So the larger the plate resistor is (compared to the anode resistance), the less effect
the nonlinear anode resistance has on the distortion. Of course, the cathode resistor
will also help with linearization, especially for smaller signals.
The mechanism is simple. The anode resistance is nonlinear, but not the plate resistor.
So the larger the plate resistor is (compared to the anode resistance), the less effect
the nonlinear anode resistance has on the distortion. Of course, the cathode resistor
will also help with linearization, especially for smaller signals.
Last edited:
You could always play with this loadline calculator 
---> Universal loadline calculator for vacuum tubes - Vacuum Tube Amplifiers - DIY
It's a great way of picturing the math.
---> Universal loadline calculator for vacuum tubes - Vacuum Tube Amplifiers - DIYIt's a great way of picturing the math.
When figuring your operating points, you must stay within the maximum ratings of the tube you're using!
For 6922 (each triode):
Max Vp = 220V (However, service life ratings are given with Vp = 100V, which is a lot lower)
Max plate dissipation = 1.5W
If your plate voltage is 200V, the max plate current would be 7.5mA (for Pdiss = 1.5W).
If your B+ is 300V, you need to drop 100V across the plate resistor, so I get 7.5k for Rp.
Does anybody out there run their 6922 tubes with >200V on the plates? Usually it's more like Vp = 100V and Ip = 8mA, or somewhere around there.
For 6922 (each triode):
Max Vp = 220V (However, service life ratings are given with Vp = 100V, which is a lot lower)
Max plate dissipation = 1.5W
If your plate voltage is 200V, the max plate current would be 7.5mA (for Pdiss = 1.5W).
If your B+ is 300V, you need to drop 100V across the plate resistor, so I get 7.5k for Rp.
Does anybody out there run their 6922 tubes with >200V on the plates? Usually it's more like Vp = 100V and Ip = 8mA, or somewhere around there.
Last edited:
@emartine:
You might want to play with Merlin's Load Line Plotter spreadsheet. It has curves for the E88CC/6DJ8/6922 tube. You can download it from this page:
The Valve Wizard
You could start with the following parameters and experiment from there:
Vsupply = 250
Ra = 27000
Rk = 1000
This will give an operating point that is near center bias and well below the tube's dissipation limits.
You might want to play with Merlin's Load Line Plotter spreadsheet. It has curves for the E88CC/6DJ8/6922 tube. You can download it from this page:
The Valve Wizard
You could start with the following parameters and experiment from there:
Vsupply = 250
Ra = 27000
Rk = 1000
This will give an operating point that is near center bias and well below the tube's dissipation limits.
Last edited:
Where is our external load??
WITH a load, THD will decline as Rp is increased, to a point, and then rise again. Hang 10k load on that 39k Rp and it will struggle.
The OP said that this stage will be loaded with a cathode follower. I think it would be safe to assume that the cathode follower will have an input impedance much larger than the output impedance of this stage, which is the plate resistor in parallel with the tube's dynamic plate resistance (about 3k).
Depending on the datasheet you are looking at, the 6DJ8 has a maximum plate-to-cathode rating of about 130 volts, and the 6922 around 180 volts. I use the 6922 a good bit and I generally operate it at a plate-to-cathode voltage between 90 and 120 volts. A B+ supply of 200 to 250 volts works for me.
Some people recommend running the 6922 fairly hot, at 8 to 10 mA depending on Vp-k. I've always gotten good results with plate current in the 3 to 5 mA range, which keeps plate dissipation low. But I admit that this is a personal preference. Others are free to disagree.
Last edited:
You could always play with this loadline calculator
It's a great way of picturing the math.
Yes. Thanks. That is what I was using. but the Excel spreadsheet from ValveWizard (I just found out) seems really great!
So many variables...
1. Do you need a lot of voltage swing from the output of the stage? If yes, then try to get close to center bias (triode can swing both as far as possible down towards 0 bias and up towards cutoff).
2. Is a smaller voltage swing from output OK, but you desire best linearity? If yes, then you may want to arrange your operating area along the loadline so that the triode only operates in the area where the plate curves are spaced most 'evenly' and symmetrically (indicating greatest linearity).
3. Do you need the stage to drive a difficult/heavy load? If so, then you may want the most power out, so Rp/rp ratio of closer to 2:1 could be better. This implies more plate current. Watch that maximum plate dissipation limit.
4. Another way to arrange for best linearity/most voltage swing is to choose as high a B+ as possible so you can get the load line as 'flat' or horizontal as possible. This implies a larger value Rp (plate load resistor) to attain a higher ratio of Rp to rp, perhaps up to 10:1.
That's not often possible, though, because it's not easy to drop enough volts across Rp (the load resistor) and still leave enough plate-cathode voltage for the tube to operate well.
In that case, you may want to look at an active load, placing a current source as the plate load (usually a depletion mode MOSFET like DN2540 or similar). This can be made even better by using a pair of these MOSFETs in a cascode arrange, making a *constant* current source (CCS). Then the choice of operating point is less critical, limited more by headroom requirements, staying with maximum ratings, achieving sufficient voltage swing out, etc.
But yeah, everything is a compromise between this and that, so there is no one easy formula.
1. Do you need a lot of voltage swing from the output of the stage? If yes, then try to get close to center bias (triode can swing both as far as possible down towards 0 bias and up towards cutoff).
2. Is a smaller voltage swing from output OK, but you desire best linearity? If yes, then you may want to arrange your operating area along the loadline so that the triode only operates in the area where the plate curves are spaced most 'evenly' and symmetrically (indicating greatest linearity).
3. Do you need the stage to drive a difficult/heavy load? If so, then you may want the most power out, so Rp/rp ratio of closer to 2:1 could be better. This implies more plate current. Watch that maximum plate dissipation limit.
4. Another way to arrange for best linearity/most voltage swing is to choose as high a B+ as possible so you can get the load line as 'flat' or horizontal as possible. This implies a larger value Rp (plate load resistor) to attain a higher ratio of Rp to rp, perhaps up to 10:1.
That's not often possible, though, because it's not easy to drop enough volts across Rp (the load resistor) and still leave enough plate-cathode voltage for the tube to operate well.
In that case, you may want to look at an active load, placing a current source as the plate load (usually a depletion mode MOSFET like DN2540 or similar). This can be made even better by using a pair of these MOSFETs in a cascode arrange, making a *constant* current source (CCS). Then the choice of operating point is less critical, limited more by headroom requirements, staying with maximum ratings, achieving sufficient voltage swing out, etc.
But yeah, everything is a compromise between this and that, so there is no one easy formula.
The stage gain requirement is at least as important as selection of an operating point. With a large plate resistor value, and especially a constant current source plate load, the gain of the stage will approach the amplification factor of the tube (ignoring cathode degeneration). This is about 33 for the 6DJ8/6922, which is likely way too much gain for a line stage, but would be appropriate if you are designing a phono preamp, as an example.
So I agree that good engineering is all about making intelligent compromises.
So I agree that good engineering is all about making intelligent compromises.
There has been a lot of good theoretical information provided here. How about a practical circuit?
Assuming that you are looking for a line amp with modest gain, attached is an example LTspice file that uses the 6922 in a single gain stage with a direct-coupled cathode follower. The gain is about x3, achieved with cathode degeneration in the first stage. This does increase the output impedance of that stage but this is mitigated by the cathode follower, which has a low output impedance roughly equal to the reciprocal of the transconductance of the 6922 (less than 100 ohms).
The operating point of the gain stage is depicted in the screen shot of Merlin's Load Line Plotter spreadsheet.
Maybe this will give you a place to start.
