1) Turned round the negative voltage sources so as not to confuse.
2) Is it better to drive the screens with high impedance by increasing R14, R15 or low impedance and use resistors to ground to reduce the screen voltage. What works better when the 6ej7's are not matched? R13, R16 would be a 100R pot for DC balance.
I choose the 6ej7 just because I have a load upstairs and I can match a few. I used them for 10.7MHz FM IF stripe before finding the gain was too high. I now use 6bw7's here. I have 6ba6, 6ac7, ef80, and a few others but don't think they are suitable - all seem lower voltage than the 6ej7 ratings some variable mu.
3) trying to find a spice model of the 6528. They don't seem easy to obtain in the UK. Does look easier to drive. Probably not as visually attractive. Given I have a collection of 6550 amps would be nice to have something look a bit different.
Tried a 12BY7 in simulation. This works too but needs the screen lowering with resistors to ground. Its important to keep R19. Without it the screen goes to near 0V and this causes a much higher saturation voltage on the plate which limits the drive range.
2) Is it better to drive the screens with high impedance by increasing R14, R15 or low impedance and use resistors to ground to reduce the screen voltage. What works better when the 6ej7's are not matched? R13, R16 would be a 100R pot for DC balance.
I choose the 6ej7 just because I have a load upstairs and I can match a few. I used them for 10.7MHz FM IF stripe before finding the gain was too high. I now use 6bw7's here. I have 6ba6, 6ac7, ef80, and a few others but don't think they are suitable - all seem lower voltage than the 6ej7 ratings some variable mu.
3) trying to find a spice model of the 6528. They don't seem easy to obtain in the UK. Does look easier to drive. Probably not as visually attractive. Given I have a collection of 6550 amps would be nice to have something look a bit different.
Tried a 12BY7 in simulation. This works too but needs the screen lowering with resistors to ground. Its important to keep R19. Without it the screen goes to near 0V and this causes a much higher saturation voltage on the plate which limits the drive range.
I would guess that driving the screens with a lower impedance would be best, since they do draw some current. You will still need comparable attenuation of the output V.
For matching, the cathode pot should work for balancing gm, but you may still find a DC imbalance. The coupling caps to the outputs should take care of that.
I curve traced an RCA 12BY7A here to see what the triode curves look like. 50V/div horizontal, 5 mA/div vertical, 2 V steps
Some obvious "rollover" in the triode curves toward the HV end, that will have some 2nd harmonic. Not a super triode, but a typical triode.
I don't think I have a 6EJ7/EF184 here to trace. Maybe I can find a cheap one on Epray. One thing I noticed on the EF184 datasheet is that it has a high plate saturation voltage in pentode mode, like 125V. This will limit the output swing some, but not sure if that would be modeled in the Spice model.
For matching, the cathode pot should work for balancing gm, but you may still find a DC imbalance. The coupling caps to the outputs should take care of that.
I curve traced an RCA 12BY7A here to see what the triode curves look like. 50V/div horizontal, 5 mA/div vertical, 2 V steps
Some obvious "rollover" in the triode curves toward the HV end, that will have some 2nd harmonic. Not a super triode, but a typical triode.
I don't think I have a 6EJ7/EF184 here to trace. Maybe I can find a cheap one on Epray. One thing I noticed on the EF184 datasheet is that it has a high plate saturation voltage in pentode mode, like 125V. This will limit the output swing some, but not sure if that would be modeled in the Spice model.
Last edited:
I've ordered up some 6C33's £160 for 4 NOS, a couple of 1650M transformers (not so cheap!), and a SMPS 12V 400W (EPP-400-12) supply for the heaters. Plan is to try something out on a steel chassis and tagboard to see what works. Love the curve tracer. The simulation of the 12BY7 is not as good as the 6EJ7 on distortion.
The saturation I see:
http://www.r-type.org/pdfs/ef184.pdf
Seem to be 30V at the currents I am using (8ma) page C7 for vc2=140v as long as you have some vg2.
You asked how the screen voltage tracks the plate voltage on the EF184. Plate is blue. The model is Ayumi Nakabayashi which are normally very good.
The saturation I see:
http://www.r-type.org/pdfs/ef184.pdf
Seem to be 30V at the currents I am using (8ma) page C7 for vc2=140v as long as you have some vg2.
You asked how the screen voltage tracks the plate voltage on the EF184. Plate is blue. The model is Ayumi Nakabayashi which are normally very good.
Attachments
Last edited:
Here is Tom Shlangen's 6EJ7 Triode 30K Loadline adjusted for reality.
In this case driving a120K Rg on the following grid. So 24K load results.
Not as much swing as before & the RHS loses more than the LHS.
That results in increased 2H, altho in this example still looks OK.
220K on the 6C33s grids seems far to high on a tube with a very large grid structure.
I'm thinking 50K max. What do the spec sheets tell us? Perhaps not much.
The 6C33s look like a PS passer tube. In that service not a problem.
On Patrick Turners plot of the 6C33 Plate Family it looks as though its mu is ~2.5.
Low mu tubes on fixed bias are always in danger of self destruction, due to current hogging.
I watched many years ago as one of the grids in a 6080 melted.
So have always used some cathode biasing as a safety measure.
In this case driving a120K Rg on the following grid. So 24K load results.
Not as much swing as before & the RHS loses more than the LHS.
That results in increased 2H, altho in this example still looks OK.
220K on the 6C33s grids seems far to high on a tube with a very large grid structure.
I'm thinking 50K max. What do the spec sheets tell us? Perhaps not much.
The 6C33s look like a PS passer tube. In that service not a problem.
On Patrick Turners plot of the 6C33 Plate Family it looks as though its mu is ~2.5.
Low mu tubes on fixed bias are always in danger of self destruction, due to current hogging.
I watched many years ago as one of the grids in a 6080 melted.
So have always used some cathode biasing as a safety measure.
Attachments
jhstewart9
That`s my opinion also , on the first place to use automatic but separate and adjustable self bias for each 6c33c tube , under such condition I will use B+ of not more than 265Vdc , expecting about 50W output from two 6c33c-b tube working in PP-A1 class , but for fixed bias operation I will stick with Penta Lab. datasheet recomendation using separate negative bias trim pot for each triode and B+ of 200Vdc max.
That`s my opinion also , on the first place to use automatic but separate and adjustable self bias for each 6c33c tube , under such condition I will use B+ of not more than 265Vdc , expecting about 50W output from two 6c33c-b tube working in PP-A1 class , but for fixed bias operation I will stick with Penta Lab. datasheet recomendation using separate negative bias trim pot for each triode and B+ of 200Vdc max.
Looks like I can get hold of 6CL6 pentode. There a simulation model too. That would allow more current in the LPT so the leak resistors can drop in value. As I said servo bias is more like cathode bias as long as you have adequate negative supply. I see max grid leak 200K and max grid current 15uA (1000h of service) so I can only go by the data sheet. That would only put 1.5V across 100k leak which the servo could easily deal with. 300uA would be 30V which is still OK.
Last edited:
Yeah and those curves aren't matched to reality, from what I have seen, as far as grid voltage vs ma. But I think what they show as far as linearity are legit.
Can you expand on this? Do you mean when both sides of the 6080 are paralleled, one side draws much more current than the other?
Here is some useful guidance when using low mu triodes in parallel.
In regulated PS passer service it is recommended to use resisters in each cathode in order to force current sharing.
The resister in the cathode is (mu + 1) times as effective compared to in the plate side.
The App Note from 1948 tells us paralleled SE operation is not recommended.
Last edited:
I'll see - depends on the input stage and the amount of GNBF. I can add a bit of bootstrap from one plate back to the other which increases the maximum drive voltage a bit and means I don't need that awkward -10V supply for the CCS. Its all simulation at the moment - need to try something out.
