JojoD818 said:
Yes, but...
It appears that the original design must have used fixed bias, since there's an extra 50V winding on the power transformer unused in the production version. Now why in the world would the engineers have changed to cathode bias?
My hunch is the power supply. The HF87 uses a doubler, which is not exactly a champion when it comes to regulation. Cathode bias conditions are closer to class A, less current swing, thus less reliance on a stiff supply. It may well be that the fixed bias version suffered from high power distortion and compression because of dropping rail voltage.
SY said:
Hi SY,
that's an excellent observation. now say that I would use a beefier tranny and caps to go with a tube rectifier (5AR4 or maybe 5U4GB), would it be a good project?
I have just finished a tube amp but my mind and body is still screaming for more tube projects.
Thanks and Season's Greetings,
JojoD
Why use a tube rectifier if you want to make the regulation better? Other than the pretty glow...
If I were doing a scratch build, I'd use a full wave circuit with moderate capacitance and a very solid regulator. But solid state doesn't put me off when pressed into the service that it renders best- constant voltage or constant current.
If I were doing a scratch build, I'd use a full wave circuit with moderate capacitance and a very solid regulator. But solid state doesn't put me off when pressed into the service that it renders best- constant voltage or constant current.
A quick sim indicates that you could get 460V at 200mA with very low ripple if the transformer is a good one. A pair of EL34s and a 6600 ohm output transformer and you've got a nice little 40 watt per channel amp.
I'm still unconvinced about the real-world merits of fast rectifiers. I can measure a difference if I look at stuff going on within the transformer-rectifier-first filter cap loop. But when I look at the output of the actual supply, the noise spectra and impedance seem the same.
Now, this may not be the case where grounding is poor and some of the rectifier switching noise gets coupled into the circuit. But the solution there is proper grounding.
I'm still unconvinced about the real-world merits of fast rectifiers. I can measure a difference if I look at stuff going on within the transformer-rectifier-first filter cap loop. But when I look at the output of the actual supply, the noise spectra and impedance seem the same.
Now, this may not be the case where grounding is poor and some of the rectifier switching noise gets coupled into the circuit. But the solution there is proper grounding.
I have the irons (6.5K it says), the EL34s are easy to find here, also have a pair of spare 6SN7GTB and a 7025 (5751 too but may not be suitable?). I have spares of UF4007s too from a former project and a bunch of 1N4007s .
How would I know the amount of initial negative voltage at the grid of the EL34s? Would it just be a straightforward transformation from a cathode bias to fixed bias?
I really appreciate your response and so many thanks.
JojoD
.How would I know the amount of initial negative voltage at the grid of the EL34s? Would it just be a straightforward transformation from a cathode bias to fixed bias?
I really appreciate your response and so many thanks.
JojoD
SY said:
Being a member for almost 3 years here has taught me a lot of valuable information regarding proper grounding. I have come to respect the importance of proper grounding for it makes or breaks a project.
I remember having lengthy hours of troubleshooting that all ends up due to a bad grounding scheme.
Jojo, typically one has a greater bias (i.e., lower idle current) in a fixed bias scheme. For a 6600 ohm plate-plate load and a 450V B+, a typical idle current would be 50mA per tube. I might be tempted to look at distortion spectra while I varied it to see how low I could go, but 50mA is pretty close to correct.
Now, the Eico circuit is connected in ultralinear, so the screen and plate voltages will be equal, and both be about 440V or so in this scenario. Turning to the EL34 data sheet, we see that this will happen with a grid bias of roughly -38V. So if you set up a voltage divider with a potentiometer in the middle, you'd want to configure it to have an adjustment range of perhaps 33-43 volts.
You might also consider using 6L6-type output tubes. In my HF87, I've used EL37 and 7027A very successfully, but they do require higher bias and drive voltages. OTOH, more power and lower distortion...
Now, the Eico circuit is connected in ultralinear, so the screen and plate voltages will be equal, and both be about 440V or so in this scenario. Turning to the EL34 data sheet, we see that this will happen with a grid bias of roughly -38V. So if you set up a voltage divider with a potentiometer in the middle, you'd want to configure it to have an adjustment range of perhaps 33-43 volts.
You might also consider using 6L6-type output tubes. In my HF87, I've used EL37 and 7027A very successfully, but they do require higher bias and drive voltages. OTOH, more power and lower distortion...
I knew someday I'll have use for these tubes.
Have seen several commercial preamps using tube rectifiers and solid state regulation. This is what I plan to do to the concertina and phase splitter sections of my Williamson inspired tube amp.
What improvements should I expect if I do this? Anyone have a first hand experience on trying this?
cheers,
JojoD
What improvements should I expect if I do this? Anyone have a first hand experience on trying this?
cheers,
JojoD
JojoD818 said:
I reduced the gain of the amp (it's too high for modern sources) by altering the feedback resistor. To keep feedback constant, I also changed the first tube to 6SL7 (actually, a 6SU7 or 5691). The 6SN7 has matched 33K plate resistors and a cathode CCS. It also has a protection diode from grid to cathode. Input stage and phase splitter have solid state regulation, a variant of the ubiquitous Maida circuit.
As it happens, I'm doing another rebuild now.
JojoD818 said:
The tube rectifier gives a slow warmup, which isn't really needed for under-400V-or-so circuits. Other than that, I don't see any advantage to it and several disadvantages (heat, reliability, cost).
SY,
Wow, that would be a huge leap for me. What I'll do is to build it as is then apply changes later to maximize my learning experience.
The schematic has uneven plate resistors for the 6SN7s, one is 33K the other 28,750 ohms. Is that a means of balance? I suppose it's like that to compensate for the 1M on the grids of the phase inverters.
I also couldn't explain the use of a 470mmf cap across the opt center tap and the plate of the top EL34?
I'm also experimenting with MJE340 as a ss regulator for my amp's front-ends.
JojoD
Wow, that would be a huge leap for me. What I'll do is to build it as is then apply changes later to maximize my learning experience.
The schematic has uneven plate resistors for the 6SN7s, one is 33K the other 28,750 ohms. Is that a means of balance? I suppose it's like that to compensate for the 1M on the grids of the phase inverters.
I also couldn't explain the use of a 470mmf cap across the opt center tap and the plate of the top EL34?
I'm also experimenting with MJE340 as a ss regulator for my amp's front-ends.
JojoD
The uneven plate resistors are indeed done that way for balance- the gain of the grounded grid side is slightly lower than the driven grid side if the resistors are equal. Going to a CCS corrects that imbalance.
The 470pF cap is an interesting tweak. The output transformer is apparently not wound in sections but as one long winding. This is evident since the outer layer has a slightly higher resistance than the inner. In any case, the capacitances also vary, so the transformer is slightly imbalanced between primary halves at very high frequencies. The 470pF cap equalizes that.
The 470pF cap is an interesting tweak. The output transformer is apparently not wound in sections but as one long winding. This is evident since the outer layer has a slightly higher resistance than the inner. In any case, the capacitances also vary, so the transformer is slightly imbalanced between primary halves at very high frequencies. The 470pF cap equalizes that.
Aha, so I was right regarding the plate resistors. An imbalance in the plate resistors will then give balance to the phase inverters. Learning CCS applications would have to wait for now though I'm sure I'll get there soon.
Thanks for the explanation on the cap, I was thinking of putting another one on the bottom side but that would have nulled the cap's purpose bigtime!
Thanks for the explanation on the cap, I was thinking of putting another one on the bottom side but that would have nulled the cap's purpose bigtime!
It is stunningly easy to implement a cathode CCS for the 6SN7. I've attached my circuit. Basically, the two LEDs are used as voltage references; the resistor string runs 5mA through them, which is a good minimum current. Clearly, the exact value of the 80K resistor is not critical- it can be 75K or 82K just as arbitrarily.
The trimmer resistor sets the tail current so that I can tune the 6SN7 plate voltage finely. A 500 ohm device is perfect. Start with it adjusted to 400 ohms. The upper transistor is an MJE340 or similar moderately high voltage NPN capable of dissipating a couple of watts. A small heatsink is sufficient. The lower transistor can be any NPN, but preferably, it will be one with a healthy beta and ft. The BC549 is a classic choice- I used an NTE device because I have a bag of them.
The entire CCS can be built on a small perfboard for about $5, less than the cost of a high quality power resistor.
The trimmer resistor sets the tail current so that I can tune the 6SN7 plate voltage finely. A 500 ohm device is perfect. Start with it adjusted to 400 ohms. The upper transistor is an MJE340 or similar moderately high voltage NPN capable of dissipating a couple of watts. A small heatsink is sufficient. The lower transistor can be any NPN, but preferably, it will be one with a healthy beta and ft. The BC549 is a classic choice- I used an NTE device because I have a bag of them.
The entire CCS can be built on a small perfboard for about $5, less than the cost of a high quality power resistor.
Attachments
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.