A fellow where I used to work is/was a brilliant circuit designer. Power supply group. EE education in Russia. I asked him if he did any audio circuits in his EE training curriculum, specifically power tube output stages. He said the only one they analyzed the theory of, was a resistive loaded plate capacitively coupled to an OPT.
I remember some hype around using PA line distribution transformers as an OPT for a single ended design. Cheap and readily available. But, no gap to allow it to handle the DC bias, as when designed, they werent expecting a DC bias.
So, what aspect of a resistive loaded, capacitively coupled to an OPT output stage is so bad, that you never see it (?) as a viable single ended design? Is it because you lose efficiency; a 10W SE amp done with the "right" gapped OPT becomes only 5 with the scheme?
So what? Low power tube amps for guitar seem to be popular; perhaps its possible to get that SE tone without having to spend $150 on an OPT designed to work in that circuit. Just move the mic a little closer to the speaker.
I remember some hype around using PA line distribution transformers as an OPT for a single ended design. Cheap and readily available. But, no gap to allow it to handle the DC bias, as when designed, they werent expecting a DC bias.
So, what aspect of a resistive loaded, capacitively coupled to an OPT output stage is so bad, that you never see it (?) as a viable single ended design? Is it because you lose efficiency; a 10W SE amp done with the "right" gapped OPT becomes only 5 with the scheme?
So what? Low power tube amps for guitar seem to be popular; perhaps its possible to get that SE tone without having to spend $150 on an OPT designed to work in that circuit. Just move the mic a little closer to the speaker.
Well, for one thing, a 400VDC supply suddenly needs to be a 600-800VDC supply. ;-) To change the load (say from 2.5K to 5K) you have to redesign the whole power supply. My Williamson amplifiers would need an 850VDC supply instead of 450VDC. You're throwing away a lot of heat and power.
Not so uncommon as you wrote ...
Capacitively coupled opt is called "parafeed", search on this forum or look at:
https://www.tubecad.com/2014/09/blog0308.htm
Resistive load would be extremely inefficient, so usually choke load or totem pole is commonly done.
Capacitively coupled opt is called "parafeed", search on this forum or look at:
https://www.tubecad.com/2014/09/blog0308.htm
Resistive load would be extremely inefficient, so usually choke load or totem pole is commonly done.
Try to do a quick design using a common 6BQ5/EL84 with 300v on the plate and biased to 50mA with a 5K load. What plate load resistor value would you use, and what primary Z for the OPT would you use? What PS B+ would you start with?
"Try to do a quick design using a common 6BQ5/EL84 with 300v on the plate and biased to 50mA with a 5K load. What plate load resistor value would you use, and what primary Z for the OPT would you use? What PS B+ would you start with?"
OK, simple design - use a 10K resistor and a 10K transformer, to get a net 5K load. Then you get half the power and the resistor gets the other half. And you need another 500 volts, for a total 800v power supply.
OK, simple design - use a 10K resistor and a 10K transformer, to get a net 5K load. Then you get half the power and the resistor gets the other half. And you need another 500 volts, for a total 800v power supply.
Why would you need that kind of voltage if you only wanted a couple of miserable watts?
70 volt transformers aren’t a 5k load, even at 5 watts. Load impedance go down, voltage requirements go down with them. The problem is then a 6BQ5 is not the right tube. Need something that can swing more current on a 250-300V power supply, to match the needed 1k plate load. It’s a job for a big fat sweep tube. Some of them can be had for cheap, just like the transformer.
70 volt transformers aren’t a 5k load, even at 5 watts. Load impedance go down, voltage requirements go down with them. The problem is then a 6BQ5 is not the right tube. Need something that can swing more current on a 250-300V power supply, to match the needed 1k plate load. It’s a job for a big fat sweep tube. Some of them can be had for cheap, just like the transformer.
I haven't seen a single high-quality line to speaker transformer. They are all designed for PA use.
You are not going to get audiophile boutique hi-fi out of it. No one expects it to. “Move the mic a little closer to the speaker” implies guitar amp. A three watt guitar amp (for practice, in an apartment) is often too loud.
Its just that using the right quality magnetic components is one of the keys to getting a good distorted guitar sound. Unless maybe you want it REALLY grungy.
It sounds like an experiment to me, not something somebody is going to gig or record the next hit rock single with.
If someone wants perfection, they just have to shell out.
If someone wants perfection, they just have to shell out.
The original premise is wrong. He's willing to get 5W from the amp and thinks that takes a $150 SE OPT.
Probably unwilling to spend $75, or even 50, too. Those cute little 5 watt open frame Hammonds are going to cost that nowadays. For 5 miserable watts and 100-10 KHz.
70-volt transformers usually have a multi-tapped primary - for example, a 16-watt transformer might have taps at 1, 2, 4, 8, and 16 watts. The impedance of the primary is then 5000, 2500, 1250, 625, and 312.5 ohms respectively. But the transformer iron can handle 16 watts at ANY tap without saturation, including the "1 watt" tap. And if it handles 16 watts at 100Hz, it will handle 16 watts at 50Hz on the 4-watt tap, and 16 watts at 25Hz on the "1-watt" tap.
Lower impedance at the same power requires higher current, and the load resistor still dissipates the same power - 25 watts in my EL84 post (50mA times 10,000 ohms equals 500 volts). No free lunch.
Higher impedance taps use thinner and thinner wire, in order to get all those turns on the bobbin. Might get away with 4 watts on the 1 watt tap before the wire has something to say about it. Don’t try it at 16, at least not with a guitar amp.
If you want a robust transformer those little still-inexpensive Antek 25 and 50 VA power trafos sound better than they have a right to - better frequency response and distortion than cheap 70V PA line trafos. Available in all sorts of impedances, as high as 18k and down in the low hundreds. The only catch is the limitation of around 240V RMS with 50 Hz operation. For 80 Hz operation you can hit it with 400 RMS if you keep within primary and secondary current limits (20-50% overload would be ok with audio). For push pull operation the DC needs to be well balanced, just plug-and-play random EL84’s or cheap sweeps with cathode bias and no adjustment at all might not be balanced enough. Any sort of balance adjustment or matched tubes any you’d be fine. The original concept of capacitively coupled removes the balance requirement and will let you get away with injecting all manner of second harmonic. Power would only be limited by the 20-ish % efficiency of such an arrangement and EL84’s dissipation limit. Go to a 20 watt sweep tube and you can get your 5 watts. 6L6’s or KT88 are possible, but then the price really ratchets up. For something that isn’t all that efficient, and is guitar amp quality.
If you want a robust transformer those little still-inexpensive Antek 25 and 50 VA power trafos sound better than they have a right to - better frequency response and distortion than cheap 70V PA line trafos. Available in all sorts of impedances, as high as 18k and down in the low hundreds. The only catch is the limitation of around 240V RMS with 50 Hz operation. For 80 Hz operation you can hit it with 400 RMS if you keep within primary and secondary current limits (20-50% overload would be ok with audio). For push pull operation the DC needs to be well balanced, just plug-and-play random EL84’s or cheap sweeps with cathode bias and no adjustment at all might not be balanced enough. Any sort of balance adjustment or matched tubes any you’d be fine. The original concept of capacitively coupled removes the balance requirement and will let you get away with injecting all manner of second harmonic. Power would only be limited by the 20-ish % efficiency of such an arrangement and EL84’s dissipation limit. Go to a 20 watt sweep tube and you can get your 5 watts. 6L6’s or KT88 are possible, but then the price really ratchets up. For something that isn’t all that efficient, and is guitar amp quality.
Some random ideas:
For SE: A SS CCS up top to balance out DC, without a lot more B+ needed.
For some cheap OT that has an extra LV winding, could use a DC balancing CCS on the extra winding.
For SE, you could series connect cheap OTs and the same for some small speakers.
A big TV Sweep tube for higher current, lower V, and then add UnSet drive to get great triode curves and a few more Watts besides.
A SS current mirror plate load (with current gain, WCG, to preserve tube sound ) could increase current linearly for a smaller tube and could give you effectively a big P type tube.
Use two Sweep tubes for output, one with a C mirror WCG up top and the other with a C mirror WCG below the cathode. Combine the current outputs for a practical OTL P-P output.
Two Sweep tubes with C mirror plate loads WCG for separate low V OTL like outputs, and drive a dual voice coil woofer. One wired for push and the other wired for pull. Could ground the B+ side and use a B- on the cathodes for safety at the speaker.
For SE: A SS CCS up top to balance out DC, without a lot more B+ needed.
For some cheap OT that has an extra LV winding, could use a DC balancing CCS on the extra winding.
For SE, you could series connect cheap OTs and the same for some small speakers.
A big TV Sweep tube for higher current, lower V, and then add UnSet drive to get great triode curves and a few more Watts besides.
A SS current mirror plate load (with current gain, WCG, to preserve tube sound ) could increase current linearly for a smaller tube and could give you effectively a big P type tube.
Use two Sweep tubes for output, one with a C mirror WCG up top and the other with a C mirror WCG below the cathode. Combine the current outputs for a practical OTL P-P output.
Two Sweep tubes with C mirror plate loads WCG for separate low V OTL like outputs, and drive a dual voice coil woofer. One wired for push and the other wired for pull. Could ground the B+ side and use a B- on the cathodes for safety at the speaker.
Last edited:
Thanks for your comments but I’m afraid most of what you have been saying has gone over my head unfortunately. Maybe I’ll understand once I have gained some experience in building tube circuits.
I would probably have more chance of understanding your ideas in schematic form. Anyway thanks again for your response.
I would probably have more chance of understanding your ideas in schematic form. Anyway thanks again for your response.
Here is a current gain mirror used to make an accurate P-tube. 3X current gain here, but can be extended with more matched transistors (or can use a MosFet mirror design too ) An N type mirror could be used to make a high current N tube also, by using cathode current to drive it.
Many circuit possibilities with a P-tube available, including OTL designs. The SS part can operate with lower voltage across it with some DC level shifting
( to get reduced power consumption for say an OTL Amp ).
A conventional tube differential stage could drive some C mirrors too. With Voltage N feedback from the Amp output to control output V. Many configurations possible.
Output transformer can be dispensed with with enough current gain. And efficiency greatly improved over a conventional OTL design by using LV power for the outputs.
Many circuit possibilities with a P-tube available, including OTL designs. The SS part can operate with lower voltage across it with some DC level shifting
( to get reduced power consumption for say an OTL Amp ).
A conventional tube differential stage could drive some C mirrors too. With Voltage N feedback from the Amp output to control output V. Many configurations possible.
Output transformer can be dispensed with with enough current gain. And efficiency greatly improved over a conventional OTL design by using LV power for the outputs.
Last edited:
