Hey All,
I have attached a draft of a schematic using Harmon Kardon A500 iron. The power supply uses a voltage doubler like the original. The screens will be regulated using one of Tom Christiansen's Maida 3.2 regulators. The output transformers have a slight imbalance in their primaries. I will use that to swap tubes until I get the best balance. The original used a balance pot for the cathodes to adjust the bias voltage. The feedback circuit is a stab in the dark and will require adjustment (original value used). The EL34's will run at a low 60ma. I'm using 5751 for two reasons, it limits gain and it sounds better than 12AX7. The output transformers shouldn't be pushed beyond around 30 watts.
Hope I haven't made any really obvious mistakes, (although that seems likely)
I have attached a draft of a schematic using Harmon Kardon A500 iron. The power supply uses a voltage doubler like the original. The screens will be regulated using one of Tom Christiansen's Maida 3.2 regulators. The output transformers have a slight imbalance in their primaries. I will use that to swap tubes until I get the best balance. The original used a balance pot for the cathodes to adjust the bias voltage. The feedback circuit is a stab in the dark and will require adjustment (original value used). The EL34's will run at a low 60ma. I'm using 5751 for two reasons, it limits gain and it sounds better than 12AX7. The output transformers shouldn't be pushed beyond around 30 watts.
Hope I haven't made any really obvious mistakes, (although that seems likely)
Attachments
For safety and symmetry you should ground the 4 ohm tap.
The 39K resistors should probably be more like 3.9K. Doubt you need that much gain? and indeed the available open loop gain is probably insufficient to satisfy the gain equation.
Not sure about the operating point of the driver stage either, if the cathode voltage is appreciably less than 1V with the 5751 you will have significant grid current.
I would recommend either fixed bias or cathode bypass caps on the 670 ohm resistors in the output stage, gain will be very low and rp extremely high.
It is not unusual for the dcr of PP primaries to differ significantly, the turns ratio match is what matters from an AC perspective. Measuring the cathode current should be sufficiently accurate for your purposes, however it does include the screen current as well.
The 39K resistors should probably be more like 3.9K. Doubt you need that much gain? and indeed the available open loop gain is probably insufficient to satisfy the gain equation.
Not sure about the operating point of the driver stage either, if the cathode voltage is appreciably less than 1V with the 5751 you will have significant grid current.
I would recommend either fixed bias or cathode bypass caps on the 670 ohm resistors in the output stage, gain will be very low and rp extremely high.
It is not unusual for the dcr of PP primaries to differ significantly, the turns ratio match is what matters from an AC perspective. Measuring the cathode current should be sufficiently accurate for your purposes, however it does include the screen current as well.
The schematic suggests that the voltage amplifiers' B+ and that of the "finals" is common. IMO, it's always a good idea to decouple small signal B+ from power O/P B+. The "check valve" technique gets the job done, without introducing a substantial voltage drop. A series combo of a UF4007 and a small resistance, say 47 Ω, will isolate the decoupling cap. quite nicely.
The 39K resistor is original. I just added a second for the other preamp side. I'm working on figuring out the actual feedback values using MJ's process outlined for the "Bevois" amplifier. As usual I'm having problems understanding where he derives some of the numbers.
I want to limit power output to about 30 watts. Jim Mcshane said that 32 watts would be the upper limit for these transformers with a full bandwidth. Since the amp was designed for 12aX7 the 5751 should provide less gain right off the bat.
The bias voltage is set at -1.5 volts should it be higher?
I want to stick with cathode bias to make the amp simple for anyone to deal with. I'll have to figure out bypass caps then.
On grounding the 4 ohm tap? You mean add a ground connection from the tap to the star ground? Not completely grounding the tap correct?
Thanks guys, I'll make the changes and post the new schematic.
I want to limit power output to about 30 watts. Jim Mcshane said that 32 watts would be the upper limit for these transformers with a full bandwidth. Since the amp was designed for 12aX7 the 5751 should provide less gain right off the bat.
The bias voltage is set at -1.5 volts should it be higher?
I want to stick with cathode bias to make the amp simple for anyone to deal with. I'll have to figure out bypass caps then.
On grounding the 4 ohm tap? You mean add a ground connection from the tap to the star ground? Not completely grounding the tap correct?
Thanks guys, I'll make the changes and post the new schematic.
It's hard to know where to begin...
Everything in a circuit -- especially a circuit with negative feedback around it -- has impacts on other parts of the circuit. It's all very interactive. Change one thing, you change its relationship to other things, which have relationships to other things, and so on. How far you want to go to control all of these interactions is dependent on how much effort you want to put in and how much you understand about how things are working in your proposed circuit.
I don't have the math skills to understand the physics of what's going on, but I try my best to learn as I go along. At this point, I know this:
- Removing the bypass capacitor from the cathode load resistor does not decrease bass response. It does change the operation of the tube, though, and by quite a lot. Internal dynamic plate resistance goes up, transconductance goes down, gain goes down, harmonic distortion goes down. If the power supply is noisy, noise at the output (plate) goes up (because the unbypassed cathode reduces the tube's power supply rejection ratio).
- A negative feedback loop forms an AC voltage divider between the series feedback resistor and the cathode resistor. Bypassing the cathode resistor shunts the AC feedback signal to ground, thus removing the voltage divider for AC (it will still work as a voltage divider for DC). Since the audio signal is AC, that removes the feedback. That's why you don't use a cathode bypass cap if you're injecting negative feedback signal to the cathode. (I hope I got that right. Anyone -- feel free to correct me.)
- There is much more going on than that. But that's a start.
--
You also have your input stage loaded pretty hard. The 5751 has an internal plate resistance of about 50k ohms. That is loaded by the 118k plate load resistance in parallel with the resistance of the grid leak resistors for your output stage (270k ohms).
270k || 118k = 82k
The approximately 50k internal plate resistance of the 5751 is working into a load of only 82k. That 82k is less than 2x the 5751's internal plate resistance (ra) of 50k. The load should be equal to or more than 4x the driving tube's ra. Increasing the 270k resistors to 470k would help. But that's about as far as you can go there, because of the maximum grid resistance allowed for a cathode biased EL34.
I could be wrong about all that, but I think that's the way it works.
--
Everything in a circuit -- especially a circuit with negative feedback around it -- has impacts on other parts of the circuit. It's all very interactive. Change one thing, you change its relationship to other things, which have relationships to other things, and so on. How far you want to go to control all of these interactions is dependent on how much effort you want to put in and how much you understand about how things are working in your proposed circuit.
I don't have the math skills to understand the physics of what's going on, but I try my best to learn as I go along. At this point, I know this:
- Removing the bypass capacitor from the cathode load resistor does not decrease bass response. It does change the operation of the tube, though, and by quite a lot. Internal dynamic plate resistance goes up, transconductance goes down, gain goes down, harmonic distortion goes down. If the power supply is noisy, noise at the output (plate) goes up (because the unbypassed cathode reduces the tube's power supply rejection ratio).
- A negative feedback loop forms an AC voltage divider between the series feedback resistor and the cathode resistor. Bypassing the cathode resistor shunts the AC feedback signal to ground, thus removing the voltage divider for AC (it will still work as a voltage divider for DC). Since the audio signal is AC, that removes the feedback. That's why you don't use a cathode bypass cap if you're injecting negative feedback signal to the cathode. (I hope I got that right. Anyone -- feel free to correct me.)
- There is much more going on than that. But that's a start.
--
You also have your input stage loaded pretty hard. The 5751 has an internal plate resistance of about 50k ohms. That is loaded by the 118k plate load resistance in parallel with the resistance of the grid leak resistors for your output stage (270k ohms).
270k || 118k = 82k
The approximately 50k internal plate resistance of the 5751 is working into a load of only 82k. That 82k is less than 2x the 5751's internal plate resistance (ra) of 50k. The load should be equal to or more than 4x the driving tube's ra. Increasing the 270k resistors to 470k would help. But that's about as far as you can go there, because of the maximum grid resistance allowed for a cathode biased EL34.
I could be wrong about all that, but I think that's the way it works.
--
Last edited:
That is correct, too, but you do need to insert the NFB somewhere
How about the grids of the drivers maybe? They are in 0 Volts anyway.
Thanks for all the help. If I've done the math correctly, it should increase the load on 5751 to @ 152K while at the same time increasing gain to cope with the negative feedback. The current decreases from .7 ma to .6ma. While at the same time allowing 2 volts pp input without going under -1 volts on 5751.
I'm having trouble uploading the new schematic. I changed the EL34 grid resistors to 470K as Rongon suggested. And increased the 5751 plate resistors to 225K. That leaves 250 volts across 5751.
I intend to use a large linear dual pot in place of the feedback resistors to tune it by ear. Once I know the feedback resistor I can add the cap.
As always, thanks!
I'm having trouble uploading the new schematic. I changed the EL34 grid resistors to 470K as Rongon suggested. And increased the 5751 plate resistors to 225K. That leaves 250 volts across 5751.
I intend to use a large linear dual pot in place of the feedback resistors to tune it by ear. Once I know the feedback resistor I can add the cap.
As always, thanks!
Last edited:
no, you cannot add the cap. The cap will eliminate the feedback signal, exactly like it would eliminate the negative .cathode feedback. It has been mentioned before.
There's a really good introductory text for designing tube audio amplifiers, by an ex-BBC engineer named Morgan Jones. The book is titled Valve Amplifiers, and it's currently in its 4th edition. The 3rd edition is available for cheap, and it's absolutely useful.
Valve Amplifiers 4th Edition on amazon
I recommend it very highly. It really helped me a lot.
--
Valve Amplifiers 4th Edition on amazon
I recommend it very highly. It really helped me a lot.
--
I second the recommendation for Morgan Jones Valve Amplifiers, I'd recommend the 4th edition unless budget is really tight. (I have both)
The cathode capacitor as implemented would shunt all local and global feedback to ground and cannot be implemented as you keep proposing.
There is an alternative way to do it, but it is doubtful I can post it as there seems to be a site bug at the moment.
Edit: Unable to post attachments
The cathode capacitor as implemented would shunt all local and global feedback to ground and cannot be implemented as you keep proposing.
There is an alternative way to do it, but it is doubtful I can post it as there seems to be a site bug at the moment.
Edit: Unable to post attachments
In the 3rd edition, there's a nice discussion of implementing a simple global NFB loop, within the discussion of the Beavois Valley amplifier. I can't find where that went in the 4th edition.
The 4th edition has a good amount of new info, which is absolutely worth having. Best is to get both, if you can!
--
PS - I'm having that same problem uploading anything. Something's definitely wrong there.
The 4th edition has a good amount of new info, which is absolutely worth having. Best is to get both, if you can!
--
PS - I'm having that same problem uploading anything. Something's definitely wrong there.
Costis, when I said I would add the cap I didn't mean a cathode bypass cap. I meant an RC filter with the feedback resistor. Although the original didn't have one.
I have MJ's third addition. The only place I've found where he actually explains how to design a feedback loop is in the Bevois valley amplifier section. I've tried off and on for weeks to transpose the process from the E88CC and EL84 to the 5751 and EL34 to no avail. For me he's like reading French philosophy. You can get the gist of what he's saying but he glosses over the basic points.
I'm looking at the section right now, page 449. He writes, "The anode signal swing for the full claimed output of 11 W is 8.636 Vrms. This means that the anode signal current must be 8.636V/47k = 0.1837mArms. I have a copy of the Mullard data sheet for EL84. There is no spec of 8.636 or 47k in the datasheet? I'm assuming the 8.636 is derived somehow but I don't see it. And I'm assuming the 47k is plate resistance? But that's not there either.
I have MJ's third addition. The only place I've found where he actually explains how to design a feedback loop is in the Bevois valley amplifier section. I've tried off and on for weeks to transpose the process from the E88CC and EL84 to the 5751 and EL34 to no avail. For me he's like reading French philosophy. You can get the gist of what he's saying but he glosses over the basic points.
I'm looking at the section right now, page 449. He writes, "The anode signal swing for the full claimed output of 11 W is 8.636 Vrms. This means that the anode signal current must be 8.636V/47k = 0.1837mArms. I have a copy of the Mullard data sheet for EL84. There is no spec of 8.636 or 47k in the datasheet? I'm assuming the 8.636 is derived somehow but I don't see it. And I'm assuming the 47k is plate resistance? But that's not there either.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.