Just a quick chip-in regarding that. Folks might recall my experiences with Quad 22 control units. All of 12 seemingly 'styrene' caps/unit associated with tone circuits had drifted up to 2,5 times off-value; with obvious results regarding sound. It is thus a good idea to check any old caps in apparently good working order for capacitance also, not just leakage. Same occasionally happened with British 'Hunts' caps of all things - no explanation, but there it is.
Glad on your behalf, Rotaspec. Such matters can be utterly irritating.
Gimp, I measured every one of the 72 LED's at 15mA and arranged them into 12 strings of 6 LED's with exact series voltage for every string at 15mA. But the bias changed during the first 20 hours or so, so maybe I should have run them for 20 hours before I matched them. The tubes have a couple hundred hours on them so I'm pretty sure it was the LED's shifting. They have settled down now. The first time I measured up all the LED's and put them neatly on an A4 paper in voltage groups, well, the cat slept on them that night and I was lucky to recover all 72 LED's as I had no spares. Next night I measured them all again and put them in little plastic bags with labels. With the LED biasing, I have had to alter the screen resistor to adjust the screen voltage to set the bias. I have 1K5 on one channel and 4K7 on the other, giving screen voltages of 242V and 223V under full power, with 9.8V rms sitting on top of the DC at 2kHz, with a 1kHz sinewave input. BTW I've been watching the Svenn thread with interest.
I have done a fair bit of searching to try to work out theoretical output power of a push-pull output stage with no success. I am getting barely 6W which is enough, but I would like to understand the factors involved, I was expecting more with 9.5W dissipation on each tube. It is obvious that power will increase with the square of the voltage on the output of the OPT, and I can't see how to increase that without raising the B+ above recommended max values. At the moment I can get a clean 7.7V rms signal onto the grid of the output tubes from a 400mV input, but the output stage is clipping with anything over 360mV at the input to the amp.
I did some measuring on the standard OPT's after removing them and fitting some OPT's from a late 1950's amp I have here. I used a 9V rms sine wave 1kHz from a sig gen fed into the primary and measured the secondary voltage with no load, I get 6K1 primary for one tranny and 6K3 for the other, presuming an 8 ohm load on the output. The trannies from the 1950's amp are 9K5 primary. Power output is the same with the 1956 OPT's, around 6W, the sound is maybe marginally better but it is difficult to describe what is better about it. When I run 4 ohm speakers on the original OPT's, it appears the loadline passes through the max power dissipation line by a considerable amount, it does not look good. The 1956 OPT's have 4 and 8 ohm taps.
Next was the square wave overshoot. Adding 56pF in series with 10K across the first input stage's anode resistor halved the overshoot and 100pF was a bit too much, but I had nothing in between so I left the 100pF in and it seems to have a beneficial effect, maybe cleaner, but it could be in my head. Resistor values between 5K and 33K seem to have little effect on the overshoot, the capacitor value seems to be the dominating factor.
The parts for the mosfet drive arrived today so I'll attempt that tomorrow, might take me 2 days, yes Gimp, it's just for the experiment to see how it goes. But it will also give individual bias on the O/P tubes and allow experimenting with triode and ultralinear modes. At the moment with LED's in the O/P cathodes, the screen voltages are holding the quiescent bias so its pentode mode only. I have nothing against pentode mode though, the amp is sounding very nice.
I have done a fair bit of searching to try to work out theoretical output power of a push-pull output stage with no success. I am getting barely 6W which is enough, but I would like to understand the factors involved, I was expecting more with 9.5W dissipation on each tube. It is obvious that power will increase with the square of the voltage on the output of the OPT, and I can't see how to increase that without raising the B+ above recommended max values. At the moment I can get a clean 7.7V rms signal onto the grid of the output tubes from a 400mV input, but the output stage is clipping with anything over 360mV at the input to the amp.
I did some measuring on the standard OPT's after removing them and fitting some OPT's from a late 1950's amp I have here. I used a 9V rms sine wave 1kHz from a sig gen fed into the primary and measured the secondary voltage with no load, I get 6K1 primary for one tranny and 6K3 for the other, presuming an 8 ohm load on the output. The trannies from the 1950's amp are 9K5 primary. Power output is the same with the 1956 OPT's, around 6W, the sound is maybe marginally better but it is difficult to describe what is better about it. When I run 4 ohm speakers on the original OPT's, it appears the loadline passes through the max power dissipation line by a considerable amount, it does not look good. The 1956 OPT's have 4 and 8 ohm taps.
Next was the square wave overshoot. Adding 56pF in series with 10K across the first input stage's anode resistor halved the overshoot and 100pF was a bit too much, but I had nothing in between so I left the 100pF in and it seems to have a beneficial effect, maybe cleaner, but it could be in my head. Resistor values between 5K and 33K seem to have little effect on the overshoot, the capacitor value seems to be the dominating factor.
The parts for the mosfet drive arrived today so I'll attempt that tomorrow, might take me 2 days, yes Gimp, it's just for the experiment to see how it goes. But it will also give individual bias on the O/P tubes and allow experimenting with triode and ultralinear modes. At the moment with LED's in the O/P cathodes, the screen voltages are holding the quiescent bias so its pentode mode only. I have nothing against pentode mode though, the amp is sounding very nice.
Ramblings:
I was about to measure the swing on the grids. Been sick so no progress. My parts are due today depending on the snow. 7,5V RMS. I see a solid state diff line driver in the future. THAT maybe.
Looking at other designs, it seems the old Quad used the paraphase, but it is still revered as one of the great ones. Hmmm. Nostalgia?
On power, it is clear this unit in not 10 Watts. Never was, never designed to be. Just another case of Chinese not understanding why we have rules against false advertising in the rest of the world. Just like the majority of e-stores I have dealt with in China. Blatant customs fraud. I am dealing with returning a defective DAC to Hong Kong. it was shipped to me as "clothes". Lead paint, fake components, chemicals in food. We know what to expect.
Wondering where the 6.3 volt standard came from. Lead acid batteries? I also wonder if the old 70V PA speakers was because of tube amps preferring to produce voltage. I bet who ever designed them did a PP with no OPT.
I was about to measure the swing on the grids. Been sick so no progress. My parts are due today depending on the snow. 7,5V RMS. I see a solid state diff line driver in the future. THAT maybe.
Looking at other designs, it seems the old Quad used the paraphase, but it is still revered as one of the great ones. Hmmm. Nostalgia?
On power, it is clear this unit in not 10 Watts. Never was, never designed to be. Just another case of Chinese not understanding why we have rules against false advertising in the rest of the world. Just like the majority of e-stores I have dealt with in China. Blatant customs fraud. I am dealing with returning a defective DAC to Hong Kong. it was shipped to me as "clothes". Lead paint, fake components, chemicals in food. We know what to expect.
Wondering where the 6.3 volt standard came from. Lead acid batteries? I also wonder if the old 70V PA speakers was because of tube amps preferring to produce voltage. I bet who ever designed them did a PP with no OPT.
1.5V Carbon Zink 1.5, 3, 4.5, 6 ...
2.2V Lead Acid 2.2, 4.4, 6.6, ...
6.6V max on 6.3V filament tubes? Max float for a three cell flooded cell battery is 6.6V. Nominal is 6.3V under load, minimum is around 6V. Matches the range for 6.3V tubes.
I've seen lots of old "Portable" radios from the early 20th century that a friend of the family had back in the late 60s and 70s (Chuck passed away in 1984.). He must have had at least 100 radios in a cinder block building that was probably 60' x 60'. It was full of surplus electronics. He showed me where the batteries went in one of the radios, and said it would take two men and a boy to move a fully loaded one. He may have exadurated slightly but I bet with a full compliment of batteries they were a job to move. It's a shame what happened to everything when he passed away. The family had no respect for electronics, or his 300 pecan trees that they had cut down.
2.2V Lead Acid 2.2, 4.4, 6.6, ...
6.6V max on 6.3V filament tubes? Max float for a three cell flooded cell battery is 6.6V. Nominal is 6.3V under load, minimum is around 6V. Matches the range for 6.3V tubes.
I've seen lots of old "Portable" radios from the early 20th century that a friend of the family had back in the late 60s and 70s (Chuck passed away in 1984.). He must have had at least 100 radios in a cinder block building that was probably 60' x 60'. It was full of surplus electronics. He showed me where the batteries went in one of the radios, and said it would take two men and a boy to move a fully loaded one. He may have exadurated slightly but I bet with a full compliment of batteries they were a job to move. It's a shame what happened to everything when he passed away. The family had no respect for electronics, or his 300 pecan trees that they had cut down.
Ian,
Measure the DC voltage drop across the 4R7 resistors and calculate your bias current to one tube. I'm betting it is less than 45mA total. Since the cathode current includes screen and plate current I expect you are probably close to 40mA plate current (I've seen over 3ma screen current on some tubes). I measure plate current by putting a 1 ohm resistor directly from the plate socket pin to the OPT wire. Measure the voltage in mV and you get 1mA/mV reading.
In order to get 10W out of a pair of 6P1P tubes in PP you need over to 50mA cathode current. In order to get 13W out I was driving near 55mA per plate IIRC and near 59mA cathode current.
The chinese 6P1 will probably not take the abuse I gave the 6P1P-EVs, so my guess is you should be able to get 10W out before the plates will start to go red. I watch the fins in a totally dark room to monitor plates for overdissipation.
I'll try to set my prototype back up and hook the LED bias back up to the cathodes and see what I get.
I don't think the transformers are the issue, as I've used some old transformers and tried different taps while taking power measurements and did not see radical changes in power out.
For instance, I have an unknown transformer marked 23A0007 1005510 which measures 7.11H Primary Inductance and 0.057 Leakage Inductance. No UL taps.
It has taps that give me 8.099Kp-p and 5.183Kp-p. The power out is almost identical with both taps when driving a 8 ohm wire wound 100W adjustable resistor as a load.
I've also got a set of transformers from an Olson AM-172A (supposidly) that measures 7.25H P-I, and 10mH L-I. The 11Kp-p sounds best, but the 3.65Kp-p taps give almost as much ouput power and have greater high frequency output.
I am running only slightly higher B+ than you, maybe 10V difference, so that shouldn't be a factor.
Running Class A like we are, the only way to get more power without exceeding plate voltage limit of 250V is to increase the plate current.
You can do that by reducing the number of LEDs in your string by one, and then adjusting your screen voltages to get the same plate current in each output tube, but it is best to measure that at the plate with a 1 Ohm sense reisstor, not at the cathode as screen current is going to throw that measurement off.
Good luck.
Steven
Measure the DC voltage drop across the 4R7 resistors and calculate your bias current to one tube. I'm betting it is less than 45mA total. Since the cathode current includes screen and plate current I expect you are probably close to 40mA plate current (I've seen over 3ma screen current on some tubes). I measure plate current by putting a 1 ohm resistor directly from the plate socket pin to the OPT wire. Measure the voltage in mV and you get 1mA/mV reading.
In order to get 10W out of a pair of 6P1P tubes in PP you need over to 50mA cathode current. In order to get 13W out I was driving near 55mA per plate IIRC and near 59mA cathode current.
The chinese 6P1 will probably not take the abuse I gave the 6P1P-EVs, so my guess is you should be able to get 10W out before the plates will start to go red. I watch the fins in a totally dark room to monitor plates for overdissipation.
I'll try to set my prototype back up and hook the LED bias back up to the cathodes and see what I get.
I don't think the transformers are the issue, as I've used some old transformers and tried different taps while taking power measurements and did not see radical changes in power out.
For instance, I have an unknown transformer marked 23A0007 1005510 which measures 7.11H Primary Inductance and 0.057 Leakage Inductance. No UL taps.
It has taps that give me 8.099Kp-p and 5.183Kp-p. The power out is almost identical with both taps when driving a 8 ohm wire wound 100W adjustable resistor as a load.
I've also got a set of transformers from an Olson AM-172A (supposidly) that measures 7.25H P-I, and 10mH L-I. The 11Kp-p sounds best, but the 3.65Kp-p taps give almost as much ouput power and have greater high frequency output.
I am running only slightly higher B+ than you, maybe 10V difference, so that shouldn't be a factor.
Running Class A like we are, the only way to get more power without exceeding plate voltage limit of 250V is to increase the plate current.
You can do that by reducing the number of LEDs in your string by one, and then adjusting your screen voltages to get the same plate current in each output tube, but it is best to measure that at the plate with a 1 Ohm sense reisstor, not at the cathode as screen current is going to throw that measurement off.
Good luck.
Steven
I ran a couple of tests tonight.
1K ohm Screen Resistors to B+
B+=264V
VKA = 253V
220R cathode bias:
VrA = 10.36/10.08
IA = 52.8/50.9mA
IS = 3.96/2.55mA
Vo = 9.4Vrms into 8 ohms - 11.45W
No red on plates in total darkness.
LED Bias (5 LEDs, one was out in two strings so I shorted enough to get two sets of 5):
VC = 9.45/9.43V
IA = 61.4/59.8mA
IS = 4.88/3.13mA
Vo = 9.2Vrms into 8 Ohms - 10.58W
No red plates in total darkness, but I only ran it long enough to take all the readings static and get a reading of peak power before distortion was noticeable on the scope.
Best to stick with six LEDs as 5 run too much plate current.
I'll have to make up new strings of LEDs as these have been abused a bit.
1K ohm Screen Resistors to B+
B+=264V
VKA = 253V
220R cathode bias:
VrA = 10.36/10.08
IA = 52.8/50.9mA
IS = 3.96/2.55mA
Vo = 9.4Vrms into 8 ohms - 11.45W
No red on plates in total darkness.
LED Bias (5 LEDs, one was out in two strings so I shorted enough to get two sets of 5):
VC = 9.45/9.43V
IA = 61.4/59.8mA
IS = 4.88/3.13mA
Vo = 9.2Vrms into 8 Ohms - 10.58W
No red plates in total darkness, but I only ran it long enough to take all the readings static and get a reading of peak power before distortion was noticeable on the scope.
Best to stick with six LEDs as 5 run too much plate current.
I'll have to make up new strings of LEDs as these have been abused a bit.
Steven, thanks for all your help and suggestions. I was running the 6P1P-EV at ~40mA cathode current, with 5 mA on the screens that leaves 35mA on the plates. I have 1 ohm resistors on top of the LED arrays to check cathode current and check the screen current by measuring voltage drop across the screen resistors which is around 5mA 1.2W close to the limit. Your experiences and testing basically answers my questions and thanks for trying the LED's, your help is very much appreciated, thank you. Hey don't go away either, might need more help!
I put the mosfets in and added a negative tripler power supply for the -130V and the +/-50V zener regulators and no smoke has escaped so far. It is a BIG improvement, in a way this is unfortunate because I like simple ccts, but my wife and I hear a lot more detail. Even the old lo-fi recordings and FM radio sound better. It is like there is an impedance mismatch between the driver stage and the output stage. I know this should not exist but it simply sounds better with the mosfet source follower. Most people would not try this because of the extra work building it, therefore they would not find out about it. That was one reason I tried it (plus it is recommended by people who have a lot of experience), the other reason was to get individual bias control of each output tube.
Amp internals are getting messy again, and here is the design for the stripboard version of the mosfet source follower I built, one per channel, if anyone is interested.
Ian.
I put the mosfets in and added a negative tripler power supply for the -130V and the +/-50V zener regulators and no smoke has escaped so far. It is a BIG improvement, in a way this is unfortunate because I like simple ccts, but my wife and I hear a lot more detail. Even the old lo-fi recordings and FM radio sound better. It is like there is an impedance mismatch between the driver stage and the output stage. I know this should not exist but it simply sounds better with the mosfet source follower. Most people would not try this because of the extra work building it, therefore they would not find out about it. That was one reason I tried it (plus it is recommended by people who have a lot of experience), the other reason was to get individual bias control of each output tube.
Amp internals are getting messy again, and here is the design for the stripboard version of the mosfet source follower I built, one per channel, if anyone is interested.
Ian.
I've been thinking about the bias voltage issue. The discrete steps afforded by RED LEDs is around 1.6-1.7V. This is too corse to allow for precise cathode (or plate) current adjustment.
Other options are (1) stay with 5 LEDs per string and increase the equalization resistor(s) values (good control over incremental voltage drop), (2) add other diodes in series with the 5 LED strings (0.6-0.7V discrete voltage drop steps), (3) stay with 5 LEDs per string with a small equalization resistor and adjust the screen voltage to set the plate current (Fine except for UL operation).
I'm also wondering why my power output went down with higher plate current (60mA vs 51mA)? I suspect this is because the output transformer is the same size as a hammond 1609, so it is probably rated for 10W adn the higher current offset coupled with saturation is what is limiting my output. I'm probably doing good to get 12-13W out of it, and it is going to be very dependent on plate current. It will peak somewhere between 50 and 60mA. The higher the plate current and PO, the more critical balance becomes.
Can someone with more insight on transformer theory confirm or deny this?
Steven
Other options are (1) stay with 5 LEDs per string and increase the equalization resistor(s) values (good control over incremental voltage drop), (2) add other diodes in series with the 5 LED strings (0.6-0.7V discrete voltage drop steps), (3) stay with 5 LEDs per string with a small equalization resistor and adjust the screen voltage to set the plate current (Fine except for UL operation).
I'm also wondering why my power output went down with higher plate current (60mA vs 51mA)? I suspect this is because the output transformer is the same size as a hammond 1609, so it is probably rated for 10W adn the higher current offset coupled with saturation is what is limiting my output. I'm probably doing good to get 12-13W out of it, and it is going to be very dependent on plate current. It will peak somewhere between 50 and 60mA. The higher the plate current and PO, the more critical balance becomes.
Can someone with more insight on transformer theory confirm or deny this?
Steven
With bias voltage and LED's, maybe try 6 LED's in each string, then use fixed bias with a trimpot tied to +5V to adjust the bias hotter. On my setup the trimpots work well, nearly 1 turn per mA on the cathode. You could get 5V using a zener from B+ as the current is miniscule. I really like the sound of LED's on the O/P tube cathode and the addition of some fixed bias gets around any disadvantages.
If you are ever in the mood to try it, I cannot recommend the mosfets enough 🙂 The power supply requirements can make it an impractical task though.
Ian.
If you are ever in the mood to try it, I cannot recommend the mosfets enough 🙂 The power supply requirements can make it an impractical task though.
Ian.
Last edited:
Hmmm. Looks like I was a little off on my power measurements. 😱
The way I was measuring it was with the scope RMS feature and a sweep that gave me this trace of five cycles indicating near 11.5 W:
Looks good, just starting to clip, nice waveform.
However if I change the sweep rate it changes to this:
I have to cut back to a bit over 8.4 W to clean it up:
It appears the scope distorts the waveform (Probably under-sampling) if the sweep rate is not fast enough. Stupid DSOs. This is a good reason to use an old analog scope for tube work.
Ian, I expect you will get close to the same PO once you get your bias up near 50mA or so plate current. I think I'll see how low I can adjust the plate current before the power out drops.
I've got 154 VRMS on the plate at clipping so I've used up the available B+ on the output tubes with this output transformer. Maybe a slightly different impedance might give slightly more power out. Otherwise I guess this is it for class A at 250V B+.
Steven
The way I was measuring it was with the scope RMS feature and a sweep that gave me this trace of five cycles indicating near 11.5 W:
An externally hosted image should be here but it was not working when we last tested it.
Looks good, just starting to clip, nice waveform.
However if I change the sweep rate it changes to this:
An externally hosted image should be here but it was not working when we last tested it.
I have to cut back to a bit over 8.4 W to clean it up:
An externally hosted image should be here but it was not working when we last tested it.
It appears the scope distorts the waveform (Probably under-sampling) if the sweep rate is not fast enough. Stupid DSOs. This is a good reason to use an old analog scope for tube work.
Ian, I expect you will get close to the same PO once you get your bias up near 50mA or so plate current. I think I'll see how low I can adjust the plate current before the power out drops.
I've got 154 VRMS on the plate at clipping so I've used up the available B+ on the output tubes with this output transformer. Maybe a slightly different impedance might give slightly more power out. Otherwise I guess this is it for class A at 250V B+.
Steven
Drawing 6P1P loadlines and trying to work out Class AB conditions is messing up my brain. Anytime the plate voltage and/or screen grid voltage changes, so do the curves. Then the loadline gradient is determined by the OPT primary impedance. I can see how the tube could be run Class AB with an anode to cathode voltage of 260V and 6K impedance a-a, but the datasheet graph is for anode to cathode voltage of 250V with the screen grid at 250V, and 260V is pushing the limits. Maybe the 6P1P is not so suited for Class AB, is this possible, that some tubes are better suited for class A? Power output is not what I am chasing, but gaining an understanding of push-pull output stages is a priority. Maybe a tube calculator like TCJ push-pull calculator would help serve this purpose?
BTW, thanks for posting the results of your tests, it all helps, a lot!
Ian.
BTW, thanks for posting the results of your tests, it all helps, a lot!
Ian.
Oops, the TCJ Push-Pull Calculator simulates triodes and triode-connected pentodes. Back to the drawing board. Steven, when you say 154V rms is that measured from anode to anode? In that last screen shot at 8.4W it is still clipping, but just marginally.
Yes as I noted it was clipping and I had 154V RMS at one anode, not P-P. Symetry was good with both peaks going into clipping near the same level.
Did you look at your anode voltage? Was clipping symetrical or did you see cutoff first before saturation, indicating class AB1 operation?
154VRMS is about 218V, add another 10V for the transformer drop and 10 v for the cathode bias and you are up to 238V. This only leaves ten to twenty volts across the tube at saturation. The voltage supply is used up.
I agree, these tubes were designed for Class A operation. This is confirmed by the 6AQ5 data sheet specifying class A1 operation. My understanding is the 6P1P is derived from the 6AQ5.
In order to go class AB1, we would need to decrease the plate current such that the tube would cut off on the negative excersion of the grid drive. With only 250V available, it would reduce power, but could be done. This would most likely increases distortion as well as reduce power out.
If it were avaialble, a 12Kohm p-p would be an interesting test. It looks to me like it would produce more power and less distiortion. This might give more headroom in AB1.
I may try boosting the B+ to see how much abuse a pair of tubes can take. It will require non-UL Operation to avoid exceeding the screen rating of 250V, and Will require screen regulators (or voltage dividers).
The 6AQ5 spec shows a maximum of 275V anode voltage. If the tubes will take it, it might be possible to up the B+ and drop the cathode current to force it into class AB1. It might make for an interesting guitar amp.
Did you look at your anode voltage? Was clipping symetrical or did you see cutoff first before saturation, indicating class AB1 operation?
154VRMS is about 218V, add another 10V for the transformer drop and 10 v for the cathode bias and you are up to 238V. This only leaves ten to twenty volts across the tube at saturation. The voltage supply is used up.
I agree, these tubes were designed for Class A operation. This is confirmed by the 6AQ5 data sheet specifying class A1 operation. My understanding is the 6P1P is derived from the 6AQ5.
In order to go class AB1, we would need to decrease the plate current such that the tube would cut off on the negative excersion of the grid drive. With only 250V available, it would reduce power, but could be done. This would most likely increases distortion as well as reduce power out.
If it were avaialble, a 12Kohm p-p would be an interesting test. It looks to me like it would produce more power and less distiortion. This might give more headroom in AB1.
I may try boosting the B+ to see how much abuse a pair of tubes can take. It will require non-UL Operation to avoid exceeding the screen rating of 250V, and Will require screen regulators (or voltage dividers).
The 6AQ5 spec shows a maximum of 275V anode voltage. If the tubes will take it, it might be possible to up the B+ and drop the cathode current to force it into class AB1. It might make for an interesting guitar amp.
Steve - at clipping, both top and bottom of the sinewave clip evenly.
Thanks for the clarification on the Class A bit. I drew a new rough load line for my amp which runs around 42mA cathode current, about 5mA on the screen grids , B+ approx 253V, grid voltage around 10.5V, that leaves around 243V across the 6P1P-EV and around 237V on the screen grids.
I finally worked out that the anode voltage on this PP amp actually goes to at least 420V DC as the valve current drops. All I had to do was set the scope channel B to "DC + AC volts". Hohoho. I was struggling with that bit, as in what happens when the grid drives negative. I found (again) some really good writings by Pat Turner that answered all the rest of the questions for me. Excellent info. A goldmine of info.
Here's the latest reflection of the current schematic.
If one wanted more Class A power it could be near doubled with around 300V B+ and EL84 or 6V6, due to Power = V squared/R. I wonder what the sound is like, with those tubes, compared to 6P1P? I will find out one day I guess.
The power limitations in Class A are, as I see it at this early stage, the anode can only swing positive and negative the same amount of voltage or current, so the bias point needs to be right in the middle. The 0V grid curve is a "road block" at the LHS of the graph. You want the bias point (in mA) to be half the current where the loadline crosses the 0V grid line, and choose the slope of the load line to give max output voltage swing, the OPT primary impedance determines the slope of the load line. Then keep the loadline within max power dissipation. Does that all sound correct? Are we there yet? 🙂 Nope probably not....
Ian.
Thanks for the clarification on the Class A bit. I drew a new rough load line for my amp which runs around 42mA cathode current, about 5mA on the screen grids , B+ approx 253V, grid voltage around 10.5V, that leaves around 243V across the 6P1P-EV and around 237V on the screen grids.
I finally worked out that the anode voltage on this PP amp actually goes to at least 420V DC as the valve current drops. All I had to do was set the scope channel B to "DC + AC volts". Hohoho. I was struggling with that bit, as in what happens when the grid drives negative. I found (again) some really good writings by Pat Turner that answered all the rest of the questions for me. Excellent info. A goldmine of info.
Here's the latest reflection of the current schematic.
If one wanted more Class A power it could be near doubled with around 300V B+ and EL84 or 6V6, due to Power = V squared/R. I wonder what the sound is like, with those tubes, compared to 6P1P? I will find out one day I guess.
The power limitations in Class A are, as I see it at this early stage, the anode can only swing positive and negative the same amount of voltage or current, so the bias point needs to be right in the middle. The 0V grid curve is a "road block" at the LHS of the graph. You want the bias point (in mA) to be half the current where the loadline crosses the 0V grid line, and choose the slope of the load line to give max output voltage swing, the OPT primary impedance determines the slope of the load line. Then keep the loadline within max power dissipation. Does that all sound correct? Are we there yet? 🙂 Nope probably not....
Ian.
Since the EL84/6BQ5 and EL90/6AQ5 are both supposed to have been derived from the 6V6, I suspect they will sound very similar in similar circuits.
I did find my asymetry clipping problem was caused by a shift in the bias point of the phase splitter when I switched from capacitively coupled to DC coupled from the input amp stage. I have corrected this and get right at 8W out before I can see any onset of clipping (Note 1). I can reduce my anode current to near 38mA (maybe lower, I didn't try it) and do not see a drop in output level, so the lower we bias the stages the better it will be in terms of tube life if this is all the power we can get.
Class A requires that the tube stay in conduction for all of it's operation. The Grid will never pull current, and will thus stay negative wrt the cathode (probably by at least 1V).
Class AB1 allows the tube to go out of conduction for part of it's cycle, but the grid must still not draw current, so it is still staying negative wrt the cathode(still -1V?). I think AB2 allows grid current to flow, so the grid may go positive wrt the cathode.
I was still struggling with my load lines because I confused Class AB1 with Class A and was using 2K5 (AB1) instead of class A 5K for the load line..
The measured anode ac voltage is 154Vrms, which translates to 218V Peak which added to the supply gives a peak excursion of about 474V. This corrosponds well to your red load line.
I tried boosting my B+ last night by putting a 24V transformer output in series with the 220V transformer. I got near 300V (cut my anode current back to 40mA to do this), but still only got 8W out.
NOTE 1: I tried four transformers last night. Hammond 1609, Olson 6AQ5, Laffiette 6BQ5, Unknown but bigger than the Hammond by 40% by weight. The Olson gave the greatest output at 8.1vrms, followed by the Unknown at 8, Hammond at 7.9-8, and the Laffiette at 7.6. These corrospond to 8.2W, 8W, 7.8 and 7.22W respectivly.
Looking at the Power vs distortion vs Ra for the 6P1P, it looks like we should be able to get 5W per tube out. I don't see what we are missing in this respect.
I did find my asymetry clipping problem was caused by a shift in the bias point of the phase splitter when I switched from capacitively coupled to DC coupled from the input amp stage. I have corrected this and get right at 8W out before I can see any onset of clipping (Note 1). I can reduce my anode current to near 38mA (maybe lower, I didn't try it) and do not see a drop in output level, so the lower we bias the stages the better it will be in terms of tube life if this is all the power we can get.
Class A requires that the tube stay in conduction for all of it's operation. The Grid will never pull current, and will thus stay negative wrt the cathode (probably by at least 1V).
Class AB1 allows the tube to go out of conduction for part of it's cycle, but the grid must still not draw current, so it is still staying negative wrt the cathode(still -1V?). I think AB2 allows grid current to flow, so the grid may go positive wrt the cathode.
I was still struggling with my load lines because I confused Class AB1 with Class A and was using 2K5 (AB1) instead of class A 5K for the load line..
The measured anode ac voltage is 154Vrms, which translates to 218V Peak which added to the supply gives a peak excursion of about 474V. This corrosponds well to your red load line.
I tried boosting my B+ last night by putting a 24V transformer output in series with the 220V transformer. I got near 300V (cut my anode current back to 40mA to do this), but still only got 8W out.
NOTE 1: I tried four transformers last night. Hammond 1609, Olson 6AQ5, Laffiette 6BQ5, Unknown but bigger than the Hammond by 40% by weight. The Olson gave the greatest output at 8.1vrms, followed by the Unknown at 8, Hammond at 7.9-8, and the Laffiette at 7.6. These corrospond to 8.2W, 8W, 7.8 and 7.22W respectivly.
Looking at the Power vs distortion vs Ra for the 6P1P, it looks like we should be able to get 5W per tube out. I don't see what we are missing in this respect.
I have no idea what I did to get the plates to glow originally.
I've had it up to 60mA and 276V anode to cathode with 10W out and no red plates after 10 minutes. So I dropped the current back to near 50mA and did some measurements.
Cathode Resistor - 250R bypassed with 500uF 50V
Ia - 49.3mA
Va-c 276V
Vs-c - 260V (2K2 screen resistor, 1u MKT at screen)
Vg-c - -12.47
Vo - 9Vrms into 8 ohm resistive load
Input - 384mV RMS
Grid drive - 10V RMS (With a g-c bias of -12.47 shouldn't this be drawing grid current?)
Added 2K2 Feedback Resistor from transformer output to cathode resistor of first stage with 137 Ohms cathode to ground un-bypassed and 47 ohms bypassed with 220uF.
Vo - 9.2V RMS
Input - 960mV RMS
Reduced Ia from 49.3 to 40.2mA
Still have 2K2 feedback resistor
Cathode Resistor - 300R
Ia - 40.2mA
Va-c 276V
Vs-c - 260V
Vg-c - -12.47
Vo - 9Vrms into 8 ohm resistive load
Input - 384mV RMS
Grid drive - 10V RMS
Vo - 9.2V RMS
Input - 1.04V RMS
Below 36mA Ia, the output starts to drop off.
I tried dropping the cathode current to 30mA, but the output started distorting badly above 6.8V rms with really strange crossover distortion.
The amp is turning into a rats nest. I'll be glad when I can build a Tube Lab.
I've had it up to 60mA and 276V anode to cathode with 10W out and no red plates after 10 minutes. So I dropped the current back to near 50mA and did some measurements.
Cathode Resistor - 250R bypassed with 500uF 50V
Ia - 49.3mA
Va-c 276V
Vs-c - 260V (2K2 screen resistor, 1u MKT at screen)
Vg-c - -12.47
Vo - 9Vrms into 8 ohm resistive load
Input - 384mV RMS
Grid drive - 10V RMS (With a g-c bias of -12.47 shouldn't this be drawing grid current?)
Added 2K2 Feedback Resistor from transformer output to cathode resistor of first stage with 137 Ohms cathode to ground un-bypassed and 47 ohms bypassed with 220uF.
Vo - 9.2V RMS
Input - 960mV RMS
Reduced Ia from 49.3 to 40.2mA
Still have 2K2 feedback resistor
Cathode Resistor - 300R
Ia - 40.2mA
Va-c 276V
Vs-c - 260V
Vg-c - -12.47
Vo - 9Vrms into 8 ohm resistive load
Input - 384mV RMS
Grid drive - 10V RMS
Vo - 9.2V RMS
Input - 1.04V RMS
Below 36mA Ia, the output starts to drop off.
I tried dropping the cathode current to 30mA, but the output started distorting badly above 6.8V rms with really strange crossover distortion.
The amp is turning into a rats nest. I'll be glad when I can build a Tube Lab.
Steven, thanks for trying that. So 10W is the limit. So it can be done but we need to bend the rules a little 🙂 Just like many others do with EL84 amps...
Have you done any listening tests of triode/pentode/ultralinear configurations? I might give that a go next weekend. I will try taking the anode dissipation up to 12W and see how it goes. I'm also waiting on parts to build voltage regulators for the screen grids. My amp too is getting crowded, I'm painting myself into a corner. Once I establish a final schematic then I'll reassess how to rebuild it properly, more than likely point to point with modules for the silicon bits.
But now I see a problem with running 4 ohm speakers which are my current favourites. The stock OPT's on my amp are approx 6150 ohms anode to anode with an 8 ohm load, and when running a 4 ohm load, it becomes a very good example of how not to load an output stage. The class B load line will shoot straight up through the max dissipation line. I have already run the amp like this and it sounded very good, but I don't think the tubes will last very long and the power output will be much lower, as I found out during testing last week. I will need some different ones like the 9K5 a-a ones I'm using now, which I would like to keep for another project. Here's some loadlines for my stock OPT's with 8 and 4 ohm speakers, and a 12K a-a line thrown in.
Looking at the ElCheapo, which uses 6AQ5/12AQ5 valves, it puts out around 8W with approx 310V across the tube.
Ian.
Have you done any listening tests of triode/pentode/ultralinear configurations? I might give that a go next weekend. I will try taking the anode dissipation up to 12W and see how it goes. I'm also waiting on parts to build voltage regulators for the screen grids. My amp too is getting crowded, I'm painting myself into a corner. Once I establish a final schematic then I'll reassess how to rebuild it properly, more than likely point to point with modules for the silicon bits.
But now I see a problem with running 4 ohm speakers which are my current favourites. The stock OPT's on my amp are approx 6150 ohms anode to anode with an 8 ohm load, and when running a 4 ohm load, it becomes a very good example of how not to load an output stage. The class B load line will shoot straight up through the max dissipation line. I have already run the amp like this and it sounded very good, but I don't think the tubes will last very long and the power output will be much lower, as I found out during testing last week. I will need some different ones like the 9K5 a-a ones I'm using now, which I would like to keep for another project. Here's some loadlines for my stock OPT's with 8 and 4 ohm speakers, and a 12K a-a line thrown in.
Looking at the ElCheapo, which uses 6AQ5/12AQ5 valves, it puts out around 8W with approx 310V across the tube.
Ian.
Ian, I think I mentioned my test into 4 ohms that went poorly. I expect you will get the same result with these opts.
I doubt you will get much more power than you have now (at least not cleanly) because the regulation of the PT is not up to it. You might get lucky, as we have found out that no two transformers seem to be built to the same specs.
The B+ started to sag on mine as soon as I tried an extra 10mA through the 6P1's. I bumped Ia to 49mA using the 317 CCS, and the plate voltage went down to 237v. It may be that the resistors on the primary are spoiling the regulation as the only way mine holds up under increased HT currents seems to be if I allow it to run the full 240v on the 220v primary, but it won't hold up for long judging by the temperature it reaches when I do that.
So that's about as far as I plan to push my Meng redesign, and it is actually quite listenable and reliable now.
Oh, and regards your musing about what it would sound like with EL84's, I can tell you that you won't go back once you have heard EL84's with suitable opts 🙂 I suspect, though, that the load presented by the Meng OPTs won't be correct for them.
My little stereo SE ECC81/EL84 amp with opts out of radios sounds better than the Meng in all respects except depth of bass, and puts out about 4W before clip on said radio's power transformer and B+ of 250v and Ia of 36mA. I can't run more current than that through the OPTs, they get hot enough with that amount.
Gary
I doubt you will get much more power than you have now (at least not cleanly) because the regulation of the PT is not up to it. You might get lucky, as we have found out that no two transformers seem to be built to the same specs.
The B+ started to sag on mine as soon as I tried an extra 10mA through the 6P1's. I bumped Ia to 49mA using the 317 CCS, and the plate voltage went down to 237v. It may be that the resistors on the primary are spoiling the regulation as the only way mine holds up under increased HT currents seems to be if I allow it to run the full 240v on the 220v primary, but it won't hold up for long judging by the temperature it reaches when I do that.
So that's about as far as I plan to push my Meng redesign, and it is actually quite listenable and reliable now.
Oh, and regards your musing about what it would sound like with EL84's, I can tell you that you won't go back once you have heard EL84's with suitable opts 🙂 I suspect, though, that the load presented by the Meng OPTs won't be correct for them.
My little stereo SE ECC81/EL84 amp with opts out of radios sounds better than the Meng in all respects except depth of bass, and puts out about 4W before clip on said radio's power transformer and B+ of 250v and Ia of 36mA. I can't run more current than that through the OPTs, they get hot enough with that amount.
Gary
rotaspec,
I just noticed last night that the cathode voltage increases with increased drive. I knew this happened with class AB1 operation, but didn't expect to see it with class A.
I can get 8W out of the Sven 6P1P-PP with 265V B+ and anode current at 33mA no drive. When I am driving it to 8W out the anode current reads 36mA. So I suspect you might be able to reduce your cathode current slightly if you are running 250V B+ and only getting 4W out. That might help your transformers run a bit cooler. In addition, you might see an increase in B+ and a slight increase in power out due to the lower static loading conditions.
I've removed the second supply that I was using to run the output tubes at 300V and am back to a single supply for the input, driver and output tubes. This is the configuration where I get 8W with 33mA drive no input and 265V on the output transformer B+ tap.
This raises the question of how much more anode current do we need (1) beyond that necessary to achieve the desired output level and (2) some slight margin to account for line input voltage fluctuation.
Is there some additional amount that is necessary which reduces distortion or some other parameter that is desirable?
It appears to me that the 6P1P/-EV/6AQ5 group of tubes is only good for 8W at 250V B+, contrary to what the data sheets say. Maybe they were testing with different impedance transformers and it is possible to get 10W under some Ideal condition, or maybe it it the theoretical max not accounting for real world losses like transformer resistance, etc.
Regardless, it appears to me that the most I'm going to get out of this amp is 8W. That is good enough for my application, so I will quit playing with power out and start looking at frequency response, feedback and changing the bias on the input stage wrt to the splitter to see if I can optimize it a bit more.
Before I make any more changes I think I'll drag one of the Kenwood KL-5050s back in the garage and see how it sounds.
I just noticed last night that the cathode voltage increases with increased drive. I knew this happened with class AB1 operation, but didn't expect to see it with class A.
I can get 8W out of the Sven 6P1P-PP with 265V B+ and anode current at 33mA no drive. When I am driving it to 8W out the anode current reads 36mA. So I suspect you might be able to reduce your cathode current slightly if you are running 250V B+ and only getting 4W out. That might help your transformers run a bit cooler. In addition, you might see an increase in B+ and a slight increase in power out due to the lower static loading conditions.
I've removed the second supply that I was using to run the output tubes at 300V and am back to a single supply for the input, driver and output tubes. This is the configuration where I get 8W with 33mA drive no input and 265V on the output transformer B+ tap.
This raises the question of how much more anode current do we need (1) beyond that necessary to achieve the desired output level and (2) some slight margin to account for line input voltage fluctuation.
Is there some additional amount that is necessary which reduces distortion or some other parameter that is desirable?
It appears to me that the 6P1P/-EV/6AQ5 group of tubes is only good for 8W at 250V B+, contrary to what the data sheets say. Maybe they were testing with different impedance transformers and it is possible to get 10W under some Ideal condition, or maybe it it the theoretical max not accounting for real world losses like transformer resistance, etc.
Regardless, it appears to me that the most I'm going to get out of this amp is 8W. That is good enough for my application, so I will quit playing with power out and start looking at frequency response, feedback and changing the bias on the input stage wrt to the splitter to see if I can optimize it a bit more.
Before I make any more changes I think I'll drag one of the Kenwood KL-5050s back in the garage and see how it sounds.
- Status
- Not open for further replies.