6CM6: Audio: 12 Watts, Vertical: 8 Watts
http://scottbecker.net/tube/sheets/137/6/6CM6.pdf
6973: Audio 12 Watts
6CZ5: Audio: 12 Watts, Vertical 10 Watts (same tube as 6973 apparently)
http://scottbecker.net/tube/sheets/049/6/6973.pdf
http://scottbecker.net/tube/photos/049/6/6973.jpg
http://scottbecker.net/tube/sheets/137/6/6CZ5.pdf
http://scottbecker.net/tube/photos/137/6/6CZ5.jpg
Some other issues to consider. Even within the same tube type, you will find different size plates, presence or absence of grid cooler fins, different plate construction/material/surface finish. The glass bulb temperature has a rating usually (different glass compositions were used), and spacing/ventilation between the tubes. TV sets usually had the tubes all crammed together tightly. Some ventilation (chassis fan, holes around sockets) should allow some uprating all by itself. Some amateur radio operators immersed the output tubes in mineral oil for cooling even.
http://scottbecker.net/tube/sheets/137/6/6CM6.pdf
6973: Audio 12 Watts
6CZ5: Audio: 12 Watts, Vertical 10 Watts (same tube as 6973 apparently)
http://scottbecker.net/tube/sheets/049/6/6973.pdf
http://scottbecker.net/tube/photos/049/6/6973.jpg
http://scottbecker.net/tube/sheets/137/6/6CZ5.pdf
http://scottbecker.net/tube/photos/137/6/6CZ5.jpg
Some other issues to consider. Even within the same tube type, you will find different size plates, presence or absence of grid cooler fins, different plate construction/material/surface finish. The glass bulb temperature has a rating usually (different glass compositions were used), and spacing/ventilation between the tubes. TV sets usually had the tubes all crammed together tightly. Some ventilation (chassis fan, holes around sockets) should allow some uprating all by itself. Some amateur radio operators immersed the output tubes in mineral oil for cooling even.
Last edited:
If I were messing around with sweep tubes, I'd keep the screen voltage low and beat the living snot out of the plate dissipation.
"Looks like maybe 20% is in the ball park."
Been waiting for George to chime in. I think he has gotten well over +50% on some exceptional tubes. But then you have to wonder how long they will last when they start glowing!
Another trick is to trade off screen dissipation for more plate dissipation. (This would require some limit on the grid drive to the tube, so the plate cannot be pulled down into saturation.) Look on page 4 of the 6JS6C datasheet below. They allow 4 more plate watts for every 1 watt less screen dissipation.
http://scottbecker.net/tube/sheets/123/6/6JS6C.pdf
Been waiting for George to chime in. I think he has gotten well over +50% on some exceptional tubes. But then you have to wonder how long they will last when they start glowing!
Another trick is to trade off screen dissipation for more plate dissipation. (This would require some limit on the grid drive to the tube, so the plate cannot be pulled down into saturation.) Look on page 4 of the 6JS6C datasheet below. They allow 4 more plate watts for every 1 watt less screen dissipation.
http://scottbecker.net/tube/sheets/123/6/6JS6C.pdf
" if you are going pp run deep into AB, then you can afford to go WILDLY over ratings, since half the time each tube is conducting stuff-all. If you are a Class A purist, you don't have that option... "
That's for sure. No rating upgrades possible if the amp's running in Class A. That would be the same mode as in the TV set. (could still up-rate with forced ventilation though) Only audio power in Class AB can take advantage of the crest factor (ratio of average power to peak power) since the tubes run cooler at low power out. The screen grid is likely the shortest fuse to overheating quickly on extended peaks, so it would pay well to avoid overdriving the tube into heavy screen current (some kind of grid drive limiter).
That's for sure. No rating upgrades possible if the amp's running in Class A. That would be the same mode as in the TV set. (could still up-rate with forced ventilation though) Only audio power in Class AB can take advantage of the crest factor (ratio of average power to peak power) since the tubes run cooler at low power out. The screen grid is likely the shortest fuse to overheating quickly on extended peaks, so it would pay well to avoid overdriving the tube into heavy screen current (some kind of grid drive limiter).
Last edited:
Just to be sure I understand. In the class AB case are you saying that the dissipation at idle can be well over spec? I was thinking that in class A max dissipation is at idle where in class AB the dissipation goes up with output.
I am planning several projects with verticals. The 6LU8 SE for my mains is already decided but I plan on at least a 6LU8 PP (got a bunch of these on hand) for woofers (not sub) and a 6GF7 SE for computer or bedroom system. On the PP it would be nice to manage 30WPC or so from single pair of bottles per channel. I can't remember the voltage available with the tranny I have on hand but I think it was 400V or better.
I am planning several projects with verticals. The 6LU8 SE for my mains is already decided but I plan on at least a 6LU8 PP (got a bunch of these on hand) for woofers (not sub) and a 6GF7 SE for computer or bedroom system. On the PP it would be nice to manage 30WPC or so from single pair of bottles per channel. I can't remember the voltage available with the tranny I have on hand but I think it was 400V or better.
Yes - largely because it will spend not a lot of time at idle (depending on how you configure obviously...)
While the instantaneous current may be very high in AB, the average will be at or below the spec'd value (again depending on your configuration). Since these are not sand items, instant doesn't matter, average does.
Huh... I guess I misunderstood the earlier comment on deep class AB. Class A runs at max dissipation all the time. So no up-rating possible. Optimum biased class AB runs at least dissipation at idle. (there is an optimum bias that gives least distortion and is usually relatively low current draw at idle). So it can take advantage of the crest factor to handle above rated power on peaks.
Deep class AB, I assume this means biased on for most of the cycle, would run hot. I wouldn't up-rate for that.
In theory, a class AB stage would again enter low dissipation at high current, since the voltage drop in the device would be low. But tubes won't work without some significant voltage drop, especially triodes. (pentodes can almost get there, but risk burning up the screen grid) Tubes are inefficient, they just get hotter the more power you try to get out of them.
Terman's book "Electronic and Radio Engineering" gives an example of optimum class AB (he calls it class B) biasing, using the projected cutoff method, on the bottom of page 355. (4th edition)
This involves sliding the two tube curves (one inverted) horizontally until their sum is closest to a straight line. He just uses the projected cutoff method to predict that point approximately.
Under biasing (deep class AB presumably) would overlap the conduction curves too much, causing "gm doubling" at low signals.
Deep class AB, I assume this means biased on for most of the cycle, would run hot. I wouldn't up-rate for that.
In theory, a class AB stage would again enter low dissipation at high current, since the voltage drop in the device would be low. But tubes won't work without some significant voltage drop, especially triodes. (pentodes can almost get there, but risk burning up the screen grid) Tubes are inefficient, they just get hotter the more power you try to get out of them.
Terman's book "Electronic and Radio Engineering" gives an example of optimum class AB (he calls it class B) biasing, using the projected cutoff method, on the bottom of page 355. (4th edition)
This involves sliding the two tube curves (one inverted) horizontally until their sum is closest to a straight line. He just uses the projected cutoff method to predict that point approximately.
Under biasing (deep class AB presumably) would overlap the conduction curves too much, causing "gm doubling" at low signals.
Last edited:
Late edit to above:
"So it can take advantage of the crest factor to handle above rated power on peaks."
Should be:
"So it can take advantage of averaging and the crest factor to handle above rated power on peaks."
Actually figuring out the amount of power handled however is complex due to the instantaneous tube power being the product of tube current and tube voltage drop. Then having to integrate that over a cycle, and then figure in the audio crest factor for non constant sine wave signals. And the crest factor varies depending on the signal source. Compressed FM radio having a low crest factor for example.
"So it can take advantage of the crest factor to handle above rated power on peaks."
Should be:
"So it can take advantage of averaging and the crest factor to handle above rated power on peaks."
Actually figuring out the amount of power handled however is complex due to the instantaneous tube power being the product of tube current and tube voltage drop. Then having to integrate that over a cycle, and then figure in the audio crest factor for non constant sine wave signals. And the crest factor varies depending on the signal source. Compressed FM radio having a low crest factor for example.
What if I told you what I observe in class A, that the max dissipation is at idle and it may decrease from idle to full signal? How can that happen?
From my analysis, the max dissipation of a class AB or B stage, for a sine wave workload, is somewhere in the middle of the output power, depending on the idle dissipation. Integrating the instantaneous volt*amps over the full cycle as Don points out. ("Instantaneous power" still doesn't make sense to me...)
Crest factor of music, even highly compressed music, is still relatively high compared with sine waves (CF of a sine wave is 1.414..) so I believe a class B or AB stage can be rated to play full volume music at clipping at a much higher level than a sine wave, based on max dissipation.
Here's an example of the crossover analysis assuming 3/2 power law, should be "gm-error":
From my analysis, the max dissipation of a class AB or B stage, for a sine wave workload, is somewhere in the middle of the output power, depending on the idle dissipation. Integrating the instantaneous volt*amps over the full cycle as Don points out. ("Instantaneous power" still doesn't make sense to me...)
Crest factor of music, even highly compressed music, is still relatively high compared with sine waves (CF of a sine wave is 1.414..) so I believe a class B or AB stage can be rated to play full volume music at clipping at a much higher level than a sine wave, based on max dissipation.
Here's an example of the crossover analysis assuming 3/2 power law, should be "gm-error":
Attachments
Last edited:
"What if I told you what I observe in class A, that the max dissipation is at idle and it may decrease from idle to full signal? How can that happen? "
You are quite right. I was earlier forgetting to subtract out the output power for the class A case. For linear tubes, the current draw from the B+ is constant, so total power draw from B+ is constant. But power output goes up as V^2, so tube dissipation would then drop as the signal gets larger. Although practically it won't fall to zero since the tube needs some operating voltage. And typical tube distortion also makes the off going tube drag its feet about turning off when it should, which usually shows up as some increase in B+ current at max signal.
This would still not be useful for any tube up-rating purposes though, unless someone likes listening to max clipped waveforms all the time.
Near class B (ideal case, linear, no bias current) as you say, increases from zero tube dissipation at zero signal, to max tube dissipation at 1/2 max signal, and back down to zero tube dissipation at max signal. Practically though, the minimum tube voltage requirements keep it from coming back down to zero at max current. And idle current biasing keeps the low end diss. above zero too.
-------------------------
Crazy off the wall idea for the day:
Suppose the amp runs in class A mode, but has an aux. switching circuit that adds in a "square" wave at the tube side but subtracts it back out at the OT secondary side. The amplitude would be adjustable (a power D/A with limited bits) so as to bring the signal at the tube plates up to near max, to get minimum tube dissipation always. Efficient class A? I guess this would really be the same as digitally adjustable tracking B+ voltage. Then again, maybe not, the digital glitches cancel here. Nah, it still the same, the tube current will glitch with the changing voltages.
You are quite right. I was earlier forgetting to subtract out the output power for the class A case. For linear tubes, the current draw from the B+ is constant, so total power draw from B+ is constant. But power output goes up as V^2, so tube dissipation would then drop as the signal gets larger. Although practically it won't fall to zero since the tube needs some operating voltage. And typical tube distortion also makes the off going tube drag its feet about turning off when it should, which usually shows up as some increase in B+ current at max signal.
This would still not be useful for any tube up-rating purposes though, unless someone likes listening to max clipped waveforms all the time.
Near class B (ideal case, linear, no bias current) as you say, increases from zero tube dissipation at zero signal, to max tube dissipation at 1/2 max signal, and back down to zero tube dissipation at max signal. Practically though, the minimum tube voltage requirements keep it from coming back down to zero at max current. And idle current biasing keeps the low end diss. above zero too.
-------------------------
Crazy off the wall idea for the day:
Suppose the amp runs in class A mode, but has an aux. switching circuit that adds in a "square" wave at the tube side but subtracts it back out at the OT secondary side. The amplitude would be adjustable (a power D/A with limited bits) so as to bring the signal at the tube plates up to near max, to get minimum tube dissipation always. Efficient class A? I guess this would really be the same as digitally adjustable tracking B+ voltage. Then again, maybe not, the digital glitches cancel here. Nah, it still the same, the tube current will glitch with the changing voltages.
Last edited:
"what class of amp did the vertical amplifier run?"
Apparently class A. I guess they had to linearly sweep both ways, where as the horiz. channel only had to linearly sweep one way? (not sure about that) Still you would think that an inductor could be linearly charged up then back down using just a switched square wave voltage? Maybe was due to the time interval (60 Hz) being slow, so would require a very big inductor? (avoiding 1000s turn winding on defl. yoke, such a big air gap for much inductance at 60 Hz)
See middle section of George's post:
http://www.diyaudio.com/forums/tubes-valves/133034-6l6gc-ab2-amp-39.html#post2119234
Some TV experts around?
Apparently class A. I guess they had to linearly sweep both ways, where as the horiz. channel only had to linearly sweep one way? (not sure about that) Still you would think that an inductor could be linearly charged up then back down using just a switched square wave voltage? Maybe was due to the time interval (60 Hz) being slow, so would require a very big inductor? (avoiding 1000s turn winding on defl. yoke, such a big air gap for much inductance at 60 Hz)
See middle section of George's post:
http://www.diyaudio.com/forums/tubes-valves/133034-6l6gc-ab2-amp-39.html#post2119234
Some TV experts around?
Last edited:
Thanks for the link. I forget what tube it was, but one of the (beam) pentode had vertical amplifier data only.
I think that is the best approach, too. The other thing to be careful of is the surface temperature of the glass. It's more important than at first blush, because the usual Barium-Aluminium getter cannot work normally at temperatures above 200 deg C.
I haven't tried it, but I imagine that the gun-type IR thermometers would give a good monitor of the temperature of the glass during varying overdissipation sessions.
"The other thing to be careful of is the surface temperature of the glass. It's more important than at first blush, because the usual Barium-Aluminium getter cannot work normally at temperatures above 200 deg C."
Getter begins outgassing?
-----------------------------
"I forget what tube it was, but one of the (beam) pentode had vertical amplifier data only. "
6HE5, 6JB5 etc?
These tubes are a bit of a mystery to me. They have nice clean pentode curves (low screen current distortion), and identical dataspecs to 6V6. (notably low gm, low DC max current, and high screen V) The 6V6 gets glam sound ratings, but its curves do show significant screen current distortion. Not really the same tube by the curves. Someone did recently think well of them (6HE5, 6JB5 ...) for a guitar amplifier they built. Called them 6V6s on steroids.
None of these look too impressive when triode wired though. Droopy curves on the curve tracer. But that is usual for small cathode tubes. Great triodes from pentodes seem to require a big flat cathode. 6CB5A a case in point. Over on the Asylum someone built a SE amp with this recently, and a group of listeners mostly gave it the winning vote versus a 300B amp.
I have to wonder what say 6LG6/6LB6 would sound like. Nice clean curves just like the 6HE5.... series, AND a big flat cathode. Well, I got a box full of 21LG6, guess I should try them out. Too many projects.
Getter begins outgassing?
-----------------------------
"I forget what tube it was, but one of the (beam) pentode had vertical amplifier data only. "
6HE5, 6JB5 etc?
These tubes are a bit of a mystery to me. They have nice clean pentode curves (low screen current distortion), and identical dataspecs to 6V6. (notably low gm, low DC max current, and high screen V) The 6V6 gets glam sound ratings, but its curves do show significant screen current distortion. Not really the same tube by the curves. Someone did recently think well of them (6HE5, 6JB5 ...) for a guitar amplifier they built. Called them 6V6s on steroids.
None of these look too impressive when triode wired though. Droopy curves on the curve tracer. But that is usual for small cathode tubes. Great triodes from pentodes seem to require a big flat cathode. 6CB5A a case in point. Over on the Asylum someone built a SE amp with this recently, and a group of listeners mostly gave it the winning vote versus a 300B amp.
I have to wonder what say 6LG6/6LB6 would sound like. Nice clean curves just like the 6HE5.... series, AND a big flat cathode. Well, I got a box full of 21LG6, guess I should try them out. Too many projects.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.