Looking through my tube manuals, I have discovered many interesting and often neglected tubes. Amongst these are Class B ("enhanced mode" in ss terminology) triodes such as the #53, #79, 6y7g. Have any of you guys ever built a push-pull amplifier using such devices? Do you think that a small forward bias might help clear up any distortion caused by Class B operation? Just curious.
Cheers
6f6
Cheers
6f6
Class 'B' audio is not so much done properly with heavy distortion, rather it comes as the result of not planning for the grids to pull current when driven into the 'B' region.
One just cannot 'cheaply' couple via RC network topology and achieve the juice necessary for low distortion operation.
Look to beefy cathode followers, or interstage transformers capable of the necessary Watts.
Many very sweet-sounding AM broadcast transmitters used class 'B' modulators and got a typical overall THD of 5% or less at full (100%) modulation with sine-wave tone tests.
Hard tubes correctly balanced & biased were always a prerequisite.
Good 'B' interstage iron was never cheap, and is tough to source today. The step-up ratio is most critical in selection as well. Best to consult your transformer company and discuss it with them...as power will be flowing in those secondaries. Think phase shift, you don't want any.
The payoff is prolific power from places you never thought, even more than AB2.
Dennis
ps...1635 dual triode. Sleeper.
One just cannot 'cheaply' couple via RC network topology and achieve the juice necessary for low distortion operation.
Look to beefy cathode followers, or interstage transformers capable of the necessary Watts.
Many very sweet-sounding AM broadcast transmitters used class 'B' modulators and got a typical overall THD of 5% or less at full (100%) modulation with sine-wave tone tests.
Hard tubes correctly balanced & biased were always a prerequisite.
Good 'B' interstage iron was never cheap, and is tough to source today. The step-up ratio is most critical in selection as well. Best to consult your transformer company and discuss it with them...as power will be flowing in those secondaries. Think phase shift, you don't want any.
The payoff is prolific power from places you never thought, even more than AB2.
Dennis
ps...1635 dual triode. Sleeper.
For superb performance at modest cost, NOTHING comes close to PowerDrive at dealing with a positive control grid current regime. Thank you George Anderson, AKA Tubelab!
If my finances ever permit, I intend to build an "El Cheapo" that employs Class "B2" #46 "finals". PowerDrive figures highly in my cogitations. The example circuitry provided by the data sheet uses a good deal of voltage gain, followed by a step down, phase splitting, interstage trafo. That sort of "iron" is damned expensive and constrains the use of NFB. PowerDrive "blocks" between the 12AT7 LTP anodes and the O/P tube grids is low cost and makes loop NFB easy.
If my finances ever permit, I intend to build an "El Cheapo" that employs Class "B2" #46 "finals". PowerDrive figures highly in my cogitations. The example circuitry provided by the data sheet uses a good deal of voltage gain, followed by a step down, phase splitting, interstage trafo. That sort of "iron" is damned expensive and constrains the use of NFB. PowerDrive "blocks" between the 12AT7 LTP anodes and the O/P tube grids is low cost and makes loop NFB easy.
Attachments
I'm working on something like that right now. Have a box of 50 x 6Н7С (russian version of 6N7GT). The goal is to get 10W out of it and only use one other tube. (12Ж1Л as voltage amplifier - CCS loaded with open loop gain about ~2000 and heavy feedback) Both tubes can be bought for less than 1$.
When this is done, it will be the daily rocker. No worries about tube prices anymore.
Had a few problems with the current driver. Need to deliver a max of about 30mA grid current for 80Vpp. Will use a floating OpAmp with driver for the next tests.
the 6N7GT is working in a 8K-Rpp load. See attachment. I'm looking to reduce the current even more. I don't need that much power here
Next thing to do is an small SMPS for it.
When this is done, it will be the daily rocker. No worries about tube prices anymore.
Had a few problems with the current driver. Need to deliver a max of about 30mA grid current for 80Vpp. Will use a floating OpAmp with driver for the next tests.
the 6N7GT is working in a 8K-Rpp load. See attachment. I'm looking to reduce the current even more. I don't need that much power here
Next thing to do is an small SMPS for it.
Attachments
The grid only draws current when it's in the positive voltage region.
This is what's designated Class (x)2, where (x)= A, B or AB
When you speak about the 'B'-region, you're talking about what negative grid voltage is sufficient to just cut-off the plate-current. Well, almost cut-off, a fraction may still be left.
In this state you'll not experience grid-current in any case.
About the second statement:
I don't think so. If AB2 is done right you'll pull A LOT of power from the tubes and at low distortion as well. You may use FET's or Cathode-Followers as DC-coupled drivers to achieve this in real-life.
The difference between B1 and AB2 is huge.
Now, if you were speaking of B2 it would be another matter but the price to pay is distortion.
rgds,
/tri-comp
Please explain why you expect to get away with a load-line so far above maximum plate-dissipation.
How much could you exceed this recommended dissipation-level and still be safe?
rgds,
/tri-comp
The modulation of a single tube in terms of voltage or current is half-sinusoidal. In terms of power it is proportional to the voltage or current squared.
When your idle operating point is at zero watts plate dissipation, then your maximum modulation can go as high as 4 times the max-dissipation of the tube. (e.g. 100W for an EL34). This works because the mean value is still below or equal 25W. Thermal inertia of the plate material guarantees the rest. This was done successfully millionth times.
The attached picture shows this effect. The area below the curves is equal, but the highest point of the red constant curve (Class-A) is only 1/4 of the blue.
As I said, I want to go down with my idle operating point to lower the idle dissipation. Interestingly I'm already lower than the datasheet states.
Hope that helped a bit. There is tons of literature which describes this operation principle.
When your idle operating point is at zero watts plate dissipation, then your maximum modulation can go as high as 4 times the max-dissipation of the tube. (e.g. 100W for an EL34). This works because the mean value is still below or equal 25W. Thermal inertia of the plate material guarantees the rest. This was done successfully millionth times.
The attached picture shows this effect. The area below the curves is equal, but the highest point of the red constant curve (Class-A) is only 1/4 of the blue.
As I said, I want to go down with my idle operating point to lower the idle dissipation. Interestingly I'm already lower than the datasheet states.
Hope that helped a bit. There is tons of literature which describes this operation principle.
Attachments
There is and always has been some confusion about class B and just what it is. The textbook definition I learned back in the 60's goes like this:
Class A - Plate current flows for the ENTIRE 360 degrees of a cycle of the test tone.
Class AB1 - Plate current in ONE output tube flows for more than 180 degrees, but less than 360 degrees of the cycle. NO grid current flows during ANY part of the cycle.
Class AB2 - Plate current in ONE output tube flows for more than 180 degrees, but less than 360 degrees of the cycle. Grid current flows for some or all of the cycle.
Class B1 - Plate current flows for EXACTLY 180 degrees of the cycle. NO grid current flows during ANY part of the cycle.
Class B2 - Plate current flows for EXACTLY 180 degrees of the cycle. Grid current flows for some or all of the cycle.
Class C - Plate current flows for LESS THAN 180 degrees of the cycle. No distinction is made as to grid current.
Note that even though the textbook definition states that a class B amplifier ceases to draw plate current at exactly the transition point, implying zero idle current, this is rarely seen in due to the high crossover distortion. Most tube data sheets that give specs for Class B operation show a non zero idle current. This is technically class AB, but called Class B in most vintage books. So, whatever you call it a small idle current is generally needed to improve performance.
The term "class B" alone does not define the occurrence of grid current. You could take a common P-P audio amp using say 6L6GC's and simply turn down the bias current (more negative grid) until the amp idles at zero (or a few mA) current and have a class B amp with no grid current. It would not sound very good, but could be called class B1. The 6L6GC has poor linearity in the region of low plate current.
There were dozens of tubes, mostly triodes, designed to operate more linearly at low plate currents. They were specified for class B push pull operation. These tubes tend to have a high Mu and are intended to operate in the grid current region. They may be called "class B" tubes, but class B2 is implied. I don't recall any common tube where class B data is given that does NOT imply some grid current. Some of these "class B" tubes may also work quite well in class A or class AB, and may be specified for such use. The 811A, and the 211 are good examples. These tubes work quite well in class A2 as well. I get 40 watts from a 211 in A2 using PowerDrive.
There was a schematic published in the late 30's detailing how to wire a pentode or a beam power tube for "enhanced triode" or "high Mu triode" mode. The schematic used an 807 with transformer drive applied directly to G2 and also to G1 through a resistor. The bias voltage was ZERO. An 807 connected in this manner does indeed behave much like an 811A.
I have taken this concept further and am still working on it. The reason for this direction is to get around the exploding screen grid scenario that happens when extracting big power from screen driven sweep tubes. Expect to see some big power sweep tube amps in the future. Se this thread for further info:
http://www.diyaudio.com/forums/tube...ed-drive-strawman-design.html?highlight=G1=G2
Actually if you are playing music, not sine waves, even these numbers are far too CONSERVATIVE!!!! Bob Carver made a career out of extracting BIG power out of small boxes.
His only real secret is the fact that most music has at least 10 db "crest factor". usually the number is closer to or above 20 db. This means that if you crank music through a 100 watt amp with the volume control set to where you just touch clipping on peaks, the AVERAGE power is between 1 and 10 watts!!!!
As the_manta states the further you reduce the idle current, the more room you have for increasing the maximum power output before the sustained AVERAGE dissipation reaches the maximum spec.
If your tubes are drawing 50 mA each at idle, from a 400 volt supply they are dissipating 20 watts each whthout making any music. If they are 25 watt tubes, you have 5 watts of headroom to make music with.
If you can redesign the circuit to get low distortion with the same tubes biased at 5 ma each at idle, your tubes are now dissipating 2 watts each at idle. You now have 23 watts per tube to make music with.
It is easy to make big power with a class B2 design. I have seen over 100 watts of continuous sine wave from a pair of small sweep tubes rated at 15 watts dissipation WITHOUT any signs of red glow. This was using pure screen drive and 5 mA of idle current.
There are two drawbacks:
Pure screen drive will abuse the screen grid and lead to meltdown if run continuously at high power with sine waves. The low crest factor in music is OK. I am working with driving both grids to eliminate this issue.
Any kind of enhanced mode (G2 only or G1 + G2) will have a high output impedance. It is even higher than pure pentode mode. Schade type feedback solves this one.
This is a different question than the one answered above. Some tubes, especially Chinese sourced audio tubes like the 6L6GC are not safe when pushed much above their published specs. Some tubes like older US made sweep tubes will operate forever at twice their published dissipation, while a different tube with the same number on it may not. The only way to tell is to test several samples of the tube in question for several hours each. I will push a tube until it just starts to show plate glow in a darkened room, leave it there for several hours and watch for signs of runaway. Then operate it at about 75% of that power dissipation. Even still I have seen a tube be stable for months then unexpectedly run away. curcuits operating "at the edge" or beyond should be fused and operate with regulated bias and screen grid voltages to avoid meltdown due power fluctuations.
Class A - Plate current flows for the ENTIRE 360 degrees of a cycle of the test tone.
Class AB1 - Plate current in ONE output tube flows for more than 180 degrees, but less than 360 degrees of the cycle. NO grid current flows during ANY part of the cycle.
Class AB2 - Plate current in ONE output tube flows for more than 180 degrees, but less than 360 degrees of the cycle. Grid current flows for some or all of the cycle.
Class B1 - Plate current flows for EXACTLY 180 degrees of the cycle. NO grid current flows during ANY part of the cycle.
Class B2 - Plate current flows for EXACTLY 180 degrees of the cycle. Grid current flows for some or all of the cycle.
Class C - Plate current flows for LESS THAN 180 degrees of the cycle. No distinction is made as to grid current.
Note that even though the textbook definition states that a class B amplifier ceases to draw plate current at exactly the transition point, implying zero idle current, this is rarely seen in due to the high crossover distortion. Most tube data sheets that give specs for Class B operation show a non zero idle current. This is technically class AB, but called Class B in most vintage books. So, whatever you call it a small idle current is generally needed to improve performance.
The term "class B" alone does not define the occurrence of grid current. You could take a common P-P audio amp using say 6L6GC's and simply turn down the bias current (more negative grid) until the amp idles at zero (or a few mA) current and have a class B amp with no grid current. It would not sound very good, but could be called class B1. The 6L6GC has poor linearity in the region of low plate current.
There were dozens of tubes, mostly triodes, designed to operate more linearly at low plate currents. They were specified for class B push pull operation. These tubes tend to have a high Mu and are intended to operate in the grid current region. They may be called "class B" tubes, but class B2 is implied. I don't recall any common tube where class B data is given that does NOT imply some grid current. Some of these "class B" tubes may also work quite well in class A or class AB, and may be specified for such use. The 811A, and the 211 are good examples. These tubes work quite well in class A2 as well. I get 40 watts from a 211 in A2 using PowerDrive.
There was a schematic published in the late 30's detailing how to wire a pentode or a beam power tube for "enhanced triode" or "high Mu triode" mode. The schematic used an 807 with transformer drive applied directly to G2 and also to G1 through a resistor. The bias voltage was ZERO. An 807 connected in this manner does indeed behave much like an 811A.
I have taken this concept further and am still working on it. The reason for this direction is to get around the exploding screen grid scenario that happens when extracting big power from screen driven sweep tubes. Expect to see some big power sweep tube amps in the future. Se this thread for further info:
http://www.diyaudio.com/forums/tube...ed-drive-strawman-design.html?highlight=G1=G2
Actually if you are playing music, not sine waves, even these numbers are far too CONSERVATIVE!!!! Bob Carver made a career out of extracting BIG power out of small boxes.
His only real secret is the fact that most music has at least 10 db "crest factor". usually the number is closer to or above 20 db. This means that if you crank music through a 100 watt amp with the volume control set to where you just touch clipping on peaks, the AVERAGE power is between 1 and 10 watts!!!!
As the_manta states the further you reduce the idle current, the more room you have for increasing the maximum power output before the sustained AVERAGE dissipation reaches the maximum spec.
If your tubes are drawing 50 mA each at idle, from a 400 volt supply they are dissipating 20 watts each whthout making any music. If they are 25 watt tubes, you have 5 watts of headroom to make music with.
If you can redesign the circuit to get low distortion with the same tubes biased at 5 ma each at idle, your tubes are now dissipating 2 watts each at idle. You now have 23 watts per tube to make music with.
It is easy to make big power with a class B2 design. I have seen over 100 watts of continuous sine wave from a pair of small sweep tubes rated at 15 watts dissipation WITHOUT any signs of red glow. This was using pure screen drive and 5 mA of idle current.
There are two drawbacks:
Pure screen drive will abuse the screen grid and lead to meltdown if run continuously at high power with sine waves. The low crest factor in music is OK. I am working with driving both grids to eliminate this issue.
Any kind of enhanced mode (G2 only or G1 + G2) will have a high output impedance. It is even higher than pure pentode mode. Schade type feedback solves this one.
This is a different question than the one answered above. Some tubes, especially Chinese sourced audio tubes like the 6L6GC are not safe when pushed much above their published specs. Some tubes like older US made sweep tubes will operate forever at twice their published dissipation, while a different tube with the same number on it may not. The only way to tell is to test several samples of the tube in question for several hours each. I will push a tube until it just starts to show plate glow in a darkened room, leave it there for several hours and watch for signs of runaway. Then operate it at about 75% of that power dissipation. Even still I have seen a tube be stable for months then unexpectedly run away. curcuits operating "at the edge" or beyond should be fused and operate with regulated bias and screen grid voltages to avoid meltdown due power fluctuations.
Thanks the_manta & tubelab.com
Your views on dissipation are very interesting and I follow.
It's what you don't find in the published literature (Arguimbau, Romanowitz, Brazee, Seely a.o.).
I asked a while ago for reflections on the attached QQE04-5 load-lines being too agressive or not and didn't get any reply.
I see now that they aren't.
rgds,
/tri-comp
Your views on dissipation are very interesting and I follow.
It's what you don't find in the published literature (Arguimbau, Romanowitz, Brazee, Seely a.o.).
I asked a while ago for reflections on the attached QQE04-5 load-lines being too agressive or not and didn't get any reply.
I see now that they aren't.
rgds,
/tri-comp
You need to consider the time AVERAGED dissipation of both the plate and screen grid since this is what raises the temperature of these components. The operating temperature is what forces outgassing and eventually contaminates the vacuum.
In the use case brought up by the_manta, he advocates operating the tube at up to 4 times the rated dissipation on peaks (assuming a sine wave). I have gone far beyond this (assuming music and brief sine wave testing), but there is another aspect that must also be considered. The tube creates a "space charge" cloud of excess electrons around the cathode. This must remain true throughout the entire operating range. In other words if you attempt to suck more out of the plate than the cathode can provide you will upset the current density and risk damage to the cathode and a possible tube arc.
If the tube was constructed perfectly the coating on the cathode would be uniform, as would the spacings of all the internal elements. Thus the current density will be uniformly distributed throughout the tube. Unfortunately this is rarely the case. If you are going to lean on a tube hard, test it first. In a dark room, increase the current until plate glow is seen. Ideally the glow should be even along the plate from top to bottom, and equal between the two sides. If one side is much brighter than the other, or it is hot at the top and not the bottom (or vice versa) then this tube is not well constructed. A tube with visible hot and cold sopts is worse case and should be avoided. The cathode will fail.
Some tubes have just enough cathode capacity to support the rated plate dissipation. Sometimes there is a peak cathode current spec. If there is, it should be respected. Most DHT's are emission limited, and are not good candidates for over dissipation. The original data sheet for the 45 tube reveals that is was intended to provide 18 watts from a pair in AB2. That might be true for a brand new pair in the 1930's, but only a few in my collection have the emission capability to support this, although all of them work fine in class A SE.
I have squeezed almost 90 watts from a pair of 6L6GC's in AB2 but it became obvious that I was getting close to the cathode's emission limits, and the plates dissipation limit. The KT88 has a bigger cathode, and can crank out 100+ watts, but the 7403 has the same plate structure as the KT88 but has a fat cathode with a 10 AMP! peak current rating. I have seen well over 200 watts flow from a pair with continuous sine wave testing. The efficiency was almost 80% so only 25 watts on AVERAGE were being burned up in each plate. The key to this kind of efficiency is a good cathode, and a low idle current. Can't get this out of a 6L6GC.
Look at the peak cathode capabilities of your chosen tube. Some sweep tubes are rated at over one AMP! If there is no rating compare the cathode size to known tubes. A good sweep tube can pull its plate down to 30 volts or so WITHOUT drawing grid current. I have seen the plate swing to ZERO volts using screen drive. Of course having the plate at zero while the screen is at +250 volts can (and did) create fireworks! The little 6BQ6's were cranking out over 125 watts from a pair WITHOUT plate glow before the screen grid failed leading to a fatal tube arc. I got the tubes for 98 cents each, so it was a worthwhile experiment.
I have recently seen 150 watts of continuous sine wave at 3% distortion flow from a pair of $1 13GB5's for about 1/2 an hour. This was in AB1. I am constructing an amplifier along these lines but it will be run at a more sane 100 WPC.
See post #66 here:
http://www.diyaudio.com/forums/tubes-valves/203110-mono-push-pull-driver-pcb.html
In the use case brought up by the_manta, he advocates operating the tube at up to 4 times the rated dissipation on peaks (assuming a sine wave). I have gone far beyond this (assuming music and brief sine wave testing), but there is another aspect that must also be considered. The tube creates a "space charge" cloud of excess electrons around the cathode. This must remain true throughout the entire operating range. In other words if you attempt to suck more out of the plate than the cathode can provide you will upset the current density and risk damage to the cathode and a possible tube arc.
If the tube was constructed perfectly the coating on the cathode would be uniform, as would the spacings of all the internal elements. Thus the current density will be uniformly distributed throughout the tube. Unfortunately this is rarely the case. If you are going to lean on a tube hard, test it first. In a dark room, increase the current until plate glow is seen. Ideally the glow should be even along the plate from top to bottom, and equal between the two sides. If one side is much brighter than the other, or it is hot at the top and not the bottom (or vice versa) then this tube is not well constructed. A tube with visible hot and cold sopts is worse case and should be avoided. The cathode will fail.
Some tubes have just enough cathode capacity to support the rated plate dissipation. Sometimes there is a peak cathode current spec. If there is, it should be respected. Most DHT's are emission limited, and are not good candidates for over dissipation. The original data sheet for the 45 tube reveals that is was intended to provide 18 watts from a pair in AB2. That might be true for a brand new pair in the 1930's, but only a few in my collection have the emission capability to support this, although all of them work fine in class A SE.
I have squeezed almost 90 watts from a pair of 6L6GC's in AB2 but it became obvious that I was getting close to the cathode's emission limits, and the plates dissipation limit. The KT88 has a bigger cathode, and can crank out 100+ watts, but the 7403 has the same plate structure as the KT88 but has a fat cathode with a 10 AMP! peak current rating. I have seen well over 200 watts flow from a pair with continuous sine wave testing. The efficiency was almost 80% so only 25 watts on AVERAGE were being burned up in each plate. The key to this kind of efficiency is a good cathode, and a low idle current. Can't get this out of a 6L6GC.
Look at the peak cathode capabilities of your chosen tube. Some sweep tubes are rated at over one AMP! If there is no rating compare the cathode size to known tubes. A good sweep tube can pull its plate down to 30 volts or so WITHOUT drawing grid current. I have seen the plate swing to ZERO volts using screen drive. Of course having the plate at zero while the screen is at +250 volts can (and did) create fireworks! The little 6BQ6's were cranking out over 125 watts from a pair WITHOUT plate glow before the screen grid failed leading to a fatal tube arc. I got the tubes for 98 cents each, so it was a worthwhile experiment.
I have recently seen 150 watts of continuous sine wave at 3% distortion flow from a pair of $1 13GB5's for about 1/2 an hour. This was in AB1. I am constructing an amplifier along these lines but it will be run at a more sane 100 WPC.
See post #66 here:
http://www.diyaudio.com/forums/tubes-valves/203110-mono-push-pull-driver-pcb.html
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.