hey guys,
woke up this morning with this itch to try a class A2 output stage.
after doing some reading, i realised that there are only a couple of tubes that are recommended for class A2, namely the 845, sv572 and the 811.
the 811 and 572 are pretty expensive and the 845 needs pretty lethal voltages.
i was wondering if i could get a taste of class A2 with more common tubes like the chinese 807 which is in abundance in this part of asia.
what do you look for in a tube for it to operate well in class A2?
thanks man!
woke up this morning with this itch to try a class A2 output stage.
after doing some reading, i realised that there are only a couple of tubes that are recommended for class A2, namely the 845, sv572 and the 811.
the 811 and 572 are pretty expensive and the 845 needs pretty lethal voltages.
i was wondering if i could get a taste of class A2 with more common tubes like the chinese 807 which is in abundance in this part of asia.
what do you look for in a tube for it to operate well in class A2?
thanks man!
Class A2 triodes
I've been thinking about class A2, too. The 811A is my first choice, and the Russian and Chinese 811As are not too expensive and seem to work well.
There is a connection of the 807 I have seen that makes them a very high mu triode - connect grid #1 and #2 together and use this as the control grid. The mu is several hundred. This should work well in class A2.
I've played around with 12AX7s in class A2 - they actually work OK, although it is easy to get the plates to run red-hot. If you want to fool your solid-state engineer friends, put a 12AX7 onto a Tek 575 transistor curve tracer with the plate going to the collector, the grid going to the base and the cathode to the emitter. You need to supply heater power. When you trace the 12AX7 as if it was a transistor, it looks like a lousy power transistor with a low hfe! This is with the grid being driven by current.
- John Atwood
I've been thinking about class A2, too. The 811A is my first choice, and the Russian and Chinese 811As are not too expensive and seem to work well.
There is a connection of the 807 I have seen that makes them a very high mu triode - connect grid #1 and #2 together and use this as the control grid. The mu is several hundred. This should work well in class A2.
I've played around with 12AX7s in class A2 - they actually work OK, although it is easy to get the plates to run red-hot. If you want to fool your solid-state engineer friends, put a 12AX7 onto a Tek 575 transistor curve tracer with the plate going to the collector, the grid going to the base and the cathode to the emitter. You need to supply heater power. When you trace the 12AX7 as if it was a transistor, it looks like a lousy power transistor with a low hfe! This is with the grid being driven by current.
- John Atwood
My choice for this is the 812, rather than the 811. The 812 has somewhat better linearity and a lower r(p).
Pretty good, I'd say.
An externally hosted image should be here but it was not working when we last tested it.
Pretty good, I'd say.
A tough control grid, more likely. It's got to dissipate power equal to grid voltage times grid current, when the drive takes you out of class A1.
So what tube? I don't know, but 6L6GC data sheets always show specs for a pair in class AB2. So they ought to be fine in class A2 SE with the right parameters.
Available from China too!
--Paul
yeah.. been searching for chinese 811 but i havent found any online. the sv572 looks like its going to cost the moon. i'd prefer to stay on the cheaper side of things.
i've been thinking if i can drive these buggas with something like an akido with a tube choice like 5687-5687, might be able to get low enough output impedance to do some work. i dont know if i can get enough voltage swing thou..
hows a low voltage 845 in A2? defeats the purpose?
i've seen the 6l6 in class a2 spec but it has never been recommended. i was thinking of the russian ones instead because they come in abundance. 807 might look alittle more snazzy but i dont know how they would measure out..
i've been thinking if i can drive these buggas with something like an akido with a tube choice like 5687-5687, might be able to get low enough output impedance to do some work. i dont know if i can get enough voltage swing thou..
hows a low voltage 845 in A2? defeats the purpose?
i've seen the 6l6 in class a2 spec but it has never been recommended. i was thinking of the russian ones instead because they come in abundance. 807 might look alittle more snazzy but i dont know how they would measure out..
Guess there are no taboos - and in source follower configuration, Cgs is bootstrapped to a small effective output capacitance. The driver "sees" this + Cgd, and although I have not tried it on a bench yet, a tube should well be able to drive a power mosfet, at least for audio frequencies. I tried several ideas using PSPICE, it is definetly worth trying it on a bench also. You only would need a +/- power supply for the driver.
Marcus
Mosfet drive 811A
I have driven a P-P 811A with mosfets, and works well...
For the 811A to draw any current at 420V plate, you'll need around 20 odd volts Positive on the grid for a bias of around 60mA, Easy to get with a mosfet and dead stable....
I used a 5K p-p transformer, and was surprised at the power, It was able to produce around 40W into 8 ohm...
Only real issue I found was hum from the AC heating of the 811A, but guess that DC heating could cure that, or even a little cunning 'hum-balancing'
Its still in bread-board at the mo, as Im re-building my OTL currently...
Ive been contemplating a low voltage 845 type A2 amp, but using the GM-70-The GM-70 shows good possibilities in A2 Worth more investigation
I have driven a P-P 811A with mosfets, and works well...
For the 811A to draw any current at 420V plate, you'll need around 20 odd volts Positive on the grid for a bias of around 60mA, Easy to get with a mosfet and dead stable....
I used a 5K p-p transformer, and was surprised at the power, It was able to produce around 40W into 8 ohm...
Only real issue I found was hum from the AC heating of the 811A, but guess that DC heating could cure that, or even a little cunning 'hum-balancing'
Its still in bread-board at the mo, as Im re-building my OTL currently...
Ive been contemplating a low voltage 845 type A2 amp, but using the GM-70-The GM-70 shows good possibilities in A2 Worth more investigation
Ganti's class A2 amp
You might want to check out Ganti's class A2 amp, described at: http://www.one-electron.com/Randall/ganti_2.html
He based this amp on Noboru Sheishido's designs in Japan. It is really excellent sounding. I wouldn't go with MOSFETs, but rather a small tube power stage driving the output tube grids through a step-down transformer.
By the way, the reason I like the 811 for class A2 is its high mu. This means that you can get the maximum grid swing without letting the grid go negative. What you really want to avoid is crossing from positive to negative grid voltage, since there is a kink in the grid current loading that is hard to get around. You want to stay completely negative (class A1) or completely positive (class A2). With lower mu tubes like the 812, you have a harder time avoiding this kink.
By the way, the 811 came out in 1939 with a maximum plate dissipation rating of 40 watts. In 1949 the 811A came out with a rating of 45 watts. You can tell the 811A (at least the American ones) by the little fins on both sides of the plate structure.
- John
You might want to check out Ganti's class A2 amp, described at: http://www.one-electron.com/Randall/ganti_2.html
He based this amp on Noboru Sheishido's designs in Japan. It is really excellent sounding. I wouldn't go with MOSFETs, but rather a small tube power stage driving the output tube grids through a step-down transformer.
By the way, the reason I like the 811 for class A2 is its high mu. This means that you can get the maximum grid swing without letting the grid go negative. What you really want to avoid is crossing from positive to negative grid voltage, since there is a kink in the grid current loading that is hard to get around. You want to stay completely negative (class A1) or completely positive (class A2). With lower mu tubes like the 812, you have a harder time avoiding this kink.
By the way, the 811 came out in 1939 with a maximum plate dissipation rating of 40 watts. In 1949 the 811A came out with a rating of 45 watts. You can tell the 811A (at least the American ones) by the little fins on both sides of the plate structure.
- John
True the grid impedance will instantly transition from megohms to ohms when the grid voltage goes positive. The only way to avoid distortion at the transition is to have a driver with a very low output impedance (a few ohms max). For some large tubes (845, 833A) I found that a mosfet works better than anything else.
The "taboo" is fading, but new ideas are adopted slowly. The capacitance VS voltage is true. The capacitance on just about any semiconductor device varies with the applied voltage. In a mosfet follower the important spec is the "reverse transfer capacitance". This is the capacitance from gate to drain, and is the effective input capacitance of the follower, which is essentially tied from the plate to ground of the driver tube. Choose a mosfet where this capacitance varies minimally over the chosen voltage range. This can usually be reduced by operating the mosfet with a lot of voltage.
See this design for ideas:
http://www.tubelab.com/845SE.htm
I operate the mosfet at 300 volts to reduce the capacitance and its change with voltage. This amp produces 40 watts from an 845 in A2 without exceeding the tubes specs.
tubelab.com said:
As the guy who probably started kicking this variable-capacitance turd around here, let me say that I'm not aware of any "taboos", just ideas and opinions exchanged. I heartily agree with George that if you are going to use a MOSFET, please do keep the drain voltage as high as possible to reduce Cdg and to flatten its variability. See Crss (=Cdg) in the image below as an example.
I happen to believe that the ear is extremely sensitive to this effect and that it goes a long way to explaining why we perceive tubes and solid state as sounding different. As always, let your own ears decide.
But I do agree with John Atwood; if you’re going to build a tube amp, then use tubes. I’d keep solid-state devices on the periphery for use in power supplies or other housekeeping functions (even in certain CCS duties with caution), where their foibles can be held at bay. My opinion doesn’t diminish George’s or anyone’s opinions – that’s what makes this fun, right?
Attachments
point taken, in that case, i am still thinking of how can i supply the 20 odd milliampere required for the grid.
i am looking for a pair of output transformers, the one electron ubt2 looks like the recommended source. but budget is tight and shipping is killer.
are there any cheaper recommendations? is 5k the bare minimum? i have been looking for 5k opts will little luck, most are 3.5k and below. would 3.5k do?
i am looking for a pair of output transformers, the one electron ubt2 looks like the recommended source. but budget is tight and shipping is killer.
are there any cheaper recommendations? is 5k the bare minimum? i have been looking for 5k opts will little luck, most are 3.5k and below. would 3.5k do?
Hey, I wouldn't call it a turd, it is a very real and measurable phenomenon. It's audible effects are as yet un-quantified, and I am looking forward to your further research into the issue.
I did some listening tests with a few volunteers on the mosfet driver and CCS load (PowerDrive) circuits. I switched the drain voltage between a +50, +150 and +400 volt supply in my 845SE amp. No one could reliably tell the difference between the +150 and +400 volt supply. Everyone could easilly pick out the +50 volt supply. I believe that the mosfet was running out of headroom since the grid voltage goes to +40 volts, but VVC effects could have been a strong contributor.
I also did some tests by inserting resistors in series with the CCS load used on the driver tube in an amplifier that did not use a mosfet buffer (effectively in series with the plate to ground capacitance). Only one listener could tell the difference when I switched the resistor in or out, but he could not reliably pick out the position with the best sound. The results depended a lot on the music being played.
Fortunately this forum (unlike that other one) seems to welcome every opinion without the criticism and "flame wars" so popular elsewhere. I will likely unveil some more sand based tube circuits in the near future. The reader is free to use what they want and discard the rest. The large quantities of often conflicting opinions does often confuse the newbies (I get lots of email), but this is still far better than being told that you are an idiot if you don't listen to (insert the name of your favorite "expert" here).
When I published the PowerDrive circuit, I stated that it was just one possible solution to the grid current - overload recovery issue common to many capacitor coupled circuits. It was based on lowering the cost of high end since GOOD interstage transformers are expensive and has its place in the database of possible circuit choices. There are other alternatives.
There is a circuit floating around that uses a 6V6 cathode follower directly coupled to the grid of the 811A with no other path for current flow. I can't find it now. The problem here is there is no way for the 811A to approach cutoff without the 6V6 being starved into a nonlinear region. I found a variation on this circuit that adds a choke, which will solve this issue, but now the DC resistance of the choke sets the operating point for the entire amp. I am sure that this design could be modified to work, and would satisfy the requirement for a pure tube design. I might even build one if I ever get the time since it is so simple. I would use a small sweep tube for the CF since the peak current capabilities are higher. My favorite for this design would be the 6AV5 but they are hard to find outside the USA.
http://world.std.com/~doyle/811Afunproject.pdf
I haven't tried them, but James transformers are highly recommended and are supposed to be available in Singapore. Search on this forum. 5K would yield lower distortion at a slightly lower power than 3.5 K transformers.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.