experimental 6L6G / 6BG6G amp design brainstorm

[chromal]

Originally posted by SY
I'm experiencing cognitive dissonance. If you want low THD, you want to go push-pull not SE. You should be able to get a solid 15 watts of class A triode from those tubes and that output transformer, though you'll need to get a better power transformer.

Well, that was always my understanding, as well. And I'm a little confused by the application data, which is where I'm getting the THD spec from. Though they say "class A P/P," which is.. again. Class A is single-ended, class-AB is push-pull. What is Class A P/P?

6BG6G TDS Tube data

[SY]

Aha, there's the confusion. Class A can be done in push-pull. All it means is that the idle current is the same as the peak current, i.e., neither tube cuts off.

[chromal]

Originally posted by SY
Aha, there's the confusion. Class A can be done in push-pull. All it means is that the idle current is the same as the peak current, i.e., neither tube cuts off.

Aaah, so this is Class-A with two tubes operating nowhere near cutoff during the entire waveform. Hrm. I wish this application data had a schematic; I guess the two plates are tied together and that hits the SET OT? If this arragement has a Zout of 8K, then a Class-A non-P/P single-power-tube arrangement would have a Zout of 16K? I don't think I've ever seen any new stock SET OTs with primary impedance so high, but I think I may be failing to grasp something...

I blame the datasheets I've found for the 6BG6G/6BG6GAs. Now, folks have compared the 6BG6G with the 807, and the 6BG6GAs with the 6L6G*s, and I do see differences in the application data.

FWIW, the 807's app data seems to match the app data for the 6BG6G, with the exception that they provide true class A app data for it:

Class: A S/E
Va: 500
Vg2: 200
Vg1: -14.5
Ia: 50
Ig2: 1.6
Ra: 39000
S: 5.7
Rk: 280
Zout: 6000
Pout: 11.5
THD: 12

Holy crap. So going from Class A P/P (triode) to Class A S/E brings the THD from 0.6 to 12. *scratches head* what's so magical about Class A P/P with respect to THD, I wonder.

[Eli Duttman]

Chromal,

PP O/P stages cancel internally generated even order harmonic distortion. There's the reduction you wondered about.

Those O/P trafos have 20% primary taps. That number argues in favor of the E-Linear topology Doug Piccard (Bandersnatch) vigorously advocates. You run in full pentode mode, with regulated g2 B+. A LTP splitter/driver made from 2X 6CA7 signal pentodes will provide ALL voltage gain needed. The LTP anode load resistors connect to the O/P trafo taps, not B+. Doug doesn't like loop NFB around the O/P trafo, but I think a few dB. are an excellent idea, given the shaky provenance of your "iron".

Don't forget to roll infrasonic noise off at the amp's I/Ps.

[SY]

I guess the two plates are tied together and that hits the SET OT?

No, each plate is individually connected to separate ends of the output transformer primary, with the center-tap being driven by the power supply (i.e., at AC ground). Topology looks identical to an AB push-pull; the loading and idle currents are different, with the plate to plate load generally being a bit higher.

[Miles Prower]

The 6BG6 is the octal version of the 807, for use as a TV horizontal deflection final. A couple of things to watch for when using these. They like to make higher order harmonics, and can definitely sound pretty nasty when run open loop. You will need local NFB to tame those beasts, and O. Schade recommended feeding back 10% of the AC at the plate back to the control grid. I tried that, and the recommendation looks to be spot on. Another 6.0db(v) of gNFB cleaned up the remaining messes nicely, and took the edge off a too-agressive sound. Used right, these can sound very good indeed.

The other thing: you will definitely be needing screen stoppers. 807s like to Barkhausen oscillate when they go into cutoff. 1K5 stoppers mounted close to the screen pins stopped that.

If you decide to run them in pentode mode, an active screen regulator is also a good idea, as pents run more linearly with the screen voltage nailed down and returned through a low AC impedance. Active regulators do both better than a series dropping resistor or a voltage divider.

[sorenj07]

Originally posted by Miles Prower
O. Schade recommended feeding back 10% of the AC at the plate back to the control grid. I tried that, and the recommendation looks to be spot on.

Interesting. I'm also sitting on an old quad of 6BG6G's I was given, and might want to throw an amp together at some point. Can you show what you mean in a schematic, though?

[Miles Prower]

"Can you show what you mean in a schematic, though?"

Vixen Main Schemo

The local feedback is from the plates of the 807s to the grid of the 6SN7 cathode followers. This is a simple voltage divider that is scaled for 10% feedback, the impedance being formed from R33-34, and the parallel impedance of R11-12, R15-16, and the r(p) of the 6SL7, this being 74.5K from the loadline of that stage. C9-10 are simply DC blocking capacitors.

[chromal]

Originally posted by Eli Duttman
Chromal,PP O/P stages cancel internally generated even order harmonic distortion. There's the reduction you wondered about.

The 'Class A P/P' really threw me for a loop. So, I can operate at class AB2 P/P with 1.8% THD at 26.5w Pout, Class A P/P with 4.0% THD at 24w Pout, or Class A P/P triode-strapped for 0.6% THD at 6w Pout. All this according to this application data I found on the net... I'm surprised the class A P/P has more THD than the class AB2. I suppose it could be because of the selected positions along the tube curves...

Those O/P trafos have 20% primary taps. That number argues in favor of the E-Linear topology Doug Piccard (Bandersnatch) vigorously advocates. You run in full pentode mode, with regulated g2 B+. A LTP splitter/driver made from 2X 6CA7 signal pentodes will provide ALL voltage gain needed. The LTP anode load resistors connect to the O/P trafo taps, not B+. Doug doesn't like loop NFB around the O/P trafo, but I think a few dB. are an excellent idea, given the shaky provenance of your "iron".

Looks like I need to read and think some more about the long-tail pair, and how NFB would interact with the CCS requirements.

Don't forget to roll infrasonic noise off at the amp's I/Ps. [/B]

Hmm, 7hz highpass to prevent DC-like OT loading? Sounds like something I read in M.Jones and since forgot.

[SY]

I'm surprised the class A P/P has more THD than the class AB2.

At full power out. I would expect that if you measured both at, say, 5-10W, the tables might be turned.

That said, I run my pentode amps in AB1, my triode amps in AB2.