6L6GC AB2 Amp

[tubelab.com]

What about pentode with local feedback?

I built a tiny little amp using 6AU6's to drive 6AQ5's about two years ago. It did sound pretty good, but I never had the time to optimize it. It is still around somewhere, and I will get back to it sooner or later.

I have not tried any form of local or global feedback on this design yet. I built a 300B push pull amp with zero feedback about 6 years ago. It had a punchy dynamic sound that I haven't been able to match with any SE amp. The 300Beast as I called it, suffered a blown power supply cap about a year ago, and has sat quietly in the closet ever since. I have built a dozen or so SE amps since I built the 300Beast. I decided that it was time to build some P-P amps, and one of my criteria was to either rebuild the Beast, or make something better. So far I have been experimenting with triode mode zero feedback designs, and a differential driver circuit. This is how the beast was made, so I started there. I started to design the "universal driver board" that began with the Beast's driver schematic. Chrish asked about a driver for P-P 6L6GC's using 6SN7's. I looked down at that box in the closet that had about 25 6SN7's in it and said why not. The circuit used here is similar except for the tube choice.

I got a pair of large by huge by immense 400 watt Plitron toroidal OPT's from their surplus page a while back, so these are a useful possibility. I also have a pair of vintage UTC LS-57's that have been on the shelf for at least 10 years. Since the budget for new stuff is zero this year, I am going to build amps with the stuff that I already have. Well there are the OPT's for two amps, and I have at least 2 more sets of lesser quality P-P OPT's.

I plan to build one "big" amp of 50 to 150 WPC using the Plitron OPT's, one "small" amp using the UTC OPT's and maybe one or two more. I am in the process of mounting two driver boards, and two output tube boards to a piece of wood. This will be used to test different output tubes and OPT's using this driver board. I have all sorts of combinations to try, so I can't say where this will end yet. I want to experiment with a fully differential direct coupled amplifier, at least once. The probability for burnt parts is high there, so I will use cheap stuff for the first one. Remember, we still have some 6FW5's too.

[agent.5]

Originally Posted by tubelab.com

Chrish asked about a driver for P-P 6L6GC's using 6SN7's. I looked down at that box in the closet that had about 25 6SN7's in it and said why not. The circuit used here is similar except for the tube choice.

I checked Ebay and sellers are selling adapters allowing loctal tubes to be used in octal circuits. I also noticed that prices of loctal tubes such as 7n7 and 7af7 are flying compared to a couple years ago.

[JoshK]

Originally Posted by tubelab.com

Remember, we still have some 6FW5's too.

Indeed and I am itching to find a good project for them. I want to build a PP amp. I just haven't decided on Screen-drive or pentode. Probably the prior given your good results with that.

[whitelabrat]

Originally Posted by agent.5

...7n7 and 7af7 are flying compared to a couple years ago.

Pssst. 14N7's

[agent.5]

Originally Posted by whitelabrat

Pssst. 14N7's

True, as most people would not know how to switch filament voltage.

[Rod Coleman]

Chrish asked about a driver for P-P 6L6GC's using 6SN7's

For a starter circuit, look no further than the 1947-8 Brimar Radio Valve Manual!

The picture tells the tale, & may be summed up as:
6SN7 x2 + 5B/250A (807) x2 = 75W in AB2

The 5B/250A is an 807, which most of us would agree is functionally the same as a 6L6G (not even -GC, which is rated for more screen voltage and power handling). If 75W is enough for you (!) we should be able to build it exactly to their spec. (to begin with anyway - I'm not sure about the phase splitter).

The only curiosity is the -78V (14mA) bias. Klug. It it made up of about 25V for biasing the 807s, and ~50V across the 10K resistor loads of the cathode-follower driver, which also give efficient Long Tail Pair operation. ~5mA standing current in the 6SN7s keeps the gm reasonable, if moderate by the standard of later valves, let alone transistors - but you can clearly build a semiconductor-free AB2 if that's your preference.

The corresponding notes in the 807 pages of the book quote:
80W AB2 for 600V anode, 300V screen, idle current 30mA each. 6400 ohm A-A. 78V grid-grid peak input. 3% distortion, PROVIDED anode & screen voltage tolerance 0/-5% & grid bias 0/-3%.

[Rod Coleman]

Instead of "Long Tail Pair operation", I should say "combination bias operation" in the above. The Follower 6SN7 has its grids biased positive (-39V) compared to the -78V common return point.

If you change anything around this driver 6SN7 (valve type, voltages resistors etc), the currents, follower biasing etc will need to be recalculated.

Also of note in the Brimar circuit is that no G1 nor G2 stopper resistors are used, but 47/1W parts are found in the anode leads. No doubt, for AB2 operation this will give the lowest distortion consistent with good stability, since grid or screen current draw will upset the bias conditions more than drops in anode voltage. It also obliges the constructor to take care with the wiring and minimize lead lengths & loop areas - or stability may be a problem. You may still end up with stability issues with othe 6L6 variants - perhaps increasing the 47Ohm value a little would be worth trying in this case.

[Miles Prower]

Originally Posted by Rod Coleman

Also of note in the Brimar circuit is that no G1 nor G2 stopper resistors are used, but 47/1W parts are found in the anode leads. No doubt, for AB2 operation this will give the lowest distortion consistent with good stability, since grid or screen current draw will upset the bias conditions more than drops in anode voltage. It also obliges the constructor to take care with the wiring and minimize lead lengths & loop areas - or stability may be a problem. You may still end up with stability issues with other 6L6 variants - perhaps increasing the 47Ohm value a little would be worth trying in this case.

I would also include active screen regulation. I think you will find screen stoppers will probably be necessary, since 807s like to Barkhausen oscillate on cut off. I would also replace the 47R plate stoppers with a stopper consisting of ten turns of #18 (AWG) wire, 7/16 in. inside diameter, space wound. Parallel with a 100R / 1W C-comp resistor conveniently mounted inside the coil. Mount that plate stopper right at the plate top cap connector.

I would also make the bias voltage at the 6SN7 grids variable so that the finals can more easily be brought into plate current balance. Better still, replace the 6SN7 cathode followers with MOSFET source followers. Source followers have way lower Zo than any cathode follower. That'll work much better.

[tubelab.com]

I would also make the bias voltage at the 6SN7 grids variable so that the finals can more easily be brought into plate current balance. Better still, replace the 6SN7 cathode followers with MOSFET source followers. Source followers have way lower Zo than any cathode follower. That'll work much better.

I would agree with Miles observations. Then I would add to make the phase splitter a true differential pair, then I would look at that 4 to 10 volts drive requirement and say that another stage is needed for the amp to be useful. If I was going to add another stage, it would be another diff pair. My observations are that cascaded diff pairs seem to work better than anything else that I have tried even if it adds more glass to the design. So what I would get is diff pair directly coupled to a second diff pair, followed by a pair of mosfet followers to insure clean AB2 operation of any output tube.

Oh, wait that looks like the board that I am using. The last schematic that I posted was shown in post #277. There have been very few changes since then, because it works. It works really good to. I did see some really impressive power readings with 6L6GC's in pentode mode AB2. The 807 is really a 6L6GB with one significant difference The plate cap allows a much higher plate voltage than the octal bese. The power level seel with 6L6GC's is limited by the 500 volt maximim plate voltage rating. I have seen tubes and sockets flash over from pin 3 to pin 2, so I am not going to bend this rating.

I have decided that for my own personal amp I want to use triode mode without any feedback. I just like the sound. I get 70 watts out of a pair of KT88's in triode at 3% distortion without busting any specs, so that is good enough for me. I have seen over 100 watts in triode mode using 7403's with 75 watts at 1.8% distortion. The sound is really good too.

I plan on building an amp that is fully differential including the output stage. Obviously the power will not be as big since the diff pair output stage forces Class A operation. Excessive grid current will steal cathode current from one of the output tubes, so A2 must be limited to low currents.

[Miles Prower]

Originally Posted by tubelab.com

I would agree with Miles observations. Then I would add to make the phase splitter a true differential pair, then I would look at that 4 to 10 volts drive requirement and say that another stage is needed for the amp to be useful. If I was going to add another stage, it would be another diff pair. My observations are that cascaded diff pairs seem to work better than anything else that I have tried even if it adds more glass to the design. So what I would get is diff pair directly coupled to a second diff pair, followed by a pair of mosfet followers to insure clean AB2 operation of any output tube.

Looks pretty much like we're working along the same lines here:

Schemo (Preliminary) Big Vix