Ultra-linear 6L6 output stage

[mnturner]

Hey guys, as my first majorish valve project, I'm planning an stereo class AB2 amp using 6P3S's in the output stage. Most of you guys probably know this but they're pretty much the same as the 6L6GB; a lower rated 6L6GC. So far I've been playing around with Spice trying to get the open loop distortion down as much as possible without using the ultralinear transformer in the simulations.

When I decided to use the ultralinear OPT, I came across a problem. In the application notes for the 6L6GB and 6P3S (which is in russian, but the values all seem to be the same), the B+ is 390V, and the screen voltage is 270V.
How can I set the ultralinear trafo up so the DC on the screens is 270V rather than the full B+ voltage?
Any help or advice would be appreciated.

[TheGimp]

Transformer with tertiary UL windings. Custom from Edcor, and possibly others.

I thought about a separate 270V supply resistivly driving the screen with capacitive coupling off the plate, but the phase shift would probably screw up the UL behavior.

Possibly a transistor driven off the plate using a resistor divider to maintain phase, and drive the transistor off the screen 270V supply.

[mnturner]

Yeah those were pretty much the solutions I came up with when I thought about how to do this; didn't think of the transistor solution though.
I'm winding the xformers myself so tertiary UL windings wouldn't be a problem. I'm assuming I'd just set the tertiary winding to have the signal AC voltage swing appropriate for whatever tap % I choose and give it a centre tap attached to a 270V supply?
Know of any popular or commercial amp designs running ultralinear with different plate and screen voltages? I wouldn't mind seeing how others have done this, alothough the tertiary winding sounds like the way to go.
Thanks for the response TheGimp.

[kstagger]

100V/5W zener in series with the screen? Results may vary

[TheGimp]

The Zener would scale the dc bias point but would give 100% feedback rather than 40% (or whatever UL percentage you wanted).

I think the transistor will have the problem with it's output being clamped by the cap that helps set the %ul.

I may try this just for S&G.

[20to20]

the B+ is 390V, and the screen voltage is 270V.
How can I set the ultralinear trafo up so the DC on the screens is 270V rather than the full B+ voltage?

I came across an RCA data sheet for the 6L6GC that has a note for using a screen tap which permits the max V. to be 500v., even though normal ops. was 350/270v. Perhaps under UL conditions there is a higher max for your tube also. At least 390v. when used with a screen tap. You might look a little deeper into the possible exceptions before engineering an unecessary workaround.

[keantoken]

Look at the CCS. The first tube will have to drive M1's gate capacitance through the 2N2222's emitter, loading it at high frequencies. Also the Zener will have a high parasitic capacitance, up to 100nF and very nonlinear and will slow down the CCS a great deal. I suggest to use a PNP and P-MOS instead. I can illustrate if you want.

For other CCS options, you may look at this thread:

SS CCS and distortion

- keantoken

[torrence]

I have also been mulling over the issue of pentode screen voltage. There is some conventional wisdom out there (I can't find the specific thread at the moment) that the screen voltage limit is strictly relevant ony for fixed screen voltage operation. The idea is to avoid the plate voltage dipping below the screen for too much of the AC cycle, and hence making the screen absorb some or all the power usually deposited on the plate.

The claim is in UL the screen voltage can not end up too far above the plate, and hence the voltage rating isn't nearly as relevant. It is really the power rating of the screen which counts. In a typical UL connection, however, the screen voltage is in fact slightly higher than the plate, so there might still be some cause for concern. Dropping resistors may help here.

I have used Zener diodes to drop 30 V from the UL tap to the screens on a different pentode, but still applying a voltage on the screens well above their rated values. So far I have had no problems, but as always your mileage might vary. I would actually love to hear a reply from somebody with more experience and/or knowledge on this topic.

[torrence]

Originally Posted by TheGimp

The Zener would scale the dc bias point but would give 100% feedback rather than 40% (or whatever UL percentage you wanted).

I don't follow this.

The zener is connected between the UL tap on the transformer and the screen, and just means that V_screen = V_UL - V_zener. The AC signal at the screen will be equal to the AC signal at the UL tap, and will still be 40% of the AC signal on the plate.

Since the plate and screen are no longer at the same voltage, the actual feedback characteristics will be slightly different, but I don't see how this would give 100% feedback.

[kevinkr]

Originally Posted by torrence

I don't follow this.

The zener is connected between the UL tap on the transformer and the screen, and just means that V_screen = V_UL - V_zener. The AC signal at the screen will be equal to the AC signal at the UL tap, and will still be 40% of the AC signal on the plate.

Since the plate and screen are no longer at the same voltage, the actual feedback characteristics will be slightly different, but I don't see how this would give 100% feedback.

I don't think he understood you were advocating connecting the zener between screen grid and UL tap.. This will work, but as pointed out in a previous post is probably unnecessary in most UL applications..

I have run 6L6G/6L6GB in MC-30s without blowing them up (look at the circuit of mac 30 and you will see exactly what I am getting at) so I think the 6L6GB will fine run straight off of the UL screen tap at or slightly above the plate voltage. Note that I also used 6P3S in these amps and they worked and sounded fine, but the bluish pulsing glow was a bit alarming..