Output section questions

[Shrapnel]

Hey everyone, it's been a while...

Okay. Two design ideas. Both supposed to be switchable between 40 and 20 watts using 6V6 output (Thank you Leo Fender for showing us these 6V6 can be pushed this hard on the HT rail.) Differences lie in the Pi section. Would like to hear the pros and cons of both of these designs, and any tweaks you might suggest. (Yyes. I know the long-tail is the most commonly used.)
Assume a B+ at Node A of 406-420vdc and plate dissipation of 65-70% for each 6V6.
Note: apparently adding the comparative image exceeds max length of post, so the image, in PDF format, is attached.

[pdf64]

There's a 100k grid to cathode resistance limit in fixed bias.
If as here 2 valves share a grid leak, it should be 50k.
So that 470k is kinda pushing it.

[Willabe]

When turning off the 2 6v6's those tubes still have their heaters hooked up, getting hot, heating the cathode. So the tubes are still cooking, not as bad as if current was running through them but still heating up.

I wouldn't do that.
 

[Shrapnel]

There's a 100k grid to cathode resistance limit in fixed bias.
If as here 2 valves share a grid leak, it should be 50k.
So that 470k is kinda pushing it.

On the 6V6 bias? If not there on the bias voltage in, where? as I'm not exactly sure as this is my first quad bottle design.

When turning off the 2 6v6's those tubes still have their heaters hooked up, getting hot, heating the cathode. So the tubes are still cooking, not as bad as if current was running through them but still heating up.

I wouldn't do that.
 

I went that way because acoustic did it years ago with some of their models. Cut heaters? Both heaters and cathodes?

[Willabe]

I went that way because acoustic did it years ago with some of their models. Cut heaters? Both heaters and cathodes?

No, I'd take the tubes out when I want 20w and put them back in when I wanted 40w.   

[pdf64]

There's a 100k grid to cathode resistance limit in fixed bias.
If as here 2 valves share a grid leak, it should be 50k.
So that 470k is kinda pushing it.

On the 6V6 bias? If not there on the bias voltage in, where? as I'm not exactly sure as this is my first quad bottle ...

The 470k grid leak resistors, that provide the negative DC bias reference for the 6V6 control grids.
For a parallel pair in fixed bias, 470k seems to be 10x the type limit.
See 'grid number 1 circuit resistance' at the bottom of p1 https://frank.pocnet.net/sheets/093/6/6V6GT.pdf

[bmccowan]

The curmudgeon in me (that's most of me) thinks that if I wanted that outcome I would build a nice 2x6L6/EL34/KT88 amp and turn it down when wanting 20W type volume. It'll sound better at the 40 watt volume and just as good at the 20 watt volume as the 4x6v6.
Another option is changes at the speaker cab.

[Willabe]

The curmudgeon in me (that's most of me) thinks that if I wanted that outcome I would build a nice 2x6L6/EL34/KT88 amp and turn it down when wanting 20W type volume. It'll sound better at the 40 watt volume and just as good at the 20 watt volume as the 4x6v6.

Yes, but 6V6's do sound different than 6L6's.

[UncleRuckus]

Aside from playing with high voltage (relay?) what would happen if cut in an 8K 20w resistor from olate to transformer center tap? Would this be seen as lower impedance, reducing the tubes output wattage and pushing more 2nd order harmonics? 

[Shrapnel]

There's a 100k grid to cathode resistance limit in fixed bias.
If as here 2 valves share a grid leak, it should be 50k.
So that 470k is kinda pushing it.

Oddly.... the AA763, AB763, ab868, a1172, and a1270 circuit uses 220k for the grid leaks of a single pair of 6V6. (Fender Deluxe Reverb)The 5F10 Harvard, and 6G2 and AA964 Princeton as well.and, yes, I do know that Leo was pushing these 6V6 in many aspects. Maybe I should go 100K based on that?  Too little if I decide I want to make it interchangeable with one duet of 6L6 or EL34 (as long as I allow enough current capacity for the heaters)?

[bmccowan]

Yes, but 6V6's do sound different than 6L6's.

Even this curmudgeon has to agree with that.

[Shrapnel]

Okay... getting some feedback in a FB group. (One even recommended that I read up on Pi designs from one of Merlin's hard to find books on preamps and PIs.) Here's a conglomeration of what I've heard so far here and there.

I'm grateful for the feedback on the power tube configuration btw.

One suggested I give up on the 6V6 idea and go straight duo 6L6, (ignoring the subtle differences in sound).... We discussed this one here a little.
One mentioned I'll only see a 3dB drop with the half-power switch. I'll agree there, but I also think there's a case where 20W may get lost in a mix, yet 40W will sit decently, all because of a measly 3dB difference.
Some look at the cathodyne split-load and mention if it overdrives, it'll sound nasty.
Some think I may have issues with the 5K pot as the cathode resistor on the gain stage that drives the cathodyne.
Some say I don't need the coupling cap between the gain stage and the cathodyne, and it will save me the grid leak on the Pi.
TBH I was concerned about using 47K for grid-leak resistors on the power tubes, but I've seen it used in another schematic on a pair of 6L6. I may go that way after all, although still wondering since Fender seemed to not pay that much attention to that on a few amps of theirs using 6V6.
Someone mentioned about in the past, batches tended to be pretty close... close enough that there was no worry about matching tubes.


Input? Opins?

[bmccowan]

I do not know about the total 3db calc, but if you are cutting through the band mix without micing the amp, having enough midrange is a key. That's part of the reason so many pros consider an EQ pedal an essential boost pedal.
I agree(d) that 6V6s sound different than 6L6s/EL34s/KTs etc. and I love 2x6v6 amps. But above the 20w range, nothing sounds better than a pair of 6L6s and their cousins - IMO. And although Leo proved that 6V6s could take the juice, their big sisters will last longer at 400+ volts.
A friend of mine used to say, "do we need to take a rocket ship, or can we just drive there?" One way to just drive there is to have a speaker cabinet that  facilitates halving the air being moved. 1 or 2 12s; 2 or 4 10s.

[pdf64]

Okay... getting some feedback in a FB group. (One even recommended that I read up on Pi designs from one of Merlin's hard to find books on preamps and PIs.) ...

The Lulu link for Merlin's excellent books can be found at his site, they're printed to order http://www.valvewizard.co.uk/

[Shrapnel]

Well... looking at the LTP power section alone.... this is the modified power section. Yes, the switchable power tubes are still there as switchable.Changes to the split-load have been made too. Here is the two schematics again (in a single file)  to download and view (again, my images (even separated) seem too big for the system.)
If I change my mind, converting to a pair of 6L6GC will be easy... for now, sticking with the 6V6 direction.

The attached schematic for the long tail has the feedback "in an odd place." Not updated here, but I've moved it to the other side of C20 of the long tail Pi on my end.

[d95err]

There's a 100k grid to cathode resistance limit in fixed bias.
If as here 2 valves share a grid leak, it should be 50k.
So that 470k is kinda pushing it.

Oddly.... the AA763, AB763, ab868, a1172, and a1270 circuit uses 220k for the grid leaks of a single pair of 6V6. (Fender Deluxe Reverb)The 5F10 Harvard, and 6G2 and AA964 Princeton as well.and, yes, I do know that Leo was pushing these 6V6 in many aspects. Maybe I should go 100K based on that?  Too little if I decide I want to make it interchangeable with one duet of 6L6 or EL34 (as long as I allow enough current capacity for the heaters)?

My rule of thumb is if it worked for Leo and Jim, it should work for me too, no matter what the data sheet says.