6V6 Plexi needs more watts

[JoeK]

I've recently built the MHuss 6V6 Marshall and it seems fine but I think the power tubes and transformers aren't pulling their weight. When I hook up the scope/signal-generator/DVM I get 8.5watts clean and 15watts flat-out into 8ohms. I would guesstimate it should be doing about 20watts plus when cranked.
As comparison I just measured my 18watt and it's doing 12 watts clean and 22 watts cranked
Any ideas to test? I've disconnected the negative feedback and also the 220 ohm safety resistor across the speaker jack to see if that was an issue but still the same.

Here's the schematic:

Here's my build:
At idle the JJ6V6S tubes are doing 20ma each with -40VDC bias, plates are 470 screens are 430VDC.

At clean max sine output: grid 25VAC plates 440VDC screens 358VDC 8ohm dummy load 8VAC.

At max volume : grid 35VAC plates 430VDC screens 318VDC 8ohm dummy load 11.1VAC.

I'm using Hammond iron PT 290BEX 660VCT 138ma; OT 1760H 20W; Choke 157J 10H 65ma.

[J M Fahey]

Almost no tube amp reaches "spec" clean power output when properly measured ... as in no Twin or Plexi reaches *measured* clean 100W and so on.
It's one of Life best hidden secrets .
Those 12W you measure on the "18" are typical.
Now this one really seems low, even with reduced expectations.
Just as a test, re measure it with twice and half load impedance (meaning you put that 8 ohms resistive load in all 3 secondary taps: 4 , 8 , and 16 ohms) to see what we have.
Later we'll re re measure with better precision but now I'm trying to find in a broad sense whether we have a drive problem or a lack of current problem.
At least, that's what I'd do if I had it on my bench.

[JoeK]

I just tried that test, thanks Juan, with 4ohm into 8ohm dummy I got 14 watts clean and 26 watts max. With 16ohm tap into 8ohm dummy I got 4.5 watts clean and 9.5 watts max.
Sitting here wondering about what voltage the screens are running at (40 volts less than plate at idle, 80 volts less than plate at clean, 110 volts less than plate at max) I believe dropping screen voltage on pentode screens is a form of volume control AFAIK.

[JoeK]

I just double checked the B+ path to the screens and it seems fine. The first 32uf filter cap (OT supply) has a 5k resistor before the choke, the choke return goes to the 2nd 32uf cap and that runs to the two 1k screen resistors. So from the first filter cap (all B+ drained) I measure 5k to choke, 5.2k after choke, 6.2k on screen pins.

[J M Fahey]

I just tried that test, thanks Juan, with 4ohm into 8ohm dummy I got 14 watts clean and 26 watts max. With 16ohm tap into 8ohm dummy I got 4.5 watts clean and 9.5 watts max.

Well, you already found it
REAL 14W clean is a lot for a "18W" class tube amp.

I have *always* measured such kind of performance.
Fact is, I started worrying after measuring , before that I was happy about the "loud" amp, go figure.
Oh well.

I think you may pull a little more, perhaps a couple watts, by fine tuning the load, but since you must use an over the counter transformer, you have the taps you have.

You already recovered a lot of power
What you found is what I did long ago:

1) tubes are not transistors

2) typical modern "wisdom" : "lower load impedance to increase power" does NOT apply ... unless you add more tubes in parallel, that is.

3) tubes are current limited ... period. SO:

> in SS amps the useful formula is : P=V^2/R which invites us to lower R for more power
Why? ... because transistors are basically good *voltage* sources.

> in tubes the useful formula is: P=I^2 * R which invites us to raise R for more power
Why? ... because tuber are basically good *current* sources.

You already confirmed that !!!!
You doubled load impedance (8 ohms into 4 ohms tap) and you raised power from 8.5 to 14W !!!!

Will you further raise power if you quadruple load?

Well, no, at some point you have no current problems but you´ll run out of voltage swing ... so the best experiment is to test many impedances until you optimize power, and in the future you order custom transformers.

Not practical for single users.

FWIW I wind my own, and my bread and butter tube amp is "50W" , actually 40W RMS which is LOUD , with any brand of 6L6 or EL34, no need to go for NOS or expensive versions.

And I wind mine for 5500 ohms plate to plate, with 420/430 V +V.

Tubes last long too .

Sitting here wondering about what voltage the screens are running at (40 volts less than plate at idle, 80 volts less than plate at clean, 110 volts less than plate at max) I believe dropping screen voltage on pentode screens is a form of volume control AFAIK.

Yes, it is a kind of volume control.
To begin with power tube "voltage gain" varies with screen voltage , although that is only noticeable in no NFB amps (think VOX, Matchless, a couple more) and in all types it works as current limiting.

Problem is that if you lower screen resistors too much, ... you can kill the tubes.

That´s why I suggest being very conservative with modern tubes, do not try to push an extra couple watts (which do not mean much to the ear anyway) for the price of halved tube life.

Our friend km6z has already commented his wonder (he lives in Russia now) when he saw tube amps used by popular musicians (those who have been playing family parties 3-4 times a week for ages) with perfectly working original tubes ... 20-30 years old !!!!!

Standard Russian tubes of course !!!!

Then he measured no more that 360V on plates, correspondingly less on screens, etc.

No mystery, just "follow the book" designs.

[R.G.]

Well, you already found it
do not try to push an extra couple watts (which do not mean much to the ear anyway) for the price of halved tube life.

Let me reemphasize JM's comment here. The nature of human hearing is that you literally can't hear the addition of a few watts in terms of loudness.

In general, it's not worth getting more watts unless you can double the power. It takes nearly 10x the power to be subjectively "twice as loud".

So unless your amp is only putting out half the power your expected, don't sweat the slightly lower power. You won't hear it. It makes a much bigger difference in loudness to change to a more or less efficient speaker.

[g1]

It would be very interesting to see if you got the same results with some "real" NOS 6V6's. The JJ's seem to be a bit of a hybrid between 6V6 and 6L6, IIRC someone else here recently had some strange results with the JJ 6V6's.
The odd thing is that they are liking a higher OT primary impedance. If they were more like a 6L6 I would expect they would like a lower impedance to run into.
(that hammond OT primary is spec'd at 6600 ohm)

[loudthud]

I think the soft screen voltage is what is holding this amp back. Short out that 5K 5W resistor in series with the choke and reset the bias. The 1K 5W resistors on each screen grid should also be reduced the 470 ohm 1W.

[potatofarmer]

It would be very interesting to see if you got the same results with some "real" NOS 6V6's. The JJ's seem to be a bit of a hybrid between 6V6 and 6L6, IIRC someone else here recently had some strange results with the JJ 6V6's.
The odd thing is that they are liking a higher OT primary impedance. If they were more like a 6L6 I would expect they would like a lower impedance to run into.
(that hammond OT primary is spec'd at 6600 ohm)

I think NOS 6V6GTs would want a higher primary impedance too. It would be nice if someone with a sim or tracer could plot anode current at screen voltages in the 350-400V neighborhood. My weak graphical extrapolation suggested somewhere in the 8k Raa range would be optimal.

It would be really nice to be able to drop that 5k resistor feeding the screens... maybe switch to 7591s...

I think the soft screen voltage is what is holding this amp back. Short out that 5K 5W resistor in series with the choke and reset the bias. The 1K 5W resistors on each screen grid should also be reduced the 470 ohm 1W.

Posted while I was writing... Dude's got 470V sittin' on his plates! At least keep the 1k screen stoppers, maybe lower the 5k to 1k but don't you think that'd be pushing it?

[JoeK]

Thanks for the posts: The B+ is quite high, I have 245VAC at the wall and that's fairly average in my country. I have some 7C5 tubes so even though I didn't plan on leaving them in this amp for a long period I'll try them out first, then I'll try to drop the 5k resistor for 1K to see what happens. Report back in about 12 hours.

[JoeK]

Ok did some testing. Put some Emitron 7C5 tubes in (with adapters) and they didn't like it at all, distorting at low volume, I have about 40 7C5 tubes Sylvania and Emitron but I think I'll hang on to them for lower voltage builds.
Next I took out the 5k resistor before the choke/screens and subbed in a 1k and like Ray Stevens said "everything is beautiful"
I now get at idle 20ma/21ma Plate 460VDC Screen 454VDC
Clean sine output 64ma/64ma Plate 428VDC Screen 400VDC Grid 31/31VAC 8ohm dummy load 11.5VAC 16 watts
Max output 90ma/85ma Plate 410VDC Screen 370VDC Grid 46/43VAC 8ohm dummy load 16.5VAC 34 watts

So it looks like the screen voltage was the cause. Have to say I can't see the "typical" Marshall hump on the scope when going past a clean sine wave but I guess this isn't a typical Marshall build. Now to check the rest of the amps voltages.

[Alan0354]

I just tried that test, thanks Juan, with 4ohm into 8ohm dummy I got 14 watts clean and 26 watts max. With 16ohm tap into 8ohm dummy I got 4.5 watts clean and 9.5 watts max.
Sitting here wondering about what voltage the screens are running at (40 volts less than plate at idle, 80 volts less than plate at clean, 110 volts less than plate at max) I believe dropping screen voltage on pentode screens is a form of volume control AFAIK.

Sounds like problem I experienced before. Looks like you found the problem. You have to make sure you match the impedance of the primary to the load line of the 6V6. For 6V6, usually you need 8K primary for typical 400V +B. too low or too high will lower the output power.

Do you have a OT with 8K primary? If you use an OT for 6L6, they are usually 4K primary. then you have to connect you 8 ohm speaker onto the 4 ohm tap to get 8K primary.

I went through the whole gantlet just to find out I mixed up the 4 ohm and 8 ohm line. I connect to the wrong tap and got much lower than 20W on a pair of 6V6.

[JoeK]

Well it seems like I need to do a bit of research on the impedance and loadlines for 460 volts with 6V6. Might try and work out how to use LTspice. Found a couple of files on the net that might help me. The screen voltage change has doubled the wattage for the amp output though. Thanks for the replies.

[JoeK]

Did a bit more tweaking with this amp and put two zener diodes in series with the centre tap 12V/5W and got the B+ down to 432VDC-plates 425VDC-screens with 11watts clean and 27watts max. They don't seem to have introduced any noise so I don't know if an electrolytic cap is required in parallel with the zeners. Then I upped the dropping resistor to the choke/screens from 1k to 2.2k and the plates/screens are now 433/418. So biased with 20ma each side I get 10watts clean and 20watts max and the tubes should last a bit longer. The higher voltages had one of the JJ6V6 tubes popping when warming up. I added a brighter cap across the "bright" volume control and settled on 50pf as a good tone value. Also put a 1meg resistor across the "fat" switch to get rid of the pops when switching it.
Can't say I've found any graphs of impedance loadlines for 6V6 on the web yet.