Screen Voltages - What's the implication (Matchless mod)

[turtle441]

After a couple years of on-again off-again work, I finally finished up my version of a Matchless 15-watter.  It's a 2-channel mod with one channel based on the Spitfire preamp, one based on the Lightning.  Miraculously (considering this has been built in spells over a couple of years), it passed the lightbulb limiter testing, and subsequent testing with guitar sounds pretty sweet.  Right now, this is the voltages that I'm getting:
          1          2          3           4          5          6           7          8          9         
V1:   127       XXX      1.39      HTR      HTR      127       XXX      1.39     HTR
V2:   195       XXX      1.64      HTR      HTR      328      196        197     HTR
V3:   250        43        68        HTR      HTR      240       40        67.6     HTR
V4:   XXX       .008      11.5     HTR      HTR      .009     351       XXX     333.2
V5:   XXX       .008      11.5     HTR      HTR      .009     351       XXX     333.2
V6: Not recorded

Using the Weber bias calculator, that comes out to 30mA plate current 12 Watts/tube.  I'm running EL84M's, so I think I'm okay there (although feel free to correct me, I'm new to the push-pull business).

I compared my voltages to the readings from Sluckey's Lightning build on his website.  My 2 big takeaways were:
1. My V2's running at higher voltages than his was.  Comparing the schematics, it looks like he went with a 22k plate resistor, while I went with a 4k7.  There are schematics and layouts out there showing both, I may play around with this eventually, but for the moment, will probably leave it.

2. My screens are running at a higher voltage than his.  Again, comparing the schematics, it looks like he went with a 22k power supply resistor while I went with a 2k2 one.  There are schematics out there showing both. 

So, my questions are:
1. What are the implications of the different screen voltages? 
2. Thoughts on running EL84M's at those voltages?  Matchlesses have a reputation for eating power tubes, I'm assuming this may have something to do with it.

 

[shooter]

I wouldn't be to concerned on the G2 volts, you can up the 100 to 470 and see what you think.  what is your cathode R, appears to be 120ohm 
if so;
11.5/120 = 96mA /2 gets you around 48mA per tube.  plate minus cathode vdc * 48 suggests ~~ 16W a tube

[turtle441]

Sorry, didn't update that on the schematic.  Cathode resistor is 150R at the moment.  Was contemplating bumping it up to 180R as I've got one around, but the amp didn't sound half bad on initial testing.

[92Volts]

Lower screen voltage is easier on tubes. With large signal, screens draw more current, eat more power, and experience more wear. By sagging when screens draw peak power, the resistor has a bigger effect than idle voltages suggest.

It also reduces output power, because steady (high) screen voltage during peak signal improves performance. This is sort of the point of a screen, though you can use it for any balance of power/tone/durability you want.

Because the (4.7k or 22k) resistor is in the cap-filtered power supply, voltage will stay high for a moment then sag. This gives a loud "attack" followed by decreased volume. Some people will like that, and it's similar to the effect of tube rectifiers.

I haven't double-checked all your math, but if current is 30ma per tube, power is only 10.2w. In many PP designs this is above the 70%ish recommended, but lots of people and amp designs abuse the hell out of EL84s.

I'm not familiar with EL84Ms. I'd be willing to run JJ EL84s in a circuit like that, not because I think they'll survive long, but because they're cheap 

[sluckey]

In many PP designs this is above the 70%ish recommended

70% is recommended for fixed bias amps only. Cathode biased amps usually run at 90-100% or even higher.

[92Volts]

70% is recommended for fixed bias amps only. Cathode biased amps usually run at 90-100% or even higher.

100% is common with cathode bias, especially claimed or true "class A" designs.

But I can't help but notice many of those amps (Vox ACs, this Matchless) have a reputation for killing EL84s

EL84 cathode resistors eat less voltage than they would for other tubes. EL84s are so often cathode biased because performance is closer to fixed bias. I wonder if this offers less protection than cathode bias would in other cases.

I've also read EL84s are just cheap and shouldn't be expected to last. They are cheap, not sure how bad their reliability is. So hot bias may not be the only reason these amps eat tubes, but I think it's a factor.

[shooter]

have a reputation for killing EL84s

the other side of the coin, they have a "sound", and that comes from tube abuse.  One of my favorite musician quotes; "there's nothing sweeter than the sound a dying tube makes" 

[92Volts]

Nothing wrong with that, as long as the owner knows what they're getting into!

I do think it's weird that "universal" bias recommendations favor tube life, except for these amps where recommendations favor tone and (seem to) risk tube life. I guess this wasn't a random decision by techs and hobbyists-- Vox ACs and others have always run hot from the factory, and that balance of tube life/tone has "worked" for decades.

[HotBluePlates]

... What are the implications of the different screen voltages? ...

... steady (high) screen voltage during peak signal improves performance. This is sort of the point of a screen, though you can use it for any balance of power/tone/durability you want. ...

"Improves Performance" is not specific enough.  High(er) screen voltage enables a higher peak plate current when you drive the output tube(s).  Compare the Vg=0v curves on pages 8 & 9 of this EL84 data sheet.  50v more on the screen yields ~40-50mA more plate current when G1 is driven momentarily to 0v.

Sadly, perhaps the relatively low maximum rated screen voltage means the data sheet doesn't explicitly show the effect of just changing the screen voltage, as is shown on the top graph on page 4 of this 6V6 data sheet.

... With large signal, screens draw more current, ... By sagging when screens draw peak power, the resistor has a bigger effect than idle voltages suggest.

It also reduces output power, because steady (high) screen voltage during peak signal improves ...

Because the (4.7k or 22k) resistor is in the cap-filtered power supply, voltage will stay high for a moment then sag. This gives a loud "attack" followed by decreased volume. ...

The EL84 data sheets don't seem to have a good representation of what happens to screen current right around a peak input signal driving G1 to 0v; this is right where we might expect screen current to grow more rapidly than otherwise shown.

You should think about the resistance from the screens back to the filter cap that feeds them, as well as the resistor between the screen & plate filter caps.  Higher resistance at one or both of those will tend to drop screen voltage during high screen current draw, and the falling screen voltage will compress plate current (and thereby output power).  The screen filter cap mitigates this somewhat, as there has to be a large enough spike in screen current to drain charge from that cap, and then cause more voltage drop across the 4.7kΩ or 22kΩ decoupling resistor.

Overall, we're talking behavior that will mostly manifest when the amp is really cranked.  You're basically deciding how much compression the output stage has, and how loud the amp needs to be before hearing it.  It's up to the designer's (your) taste whether more/less compressed & at-what-volume is a good thing.

... But I can't help but notice many of those amps (Vox ACs, this Matchless) have a reputation for killing EL84s

EL84 cathode resistors eat less voltage than they would for other tubes. ... I wonder if this offers less protection than cathode bias would in other cases. ... So hot bias may not be the only reason these amps eat tubes, but I think it's a factor.

I think the high transconductance of the EL84 may offset the small cathode resistor value as a source of trouble.  Meaning the tube should receive sufficient corrective voltage-change even with the small cathode resistor.

But I do wonder about the 300v max rating for plate & screen, and how rarely I see amps respecting that limit.  Maybe that rating isn't as conservative as with other tubes...

Or maybe it's just the physical structure of the EL84 and small bulb size relative to the 6V6?

[66Strat]

The attached Mullard EL84 datasheet has good representations of screen grid current vs plate voltage. See pages 7 and 9 respectively for 250 volt and 300 volt screen grid supply voltages. I would maybe go as high as 1 K ohm screen grid resistors for EL84.

[HotBluePlates]

The attached Mullard EL84 datasheet has good representations of screen grid current vs plate voltage. ...

Nice!  I probably picked the wrong sheet to look at!

The trick to using those curves is to also look back at the regular plate curves (with your loadline drawn) to figure out how low the plate voltage will really be pulled in your particular amp's circuit.  There's a big difference in possible screen current draw if your amp can never pull the plate below 50v, or if it can...  And that difference in screen current will materially alter your choice of "best" screen/decoupling resistance.