El84 Max screen grid dissipation

[Gaz]

Been losing sleep over screen R values in an AC30 type build. It's got relatively low plate and screen voltages at about 310 and 304 with respect the cathode at about 10.5 volts.

I've found some old Brimar datasheets that suggest 2W max at idle, and 4W at max signal. The JJ EL84's, the tubes that I'm using, datasheet says 2W max with no other information. I'm not sure what go with.

The screens glow with resistors from 100R to 1k, so I can't go by a visual indication of over-dissipation alone. I won't go over 1k because I believe the tone suffers too much. I've measured dissipation with both guitar signal and a sine wave. Dissipation with all tested values (100R, 470R, and 1K) is lower with the sine wave, all under 4W. When testing with guitar, I can get about 5.5W dissipation with the lowest 100R value, and 3.8W with 1K.

Which test method is more conclusive, you guys think? What dissipation do you shoot for with max signal? I'm really surprised to see modem amps (i.e. Peavey and Vox) with higher voltages, and even lower screen resistors. Seems pretty stupid to me!

Thanks for any experiences or advice.

[jjman]

I've been contemplating the screens on my 2xel84 amp the last few days. I have it on my bench and never considered the screens which are fed by a shared 560ohm resistor off the B+1. It's a 1967 Harmony H415.

I'm seeing ~4ma on each at idle @ ~360 net volts for 1.5watts each. Of course all the voltages are higher than the scheme. I also have JJs. I see some blue near the screens but I think they are ok. I'm more used to 6v6 tubes that idle around 3ma, and at higher voltages.

Guitar amps with el84s appear to break the "rules" more than any other. I haven't scoped this one yet and am installing individual cathode bias adjusters and the 1ohm bias/scope taps.

I wonder if your tubes are bad since they are glowing the screens and conducting more screen current, all at a lower net screen voltage than mine. Although I did increase the shared cathode resistor to 200 from the stock 130 to bring back down the idle watts in reaction to the overall higher voltages. Mine are probably 7-10 years old since I don't use that amp much.

[Gaz]

Sorry if I was unclear in my post, but those measurements were taken with max signal, not at idle. At idle, None of the aforementioned values dissipated over the 2W watt max. If I were you I'd check the dissipation at max signal by checking the voltage across the resistor, divide by the resistance as usual, and multiply that times the voltage from cathode to screen with at max signal (not idle signal). I think you may be unpleasantly surprised by the results wth only a 560R shared resistor at those voltages. I'm curious to see what you come up with, as I've never seen blue screens myself.

[Alexo]

Do you have a scope? If you are getting higher current draws with a guitar than with a signal generator, that just suggests that you need to increase the output of your signal gen. a little to match the guitar level.

So, obviously if your screens are glowing red and you are measuring 5.5 watts across them, you are hitting them too hard.

Don't judge other amp's settings until you know their load! Lower loads = less screen current under signal. 4K is kind of a hard load on the screens in quads of EL84's when the screen supply is almost equal to the plate voltage (a la Vox).

So I hear what you're saying about giant screen R's hurting the tone... there are other ways to drop the screen voltage which will shift the knee down and draw less screen current. If you could find a 10 watt, 20 volt zener diode or two and lay it 'backwards' in series with the choke in the power supply line, that will drop your screen voltage without giving you the extra compression and token crossover distortion/shifting operating point you get with huge screen R's. Otherwise, you just have to choose between killing your tubes for the preferred tone and saving your tubes and compromising. Always a trade-off!

[Gaz]

Alexo, you're totally right about the output level of my scope! I totally overlooked that while in testing frenzy. Thanks.

Like I said, the plate voltage is only 310vdc with respect to cathode, which is quite conservative for EL84s it seems. I've seen amps, namely Peavey and Carvin, with 400vdc on the plate and screens with only 100R screen ressitors!

I think I'll end up just using the 1ks for peace of mind, but my question still stands about what the max screen dissipation with max signal. I wrote JJ in the meantime...

[Ian444]

It's got relatively low plate and screen voltages at about 310 and 304 with respect the cathode at about 10.5 volts.

The screens glow with resistors from 100R to 1k, so I can't go by a visual indication of over-dissipation alone.

Gaz, the screens shouldn't be glowing with those voltages on them, should they? Are you sure you can actually see them glow?

...but my question still stands about what the max screen dissipation with max signal. I wrote JJ in the meantime...

Thats a very good question and I've been thinking a lot about it of late with some EL34's. It's interesting to put a scope across the screen resistor itself, you get something like a sine wave but only the top half or bottom half depending on polarity of test leads. I use the scope to measure RMS voltage across the screen resistor and calculate power using Ohms law, dunno if that's the best way or not. An EL34 amp I built exceeds screen dissipation by 25% at max but I could not see screen grids glowing so left it at that. If it blows up I will regulate the screen voltage.

Regarding screen grid resistor size, from 470R to 1K there is some local negative feedback happening with EL84 IME. Depends on the amp if this is good or not or what your ears want to hear, if your amp already has negative feedback then the extra local feedback might not be good. Need to look at the whole picture. What sort of amp are you working on?

[Alexo]

RE scoping out the screens - yes, your screens are only conducting on half-cycles. I suspect that this really makes a meter-based RMS reading inaccurate, since that non-conducting half-cycle while bring the average power way down.

Having said that, hooking up a couple meters - one to measure current and one to measure voltage - seems to give a reliable figure for checking how far you are from the limits. Maybe the fact that it's only conducting half-cycles lets it peak higher because it has a half cycle to cool off - i.e. maybe the average IS what counts, not the peak.

...I've had EL84 screens glow at those voltages too, btw.

[Philoffline]

Any chance you could post a pic of the glowing screens in question? I've been wondering about my screen current too.

This probably sounds stupid, but I'm not sure I can tell what's the screen and what's the filament in there. It all looks very tiny compared to the 6V6s I'm used to.

[Gaz]

It'll be a little bit until I could post a picture, but there should be a little window in the plate structure (near the middle) where you can see the screen grid. If it glows, it should do so when more signal is applied, usually past clipping. It sometimes takes a few minutes for them to get warmed up (see: stressed out.)

I'm not surprised at all by the glowing, and have experienced it often in El34s as well, even at lowish voltages (420vdc) with 1k screen stoppers. I have a feeling it would take a lot to get the glowing to stop completely, which would probably 'kill the tone' and headroom by then. Because of poor screen alignment, I kind of think the tubes are to blame at certain point, however no tube is really designed to hard clip it's whole life.

So here's the exchange with JJ (LOL):

Me: "I've found the datasheet for your EL84 tube, but it only has information for Class A operation. I'm looking for the Class AB information, namely the max screen dissipation at idle as well as at max signal. Please let me know if this information is available."

JJ: "we don't have requested information.

Best regards,
Julia Jurcova
sales"

[Ian444]

Gaz, I believe you know what you are talking about when you see screens glowing. I was under the impression that JJ EL84 can easily survive 340V B+ and 340V on the screens up to max power without glow, but have not tried it myself. Maybe not all the JJ EL84's are the same...

Only thing I can think of to try to stop the glow and keep the tone, is to regulate the screen voltage using a circuit like the VVR, but just use it to control the screen voltage only. This is not uncommon in hifi amps (where they are aiming for max clean power without blowing anything up). Be careful though, since changing the regulated screen voltage will also change the output tube cathode current.

[Gaz]

Yes, unfortunately I've seen them glow too many times. IME, however, doesn't mean the tube's gonna fail imminently, as long there are sizable screen R's and dissipation isn't through the roof. Still does bug me, as can be seen by my posting in the first place.

I don't really pay attention to hearsay about pushing tubes anymore because it's easy enough to test for dissipation, and look at 'proven' designs. Many of those designs torture tubes, of course, and it's funny to read biased reviews like: 'Matchless runs their tubes hard, which gives them amazing tone!' compared to 'That Peavey eats tubes, take it to the pawn shop!' If you've got clout and mystique you can do whatever you want!

I'm always impressed how Ken Fischer, whose amps are considered some of the best, ran tubes at relatively conservative voltages. I think he's the exception.

[pdf64]

Regarding screen grid resistor size, from 470R to 1K there is some local negative feedback happening with EL84 IME.

That's my understanding also.
My thinking to limit screen dissipation is, rather than fit such large screen grid resistors to each tube, to 'common' them up then bypass to ground and use regular 100 or 220 ohm individual screen grid resistors.
eg just add a 470 ohm resistor in series with the choke, as if the choke is very high resistance.
The bypassing gets rid of the local ac neg feedback, which reduces gain and output. My feeling is that is what makes large individual screen grid resistors sound bad.
Pete.

[Gaz]

Never heard of 'local NFB' being the cause for lower output and less desirable tone. the choke i'm using already has a 750R DC resistance, which I thought would be enough. What do you mean by "bypassing to ground?"

[Andy Le Blanc]

screen grid resistors set the difference between the plate voltage and screen voltage.

feedback will effect this difference. You can shut off the tube with the
screen grid voltage the same way as with bias, large screen resistors
can begin to approach the shut off, and sound bad if your not careful.

The old fender 6l6 design set the screen + about half a volt
so you have to bias cold to prevent the fizz, but the design was never
ment to distort...

so anything fender based, even a wreck, tends to use a conservative bias.

before the screen grid resistors, try adding an extra bleeder and cap
after one extra resistor from the PS rail, this extra resistor will allow you to
precisely set the screen voltage without using excessive series screen grid resistors .

Your setting the difference between the plate and G2 at what ever bias
point your set on, bias is the PS current irregardless of the tube

which is why you see so much variation with design implementation.

So... keep your plate voltage in line with the max V for the tube type.
keep the heater voltage at the published v for the type and your rig won't eat tubes.

The extra bleeder will provide some regulation for the screens. you can set the screen voltage
relative to the plate, and then you can push the tube type

I have a deluxe based thing that can run anything from 6v6 to KT90
the screens are set at a bias point that is a % of the PT rating.
when that bias point is maintained irregardless of type...

I know exactly what the circuit voltages are, and all tube types provide
the same performance whether they'er 6v6 biased to 110% for the type
or kt90 biased to only 35% for the type.

Going on a bit.... sorry

Try measuring the V difference between the plate and G2 at the tube socket (with the tube).
This is what lead me this way, you can do it with a regular DMM meter
Your not referencing the screen to ground or measuring the v drop
over a series resistor, so the measurement doesn't suffer the accuracy issue ,
and can be a simple static measurement.

[Structo]

Marshall did something like this on the early amps.
Here is a JTM45 showing the 1K floating resistor as they call it.

The red arrow points to it.