2W screen R's for EL34

[g1]

Any problem with using 2W 1K screen resistors for EL34? I only ever see 5W. I figure if Fender uses 1W for 470R, at equivalent current I squared R would require double the wattage.
It's very rare to see burnt screen resistors with EL34 when they are 5W, but I'd prefer the fusing action. Currently replacing a PT in a 50W EL34 amp that had a bad EL34 and the PT had meltdown before the 3A fast-blow opened. Thinking about all the possible scenarios, including over-fusing or tin foil.

[The Dude]

I don't think this can be answered without more info on a specific circuit, voltage measurements, etc. I've read pages and pages of debates on the subject and even reputable engineers/builders seem to have varying opinions. There are even those who recommend 10W screen resistors. I think rule of thumb is: Crank the amp to clip, measure the voltage drop across the resistor, calculate dissipation using ohms law and double it for recommended resistor wattage.
Generic answer: I think your 2W resistors would run pretty hot if the owner runs the amp at all cranked and the amp is somewhat typical EL34 design

[Tom Phillips]

It would be interesting to run an EL-34 amp up to full clipping and measure the voltage across the 1K screen resistor and calculate the power dissipation. (At least for the particular amp tested). I'm thinking I have done that but I can find the data right now.

[The Dude]

I remember working on a Carvin EL34 amp a few years back that had 2 or 3 watt screen resistors. Don't remember which model. They got so hot that they melted the solder and fell off the board. I found them rolling around in the chassis. Installing 5W solved the problem.

[g1]

I don't think this can be answered without more info on a specific circuit, voltage measurements, etc. I've read pages and pages of debates on the subject and even reputable engineers/builders seem to have varying opinions.

It's good that you ask this as I realized it's not 1K but 470R in this one. It's a Garnet that has pretty much the same arrangement as attached, but with 5W screen resistors. Attached would be newer and he has changed to 2W so I think I should be ok with that.
Tom, I will try to do the test you suggest, maybe with both 470 and 1K values.
Still have to order PT so it might not be this week.

[Purgasound]

The lower wattage ratings will not be able to dissipate heat as well. They will overheat and melt the solder holding them in. Even certain 5W resistors can get too hot. Some of the metal oxide ones will smoke when certain amps are cranked. It's a shame some of the axial cement wirewound resistors are getting harder to find. Those work really well.

[Greg Robinson]

I figure if Fender uses 1W for 470R, at equivalent current I squared R would require double the wattage.

You're forgetting that EL34's being true pentodes vs the beam tetrodes of 6L6's, draw significantly higher screen current.

IMO, relying on resistors as fuses is poor practice. A vaguely competent end user can replace a fuse, you need a tech or some higher skill level and access to appropriate tools to replace screen grid resistors, not to mention the collateral damage it can cause. Of course not every amp is appropriately fused, so it's always a good idea to use fusible/flameproof resistors, but note that the term "fusible" only means that the part should not catch fire, not that it has any fuse-like characteristics like repeatable clearance times. Hell, even cartridge fuses are only vaguely repeatable, with clearance times usually being only within a 10x window, only guaranteeing MAX times.

[The Dude]

You're forgetting that EL34's being true pentodes vs the beam tetrodes of 6L6's, draw significantly higher screen current.......

I hadn't even thought about that. Good point!

[g1]

I understand that there are examples of amps where the screen R's overheat and melt their solder. But that does not seem to be so common. Fender runs 1W mounted on the board with their new re-issues and this is not a problem. It's definitely not a problem in classic Fender circuits that have the 1W mounted at the sockets. EL34 screen current seems to be in the ballpark of 150% of 6L6. 150% current makes twice the wattage. So by that calculation, a 2W like in the schematic posted should be sufficient. Maybe it isn't.

It is interesting that certain amps do have the melting solder issue. Not sure why it would occur in some amps but not others. Maybe amps run with heavy clipping would be more prone to this?

[The Dude]

There are many variables- B+, screen voltage, screen resistor value, OT impedance , etc., etc. All EL34 amps are not the same. Stating the obvious (exaggerated for effect), an amp with 200V supply is not going to do what an amp with 600V supply will do using the same tubes. I think, as I said in post #2, the only way to know for sure with a particular amp is with measurements. Otherwise, we just ballpark it and call it good.

[g1]

IMO, relying on resistors as fuses is poor practice.

I agree, HT fuse would be better. But owner does not want to do that in this case. And unless it's buried in the chassis, user can 'tin-foil' that one too. As far as collateral, PT is the worst.
Sorry I pulled a switch mid-thread, it turned out these resistors are 470R, so with extra EL34 screen current, I figured 2W (double the Fender original 1W rating) should suffice. And the designers later model (post #5 attachment) seems to concur.

[Helmholtz]

. EL34 screen current seems to be in the ballpark of 150% of 6L6. 150% current makes twice the wattage. So by that calculation, a 2W like in the schematic posted should be sufficient. Maybe it isn't.

Comparable datasheet examples for 55W designs give a full power (without clipping) screen current of 11mA per 6L6GC and 25mA per EL34.
So the current ratio is 2.3 resulting in a power ratio of 5+ for the screen resistors.

Otherwise I'm with The Dude, that measuring in the amp is best.
Screen currents will increase with clipping and at higher than nominal load impedance.

Generally a resistor having a larger surface area will run cooler.

[loudthud]

It would be interesting to run an EL-34 amp up to full clipping and measure the voltage across the 1K screen resistor and calculate the power dissipation. (At least for the particular amp tested). I'm thinking I have done that but I can find the data right now.

It's an important point that Average Voltage that your meter would measure does not equate to Average Power Dissipation.

Consider a 1K resistor with 25V across it. DC dissipation would be E²/R = 0.625 Watt. If the same resistor has 50V across it for half the time and zero across it for the other half of the time what is the dissipation ? The Average Voltage is still 25V...

When the Voltage is 50V, the dissipation is 2.5 Watt, taking into account the half of the time when power is zero, the dissipation is 1.25 Watt, twice the dissipation with the same average Voltage.

If you have the right scope probe, look at the screen Voltage when the amp is rail to rail clipping and calculate the dissipation from that waveform. If you don't have the right scope probe, build a Voltage divider with two 100K 2W resistors. Compare that number to what a "True RMS" meter might read. They might not be the same because of an excessive Crest Factor that limits accuracy of some meters.

[Tom Phillips]

...It's an important point that Average Voltage that your meter would measure does not equate to Average Power Dissipation...

Good point. Thanks for the reminder. I should have mentioned that in my previous post. Fortunately, I have some good True RMS meters so I don't often think about that.

[Helmholtz]

Agree that loudthud made a very good point.
The screen current is unipolar (no zero crossings as electrons don't fly back to the cathode) but consists of a DC current superimposed with an asymmetrical AC current.
For screen resistor dissipation calculation one needs to determine the total RMS voltage across it.
The total RMS voltage is the vector sum of average DCV and RMS ACV.
It calculates as square root of (Vdc˛ +Vac˛).
Unfortunately most so-called True RMS meters are AC coupled and only read the AC part, so it takes 2 measurements (AC and DC).

Problem with using a scope is that the voltage is floating on the screen supply.