6V6 Screen Grid Dissipation - Is it an issue?

[Littlewyan]

I've been tinkering with an amp build of mine and found the screen grids appeared to be running a bit hot. It's 6V6 Cathode Bias, 8K OT, 340V on the anodes, a big 400Ohm choke between OT and screens and 470Ohm screen grid resistors. I tested it at idle, max clean (just before clipping) and then just flat out hammered it with a 1V input signal. I'm using Tungsol 6V6RI valves and the screens would dissipate up to 4.4W! After much tinkering I've upped the screen grid resistors to 1K and added a 1.5K resistor inline with the choke and now they sit at around 2.6W flat out. Tungsol 6V6RI datasheet states 2.2W max for screen grids and 13.2W max for anodes.

I'm wondering though, how much of an issue is this? I've watched the valve at various conditions and it seems the screen grid glows even at 2.2W, it just glows a bit brighter when I'm hammering the amp. If you look at a Fender Deluxe AB763 they just have a small choke (around 100Ohms) and 470Ohm screen grid resistors. A Suhr PT15 has a 1k choke resistor but 470Ohm screen grid resistors. Both of these amps run at 420V! (Although Suhr recommends the JJ 6V6 for his) Is it just accepted that the screen grid will get hammered under overdrive conditions? I'm wondering if I'm just neutering the amp by trying to get this under control. Or maybe it's just a slightly dicky valve, as the dissipation of both the anodes and screens seem to vary a bit each time I test them and it seems the screen grid on just one of them actually glows bright each time. Interested in people's thoughts on this.

[Littlewyan]

I should mention, I have done the loadlines for this amp and they don't indicate that the screens would be drawing a lot of current. I'm operating just a tiny bit below the knee.

[martin manning]

Looks like your 8k load line is a bit above the knee, which reduces the tendency to draw screen current. How are you measuring the screen dissipation?

[pdf64]

Yes, the screen to cathode voltage will reduce, probably significantly so, when cranked.

[Littlewyan]

I’m measuring voltage across the screen grid resistors using my meter on vdc. I think I’m probably being a bit too cautious with the screens. I was just thinking, datasheet values will be for getting maximum clean output, the engineers who came up with them wouldn’t have thought about guitarists pushing the valves into heavy distortion. So maybe I should just ensure the screens aren’t over dissipating at max clean output and just accept they’re going to get hot if I push them into overdrive.

[pdf64]

I think most amp designs probably do result in their screens overdissipating when the output is pushed to max squarewave.
I haven’t checked, but a typical Express might not.
It seems a good idea to try to avoid pushing the dissipation too far over the limit, in the worst case condition thst the amp’s expected to be operated.

[Littlewyan]

I do remember another amp of mine does overdissipate the screens. EL34s with 6.6K OT, 470V HT, Choke but 2.2K screen grid resistors. I think it runs the screens at 12W flat out and the limit for EL34s is 8W. Komets must be even worse given they use 1K screen grid resistors with a choke and 5.2K OT!

I might just bring the choke resistor down to 470Ohms or remove it altogether. Then maybe up the screen grid resistors to 2.2K. I'm wondering if I've either got a dicky valve or the signal at the screen grids is messing with my meter, as I get slightly different readings every time. Plus, my test flat out isn't really an accurate simulation of someone playing guitar through the amp, unless they're just hitting heavy power chords continously!

I should also say, the PT for this amp has been overspec'd, so there is no sag at all. Which could be another factor. I know Fender Deluxe AB763 amps sag under load, not sure about other 6V6 amps.

[LOUDthud]

When you calculate dissipation, you are making an error if you are using average Voltages read from a meter. Allow me to illustrate.

100VDC across a 1K resistor is 10W. If the resistor is seeing 200V for half the time and zero for the other half, what is the dissipation ? The average Voltage is still 100V right ? When the 1K resistor sees 200V, the dissipation is 40W for that amount of time. Averaged out over the time when dissipation is zero, the average dissipation is 20W.

To measure screen dissipation, set up a dual trace scope to look at both sides of the screen resistor, Then look at the difference between the two traces to get the Voltage across the resistor under signal conditions. The screen side probe will give you the screen Voltage and you can calculate the instantaneous screen dissipation, then average that out over an entire cycle to get actual dissipation.

[brewdude]

I am no expert, but I can’t help but think,“why bother?”… I mean, worrying and calculating so precisely. If you have not strayed too far from typical operating conditions, you have screen grid resistors and are not over voltage at idle, are you in “safe” enough territory?

Is there a problem you need to correct? Or, is it just educational curiosity?

Am I foolish for not being concerned enough to check screen dissipation?

I have noticed that adjustments of grid bias can effect current screen voltages, I assume it effects screen current drawn as well.

[pdf64]

When you calculate dissipation, you are making an error if you are using average Voltages read from a meter. Allow me to illustrate…

I don’t see what’s wrong with using the VDC across the screen grid resistor to derive the current etc?
Surely with heating, the time averaged reading that will be displayed is exactly what’s wanted?

[LOUDthud]

I don’t see what’s wrong with using the VDC across the screen grid resistor to derive the current etc?
Surely with heating, the time averaged reading that will be displayed is exactly what’s wanted?

If you're OK with the actual dissipation under overdrive conditions being TWICE what you calculate, go ahead. As illustrated above, the Average Voltage across the screen resistor, does not equate to the Average Power dissipated by the Resistor or the Screen. The reason is when the Voltage is TWICE, the power is FOUR times as much.

[bepone]

formula is always U x I, so need to find average current and average voltage under heavy load.. scope is ok for that. big screen resistor help to protect grid2 from over current of course.

[bepone]

big screeen current is sign of oscillation also, i see nobody mentioned here this .. so it is good to check output for oscillations with the scope and speaker attached, reactive load not resistor..(search for many 10ths or hundreads kHz)

[pdf64]

… the Average Voltage across the screen resistor, does not equate to the Average Power dissipated by the Resistor or the Screen. …

My thinking is that the VDC across the screen grid resistor (at max squarewave output into a resistive load), divided by the resistor value, will be the average screen grid current under this worst case condition.
A screen grid to cathode VDC measurement in that scenario will be the average screen grid voltage.
Hence the worst screen grid dissipation will be the product of the above average current and voltage.

[LOUDthud]

formula is always U x I, so need to find average current and average voltage under heavy load.. scope is ok for that. big screen resistor help to protect grid2 from over current of course.

Average Voltage is not the same as average power.

Av_E_Av_P.GIF