Why should I run screen resistors?

[Wombaticus]

You don't "design" a tube output stage, you just get the Philips or GE datasheet and pick the nearest set of example conditions to the transformers you managed to scrounge. If it doesn't work, you fiddle with the screen resistors and blame "modern tubes". It's not rocket science.

Even if you wanted to use some weird tube whose datasheet doesn't specify audio service conditions, or God forbid build a tube hi-fi amp, the factors you mentioned make it loose enough that you can just kind of squint at the characteristic curves and imagine where the overdissipation region would be.

To restate the point about the average dissipation: It's OK for part of the load line to be in the overdissipation region, since other parts of the load line are far away from it, so it averages out. But how much is it allowed to poke in there? Do you know how to calculate how much? Graphically? For a sine wave? A clipped square wave? What about taking idle dissipation into account? I rest my case.

Seriously?

Richard Kuehnel in his book "Guitar Amplifier Power Amps" describes a 19-point graphical extension of the old RDH4 9-point method that allows you to pretty accurately estimate average plate dissipation, as well as getting at the average current for a sinusoid at full power. And the method also works for screen dissipation. Simulation can also be very helpful, especially if you wish to explore a variety of scenarios. And idle dissipation is pretty straightforward. So yes, I guess I basically do know how to approach this, although I recognize that it's mostly an exercise in assuring the "reasonableness" of a design. Randall Aiken has also posted some very informative material about load lines and where plate dissipation maxima occur under different conditions. Among other things, he's pointed out how Marshall basically got it wrong with their old high voltage 100W amps. More than one cloner has been bitten on the rear because they used their crazy load line, but with a PT with insufficient sag to keep things marginally workable. All of which followed more-or-less from Marshall coping stuff from datasheets but not bothering to do the maths to compensate for increased voltages.

Anyway, the bottom line is that if you wish to adhere to conservative engineering practices, it's probably a good idea to at least have an idea where your load lines lie relative to the maximum plate dissipation line, and probably avoid exceeding it (or 2X for the class B line). By all means though, continue to squint away. It works -- most of the time.

[Steve Conner]

So, maybe my reply was a little tongue in cheek, but come on, who ever does a 19-point graphical calculation?

Certainly not the designers of the classic guitar amps that are still held up as oracles of tone, even though, as you correctly point out, they are designed wrong and can barely make it through a gig without melting down. (I once saw Marshall's Hendrix signature Plexi reissue at a Marshall Roadshow event, and it failed, presumably for vintage-correct reasons. It died after the first couple of power chords.)

If you try to re-engineer these classic designs so they actually work reliably, the DIY community will just ignore you and copy the originals, as they supposedly contain some sort of organic wisdom that mere eggheads like ourselves can't hope to comprehend. At least, this is what boutique builders who dropped out of EE school tell themselves and any customers "organic" enough to listen.

Making the screen resistors a bit bigger helps, so this is kind of on topic!

[imaradiostar]

Making the screen resistors a bit bigger helps, so this is kind of on topic!

And the truth will set you free...

I recently (a year or two ago) went about torturing a variety of tubes in different over voltage conditions on the screen and plates and in the end I concluded a few simple things. I would imagine Steve or Enzo or Bruce or whoever would agree with most or all of these:

1)use big enough screen resistors to prevent the screens from melting- the tiny gains from running dangerously low screen resistors isn't worth it.

2)use a quality output transformer to ensure decent high and low frequency extension. No amount of careful tube or voltage selection will matter if your OT is pathetic. I was surprised to see how ugly the output from some of my OT's was. My ears seemed to agree with my eyes which was a surprise!

3)if in doubt, drop the screen voltage. Your ears won't notice the lower wattage but your wallet will- it'll thank you. This especially applies in cases where the screens are within a few volts of the plates and close to or over 500 volts!

4)make sure you have a solid bias supply.

5)don't over-think it. you're not putting man on the moon. Tubes are so incredibly forgiving! Steve's advice about picking something that works for the parts you have is a great way to put it.

Many of the classic amps were built with the junk they could buy cheaply. They made it work and now it's become "mojo."

I'd like to think I can build in my own mojo by using my ears, a soldering iron and careful parts value (not even type!) selection. (resistor type arguments need not apply)

jamie

[barfoden]

I heard that screen resistors can limit output for beam forming tubes.. That is why you don't go to high with these.. 1K screen resistors in a fender deluxe reverb will give a little more compression and slightly less total power..
True pentodes with a G3 surpressor draw more screen current in relation to plate current compared to beam forming tubes and hence need more screen resistance when you treat them rough..

My new Laney LC50 runs with plate voltage at 400V and screen voltage slightly lower with a 270 ohm dropper in the supply. Since this limits the amount of ripple filtering compared to a choke Laney added a huge 100µf screen cap and fitted 5W 1K screen resistor on each EL34.

But sometimes I get amazed when looking at modern schematics..

The traynor YCV40 runs its 6L6GC tubes at 410V plate, a 390 ohm screen dropper and small 1W 220 ohm screen resistors. This is fine for 6L6GC tubes as they are run at a "modest" 410V plate and 40w output and don't draw much screen current at a primary impedance of 4-4.2K (my guess) for a pair of 6l6GC.

The "british" counterpart, the YCV50B is equipped with two EL34 tubes for 50W.
According to the schematic it has the exact same setup.. 410V plate, 390 ohm screen dropper and 1W 220 ohm screen resistors.
The screens are supplied from a smaller 15µf so I guess less power is available from that nod... but still,, 1W 220 ohm resistors seem small..

[Chuck H]

Excellent observation and definitely gets us away from the generic considerations. I still contend that bad things can happen without individual screen grid resistors when an amp is clipping. I don't have nearly enough tech savvy to run the math for beam tubes vs. straight pentodes so I just use 470 ohm for beam tubes and 1k for everything else Actually I've used 470 ohm for el84's too when HV is closer to 350 than the near 400V that a lot of amps run. Any lack of power using 1k screen grid R's is really insignificant WRT guitar amps. Though the difference in feel/tone may matter more.

[jazbo8]

The "british" counterpart, the YCV50B is equipped with two EL34 tubes for 50W.
According to the schematic it has the exact same setup.. 410V plate, 390 ohm screen dropper and 1W 220 ohm screen resistors.
The screens are supplied from a smaller 15µf so I guess less power is available from that nod... but still,, 1W 220 ohm resistors seem small..

I think your observation is spot on, the YCV50 is running the EL34's screen way too hot if the values you showed are correct, perhaps the actual values of the resistor in the amp are different than those shown on the schematic, I can't image Traynor would push it that hard, below are the operating conditions/load lines for the two amps, as you can see on the bottom of the charts, the estimated Pdg2 for YCV50 is 11W, which is way over the max Pdg2 per the datasheet.

[mhuss]

The Hiwatt 100 watt amp (4 x EL34) used a shared 470 ohm 20w screen resistor for all four tubes, plus an individual 100 ohm 1w on each tube. Dave Reeves was a Mullard man, so perhaps that's where he got the idea. The 470 limits the overall screen current, and the individual 100R play the fuse role (and I've seen quite a few 'blown fuses.')

In addition, he used to fit an optional "half power" switch that just added a largish value of resistance into the screen circuit. I never cared for the sound of this setting myself.

[elipsey]

You mean planet long division?

lulz... i was marooned there for several months in the third grade before being rescued by the Calculator Air Patrol.

[elipsey]

You don't "design" a tube output stage, you just get the Philips or GE datasheet and pick the nearest set of example conditions to the transformers you managed to scrounge. If it doesn't work, you fiddle with the screen resistors and blame "modern tubes". It's not rocket science.

Even if you wanted to use some weird tube whose datasheet doesn't specify audio service conditions, or God forbid build a tube hi-fi amp, the factors you mentioned make it loose enough that you can just kind of squint at the characteristic curves and imagine where the overdissipation region would be.

To restate the point about the average dissipation: It's OK for part of the load line to be in the overdissipation region, since other parts of the load line are far away from it, so it averages out. But how much is it allowed to poke in there? Do you know how to calculate how much? Graphically? For a sine wave? A clipped square wave? What about taking idle dissipation into account? I rest my case.

I have the habit, when designing my own amp, of drawing a load line and a max dissapation curve and making sure they don't cross. Are you saying I don't need to do that?

This seems like it could result in a configuration that dangerously exceeds max dissapation under certain conditions; I think Aiken shows what those conditions are when he explains 70% rule of thumb for bias.

[barfoden]

Hi Mhuss

A shared 470 ohm resistors on 4 EL34. What would that correspond to if each EL34 had its own screen resistor.??

My guess would be a 1.88K resistors on each screen grid.. is that correct...

I found another amp that has very similar EL34 operation.. A Laney GH50L.. 460V B+, 100µf main filter,, -35V bias and a ~3.5K output transformer..

My new cheap amp has the EL34 at 400V plate and 100µf plate and screen filters and a 270 ohm screen dropper with 1K screen grid resistors.. Right now I have it biased at -35V which should be a cold bias, but since there is no 1 ohm testpoints and I am not familiar with the OT resistance and have no bias probe for octal tubes I have no idea what the idle dissipation is..

Bogner claims that their Shiva amp with two EL34 is a 80w amp.. Hmmmmm,,

Is this possible with a conventional guitar amp power supply and around 515 volts on the plates.. Maybe the 80 watts are with a great deal of THD (not less than 10%)..

Mullards data says something like 55W at 5% THD and a plate voltage of 425V when the power supply has no sag and endless amounts of current to give..

[g1]

I think your observation is spot on, the YCV50 is running the EL34's screen way too hot if the values you showed are correct, perhaps the actual values of the resistor in the amp are different than those shown on the schematic, I can't image Traynor would push it that hard, below are the operating conditions/load lines for the two amps, as you can see on the bottom of the charts, the estimated Pdg2 for YCV50 is 11W, which is way over the max Pdg2 per the datasheet.

Not being too big on the theory, I'm just wondering how the Pdg2 numbers are estimated, and how you arrived at those g1 voltages? Are you taking into account that the tubes are biased at 37mA, for about 15W, for both amp models?

[mhuss]

Hi Mhuss

A shared 470 ohm resistors on 4 EL34. What would that correspond to if each EL34 had its own screen resistor.??

My guess would be a 1.88K resistors on each screen grid.. is that correct...

Yes, essentially 4 x 470. However, it's still a good idea to have the additional 100R in each individual circuit to account from tube to tube variations,

[es345]

In a PP with 4 tube always basically only 2 tubes are active at a given time.
If you want to have comparable separate screen resistor per tube its 2*470+100= 1040Ohm >> use 1KOHm per tube.

[jazbo8]

Not being too big on the theory, I'm just wondering how the Pdg2 numbers are estimated, and how you arrived at those g1 voltages? Are you taking into account that the tubes are biased at 37mA, for about 15W, for both amp models?

Pdg2 estimates are based on the formula from RDH4, which is (0.5Ig2 idle + 0.25ig2 max) x Eg2. Eg1 voltages are based on the figures shown on the schematics, the two models are biased differently as shown.

[Tooboob]

The Carvin X100B makes an interesting reference for screen treatment. The basic architecture appeared in '81, as a 100W 6L6 amp, the design stabilized in '84, and converted to EL34 in '87 or '88, with production continuing until '92. They never achieved a shipping volume anything like Fender, but over time, quite a few amps got out there, and Carvin does lots of their own service in their factory. Let's look at the screen resistors.

The proto-Beasts in '81 used a 350 Ohm 10W(!) dropping resistor and individual 470 Ohm 2W screen resistors, with 40uF after the dropping resistor. Then in '84, they dropped the screen resistors to 1W. Shortly thereafter, they did a version that dropped the dropping resistor to 5W, and used 22 Ohm 1/2W screen resistors. 1/2W seems insane until you realize that with 22 Ohms, power won't be much. Around '87/'88, they switched to EL34s, and they had 350 Ohm 5W in all slots.
The amp went away 5 years later, but it came back in '08(?), this time with a 10W dropping resistor.

It's pretty obvious that Carvin doesn't mind 5 power resistors, though they'd prefer them to be the same value, they liked low-value screen resistors, and they had problems with molten amps.

All these amps had non-adjustable bias for the output tubes, and they were nominally set for 25mA of bias current per tube. The schematic asks that the bias resistor be adjusted in production, but I can't see that it ever happened. Looking at used units, t's not uncommon to see a bias pot retrofitted in by an owner. Even on unmodified amps, it's not unusual to find badly scorched screen resistors or PC boards among the early models.

If you just eyeball the curves for 6L6, EL34, and EL84, you can see that transconductance is fairly constant for a 6L6, increases with grid voltage a bit on an EL34, and increases significantly on an EL84. This is the opposite of compression. It's a guitar amp, and I don't mind a bit of compression - the NFB can handle it anyway if you have one, at least when things aren't overdriven, and one man's "cuts through the mix" is another man's "can of bees", but look at what a UL transformer does, and that's hi-fi. If screen voltage drops a bit with screen current, I count it as a good thing. So I see no fundamental reason not to take screen resistors up to 1K, or beyond, especially for EL84. You'll get a bit more voltage drop, especially if you overdrive the tube, but when you overdrive the tube, the PI loses control and increases drive to the tube dramatically - so much so that it's hard to tell by inspection whether the PI or power tubes are clipping first. This is the Marshall version of the sound of rawk, but there are nice sounds if you can make it a bit less abrupt. Rumor has it that high value screen resistors and grid stops kill the bees in EL84s too, and, to me, that's a public service.

So, you come up with a sonically optimal value for screen resistance that doesn't make toast. You still need to think about what happens if a power tube develops a screen short. If it's not addressed, you get a few amps that go pyrotechnic. Up the screen resistance and wattage, and the resistors lose the capability to act as a fuse (admittedly one that's hard to replace, but it beats patching a PC board hole). It puts the responsibility on the AC fuse, and you've boosted the screen resistance, so the current isn't very high, so the fuse might not blow immediately or ever, and the screen resistor or dropping resistor winds up frying anyway, taking long enough to do it to fry anything nearby.

I've designed lots of products used by normal people, and though I'm usually somewhat insulated from the customer, I get the distinct impression that customers get all tweezed when their stuff emits smoke and flame (ever look on a computer circuit and see a little black Sharpie dot on all the tantalums?).

It's not an easy problem, amenable to "stick a fuse here", but it's worth investigating, especially if you are using power tube grid buffering, zoonie B+, etc. All the solutions I can think of that don't work by failing on failures won't work with parts suspended in space, connected with push-back wire, and certainly not without late-model HV silicon, though the traditional construction methods are much easier to repair once the smoke clears. The best solution for a fixed-bias experimenter is probably low-wattage low-value cathode resistors, since they also provides a sensing point for biasing, but resistor manufacturers don't exactly get depressed when their resistors can take more than their wattage rating.