Adding screen resistor where there was none before?

[CitizenCain]

On my Heathkit amp conversion project, the 7591 power tubes don't have a screen resistor in the original circuit. I'm supposing it would be good to add. Is there way to figure a proper value, or just use something standard like 470R or 1K?

[BackwardsBoB]

It's possible that the screen supply is high enough impedance that if a tube faults, you'll be OK. But I doubt it. Worth checking. Figure 100mA fault load.

If you're trying to get a nice, linear output stage, it's nice if the screen voltage is constant, so, in theory, you want to hook it up to a low-impedance (low resistance) voltage source. This ignores the fact that power tubes like to fail and draw a high current on the screens. If you have a high impedance (high resistance) on the screens, the screen voltage will drop in this case, reducing the current through the tube and avoid untold destruction of the circuit. On the other hand, screen current varies with tube anode current (figure maybe 10% of anode current in normal operation), so if you have a high impedance on the screen, the voltage will drop on the screens when the tubes are sinking lots of current. In normal operation, this peak screen current is around 10mA (it probably isn't, but it will get you in the ballpark), so if you put a 1K screen resistor, the voltage will drop with the impact of 10V during peaks when the tubes are sinking drive current. This ain't much, but it will slightly reduce the gain of the tube. If you're trying for great linearity, that doesn't sound so good, but neither does great linearity. Your feedback loop will counter this effect anyway. You can add 1K screen resistors, and all you will get is a very little bit of compression and a very little bit of harmonic distortion. Those are good things, right?

If you don't want this effect, you can always add a big cap right on the screen too, but...

Another, perhaps more important purpose of a screen resistor is to crash the screen voltage in the case of a tube fault. A cap will fight you a bit on this. Let's go back to our 1K screen resistor with 10mA peak current. If the tube faults, and the current goes to 100mA, the 1K screen resistor will drop the screen voltage 100V, which makes a big difference in limiting the current. Now this only happens when the grid voltage is high, maybe 30% of the time (or 50% if you're feeling enegetic and you have a high-output PI circuit, or even more if you have some kind of sick grid fault). The cap will average the current, and now you're only dropping the screen 50V.

If you've been paying attention, you heard Mr. Moore state that he like >4K screen resistance (at several Watts). He's dropping maybe 40V peak screen voltage in normal operation, and smiling. For a guitar amp, especially if you like to make it cry and sing, 1K 5W isn't silly, and you could really go much higher. Traditional output circuits don't generally have this feature, for unknown reasons. Could be spooks.

If you look at my favorite Carvin X-series amps, over 14 years, the screen resistors kept getting bigger. Carvin does its own repairs.

Now, you've got to ask yourself, if there's 100mA of screen current, what else is going to blow? Trace back through your circuit, find anything that's waiting to fry, and beef it up, or make sure that something cheap and stood off from the board will fuse and save you. Then if a tube blows messy, and you keep playing because it sounds kind of cool, you'll be back in action with a tube swap, instead of waiting 2 weeks for that part order to come through.

Some folks like "ultra-linear" taps on their OTs to improve the sound. These vary the screen voltage with tube plate voltage (sort of). Screen resistors actually get you part way to that sound.

If your tubes are running happy, you can try it both ways and decide for yourself.

If your amp sounds funny, stop playing and turn it off quick. It takes a while for the smoke to leak out, and the amp runs on smoke, so if the smoke leaks out, you're going to have to pay for an expensive smoke recharge.

But that's just me. Perhaps some wizard will refine my answer.

[kg]

pretty good answer, bob.

large screen stoppers will improve longevity, but sacrifice the "stiffness" of the amp as you crank it up, and reduce output power. sometimes the softer response is welcome, sometimes not.

smaller value or no stoppers will make more power, and provide a very hard edged response, but your screens are going to take a beating. i wouldn't suggest it these days, though in the past with good NOS it wasn't nearly as big of an issue as it is now (mostly due to attention to detail in tube construction).

personally i don't like to go higher than 2k2, but i do my own repairs. for someone who just wants to crank the crap out of the amp and not worry about it, if they don't mind the tone with higher value stoppers it certainly will help prolong tube life, and possibly the amp's in the process.

ken

[BackwardsBoB]

Pondering further...

At the risk of sleep inducement,

If you look at the curves for a pentode for a while, you'll see that for a constant plate voltage, transconductance increases as you approach staturation (higher current). A drop in the screen voltage can compensate for this, and the unavoidable drop in plate voltage that makes the OT work also helps. So you may actually be more linear with a bit of screen voltage drop due to a screen resistance.

That said, your circuit probably has a nice negative feedback loop or cathode bias resistor that's going to take care of the whole non-linearity thing anyway, until you get to saturation, if you can even get there with your drive circuit.

I'm used to amps that have wimpy screen resistors, and I like to beef them up to prevent future heartache. Since you don't have separate screen resistors, you'll be cutting the traces to the screens or maybe sky-wiring in screen resistors to make the mod. This would be in order to add protection from tube faults, and all those other hundreds or thousands of units of the amplifier in question work pretty well without it.

Waxing philosophical, every production circuit has a daddy. In every case, to reach production, that daddy has to say, at the risk of his reputation, "Yeah. Sounds good. Works good. I think you should spend lots of money and build bunches." Some guy said that with your amp, and some guy with lots of money trusted him enough to believe him. It may not be the amp you want, but it's probably a pretty good amp, and you should only modify it with good reason, and if the reason is that you don't agree that it sounds good, your opinion might be subjective, and maybe you should try another amp.

I'm not a tube amp designer, but I am a circuit designer, and I know that this is true.

Of course, the daddy is question didn't put in any screen resistors, so it's possible that the guy with alot of money was stupid to trust him. I have met many people with money that can't tell a wizard from a hack.

[mooreamps]

I have always spec'ed SGR's for specific tubes, for specific tones. The added benefit was that it helped keep screen current in check. If you have a tube that is smoking a SGR, more than likely it's time to replace the tube. However, for the 7591, I would start out at 1K 5 Watt, flame proof. One each per tube.

-g

[CitizenCain]

Hahaha, thanks for the answers guys

Bob, I'm converting a Heathkit audio amp into a guitar amp, so extreme linearity is not on the top of the wish list

Sounds like the 1K 5W is a good place to start.

[kg]

If you look at the curves for any tube ever made for a while, you'll see that for a constant plate voltage, transconductance increases as you approach staturation (higher current).

fixed for greater truth.