What would happen if you connected screen grids to constant current source?

[Austin]

Instead of a regulated voltage source where the current varies to hold the screens at a certain voltage, what would happen to the sound if you connected the screens to a constant current source (via another tube) so the voltage varied to keep the current the same all the time? This might be a silly question but I have been curious about this and figured you guys would know. I imagine that when the plate voltage is high the screens would draw very little current and the screen voltage would increase to a point where it was drawing the same current again up to the the supply voltage, and when the plate voltage hit a low point the ccs voltage would fall to a point to limit the current, but what effect would this have on the sound? Anybody know? The simplest ccs is a series resistor, some amps use a simple resistor in series with the b+. So, what if the high initial voltage on the screen caused an increase in sensitivity or feel when you plucked a string but was still safe for the screen max power? I understand you want to limit the power on the screen so for instance to limit the power to 5 watts on the screens maybe you would make a .01 amp ccs and give it a 500 volt supply voltage and then feed the screens with that. Thanks in advance, I have learned very much from this forum and appreciate you guys very much

[es345]

Let me try a short answer without theory details, please have a look to tetrode/pentode theory in well known books.

G2(screen grid) was invented to reduce the influence of plate voltage to the plate current at a given G1 voltage.
This is what you see if you compare the diagram of a pentode with the diagram of the pentode connected as a triode. If you now limit the screen current as you have described G2 will follow basically the anode voltage - which result in something which is more triode behaviour.

[woodyc]

Just agreeing with es345. You'd end up with something vaguely like, but even weirder than, triode or UL mode, namely un-pentode-like characteristics and significantly reduced output. Its a really subversive idea. Might be fun to try, especially if you have a curve tracer.

[jazbo8]

Have not tried this on a curve tracer, but the sim showed largely pentode characteristics with VERY compressed operating range for Vg1 as compared with normal operation. I ran the sim at the max signal Is recommended for the 6L6 of 7mA, varying the CCS current simply move the curve family up and down. Ok for an academic exercise, but don't think there is much use for such application.



Jaz

[woodyc]

Have not tried this on a curve tracer, but...

Thats really interesting. But I think you might be a little outside of the working limits of the model. In reality, a sufficiently negative grid should be able to shut off the current regardless of what is connected to the screen.

[jazbo8]

Thats really interesting. But I think you might be a little outside of the working limits of the model. In reality, a sufficiently negative grid should be able to shut off the current regardless of what is connected to the screen.

Damn models! What you said made perfect sense, shutting off the grid should result in little plate current... Should have read this http://music-electronics-forum.com/t21787/ first, the subject has already been covered. In any case, I guess we just discover another weakness in the pentode SPICE models...

Jaz

[exclamationmark]

Damn models! What you said made perfect sense, shutting off the grid should result in little plate current... Should have read this http://music-electronics-forum.com/t21787/ first, the subject has already been covered. In any case, I guess we just discover another weakness in the pentode SPICE models...

Jaz

Spice does all sorts of weird stuff... I was simulating the transient response of some tube switching circuits (so they wouldn't melt...) and it came up with lots of errors. According to spice my cathode capacitor could source several hundred milliamps to charge a newly switched in coupling capacitor!

[woodyc]

Damn models!...

It's not really a spice problem. It's that the screen always having 7ma is inconsistent with the idea of the grid controlling the tube current. Maybe a more realistic scenario would be to have some kind of constant current with a ceiling on screen voltage? I assume the screen had to be pretty high voltage to pull 7ma with the grid way negative?

[Steve Conner]

In beam tetrode tubes, screen current is an error current, like base current in transistors. The magnitude of the current is quite small, and varies widely between tubes, because it depends critically on the alignment of the screen grid and control grid wires. (If the wires were perfectly aligned, it would be zero.)

Basing your circuit design on an error current is a bad idea, reminiscent of the dark ages when people would bias transistors with a single high value resistor to B+. (Or -9V since they would most likely be PNP germanium devices. )

In true pentodes, the screen current is larger and more predictable, but the idea still makes me uneasy.

Now, feeding the screens with a current source set to the current that corresponds to maximum safe screen dissipation, that would be a good idea. It would normally just peg the rail and act as a constant voltage source, but would transition to constant current and limit the screen current when needed.

[Merlinb]

If you used a perfect current source then you would get essentially no gain from the pentode (it's like having an infinitely large unbypassed cathode resistor). To get any gain you would have to bypass the screen with a capacitor. But that would defeat some of the advantage of current limiting, since the screen could still suck current pulses from the capacitor instead.

On the other hand, the capacitor wouldn't have to be very big, so maybe it would still be safe.

[Merlinb]

Thats really interesting. But I think you might be a little outside of the working limits of the model. In reality, a sufficiently negative grid should be able to shut off the current regardless of what is connected to the screen.

Not if the current source can produce an infinitely high voltage, which is what jazbo8's simulation can do!