Well this is probably a real stupid question -- but here goes.
As I look at datasheets for output tubes - they often list a dissipation value for the screen grid. Presuming that this dissipation is indeed amplified audio signal (screen acting as a "plate), are there uses for this signal (seems a shame to my tiny mind to waste this good stuff)?
I suppose that when a pentode is connected as a triode this signal gets "summed" into the plate signal. However, are there any other uses for it?? Could it be used for either local or global feedback? What would be the effect on the operation of the tube (bias and such) if one coupled this signal? Would it need to be cap coupled? And so on. thx.
As I look at datasheets for output tubes - they often list a dissipation value for the screen grid. Presuming that this dissipation is indeed amplified audio signal (screen acting as a "plate), are there uses for this signal (seems a shame to my tiny mind to waste this good stuff)?
I suppose that when a pentode is connected as a triode this signal gets "summed" into the plate signal. However, are there any other uses for it?? Could it be used for either local or global feedback? What would be the effect on the operation of the tube (bias and such) if one coupled this signal? Would it need to be cap coupled? And so on. thx.
Last edited:
Adding a resistor in series with the screen is local feedback.
If you look through the PP section of RDH4 you will see an output stage where the signal at the screen grid feeds the grid of the second tube, so the output stage does its own phase inversion.
If you look through the PP section of RDH4 you will see an output stage where the signal at the screen grid feeds the grid of the second tube, so the output stage does its own phase inversion.
g2 can be used like another control grid if you are careful. It's not high Z like g1 normally is. If the plate V falls below it, it will start drawing large non-linear currents (even destructive possibly). It gets used as a control grid sometimes for g2 drive "sweep" tubes where it seems to be more linear than g1 (but less sensitive). An important parameter is the tube's g2/g1 "Mu", this determines how effective the g2 grid is versus g1. gm2 = gm1/Mu. It sets the voltage gain ratio for triode configured pentodes, and for g2 feedback applications.
g2 can be used for feedback control in some situations, output stage UL being the most common. The trick there is to have the g2 grid roughly track at some % below the plate voltage ideally, then it looks like a constant impedance load (scrapes off a constant fraction of plate current). Otherwise, large non-linear variations in g2 current can set off oscillations if the g2 controller is not low Z. Usually a Mosfet source follower is called for to control g2.
UL or Schaded pentode driver stages can be made to work for example. However, one needs to be careful where the feedback is taken from also. For example, in a P-P setup, the output plates back to the driver screens is a natural feedback path (requiring less fdbk attenuation than to g1, and not loading down the driver's g1 input Z either). But due to inversion, the feedbacks have to be cross-coupled from the opposite output plates. In class AB, this means the feedback has to traverse the output xfmr when one output tube cuts off. Then you have phase issues thru the xfmr. A g2 Mosfet driver helps then to lower loading on the xfmr series leakage inductance.
Feedback from a CFB output stage cathode to the driver screen does not suffer inversion, so should be workable (even for SE). The driver screen will be varying less than the driver plate though in that setup, which might provoke oscillation or distortion if the driver plate V dips below the screen.
Then there are a number of circuits around like splitters and input stages where the g2 has been used as a DC only fdbk/bias control loop for DC coupled input stages. Some have used output screen current to trigger clipping LEDs.
One can also "return" the screen current to the plate with a HV Mosfet follower controlling the g2 voltage. The Mosfet's drain then gets connected to the plate. This only works if the plate V stays above the screen V however. Trickier circuits with bootstrapping caps can overcome that limitation. The tube then develops very square knee curves. Requires very high V Mosfets for xfmr loaded stages though. Some 1700 V IGBTs come in handy there.
Putting some resistance in series with the g2 (more than just a stopper) can be used to round off the knees of tube curves also. This could be used to remove some 2nd harmonic in a SE output stage. RH84 comes to mind.
Another trick is to use a floating power supply for the screen which floats with the cathode. That keeps screen currents out of the cathode circuit then (useful for curve tracers mainly).
g2 can be used for feedback control in some situations, output stage UL being the most common. The trick there is to have the g2 grid roughly track at some % below the plate voltage ideally, then it looks like a constant impedance load (scrapes off a constant fraction of plate current). Otherwise, large non-linear variations in g2 current can set off oscillations if the g2 controller is not low Z. Usually a Mosfet source follower is called for to control g2.
UL or Schaded pentode driver stages can be made to work for example. However, one needs to be careful where the feedback is taken from also. For example, in a P-P setup, the output plates back to the driver screens is a natural feedback path (requiring less fdbk attenuation than to g1, and not loading down the driver's g1 input Z either). But due to inversion, the feedbacks have to be cross-coupled from the opposite output plates. In class AB, this means the feedback has to traverse the output xfmr when one output tube cuts off. Then you have phase issues thru the xfmr. A g2 Mosfet driver helps then to lower loading on the xfmr series leakage inductance.
Feedback from a CFB output stage cathode to the driver screen does not suffer inversion, so should be workable (even for SE). The driver screen will be varying less than the driver plate though in that setup, which might provoke oscillation or distortion if the driver plate V dips below the screen.
Then there are a number of circuits around like splitters and input stages where the g2 has been used as a DC only fdbk/bias control loop for DC coupled input stages. Some have used output screen current to trigger clipping LEDs.
One can also "return" the screen current to the plate with a HV Mosfet follower controlling the g2 voltage. The Mosfet's drain then gets connected to the plate. This only works if the plate V stays above the screen V however. Trickier circuits with bootstrapping caps can overcome that limitation. The tube then develops very square knee curves. Requires very high V Mosfets for xfmr loaded stages though. Some 1700 V IGBTs come in handy there.
Putting some resistance in series with the g2 (more than just a stopper) can be used to round off the knees of tube curves also. This could be used to remove some 2nd harmonic in a SE output stage. RH84 comes to mind.
Another trick is to use a floating power supply for the screen which floats with the cathode. That keeps screen currents out of the cathode circuit then (useful for curve tracers mainly).
Last edited:
This is something that has been puzzling me for some time, being the newbie that I am. In say for example a PP Class A1 or AB1 amp running in pentode mode, what happens when the plate goes down to say 100V but the grid2 is still sitting near B+ via say a 1K resistor? I know I'm missing some basic point but I don't know what it is. Why doesn't the grid2 self-destruct?
Ian.
The 1k resistor limits the screen current. As the screen draws more current, the screen voltage falls to a safe level.
Datasheets usually show maximum dissipation for g2. Once it is exceeded, g2 will self-destruct.
You can also find Ig2(Va) curves and calculate the value of "screen stopper" resistor from those figures: it has to drop enough voltage when screen current goes up to keep the screen at safe level. The lower the screen voltage, the lower the screen current.
This resistor will drop some voltage during normal operation (= past the knee of anode curves) but this drop is relatively small as Ig2 is small. Once Ig2 skyrockets (with screen acting as anode and drawing pretty much all of cathode current), voltage drop across this resistor increases, steering tube clear of destruction.
Thanks Frank, but I still don't get it. Say with a little tube like a 6P1P, the screen is rated at 2.5W. At 2.5W the max voltage dropped over the 1K screen resistor is 50V, which puts it still 100V higher than the plate voltage given a B+ of 250 and that the plate voltage drops to 100V. That is, using Power = current squared times resistance, 2.5W = current squared x 1000, current = 50mA, and 50mA x 1000R = 50V drop. I am missing something fundamental...
Edit: oops missed your post Arnulf
Edit: oops missed your post Arnulf
Not really. If the tube was operated in such a way that the plate voltage was lower than the screen voltgae continuously AND the tubes characteristics were such that G2 drew enough current for it to be over dissipated, it would glow red, fail, and probably short to G1 or G3. Trust me I know! Fortunately this is rarely the normal situation.
Music has transients that are 10 to 25 db higher than the average signal. If you have a 100 watt amplifier cranked up to the point of distortion on signal peaks, the average power being produced by the amp is in the 2 to 10 watt range. Cranking a sine wave through this amplifier at the 100 watt level will subject the amp to more stress than typical music but the screen grid does not see an overload situation except on the sine wave peaks which are still only a small fraction of the total time. The dissipation rating for a screen grid is based on thermal effects (it will glow red hot if overloaded) which are time averaged over many complete audio cycles. If the dissipation, averaged over time does not exceed the spec, the tube will survive. Some tube specs quote an average G2 dissipation and a peak G2 input rating for this reason.
Over the years some amplifier designers have realized that music is less demanding than full power sine wave testing, and designed amplifiers to operate near the limit for music. These amplifiers may not survive full power sine wave testing. The early Carver solid state amps were an example of this. I have a Carver M-400 and this is clearly stated in the manual. There were plenty of tube amps that did the same thing but didn't carry the warning.
Most of the older EL84 amps that ran the tubes at 350 volts or more will play music just fine. Crank a sine wave through them and the screen grids will glow. Do it for ten minutes and the screen grids will blow!
I have a radio that uses the screen current as a signal for a self splitting PP arrangement. The voltage from the screen resistor on tube 1 feeds the grid of tube 2.
Thanks George, now I understand, and thanks to Frank and Arnulf. All very interesting posts in this thread. Often the "what if" scenarios help me to better understand how valves operate than the theory books.
Tubelab - The "bench perspective" you bring is both a comfort *and* inspiration to .... 'boldy go where no man has gone ...'
thx much!
Smoking amp -- thanks much for this terrific bag of 'tricks' - just what I was after --thx much! 😀
- Status
- Not open for further replies.