Best gurus:
Both Gary Pimm and J.C. Morrison have posted schematics of CCS assisted pentodes with the load resistor in parallel with the tube.
I got to wondering: could the load resistor be tapped, to supply both screen voltage and ultra-linear negative feedback?
What do you think? Regards, Bill
Both Gary Pimm and J.C. Morrison have posted schematics of CCS assisted pentodes with the load resistor in parallel with the tube.
I got to wondering: could the load resistor be tapped, to supply both screen voltage and ultra-linear negative feedback?
What do you think? Regards, Bill
The voltage divider does lower the screen voltage substantially, but as the plate voltage fluctuates I fancy that the screen would fluctuate about half as much, as if it were connected to an ultra-linear output transformer tap - providing about 50% NFB as opposed to the 100% of a triode.
Your connection is somewhat different than ultra-linear, since it is dividing from plate to ground, while UL mode divides from plate to B+. In UL mode the plate V can drop below the screen V, causing significant non-linear screen current variation.
In your scheme, the plate V will always stay above the screen V. Since the screen also draws some current, it will alter the divide ratio from that of the resistors alone. Since the screen V is approx a constant fraction of the plate V, it hopefully will draw near a fixed fraction of the plate current, so the screen may behave close to a constant resistance, which is good, since it will maintain the R divider linear (but you have to figure in all three resistances to get the divide ratio).
A variant, which has been around, is to use an N type Mosfet follower from off the plate (plate to gate), with the R divider going to ground from its Source terminal. That provides a low impedance drive for the next stage from the Source, and a lower resistance R divider can now be used as well for feeding the screen. It gives quite linear results, since the plate is unloaded (CCS) and the loop gain is high back to the screen. Essentially it gives CCS loaded triode results, with no plate loading, and a low output impedance. Best of all worlds. One could of course use a tube follower instead, if SS phobic, but most of the capacitance issues of Mosfets are handled well in a follower mode.
In your scheme, the plate V will always stay above the screen V. Since the screen also draws some current, it will alter the divide ratio from that of the resistors alone. Since the screen V is approx a constant fraction of the plate V, it hopefully will draw near a fixed fraction of the plate current, so the screen may behave close to a constant resistance, which is good, since it will maintain the R divider linear (but you have to figure in all three resistances to get the divide ratio).
A variant, which has been around, is to use an N type Mosfet follower from off the plate (plate to gate), with the R divider going to ground from its Source terminal. That provides a low impedance drive for the next stage from the Source, and a lower resistance R divider can now be used as well for feeding the screen. It gives quite linear results, since the plate is unloaded (CCS) and the loop gain is high back to the screen. Essentially it gives CCS loaded triode results, with no plate loading, and a low output impedance. Best of all worlds. One could of course use a tube follower instead, if SS phobic, but most of the capacitance issues of Mosfets are handled well in a follower mode.
Last edited:
Feeding a screen from a high impedance can introduce re-entrant distortion.
I'm not sure what the point of this circuit is. It may have slightly higher gain than triode-strapping, but it will also have higher output impedance.
I'm not sure what the point of this circuit is. It may have slightly higher gain than triode-strapping, but it will also have higher output impedance.
It is mainly a flight of idle fantasy. If it has any reason to exist, it would be the low B+, and the short signal loop.
Another possible way to try this idea out...
Setup the circuit with the plate resistor going to ground and the CCS supplying all the current for the plate resistor and pentode. Take the signal output from the MU output. Connect the voltage divider for the screen to the MU output of the CCS.
This will do several things-
You will get the high gain offered by the CCS assisted pentode configuration.
Taking the signal from the MU output provides low driving impedance for the next stage.
The CCS will provide very high power supply isolation.
The screen divider can be lower impedance and won't effect the gain of the circuit.
As an additional bonus, the current drawn by the screen divider lowers the output impedance of the MU ouput.
Gary
Setup the circuit with the plate resistor going to ground and the CCS supplying all the current for the plate resistor and pentode. Take the signal output from the MU output. Connect the voltage divider for the screen to the MU output of the CCS.
This will do several things-
You will get the high gain offered by the CCS assisted pentode configuration.
Taking the signal from the MU output provides low driving impedance for the next stage.
The CCS will provide very high power supply isolation.
The screen divider can be lower impedance and won't effect the gain of the circuit.
As an additional bonus, the current drawn by the screen divider lowers the output impedance of the MU ouput.
Gary
I made a couple of LT Spice simulations with 6SJ7. The DC operating point and g2 current is a bit different than in original schematic.
The gain, output voltage and THD is shown at both schematics.
It seems that triode connection gives higher gain, obviously due to higher andode current, but it is also more linear.
Fig. 1 6SJ7 UL
Fig. 2 6SJ7 triode
The gain, output voltage and THD is shown at both schematics.
It seems that triode connection gives higher gain, obviously due to higher andode current, but it is also more linear.
An externally hosted image should be here but it was not working when we last tested it.
Fig. 1 6SJ7 UL
An externally hosted image should be here but it was not working when we last tested it.
Fig. 2 6SJ7 triode
Thanks to all for your thoughts, effort and interest. Here's another concept that crept into my mind:
If the load resistor was terminated at the cathode instead of ground, all current would pass through the bias resistor - effectively locking in the bias voltage, just as if we had placed a huge capacitor across it?
If the load resistor was terminated at the cathode instead of ground, all current would pass through the bias resistor - effectively locking in the bias voltage, just as if we had placed a huge capacitor across it?
Last edited:
You now have a small amount of positive feedback, which will partly counteract the negative feedback to g2 and cathode degeneration. The effect will be to stabilise the voltage at the cathode, but you now have to arrange a floating load - easy if it really is just a resistor; hard if it is really the next stage.
If your aim is stable bias then why not use an LED or bypass cap?
If your aim is stable bias then why not use an LED or bypass cap?
i still can't see a image, but, if i read correctly, it's named "ultrapath circuit" where the load does not go to circuit ground but to the cathode, by Western Electric.
As soon as you add an output the balance is disturbed, unless the load has infinite impedance. It turns out that this is exactly the same situation as you would have without your anode-cathode resistor, so all it does is increase the gain and distortion by a small amount while wasting a little power.
Wouldn't the CCS force the current through the bias resistor to be constant? From signal ground, current would flow thru the bias resistor and then A) cathode to plate, B) cathode to screen (not drawn), and C) through the shunt resistor, which I perhaps too hurriedly labeled "load", the equivalent of the plate resistor in a conventional circuit. All three paths going to the CCS have the bias resistor in common, so their currents should sum through it.
- Status
- Not open for further replies.