You should be able to "tune" a "trioded" Mosfet device (trioded = Drain Fdbk) by putting a source resistor Rs in series with the device source terminal. With high Rs, the device is near linear. With no Rs the device is square law. With Rs = 1/gm the device should be about half way between a square law device and a linear device. Ie, a 3/2 law device, like a tube. Of course gm will be halved around there with that Rs addition.
------------------------------------------------------------------------------------------------------------------
The real tube however still reigns supreme in a modified tube circuit that seems to be unknown:
You take a pentode and put in the resistors for Crazy Drive. (R1 from grid2 to grid1, 4K say. And R2 from grid1 to cathode, 4K say. Also try R2 from grid1 to Gnd or some - Bias V, may simplify biasing. ) Then we add UnSET to it. Which is resistive Schade Fdbk plus a P channel follower to drive the cathode. The resistive Schade would be a high value resistor from plate to grid2 in this case. (normally to grid1 for Unset) Say 100K from plate to grid2. And then a P channel follower goes under the cathode to drive it. With everything combined (some R value tweaking for the specific tube likely), you can get linear mode triode curves. Or a "Wire with Gain". Here is a curve trace: (been mentioned several times, but never gets any interest)
Here is Crazy Drive alone:
And here is UnSET alone:
------------------------------------------------------------------------------------------------------------------
The real tube however still reigns supreme in a modified tube circuit that seems to be unknown:
You take a pentode and put in the resistors for Crazy Drive. (R1 from grid2 to grid1, 4K say. And R2 from grid1 to cathode, 4K say. Also try R2 from grid1 to Gnd or some - Bias V, may simplify biasing. ) Then we add UnSET to it. Which is resistive Schade Fdbk plus a P channel follower to drive the cathode. The resistive Schade would be a high value resistor from plate to grid2 in this case. (normally to grid1 for Unset) Say 100K from plate to grid2. And then a P channel follower goes under the cathode to drive it. With everything combined (some R value tweaking for the specific tube likely), you can get linear mode triode curves. Or a "Wire with Gain". Here is a curve trace: (been mentioned several times, but never gets any interest)
Here is Crazy Drive alone:
And here is UnSET alone:
Last edited:
I've searched for that UnSET a couple times and never have seen schematics or an amplifier design using them. EDIT: Over in the tube lab forum I see, not sure if there's any other examples of people implementing it, I'd be curious.
I had messed around with Gary Pimm's HV cascoded mosfet amplifier stuff in the past but have had too many stability problems with current production mosfets.
I had messed around with Gary Pimm's HV cascoded mosfet amplifier stuff in the past but have had too many stability problems with current production mosfets.
Last edited:
The UnSET and related stuff is over on Tubelab's (George's) forum. The idea originated ( I think, at least for me ) from the realization that ordinary Resistive Schade Fdbk (resistor from plate to grid1, and the input current drive on grid1 ) has a flaw in it. The resulting grid1 voltage variation corrupts the Schade resistor Fdbk current. Resulting in cramped triode curves at the HV end. This can be fixed conveniently in two ways. A grounded gate P Channel Mosfet can be placed at the bottom end of the Shade Fdbk resistor to convert the N Fdbk to isolated current at the Drain end for grid1 (so un-affected by grid1 V variation there) OR one can just move the drive signal to the cathode instead, providing isolation. (with an R divider resistor added from grid1 to gnd )
That also requires a P channel Mosfet to drive the cathode, but in V follower fashion. The tube gm still determines cathode current, so tube authenticity is still preserved. This also has the effect of effectively increasing tube B+ when the cathode is pulled negative. So power output is increased (sum of tube power plus driver power ) Clearly of some interest to George.
The lowish screen V for TV Sweep tubes can be accommodated as well, unlike the typical triode wiring of pentodes.
Now called UnSET and various derivatives for P-P too. I think George has a universal PC board under development to use the UnSET like arrangement for a multiplicity of tube types and for each stage of an amplifier. I probably would simplify to just using it for the output tubes for homebrew. The triode curves obtained from cheap pentodes can be very high quality using modest B+ voltages. (depends on N Fdbk level in the local Schade loop )
That also requires a P channel Mosfet to drive the cathode, but in V follower fashion. The tube gm still determines cathode current, so tube authenticity is still preserved. This also has the effect of effectively increasing tube B+ when the cathode is pulled negative. So power output is increased (sum of tube power plus driver power ) Clearly of some interest to George.
The lowish screen V for TV Sweep tubes can be accommodated as well, unlike the typical triode wiring of pentodes.Now called UnSET and various derivatives for P-P too. I think George has a universal PC board under development to use the UnSET like arrangement for a multiplicity of tube types and for each stage of an amplifier. I probably would simplify to just using it for the output tubes for homebrew. The triode curves obtained from cheap pentodes can be very high quality using modest B+ voltages. (depends on N Fdbk level in the local Schade loop )
Last edited:
In reality the whole UNSET thing came from an attempt to remove the negative voltage supply needed in fixed bias tube amps. Somewhere about 8 to 10 years ago I made a simple test board with a 12 pin compactron socket with a Pch / Nch mosfet pair on the cathode, control grid, and screen grid with the capability to sink or source current through each of the three pins while simultaneously applying in drive from either phase output of a push pull driver board. These experiments were all single ended, but the first two working amplifiers were both push pull. One squeezed 20 watts from a pair of 50C5 radio tubes in a guitar amp design. The other got 50 to 80 watts from a pair of 6GF5 sweep tubes. The 6GF5 is a shrunken 6GE5 which is a 6DQ6 stuffed into a compactron bottle without a plate cap. They were once on the $1 list, so I got a bunch.
I did create an UNSET output board that can be used for a single channel push pull amp or a stereo SE amp. I have boards for the octal tubes, and have a layout done for 12 pin Compactrons but have not sent it out for fab yet. So far, I have seen a single pair of 26HU5 tubes hit 250 watts and two pair make well over 500 watts.both of these test amps used the Universal Driver Board feeding one or two octal UNSET output boards.
I also made an all mosfet SE amp that makes 20 watts in SE through an OPT on 250 to 300 volts. The board layout is about half done.
I am typing this from a hotel room in Tennessee, so I don't have access to my main computer for schematics, data, and pictures. I should be back Sunday but will be met with an acre that has not been mowed in two weeks.
I did create an UNSET output board that can be used for a single channel push pull amp or a stereo SE amp. I have boards for the octal tubes, and have a layout done for 12 pin Compactrons but have not sent it out for fab yet. So far, I have seen a single pair of 26HU5 tubes hit 250 watts and two pair make well over 500 watts.both of these test amps used the Universal Driver Board feeding one or two octal UNSET output boards.
I also made an all mosfet SE amp that makes 20 watts in SE through an OPT on 250 to 300 volts. The board layout is about half done.
I am typing this from a hotel room in Tennessee, so I don't have access to my main computer for schematics, data, and pictures. I should be back Sunday but will be met with an acre that has not been mowed in two weeks.
Thanks to both of you, that answers better the question about the theory behind it that I hadn't properly expressed.
The reason this thread caught my eye was I was thinking about building Pete Millet's e-linear that uses the UL tap as the supply for the driver, to provide feedback that way. Since d3a are harder to source I've been curious about other interesting configurations for feedback for pentodes or unusual ss stages. At one point I was considering Gary Pimm's ccs-as-ss-pentode for a first stage in e-linear but the aforementioned stability problems made me pump the brakes.
The reason this thread caught my eye was I was thinking about building Pete Millet's e-linear that uses the UL tap as the supply for the driver, to provide feedback that way. Since d3a are harder to source I've been curious about other interesting configurations for feedback for pentodes or unusual ss stages. At one point I was considering Gary Pimm's ccs-as-ss-pentode for a first stage in e-linear but the aforementioned stability problems made me pump the brakes.
I did try my hand at a few solid state amplifiers decades ago.
The Southwest Technical Products 'Little Tiger' had terrible crossover distortion.
I used a JFET Op Amp, and applied negative feedback. That fixed the crossover distortion.
A built a Dyna SCA80 stereo amplifier using 2N3055 quasi complimentary outputs. That worked well.
Then there was a 12, or 14 watt Heathkit stereo amp using TO-66 outputs. That also worked well.
I wanted to make a working MOSFET Source follower amplifier.
A tube drove the gate, and the Drain was cap bypassed to ground.
The Source drove a very low DC resistance, medium inductance choke, and the output's DC offset voltage was very low.
The problem was that no matter what resistor, or ferrite was added to the gate, source, drain . . . the MOSFET was unstable.
Project abandoned.
Then, I remembered the vacuum tube radios and amplifiers I built in the late 50s and through the 60s.
So about 1995, I went back to vacuum tubes, and I never returned to solid state.
I limit my use of solid state parts to CCS, and B+ rectifier diodes.
The Southwest Technical Products 'Little Tiger' had terrible crossover distortion.
I used a JFET Op Amp, and applied negative feedback. That fixed the crossover distortion.
A built a Dyna SCA80 stereo amplifier using 2N3055 quasi complimentary outputs. That worked well.
Then there was a 12, or 14 watt Heathkit stereo amp using TO-66 outputs. That also worked well.
I wanted to make a working MOSFET Source follower amplifier.
A tube drove the gate, and the Drain was cap bypassed to ground.
The Source drove a very low DC resistance, medium inductance choke, and the output's DC offset voltage was very low.
The problem was that no matter what resistor, or ferrite was added to the gate, source, drain . . . the MOSFET was unstable.
Project abandoned.
Then, I remembered the vacuum tube radios and amplifiers I built in the late 50s and through the 60s.
So about 1995, I went back to vacuum tubes, and I never returned to solid state.
I limit my use of solid state parts to CCS, and B+ rectifier diodes.
Well, that one 😄, because it has a dual gate which can be used to simulate a triode internal NFB.
But not 99.998% of Fets or MosFets out there, including the new darling being discussed here: LND150 😫
1. I was pretty sure that the dual gate MOSFETs now are either Un-Obtain-ium, Obsolete-ium, Rareified-ium, or only found in Museums.
2. In most UL transformers, the B+ voltage to the screens is higher than the B+ to the plates (DCR x current = Voltage drop).
3. Like all transformer windings (especially if they are not bi-filer), there is leakage inductance between the UL Tap to the Plate Tap.
2. In most UL transformers, the B+ voltage to the screens is higher than the B+ to the plates (DCR x current = Voltage drop).
3. Like all transformer windings (especially if they are not bi-filer), there is leakage inductance between the UL Tap to the Plate Tap.
Last edited: