Tweet
Hi everyone. This is my first post here on Ampage I think. I hope that this is an appropriate place to discuss my various ideas and projects - I think it probably is, I've been reading the old posts. I'm somewhat of a hobbyist - I've designed and built quite a few guitar amps in my life, some transistor, some tube, some of them even work and see regular gigging and practice. I've also done a few effect projects and I do the occasional repair for people. So that's why I'm here, in the hope of meeting like-minded correspondents.
Anyway, I've had this idea for building a 'tubelike' MOSFET output stage using an autotransformer type of output topology, to obtain similar electrical characteristics to the classic push-pull transformer output stage with conveniently available components in a solid-state design.
Suppose I just tried to put FETs instead of tubes, I could probably do that:
Now I'd have to find an appropriate audio-band transformer to appropriately match the voltage and current swings. Hard enough to find tube transformers, really. Probably a 25V or 70V line transformer could be made to do the job. But as we're talking low-impedance devices (FETs) and low-impedance load (loudspeaker), a transformer here seems somewhat gratuitous. Could I get by with just a choke? Chokes for parafeed triode amplifiers and the like have to be rated in the Henries, because they work in high-impedance tube circuits. In a low-impedance solid state circuit, I could get by with millihenries, much easier to find or make. So here's what I'd do with a center-tapped choke load replacing the transformer:
Sort of looks like 2 choke-loaded single-ended stages in a bridge conformation - the only difference is that the 2 chokes are coupled into one. This allows the stage to be driven into class AB as a classical transformer-output stage could be.
What's interesting is that, when I SPICE it, this circuit can actually approach putting 4 x B+ peak-to-peak into the load! Could think of it as the same phenomenon wherein a single-ended choke-loaded stage can swing nearly 2 x B+ peak-to-peak, but it's really an autotransformer stepping up the voltage and stepping down the current. The transistors actually pull twice the current as the load gets, but at a lower voltage. This is kind of cool, really, because the less current the choke has to carry, the cheaper it can be. Very much like how 120:240 line 'transformers' sold for travelers are all really autotransformers. It's not a conspiracy to deny us cheap B+ transformers although I used to think so, it's that the autotransformer is cheaper because it carries less current. The 'primary' and 'secondary' currents circulate in different directions and partially cancel as shown here:
If I were to really build this, I'd definitely have to make it a bit more complicated. MOSFETs are different from tubes in that they do not have an analog of the zero-bias line where the grid starts to draw current, providing a sort of maximum current limit for the tube when driven by a high-impedance driver circuit. Adding a small source resistor to 'sense' the current flowing and a small signal NPN transistor lets a current limit be set for the MOSFETs and even causes 'grid current' to be drawn from the phase inverter when driven to this point of clipping, just as a power tube would:
It might also be a decent idea to put a couple fast diodes in reverse between the FET drains and ground, to catch any flyback from running this thing loud with no load or such situations.
I guess I wonder - would this work? And would parts be easily available? I don't think the choke would be a huge problem - a centertapped 40 millihenry choke would have an impedance of 20 ohms at 80 Hz (guitar low E) and while I haven't seen such a choke for sale I imagine it'd be pretty much equivalent to a common mode dual 10 mH choke, which I have seen for sale. I do wonder where to get FETs - they'd have to be somewhat matched for this to work well and it seems that the common switching FETs don't do (although some people seem to use them anyway) - I'm a bit confused as to what's really out there for audio MOSFETs in terms of price and availability. But I think this might work. SPICE simulations show a somewhat tube-like power transfer, with ideal power transfer at some load impedance rather than the perfectly damped output of the traditional transistor amp. Magnetic saturation might well show up, with a somewhat undersized choke. And the current limiting transistors would allow some simulation of grid current, with corresponding bias shifts being possible from charging of coupling caps from the PI. The step-up of the B+ voltage is pretty nice, allows consideration of a fairly low rail voltage and readily available power supplies.
Has this topology seen any use in the past? It seems pretty simple. I'm interested in hearing what people here might think of it.
Anyway, I've had this idea for building a 'tubelike' MOSFET output stage using an autotransformer type of output topology, to obtain similar electrical characteristics to the classic push-pull transformer output stage with conveniently available components in a solid-state design.
Suppose I just tried to put FETs instead of tubes, I could probably do that:
Now I'd have to find an appropriate audio-band transformer to appropriately match the voltage and current swings. Hard enough to find tube transformers, really. Probably a 25V or 70V line transformer could be made to do the job. But as we're talking low-impedance devices (FETs) and low-impedance load (loudspeaker), a transformer here seems somewhat gratuitous. Could I get by with just a choke? Chokes for parafeed triode amplifiers and the like have to be rated in the Henries, because they work in high-impedance tube circuits. In a low-impedance solid state circuit, I could get by with millihenries, much easier to find or make. So here's what I'd do with a center-tapped choke load replacing the transformer:
Sort of looks like 2 choke-loaded single-ended stages in a bridge conformation - the only difference is that the 2 chokes are coupled into one. This allows the stage to be driven into class AB as a classical transformer-output stage could be.
What's interesting is that, when I SPICE it, this circuit can actually approach putting 4 x B+ peak-to-peak into the load! Could think of it as the same phenomenon wherein a single-ended choke-loaded stage can swing nearly 2 x B+ peak-to-peak, but it's really an autotransformer stepping up the voltage and stepping down the current. The transistors actually pull twice the current as the load gets, but at a lower voltage. This is kind of cool, really, because the less current the choke has to carry, the cheaper it can be. Very much like how 120:240 line 'transformers' sold for travelers are all really autotransformers. It's not a conspiracy to deny us cheap B+ transformers although I used to think so, it's that the autotransformer is cheaper because it carries less current. The 'primary' and 'secondary' currents circulate in different directions and partially cancel as shown here:
If I were to really build this, I'd definitely have to make it a bit more complicated. MOSFETs are different from tubes in that they do not have an analog of the zero-bias line where the grid starts to draw current, providing a sort of maximum current limit for the tube when driven by a high-impedance driver circuit. Adding a small source resistor to 'sense' the current flowing and a small signal NPN transistor lets a current limit be set for the MOSFETs and even causes 'grid current' to be drawn from the phase inverter when driven to this point of clipping, just as a power tube would:
It might also be a decent idea to put a couple fast diodes in reverse between the FET drains and ground, to catch any flyback from running this thing loud with no load or such situations.
I guess I wonder - would this work? And would parts be easily available? I don't think the choke would be a huge problem - a centertapped 40 millihenry choke would have an impedance of 20 ohms at 80 Hz (guitar low E) and while I haven't seen such a choke for sale I imagine it'd be pretty much equivalent to a common mode dual 10 mH choke, which I have seen for sale. I do wonder where to get FETs - they'd have to be somewhat matched for this to work well and it seems that the common switching FETs don't do (although some people seem to use them anyway) - I'm a bit confused as to what's really out there for audio MOSFETs in terms of price and availability. But I think this might work. SPICE simulations show a somewhat tube-like power transfer, with ideal power transfer at some load impedance rather than the perfectly damped output of the traditional transistor amp. Magnetic saturation might well show up, with a somewhat undersized choke. And the current limiting transistors would allow some simulation of grid current, with corresponding bias shifts being possible from charging of coupling caps from the PI. The step-up of the B+ voltage is pretty nice, allows consideration of a fairly low rail voltage and readily available power supplies.
Has this topology seen any use in the past? It seems pretty simple. I'm interested in hearing what people here might think of it.
Comment