The input impedance to the output state is high except for the capacitive load of the output FETs. In simulations, adding a 100R resistor between the FE and the OS increases the output distortion, increasing with frequency. It will also vary with the choice of FETs.
I figured out why the FQA28N15/FQA36P15 FETs have such low distortion in my output stage. They are nearly perfect compliments and exhibit a high degree of square-law cancellation.
Dumb luck! By accident, I chose these 4 FETs at random from a small collection of various FETs that had be used in previous projects. Now, looking at the measurements data from about 10 years ago, these FETs had apparently been matched for transconductance at 1.25A.
Now I have made measurements of 37 each FQA28N15 and FQA36P15, and figured out how to match them for best harmonic cancellation. Out of the lot size of 37 each, I can get about 8 very good pairings, enough for 4 stereo amplifiers. If the matching specs are relaxed a bit I can get more channels.
I found a source for more of these FETs and will purchase them if enough DIYers are interested in building this amp. I would offer the FETs and the PCB.
I do not know how many are interested. I do not know how long these FETs will be available.
Dumb luck! By accident, I chose these 4 FETs at random from a small collection of various FETs that had be used in previous projects. Now, looking at the measurements data from about 10 years ago, these FETs had apparently been matched for transconductance at 1.25A.
Now I have made measurements of 37 each FQA28N15 and FQA36P15, and figured out how to match them for best harmonic cancellation. Out of the lot size of 37 each, I can get about 8 very good pairings, enough for 4 stereo amplifiers. If the matching specs are relaxed a bit I can get more channels.
I found a source for more of these FETs and will purchase them if enough DIYers are interested in building this amp. I would offer the FETs and the PCB.
I do not know how many are interested. I do not know how long these FETs will be available.
Sometimes a little bit of distortion (the right kind) can be a fun thing.
I have a few FQA28N15 on their way, and a small tube of IXTQ36P15 already on hand. There are other possible pairings with other parts.
What would be really valuable at this point would be a description of the test jig and the parameters to test.
Channel boards of course would be convenient as well.
I have a few FQA28N15 on their way, and a small tube of IXTQ36P15 already on hand. There are other possible pairings with other parts.
What would be really valuable at this point would be a description of the test jig and the parameters to test.
Channel boards of course would be convenient as well.
The H2/H3 ratio can be changed by how the FETs are matched. The THD can be changed by the amount of global feedback used, without significantly changing the shape of the spectrum.
I would be matching them. There are there are alternative ways to deploy the output stage:
With or without global feedback. Most of my measurements and listening has been done with 12dB global feedback.
There has been one ceramic capacitor added to the schematic of post #21. The some parameter values depend on rail voltages, choice of output FETs, amount of global feedback, closed-loop gain, and other.
Are you going linear or SMPS for the PSU?Interested too, have the chassis and power supply ready and waiting