Connecting a capacitor parallel with R14 results in to much gain swing provided to M1 (over-compensates).
The only way I see of sziklai pairing the N-chan fet results in to much voltage loss due to gate-to-source and would need boot-strapped somehow 🤔
The only way I see of sziklai pairing the N-chan fet results in to much voltage loss due to gate-to-source and would need boot-strapped somehow 🤔
I have this running in sim, seems to work well with a heavier bias, but R455 and R457 are very sensitive to current adjustment 😱
In the original circuit, instead of R455, I used two diodes. C163 is not needed in this case
PS We are slowly but surely approaching JLH 🙂
PS We are slowly but surely approaching JLH 🙂
It seems to me that resistor R14 in the circuit from post #19 should be a trimmer. It needs to be tuned to minimize amplifier distortion.
If this is confirmed, it will be inconvenient when repeating the design, since it requires certain equipment to carry out measurements.
If this is confirmed, it will be inconvenient when repeating the design, since it requires certain equipment to carry out measurements.
First the BJT version, current is now set to 1A:
FFT 12Vp-p (0.012%):
FFT 20Vp-p (0.021%):
The design seems quite tolerant of different transistors and seems to work well with a nice 2nd distortion profile.
It put's out 1.5A while the standing idle current is 1A. Is there any way to bring down the current and improve efficiency?
though it will likely make a good headphone amplifier 👍
I will need to buy the output power transistors so then I will be able to test.
FFT 12Vp-p (0.012%):
FFT 20Vp-p (0.021%):
The design seems quite tolerant of different transistors and seems to work well with a nice 2nd distortion profile.
It put's out 1.5A while the standing idle current is 1A. Is there any way to bring down the current and improve efficiency?
though it will likely make a good headphone amplifier 👍
I will need to buy the output power transistors so then I will be able to test.
Thanks steveu I can see that now 👍 it will be interesting to see how this handles in real life. Do you think is worth testing as it is now? I just hope I don't build another big oscillator, kinda find those boring 😛 so I just need output transistors. maybe 2n/tip 3055 in TO-247?
Be sure you heat sink the OP's well. 16Watts is a lot of heat and without a heat sink, even TO-247's will melt/short in a couple seconds. I have burnt myself on resistors running ~10 Watts.
I have some reasonable size sinks I use for testing and PC fans work well too if needed. Its usually the load resistor that catches me out 🙄
Your scheme is very similar to the one described in patent US6434243. Perhaps the information provided in the patent will allow you to improve the parameters
This illustrates the problem with Patents. This patent is from 1998 but I first saw the idea in a Canadian "Electron" magazine about 1970. The idea is part of what is used in the circuit I shared in post #14, but with a couple a couple improvements and simulated.
Good find! yes it is very similar, just flipped..
"Let the quiescent current be set at one tenth of the peak output current, so for an output of 3 amps the quiescent is 300 mA giving R1=2 ohms. As R2 will drop 6 volts at 3 amps the supply will need to be 30 volts to be able to drive an 8 ohm speaker at 36 watts. The dissipation at quiescent is therefore only 18 watts compared with 144 watts for a conventional class A circuit."
Yes please! I'm not sure if the power savings could be that good though, more reading is needed...
"Let the quiescent current be set at one tenth of the peak output current, so for an output of 3 amps the quiescent is 300 mA giving R1=2 ohms. As R2 will drop 6 volts at 3 amps the supply will need to be 30 volts to be able to drive an 8 ohm speaker at 36 watts. The dissipation at quiescent is therefore only 18 watts compared with 144 watts for a conventional class A circuit."
Yes please! I'm not sure if the power savings could be that good though, more reading is needed...
Very interesting. Can you post this circuit as image?
I have seen names like "economy class A" for such circuits. They are usually characterized by a higher quiescent current than class AB and have increased efficiency compared to pure class A. Some kind of intermediate position in the classification.