Just some fast calculation:
The peak output voltage is 75V, the peak load current is 19A. It means 1,9A/output device. The Vds can be something like 5-6V, the driver transistor needs some 0.7V Vbe, that is ~7V, so the driver stage must provide 82Vpeak.
Using 85V rails the output looks realistic.
Anyway I see some strange things, such as one output device is not connected on the posistive side, and I would mount the Vbe mulitplier on the heatsink for thermal compensation.
Sajti
The peak output voltage is 75V, the peak load current is 19A. It means 1,9A/output device. The Vds can be something like 5-6V, the driver transistor needs some 0.7V Vbe, that is ~7V, so the driver stage must provide 82Vpeak.
Using 85V rails the output looks realistic.
Anyway I see some strange things, such as one output device is not connected on the posistive side, and I would mount the Vbe mulitplier on the heatsink for thermal compensation.
Sajti
The schematic (#2) shows R17 and R18 connected to the bases of Q12 and Q14, instead of to the emitters. FET gates do not draw any DC current.
Besides that, you need a capacitor, about 10uF, between the emitters of Q12 and Q14. That may solve your instability. FETs need to be driven both on and off. They do not turn off by themselves.
Besides that, you need a capacitor, about 10uF, between the emitters of Q12 and Q14. That may solve your instability. FETs need to be driven both on and off. They do not turn off by themselves.
That schematic was drew wrong
here's my latest schematic :
New2.png - Google Drive
So as above schematic shows I short Circuit those 1N4148 Diodes and Now I get much cleaner output. I also tried replacing them with 1N4007 & MUR460 and even with a 82 ohms resistor to form a 100 ohms resistors and there were no difference in output signal.
Here's a shot of current output:
IMG_20190410_194055.jpg - Google Drive
PS
I used that 10uf cap across driver's emitters but it made no difference
Last edited:
You could have grounding problems: I would connect the feedback decoupling caps C3,C9 to the other side of R43 instead of directly to ground, and/or be sure C3,C9 connects to the R43 ground before both are connected to the power supply ground.
The output of this amp is a follower, not a quasi circuit, so the RC network R38+C6 may be not required and could even cause problem ground currents. In any case C6 should have it's own ground wire to the power supply.
Q10 can be unstable without a series base resistor (100-1K) or a CE shunt capacitor (about 10nF).
I would cross couple the drivers by using a single resistor between driver emitters instead of R17, R18, and use a cap as I mentioned before.
470 Ohm gate resistors is a bit high and can cause unnecessary delay and therefore feedback instability. I would use about 100 Ohms, but not less than 10 Ohms. Since you have 20 such resistors, that would be a last resort.
Good luck
The output of this amp is a follower, not a quasi circuit, so the RC network R38+C6 may be not required and could even cause problem ground currents. In any case C6 should have it's own ground wire to the power supply.
Q10 can be unstable without a series base resistor (100-1K) or a CE shunt capacitor (about 10nF).
I would cross couple the drivers by using a single resistor between driver emitters instead of R17, R18, and use a cap as I mentioned before.
470 Ohm gate resistors is a bit high and can cause unnecessary delay and therefore feedback instability. I would use about 100 Ohms, but not less than 10 Ohms. Since you have 20 such resistors, that would be a last resort.
Good luck
Hey guys
thanks again for all your helps
I test this amp with 80 & 100 Hz signal for more that an hour at 800W output power with a 3.6 ohms resistor ( dummy load) and everything was fine
I wondor if this is the worst situation that I could test my amplifier with?
this amp will power up a subwoofoer and there's a low pass filter at the input with 80Hz cut-off frequency
thanks again for all your helps
I test this amp with 80 & 100 Hz signal for more that an hour at 800W output power with a 3.6 ohms resistor ( dummy load) and everything was fine
I wondor if this is the worst situation that I could test my amplifier with?
this amp will power up a subwoofoer and there's a low pass filter at the input with 80Hz cut-off frequency
18 years ago....
have start DIY with friend and similar problem with oscillation IRFP9240 / 240
to fix it we have cut all Mosfet Gate Pin up to 3mm from Mosfet casing,
- soldering gate resistor (cut resistor Pin to 2 mm not more) direct at gate Pin from Mosfet housing
finished, poblem with oscillation is solved
try it and give feedback
have start DIY with friend and similar problem with oscillation IRFP9240 / 240
to fix it we have cut all Mosfet Gate Pin up to 3mm from Mosfet casing,
- soldering gate resistor (cut resistor Pin to 2 mm not more) direct at gate Pin from Mosfet housing
finished, poblem with oscillation is solved
try it and give feedback
after fix oscillation problem, amp run many years in Disco PA application +/- 86 V Rail
be careful this design will running very hot
This mosfet amp is much better und running not very hot @ 90 V DC
Another quasi-complementary design
be careful this design will running very hot
This mosfet amp is much better und running not very hot @ 90 V DC
Another quasi-complementary design
Thanks for your help
I will do that and report back
I know about the heating and there's 3 fans on each rail's heatsink
I test this amp with full power (800W RMS) with 100Hz sine and with all fans running the temprature on mosfet's cases didn't go higher than 65'c (the test run non stop for 3 Hours)
IMG_20190426_224623.jpg - Google Drive
Last edited:
I'm planning to build a class D amp as well
can you recommend any working and reliable schematic?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.