I've been working on an old International Rectifier design from Ap Note AN-948.
Link: http://www.cieri.net/Documenti/Altr...ifier Using Complementary HEXFETs (AN948).pdf
The following semi's were used. Q1, Q2 ; 2N5087. Q4 ; MPSA06. Q3 ; PN2222A. Q5 ; IRF630 Q6 ; IRF9630
The following changes to the schematic: R10; 220 Ohms, added gate stopper to Q6, 220 Ohms.
Source resistors added to Q5 and Q6, 0.22 Ohms 5W.
Q5 and Q6 are on a common heatsink with Q3 held face down between them. Initially Q3 was not touching the heatsink. At 28.3C bias was set to 0.45V across the two source resistors, ~ 102mA. At 56.5C the bias current was 68mA. After I made a little clamp to hold Q3 to the heatsink, the change in bias was worse. At 31.5C bias was 116mA and at 56.8C it shrunk to 47.7mA.
Looking into things further, the Q6 gate to Q5 gate Voltage was 7.83V at 100mA (at 28.8C) and shrunk to 7.13V at 43.2mA and 53.3C. While the heatsink was at 53.3C the bias was adjusted to 100mA and the gate-gate Voltage measured 7.67V.
Note: The above measurements were made without disconnecting the 4 Ohm dummy load so load current could have corrupted the idle current measurements.
Two solutions come to mind. First, insert a stable Voltage in series with the Vbe multiplier like a TL431. It looks like the TL431 needs to be 5.075V and the Vbe multiplier at 2.755V so that the temperature coefficient of the multiplier matches the output MOSFETs.
Second, use a small MOSFET like an IRF510 as a Vgs multiplier. Not sure if this will work because the current in the Q4 is only 5mA.
Has anyone tried these solutions or come up with something else ?
Link: http://www.cieri.net/Documenti/Altr...ifier Using Complementary HEXFETs (AN948).pdf
The following semi's were used. Q1, Q2 ; 2N5087. Q4 ; MPSA06. Q3 ; PN2222A. Q5 ; IRF630 Q6 ; IRF9630
The following changes to the schematic: R10; 220 Ohms, added gate stopper to Q6, 220 Ohms.
Source resistors added to Q5 and Q6, 0.22 Ohms 5W.
Q5 and Q6 are on a common heatsink with Q3 held face down between them. Initially Q3 was not touching the heatsink. At 28.3C bias was set to 0.45V across the two source resistors, ~ 102mA. At 56.5C the bias current was 68mA. After I made a little clamp to hold Q3 to the heatsink, the change in bias was worse. At 31.5C bias was 116mA and at 56.8C it shrunk to 47.7mA.
Looking into things further, the Q6 gate to Q5 gate Voltage was 7.83V at 100mA (at 28.8C) and shrunk to 7.13V at 43.2mA and 53.3C. While the heatsink was at 53.3C the bias was adjusted to 100mA and the gate-gate Voltage measured 7.67V.
Note: The above measurements were made without disconnecting the 4 Ohm dummy load so load current could have corrupted the idle current measurements.
Two solutions come to mind. First, insert a stable Voltage in series with the Vbe multiplier like a TL431. It looks like the TL431 needs to be 5.075V and the Vbe multiplier at 2.755V so that the temperature coefficient of the multiplier matches the output MOSFETs.
Second, use a small MOSFET like an IRF510 as a Vgs multiplier. Not sure if this will work because the current in the Q4 is only 5mA.
Has anyone tried these solutions or come up with something else ?
I use the 9630 and the 630 as outputs in one of my amplifiers - I use a 610 - suppose you could use a 510 for the bias transistor - i mount it on the heatsinks as well - works quite well
Good Listening
Peter
Good Listening
Peter
I tried the Vgs multiplier (IRF510 is what I have) and it worked pretty good. Still need to change one resistor to improve the adjustment range. Bias current went from 74.7mA at 24.9C to 70mA at 56.4C.
Im surprised, decades ago IRF provide Hexfet Mosfets for linear Mosfet Amplifier
later IRF refused to do it, Hexfets desigend as switches and not for linear Mosfet audio amplifier
I have build 20 years ago similar amp with IRFP240 / 9240
but later I changed to quasicomplementary Hexfet output stage
Great sounding Mosfet amp, will easy outperforn BJTs amplifier in mid and highs clarity.
BJTs amplifier have no choice in clarity and natural sound v/s Class AB mosfet amplifier
If you want improve Hexfet amplifier sound in clarity, you need UcD discret class D with GaN output devices
later IRF refused to do it, Hexfets desigend as switches and not for linear Mosfet audio amplifier
I have build 20 years ago similar amp with IRFP240 / 9240
but later I changed to quasicomplementary Hexfet output stage
Great sounding Mosfet amp, will easy outperforn BJTs amplifier in mid and highs clarity.
BJTs amplifier have no choice in clarity and natural sound v/s Class AB mosfet amplifier
If you want improve Hexfet amplifier sound in clarity, you need UcD discret class D with GaN output devices
Last edited: