A quick look at the datasheet and you may get lucky using Vishay IRFP240 & IRFP9240 MOSFETs - but you may have to work on BIAS voltages here. I have never played with the MLE20/MJD20. The IRPs normally have a 3.6V gate bias voltage to idle at ~ 150mA at 60V. But each setup will vary. Swapping MOSFETs always has extra work associated with it.
What I did with the amp was turn bias down to zero.
Apply a small sine wave to input.
Monitor output to speaker on the scope.
I turned up bias until cross over distortion went and the result was about 7mA bias current.
7 mA bias is way too low if you want low crossover distortion. You really need a distortion analyzer to see crossover distortion. Unfortunately, most people don't have access to one. This also goes for setting the bias for BJT power amplifiers.
In general, you want the largest class-A region that is consistent with your heat-sinking limits. However, with BJTs, too much bias current leads to so-called gm doubling distortion, so for BJTs there is an optimum value, usually in the range near the current that drops 26 mV across the emitter resistors of the output transistors. If the emitter resistors are 0.22 ohms, this comes to about 120 mA. In very rough terms, this will lead to a class A region that extends to very roughly 240 mA into the load, assuming a single output pair. In the real world, the optimum number tend to be less than 26 mV due to device characteristics of real BJTs, more like 20 mV.
The situation with MOSFETs is very different, especially with lateral MOSFETs. You want high transconductance in the output stage, and with MOSFETs (as with BJTs). Transconductance of MOSFETs is lower than that for BJTs at the same operating current. In fact with MOSFETs you will generally not get gm doubling no matter how high the bias current is. This means, especially with laterals, you can run them as hot as you want as long as you stay within device limits and temperature limits. You can have a good-sized class A region with smooth transitions into the class B region at higher signal currents. With lateral MOSFETs, their current tends to be self-limiting, at currents around 100-200 mA, so biasing is not as tricky as with BJT devices. Vertical MOSFETs can have higher transconductance and source a lot more current, but their self-limiting operating current is much higher - usually several amps. So they are not completely immune to thermal runaway, but far less likely to runaway than BJTs.
In the Hafler DH-220C redesign, we run each of the two pairs of lateral MOSFETs at about 150 mA if I recall correctly, which leads to a class A region that extends to very roughly 600 mA of peak signal current, and which transitions smoothly into the class B region.
Cheers,
Bob
In general, you want the largest class-A region that is consistent with your heat-sinking limits. However, with BJTs, too much bias current leads to so-called gm doubling distortion, so for BJTs there is an optimum value, usually in the range near the current that drops 26 mV across the emitter resistors of the output transistors. If the emitter resistors are 0.22 ohms, this comes to about 120 mA. In very rough terms, this will lead to a class A region that extends to very roughly 240 mA into the load, assuming a single output pair. In the real world, the optimum number tend to be less than 26 mV due to device characteristics of real BJTs, more like 20 mV.
The situation with MOSFETs is very different, especially with lateral MOSFETs. You want high transconductance in the output stage, and with MOSFETs (as with BJTs). Transconductance of MOSFETs is lower than that for BJTs at the same operating current. In fact with MOSFETs you will generally not get gm doubling no matter how high the bias current is. This means, especially with laterals, you can run them as hot as you want as long as you stay within device limits and temperature limits. You can have a good-sized class A region with smooth transitions into the class B region at higher signal currents. With lateral MOSFETs, their current tends to be self-limiting, at currents around 100-200 mA, so biasing is not as tricky as with BJT devices. Vertical MOSFETs can have higher transconductance and source a lot more current, but their self-limiting operating current is much higher - usually several amps. So they are not completely immune to thermal runaway, but far less likely to runaway than BJTs.
In the Hafler DH-220C redesign, we run each of the two pairs of lateral MOSFETs at about 150 mA if I recall correctly, which leads to a class A region that extends to very roughly 600 mA of peak signal current, and which transitions smoothly into the class B region.
Cheers,
Bob
I am thinking irf9610 610 (or 620 9620) driving 2Sc5200 2sa1945 bias at 6.2v is going to beat latfets and both are easy to get.
The irfp240 9240 have a bit of a knee with higher current. 0v bias can drive the 5k feedback resistor and deliver a perfect sine at 20k on the scope. Load to 100 ohms and distortion appears. Load to 8 ohms and alot more appears.
4v ended the visible distortion on the scope with my amp, the gate waveform showed alot of correction. I finally locked it in at 6.5 volts
Fine details and powerful bass audio nirvana.
4v ended the visible distortion on the scope with my amp, the gate waveform showed alot of correction. I finally locked it in at 6.5 volts
Fine details and powerful bass audio nirvana.