You can use LM317 but the power supply to be 51 volt and adjust C2 26v instead of Vcc/2. I verified on simulator.
Lateral MOSFETs are self-stabilizing above a fairly small current value, but VFETs may have this threshold of thermal stability way up at tens of amps - basically they need thermal compensation typically.
There is a point of inflection in the response curves below which the tempco is negative, and where this is matters.
https://www.onsemi.cn/PowerSolutions/document/AND8199-D.PDF
There is a point of inflection in the response curves below which the tempco is negative, and where this is matters.
https://www.onsemi.cn/PowerSolutions/document/AND8199-D.PDF
This mosfet has more than twice the input capacitance but less than twice in transconductance , which results slower response. It needs to redesign with lowering the gate resistors 220 ohms to maybe 180 ohms and increasing the CCS . The bias current needs to be determined by listening for fatigue free.
The irfz44 is very popular and costs very very cheap. It is identical to irfp150 but with lower Vceo,60v and Pdmax, 150w.
Some call it bastode . NAD3020 uses it also in low level. According to our historian PRR, he has a book dating the 60's where such circuit is described.
I had a go at simulating this topology with a few variations. I find the feedback network resistors R4/R6 can be increased in value by at least an order of magnitude, the gate biasing resistors R3/R7 can be increased and the constant current source reduced to match, again by at least an order of magnitude, and this seems to remove a strong peak in the frequency response at HF.
The distortion residual with negative feedback looks like mainly crossover distortion to me - certainly not the 3rd harmonic of the open loop.
The distortion residual with negative feedback looks like mainly crossover distortion to me - certainly not the 3rd harmonic of the open loop.
To design this amp you need to understand how it functions . The driver made of Rush cascade and CCS pours it's power to the load via zener , 220 ohms and two diodes . The current is sensed by the resistors and command the mosfet to pour it's current to the load and unload the class a driver. This type of amp is immune to crossover distortion as the class a amp fills the blanks while the outputs need time to switch on. On simulator with outputs biased just on cut off with leakage current has no crossover distortion at 100mw. But in real world the gate voltage of the outputs do not cutoff so sharply bellow 20ma . I could reduce it down to 5ma by 17ma on CCS , the sound is exactly the same as biased 120ma . Only by listening several days that 20ma bias by 20ma CCS I got the best fatigue free listening.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.