AB-dynamic

[anli]

Did you have problems with DC stability?

I was forced to choose between mosfets current stabilty and output drift stability. At current variant mosfets' current changes ~5-8% during warming up, and output offset changes ~40-50mV (I don't remember exactly measured values now ).

How did you determine the ratio between R10 and R13?

It is more correct to say about R13*R6, and load, and M1 transfer characteristic ratios - just to make dynamic load able to supply current to load at positive period half. Absolute R13 value is in sync with R8.
As for R10 - if you have found first picture descibing the idea, you can guess I have realized some kind of hooliganism - Vgs and R10 are used to provide a (pseudo-constant) current of left bjt used as diod. This step was done in favour of simplicity, and R10 value was selected experimentally.

 

[Lumba Ogir]

Thanks for the information, anli. I`m just curious about the thermal stability of your arrangement. I am going to make experiments with some hard-to-tame vertical MOSFETs in a few weeks.

 

[anli]

Thanks for the information, anli. I`m just curious about the thermal stability of your arrangement. I am going to make experiments with some hard-to-tame vertical MOSFETs in a few weeks.

Good luck! Will be glad to know your listening experience.
It's funny, dynamic load (M1 and around) demands on using of vertical rather lateral "for audio" mosfets: initial Vgs must be rather high to take "voltage air" for X2,X4.

BTW, R10 (instead of CCS) "softens" NFB existing in the dynamic load itself - M1 transfer characteristic (dI/dV) is limited, and, as a result X2 current is mdulated by Vgs, slightly reducing dynamic load's NFB loop gain. It is, probably last "most secret place" in the schematics

 

[anli]

About thermostability, more strictly - about thermoenviroment. If you have looked at a picture with showing the amplifier inside, you, probably, noticed uncommon realization details. Among others (say, frontal input RCA connectors), you can see, I have rejected common practice to mount pcb to heatsink. I'm strongly sure, low-current details (i.e. placed on PCB) must be far from heartsinks. Top and bottom case's plates perforation serv this aim also.

 

[Lumba Ogir]

Thanks for the tips. Normally, I would have proceeded a from symmetrical CFP topology (like your input stage with M2) to linearize the MOSFETs and to achieve DC stability. Single-ended input is preferable.

 

[Bigun]

I'm curious if anybody else has built this amplifier - I've not heard much about it and when I look at the goals listed in post no.1 they look very good ?