I'm certainly stumped.
The mosfet is supposed to fully conduct current from drain (pin2) to source (pin3) when the voltage from gate (pin1) is around 5V or more higher than the source (pin3)
ie. when Vgs > 5V or so, the resistance thru the series pass mosfet approaches the minimum and current the maximum
That means that when the gate voltage Vg > Vout + 5V the thing is turned on
Before power on, mosfet Vg and Vs are at 0V of course.
At power on, mosfet Vdrain is the first thing that hits full B+ voltage.
The power up sequence (I think) is firstly D5,D6 -> R3, R2 -> 100V zener
Around the same time, Q1 collector voltage begins to rise > 0, turned on as it is by D5, D6 voltage drops.
So, there begins some Vg on the mosfet, which gets to 5V greater than the Vsource, which starts it conducting leading to mosfet Vs ramping up.
Both the Q1collector and mosfet Vg ramp up, higher than Vs and the ramp of all three continues in step.
When the Vs at the mosfet exceeds the 100V zener voltage, resevoir cap C6 starts to charge up and hold voltage, as well as Q3 turns on which means Q2 can conduct and the whole the 'regulation part' starts rolling.
ie. Q1, Q2, Q3 are sampling the voltage at the mosfet source Vs (via R5/R6 divider) and comparing to a reference voltage in order to adjust the Q1 collector/mosfet Vg voltage. mosfet source Vs follows (5V less) , thereby regulating the Vout.
D8 is a protection diode to prevent conduction across source-drain. There should probably be a zener across the first mosfet g-s similar to the second one to prevent the voltage from getting too high.
--
In Matta's case, even when we modified things to ensure Q3 turn on, the output voltage seems to ramp up but falter at 140V or so under a very modest load, even though it is much higher at no load.
That suggests that even a tiny load is enough to cause major 'drooping' of the Vout.
That could be because the mosfet can't supply even a tiny amount of current
ie. it's not sufficiently turned on because it's Vgs is not sufficiently near the required 5V
OR
The mosfet gate voltage Vg is only making it to 145V or so, meaning it can supply current but only at a voltage 5V less than the gate ie. Q1 collector voltage only makes it to 145V under load.
I'm not sure which - would need to measure mosfet gate and source voltages under load again to know 'how turned on' the mosfet is.
If the mosfet IS turned on, then the problem is with the developed Q1 collector voltage being plain too low.
If the mosfet IS NOT turned on, why not? It is likely borked.
Now I've never been sure what sets the Q1 collector voltage ie. is it Q1 Vce doing it, biased as it is here, it or is it the mosfet gate somehow setting it?
It's far closer to to Q2's emitter voltage (set by R2/R3 divider) than it's own emitter voltage (set by B+), so ....
In any case, it seems to suggest that Q1, Q2 are NFG if Q1 Vce is that low.
------
Conclusion - could be any/all of Q1,2,3 and even mosfet Q4 that are nfg.
I am pretty sure D5,D6 are OK. R1, R8 are certainly OK and I'm pretty sure R5 is OK.
FIX : verify the working components above and change ALL the &**&&$## components.
-----
The business with the R6-R7 voltage being too low to turn on Q3 is due to the lack of voltage at mosfet Q4 source.
This happens even at no load, let alone with tiny load (leading to major droopage)
It seemed that mosfet Vs ramped up to 216V or so then choked up - perhaps it was throttled back by Vgs dropping less than 5V and the ramped voltage just drained away. Why the throttle back?
Again - could be a borked mosfet or Q1 throttling back, hobbling it to that sad and pitiful voltage.
In any case, we were attempting to force fix a symptom and not the cause of the problem. This issue would affect regulation and not basic ramping to HV. (I think :
)
ONE other thing is that the second regulator section could be nfg and 'dragging' the first regulator down.
ie. mosfet 2 goes short circuit and your get 'blow-back' into mosfet1's source. Hmm.....
----
PS - in any case it's not really that important the exact cause - once you have the PSU board out, it's just as easy to replace all Q1,Q2,Q3 and I would say may as well go to IRF840 too at that time.
PPS - I have had ALL of the above components go in one board and the ONLY thing that got me out of the vicious cycle was member 'ppa' mods.
PPPS - there does seem to be a thing whereby if you replace one or two but not all faulty components, the replaced ones die again! That's where replacing ALL the f*^&ckr components AND the 'ppa' fix is the only thing that helps
----
Now, when desoldering the @$%%!& components Q1,2,3, (if you don't have a desolder station), put a 'yanking clip' on the transistor, use lowest heat and put MORE solder on the 3 pins of each to get an even melt at the pins, simultaneous.
Gently, gently yank the sucker out. Too much heat and you will burn the pads which is a real pain. Too much yank and you will pull the thru-hole vias out. Way-frtized transistor will just crumble
; if so, do each pin yank in turn.
(in fact I almost recommend crushing the thing and doing each pin in turn - better than a poorly executed 3-way yank!)
Once the transistor is out, clean up GENTLY with heat and solder wick (small width). I recommend clean up resulting mess with flux remover.
DO NOT attempt to remove solder first before yanking - you WILL f&** it up.
---
So - anyone else PLEASE feel free to correct this description if you wish
I'm still mystified by the cause-and-effect sequence of operation at power up/down and failure modes ??? (I am a spice free zone)
Good luck to anyone suffering psu issues here.