Simplistic Mosfet HV Shunt Regs

[AndrewT]

Salas,
I see many builders are using a low voltage device for M1.
M1 takes enormous stress at start up for a few us or ms depending on the load characteristics.

Is there a history or good, or poor, or bad, reliability for the M1 devices?

[Salas]

Although it started as a what I had in the parts bin thing, very good M1 reliability. Seems that the small Vref filter cap (1uF) does not stress them long enough at start up.

[AndrewT]

Yes, there is only 1uF and 1uF+3r3 as a start up load inside the regulator.
What is the external load capacitance?

I will point out that I believe a lot of the good performance of these Salas Style regulators comes from the fact that very few capacitors are used.

[AndrewT]

Back to #3118,
Why is R3 so large?
Would a much lower value suffice? Like 220r or 470r?

[Salas]

I don't remember why that value, really. On the lower range circuit it was sub 1K indeed.
Its 4 years now that initial circuit and I arrived there on test, it wasn't simulated before.
*Few caps and no significant lytics was a a quality and economy goal.
**Maybe he needs an input decoupling film if he has got long wires.

[triode_al]

Just a question: as I see it, the collector of Q3 is kept at 1,2 volt.
Then I see a mismatch with the base of Q1 at [at least] 0,6 volt.
So maybe this design needs a darlington at Q1? on the other hand, this idea could work when using a NPN pair current source and proper termination at 'the top'..
apologies for a maybe stupid question.

[AndrewT]

No,
the collector of Q2 is @ ~ 1.3V below the output rail.
The base of Q1 is @ ~ 0.65V below the output rail. This forces Q3 to a Vce of <0.1Vce.
i.e. Q3 is saturated.
Not a good way to ensure good performance from a two transistor CCS. That CCS determines the current and thus the voltage drop across R9.

Your idea of a darlington for Q1, or maybe a mosFET, would help get Q3 Vce back up to at least 0.5V and maybe towards a couple of volts.

[eboz]

Originally Posted by Salas

Make R10 4K7, see if its stabler.

Hi Salas,

The regulator is stable after making R10 as 4k7. Thanks!!! Q1 will now also dissipate less heat. Is there a drawback to this?

---
How hot does R9 become? Be careful it has a lot of volts across it !

AndrewT, it is not hot, only a little warm.

Regards,
-MW

[Salas]

eboz, nice. No problem, its better gain and less dissipation that way. I had revised that when I had used it near max Vo a long time ago, its 4k7 on the schematics from back then also. So you are at standard.

[Salas]

Originally Posted by AndrewT

No,
the collector of Q2 is @ ~ 1.3V below the output rail.
The base of Q1 is @ ~ 0.65V below the output rail. This forces Q3 to a Vce of <0.1Vce.
i.e. Q3 is saturated.
Not a good way to ensure good performance from a two transistor CCS. That CCS determines the current and thus the voltage drop across R9.

Your idea of a darlington for Q1, or maybe a mosFET, would help get Q3 Vce back up to at least 0.5V and maybe towards a couple of volts.

There is the JFET Vref standard one, the SSHV1 one to make, instead of head scratching revising the initial idea one, although some still find synergistic qualities in it for some circuits and like it as is...