PSU with regulated Supply for front-end

[MiiB]

Idea was to use a voltage doubler for the higher Voltage supply for the front end...Use some CLC filtering and the chop it down with a Salas Like shunt...to get app +-90 V...

Mirroring the Salas Shunt does not give the desired results at the voltages end up a few V apart....95 V vs 102 V.. Is this due to differences in the shunting mosfets...or am I missing something else..??

[AndrewT]

It is not close to a Salas style shunt regulator.
It has omitted the CCS stage.
Scrap it and adopt a real Salas Shunt.
Use 150V or 200V Vfets for the 4 active power devices.

Going back to your schematic.
The output voltage is Vbe Q3 plus V across R9 (82k).
The Vbe has only a tiny change for operational conditions.
Your "error" is due to the simulator selecting a different current passing through the two 82k resistors.

It's another example of asking the simulator the wrong question and as a result getting a wrong answer. You must understand the operation and be able to check the results before you adopt software to save on the arithmetic.

If you cannot check the results then you should not be using the software.

[MiiB]

How nice....

Thanks...You realize i am asking questions to learn, not to be ridiculed..???

[AndrewT]

You are closing your mind to learning.
There are a number of lessons in my reply.
"How nice" implies you will ignore at least some of them.

I will repeat the most important lesson.

You must be able to understand what the software is doing and be able to check the results before ever relying on any questions/answers of that software.
Learn the operation first. Be able to do it with pencil and paper. Then you are ready to take shortcuts with the assistance of software.

[GoranB]

Originally Posted by AndrewT

It is not close to a Salas style shunt regulator.
It has omitted the CCS stage.
Scrap it and adopt a real Salas Shunt.
Use 150V or 200V Vfets for the 4 active power devices.

Going back to your schematic.
The output voltage is Vbe Q3 plus V across R9 (82k).
The Vbe has only a tiny change for operational conditions.
Your "error" is due to the simulator selecting a different current passing through the two 82k resistors.

It's another example of asking the simulator the wrong question and as a result getting a wrong answer. You must understand the operation and be able to check the results before you adopt software to save on the arithmetic.

If you cannot check the results then you should not be using the software.

Calm down man

[Salas]

Originally Posted by MiiB

Idea was to use a voltage doubler for the higher Voltage supply for the front end...Use some CLC filtering and the chop it down with a Salas Like shunt...to get app +-90 V...

Mirroring the Salas Shunt does not give the desired results at the voltages end up a few V apart....95 V vs 102 V.. Is this due to differences in the shunting mosfets...or am I missing something else..??

There must be Vbe differences between Q2 & Q7 models and the resulting currents via R7=R11. That is I to V translated on the R9, R13 82K Vrefs. Try differentiate R7, R11 on the simulator to see if it works bringing the power lines closer. Also no damping resistors on C22,26 may prove oscillation prone... The chosen Mosfets have low Yfs for best output impedance in this application and are marginal on VDSmax. Very good components for power amplifiers, don't waste them or endanger them here IMO. I suggest you borrow from the following attached CCTs, which have better LF characteristics than the BJTs Vref first version. Those have been used as symmetrical HV regs in two builds for a +/- 300V powered headphones amp as reported by members, scaled up in components ratings of course. The trimmers can be substituted for resistors after the values are trimmed. Using Toshiba 2SA970 for Q2 MJE350 shows no problem with loop gain's phase margin, there is chance is going to prove problematic for Vmax in on/off cycles though, but dissipation is adequately low for your Vo specs. Better use MJE350 to be sure, or check with the 2SA970 in practice. Its a better Noise Figure and Ft part and I understand its appealing.

[MJL21193]

Drawing excessive current from a doubler supply (via a shunt reg) is probably not the ideal. Using a linear regulator makes more sense.

[Salas]

Shunts are not power efficient, no question about it. For what application is that? What will be the demanded average & peak current?

[MiiB]

Salas

Thanks for the suggestions....

Its for powering the fronted of my amplifier..I use app 85-90 V rails, consumption is app 40 mA...making for a total of 80-90 mA for both channels....

[Salas]

If you will go such super regs (i.e. CCSed shunts) in the end, you got to run them +23mA on top of your max consumption per rail. Check what is drawn on max swing of your VAS. ICCS=(VF3xLEDS-VGSQ1)/R1. Those attached are the simulated loop gain and Zo characteristics. Vin-Vout 20V will be fine, safe from mains and toroid's sags. Keep short output cables, and have good luck.