The simplistic Salas low voltage shunt regulator

[Salas]

That one with layout drawn by Disco is fine for you then. Just use a 18V Zener, or follow my recommendation for a 8-9xLed string around 2Vf each, cathodes facing down. Maybe on a vertical sub board. Does better. In any case up the Vref bypass cap to 1000uF. 26-28Vin.

Regards

[DougL]

That one with layout drawn by Disco is fine for you then.

Perfect.

Thank you.

Doug

[Salas]

One single rail schematic for up to 60V DC out. Two symmetric schematics for up to +/-22V DC out. 6-10V DCin-DCout is enough. 28V Max in for the symmetric. All tried and tested. Two beta tests. The single in Athens, the symmetric in Sydney. Performance seems adequate.

[Salas]

Both the noted TO-247 and TO-220 Mosfets can work on all schematics. Both types tested. 2N5457, BF245A for Q6 exchangeable. 2SK170 Q6 for the lower voltage range symmetric also. Up to 10uF MKT or MKP for C2. 47uF if with chemical cap. Output RC Zobel values are standard. Film cap there.

[Salas]

Some sim graphs.

[Salas]

Would like to especially thank Bill (HiFiNutNut) for building and running the symmetric lower voltage beta. Cheers mate!

[massimo]

Big news then!
That "TO-240" is most probably a typo for TO-247.....

I do hope IRFP9140 will work well in the 1st schematic (Single rail up to 60V DC out) or do you prefer TO-220 now?
Output C3 preferably MKP, isn't it?
While C2 can be also a premium elcap (i.e. Silmic II), according to the polarity sign.

I like your assessment about performance....like a Rolls Royce engine.....hp: adequate!

[Salas]

Yes typo, I changed it now, thanks.

IRFP9240 or 9140 OK. No I don't prefer them TO-220s, I just tried to see if there will be problems or not because I had some. Heat is high on the single above 40V on my sink, I would rely more on TO-247. The 9610 gives more PSRR higher band in sim, but the performance threshold is excellent already with IRFPs. Another two singles have been made for beta with the heavier Mosfets by a friend here.

47uF the most if you will use a lytic for bypassing the Vref as I mentioned, but an MKT or MKP is advised. The testers will surely have to say more on such.

Output C3 must be film, MKT or MKP. Or 22-47uF medium ESR lytic instead of the RC across out. 2N5459 in CCS tail 7mA Idss.

Here's Bill's layout BTW. He uses low value Vref caps now because he has it on the higher portion of his active system after he tested with lytics also. No less than 4u7 C2 for a full range application, preferably 10uF. I used a Philips MKT 10uF which was relatively compact and bypassed with a Vishay MKP1837 in that photo for example.

[DougL]

Salas,
Can you comment a bit about the design choices?

The differences I see are:
In the series section, PNP replaces 3 LED reference
In series section, Cascade or Ring of 2 NPN replace 2SK170 jfet css
In Shunt section, Resistor replaces Zener String
In Shunt section, added RC feedback on Ref driver.
IN shunt section, added NPN follower.
In shunt section, added remote sense "requirement"

This is in contrast to Post 501.

Just Curious of your perceived merits of the changes?

Doug

[Salas]

Its about the performance things discussed in this thread.

The cascode tail increases the psrr of the ccs. Vbe ref for no Rset hassle with LEDs. Cascoding brings it up to par.
The resistor comes handy for an adjustable output. Its low noise enough but leds are better, and stiffer for drift btw.
The RC feedback is a compensation to keep oscillations away.
The buffer helps open up Zo performance in frequency.
Rsensing is a good practise not to throw the better performance measures out of the window.

About subjective merits, I hand it over to the testers.