John Curl's Blowtorch preamplifier part II

[janneman]

Originally posted by kamis

Open-loop regulators, series or shunt, is here accepted name for no-feedback regulators. Mr. Perez regulators, showed here, are open-loop types. They consist of zener voltage source and shunt transistor, without any error amplifier. Constant current source or resistor are used with both feedback and no feedback shunt regulators. No feedback regulators have very high output impendance . Jung super regulators current/sense connecting method was developed to eliminate mutial influence between stages you have mentioned. Blowtorch preamp uses LM317/337 series IC regulators and then feedback-free shunts to improve HF regulation, and even cap. multiplier near the audio circuits.
But Jung/Didden regulators are much better, especially in current/sense connecting mode.

Well, the shunt 'regulators' that were shown before don't regulate as far as I can see. They are just a constant current source. Calling that an 'open loop shunt reg' is very creative, I admit that.

Jan Didden

[janneman]

Originally posted by bear
Jan said: "The ground current with a shunt back to the supply cap is constant (DC) because the sum of load current and shunt current is constant (= the CCS current). So mutual influence between stages is much less. But it anyway depends a lot on the implementation, you really have to know what you are doing of course."

Bear wonders out loud: "well sure, on paper... but what if the tracking between the two is less than ideal in the real world? ...what would the 'difference' waveform look like, would it be nasty higher order stuff? And, of what magnitude? If the series regulator has more sinusoidal ground currents, might that end up sounding better? So, in practice how good is the tracking of the shunt requlator, and what sort of HF response and bandwidth does it need to have?..."

_-_-bear
[snip]

Yes you're right. But I would think that it would be a really badly botched job if that shunt wouldn't be orders of magnitude better than the series reg in avoiding ground ripple currents. But yes, not zero.

Originally posted by bear
[snip]PS. Jan I think your diagrams alone are not accurate, they really need to reflect the constant DC claim for the shunt going to the main return, and the AC varying current with the series regulator??

This is to illustrate the principle. It mainly shows the AC component differences between series and shunt. I have no intention to present a cookbook-ready design for anuone to copy

Jan Didden

[janneman]

Originally posted by KSTR
The shunt regs provide the low Z and the stable output voltage, the ccs in front of them try to isolate AC components to the amount they act as ccs (vs. freq), and the series regs bootstrap the ccs to help them isolate even more.

Basically a shunt reg also works with a simple series resistor, fed from noise supply... but not that good.

BTW I don't quite get is why people don't use like circuits instead of complementary ones, when using two-pole designs.

And as for the Perez schemo, don't chop the mains xformer capacitance with individual bridges, even with common-mode chokes. Workaround: 4 secondaries or capacitive bypasses at the right place.

- Klaus (off to lunch now)

Well, I may be missing something, but a series CCS followed by a shunt CCS, seen from the load, has a very high Z, so any varying current drawn by the load will lead to huge ripple and signal components on the supply. About as bad as you can think of.

Jan Didden

[KSTR]

Jan,

the last reg in the chain is supposed to be a shunt-type *voltage* regulator, that's what you were missing, I think.

- Klaus

[janneman]

Originally posted by KSTR
Jan,

the last reg in the chain is supposed to be a shunt-type *voltage* regulator, that's what you were missing, I think.

- Klaus

OK, are we talking about this post (and following):

http://www.diyaudio.com/forums/attac...amp=1246503992 ?

I see a constant current source as the last element. A fet with a constant Vgs drawing a constant current?

Jan Didden

[KSTR]

Originally posted by janneman
Well, the shunt 'regulators' that were shown before don't regulate as far as I can see. They are just a constant current source. Calling that an 'open loop shunt reg' is very creative, I admit that.

They are "reversed followers" being their own (bad) error amplifiers of sorts. Very local feedback aka "open-loop", not a classical multistage feedback loop, that's true, just the next simplest way of a shunt reg instead of a zener diode.

- Klaus

[KSTR]

Originally posted by janneman
I see a constant current source as the last element. A fet with a constant Vgs drawing a constant current?

No, check the polarities of the MOSFETs and the way they are hooked up (ignoring the source resistors for a moment). Best seen on the neg side, the source follows the drop on that R to gnd, impessed by the JFET ccs.

- Klaus

[syn08]

Originally posted by KSTR
They are "reversed followers" being their own (bad) error amplifiers of sorts. Very local feedback aka "open-loop", not a classical multistage feedback loop, that's true, just the next simplest way of a shunt reg instead of a zener diode.

Correct. They have as much feedback as a source follower.

[Justcallmedad]

First schematics, those posted by flg where only intended for learning, trying and illustration purposes…

Last ones in post 93 are fully functional Zout is about 0.5R from 0 to a few tenth of Mhz(I cant measure beyond) noise is -150 db from 100hz to 100 kHz (can't measure beyond neither ) and decreasing to -130 db at 10Hz of course there is no feedback "loop", for me the shunt bjt is 100% “feedback”.

New PSU

[KSTR]

Originally posted by bear
So I am thinking why not a choke or a pi filter in the same spot?

Yes, that would work also, the choke being a non-perfect constant current source for AF only (not at DC and not at HF), and you'd need one with a high-ish DCR, otherwise the series reg would only fight the shunt reg (the shunt would win probably, forcing the series reg into current limit).

- Klaus