My open loop shunt regulator

[Onvinyl]

Hi,
I've continued working on my ol shunt reg.
It is attached, and I tried to show you the actual wiring (point-to-point on dotted protoboards) in the schematic. The current sinking as shown is 200mA, and you may adjust the V(out) by chosing an approriate zener.

When I apply mooly's test method with a drain load of 100R and an applied gate square wave with 5V, the sim only shows a corresponding square with 30uV. Since I do not know the parasitic values, it's off to reality, where a square of ~ 1mV results with spiking of about 10mV. I did not sim parasitic inductance(s) so here might lay the difference.

Additional ringing may be obtained when chosing inapproriate output caps...
This square is, however, consistent in magnitude between 50Hz and 2MHz (the end of my FG).

Anything wrong or (easily) improvable with this design?

Some reasoning concerning various desing decisions:

- Input CCS. Among several I tried, this is the one which keeps sourcing 200mA even when the output is shorted (as long as you keep device temperature in order, grmpf...). Care has to be taken to place Q1 and Q2 close together (not that easy with bulky heatsinks)

- Zener voltage reference. Very quiet in that voltage region.

- Mosfet output device makes good high frequency behaviour

- gain stage for output device. The best I could do regarding open loop. (OK, it's debatable, but open loop in respect to the voltage ref.). Physical layout is difficult as hell, at least for me.

What do I like about shunt regs, apart from sound?

- They decouple effectivly the current loops between the load(s) and the psu

- any voltages for the voltage reference and additionally circuittry are already clean.

The next post shows the regulation vs. freq

[Onvinyl]

here is the plot:

[Onvinyl]

Here is an output plot of impedance vs. frequency. Real world impedance would be higher due to wiring and other parasitic inductances.
Rüdiger

[lineup]

Onvinyl

like you and Nelson Pass & others
I often prefer shunt regulation.
Mostly for my preamplifiers and for Power amp Input/VAS stages.
(Power/Preamp Output stage, I prefer un-regulated)

You circuit is interesting.
I have never tried that way before.


One question though:
What is the voltage output?
This is a value I always want to know from any regulator
I guess it is, in your attached circuit, like 12 V (zener) + 3 V (VGS of PMOS)
which would render like 15 VDC

[Eva]

This circuit is not open loop at all. Its operation is based on strong global NFB enclosing 3 stages.

[Onvinyl]

Hi,
lineup, with mosfets I have I get 15.3V. I will use a slightly different version with the cfp-triple swapped (n -- p) for both polarities which gives me a bit better regulation and more consistent V(out) (~14.6V with nmos).

Eva, yes, the output triple is 'bonded' by feedback, but not in respect to the output voltage, unless I have a serious blind spot. This idea emerged from a simple voltage reg you onced proposed in a recent thread.

Rüdiger

[Elvee]

I suppose this:

I use to feel like the small child in The Emperor's New Clothes tale

explains that:

This circuit is not open loop at all. Its operation is based on strong global NFB enclosing 3 stages.

You shouldn't destroy people's fantasies. After all, has anybody cared to tell Susan Parker the truth about her "Zero feedback impedance thingie"?
The fact is, that most designs labelled "open-loop" or "zero feedback" do have some form of NFB hidden somewhere. At least, the ones displaying half decent specs.

[Onvinyl]

Hi,
I'm not against feedback. It's one of the building blocks of the universe.

Rüdiger

[lineup]

even if we do not want or like to call it feedback (or even local feedback )
we may call it something like:

-> controlled voltage gain

without some control of transistor gain
we wouldnt be able to do good audio or good voltage regulation
because of un-linear transfer function

[Eva]

You would have to connect C1 to the output voltage rather than to ground in order to get an open-loop like response. But then, performance would be unacceptably bad.

You seem to be missing that X2 is directly sensing output voltage, through C1 at high frequencies and through R1/D3 at low frequencies. In fact, this C1 to ground is an usual regulator trick to provide extra feedback at high frequencies and improve transient response.