No, not at all: you seem to envision a kind of gyrator circuit for the input (which could also be an alley worth exploring), but the topology I created, or discovered (or perhaps rediscovered, you never know) is "straight" tandem, having negligible internal time constants, but combining positive and negative feedback in a way that cancels all first & second order errors; in sim, at least, because in reality, self-heating effects of the active elements slightly degrades that perfection.
From a more philosophical POV, these regulators can be viewed in a variety of ways: Osvaldo rightly noted their (apparent) similarity with class AB amplifiers:
And it is true that such an amplifier can work as yet another class of tandem regulators, and most tandem regulators could also in principle act as class B buffers: it is the case with my latest implementation: if the reference voltage is replaced by an input signal, they work as ideal textbook buffers, with zero output impedance and zero distortion.
The concept of class AB amps used as regulators isn't new, they are generally called 2-quadrant regulators, and there are much earlier examples than the one provided by CFT.
The purpose of 2-Q regs is generally different though: it helps the transient performances by including an active output bleeder.
Regarding "true" tandem regulators, there is a subtle distinction: class AB amps combine two unity gain buffer of opposite polarity and drive them in phase, perfectly synchronously and with an offset carefully adjusted to avoid a dead-band or a thermal runaway, but a tandem regulator (yours and mine anyway) uses the shunt regulator as the master device, not only providing regulation but also steering the slave series regulator.
This series regulator can normally not be a pure voltage regulator, because it would conflict with the shunt regulator, (unless some independent mechanism manages to synchronize them to the perfection, which I have achieved) and has to be some kind of CCS.
Yeah, I got it!
keep in mind, for those not skilled in tube art, that there aren't tube P devices.
Attachments
Member
Joined 2006
The tube circuit Osvaldo showed is not about shunt regulation.
It's a series regulator with a constant current load.
I have a problem with the circuit in post 1.
The reference is formed by the drop on R15+R16 from the Q9 CCS + Vbe(Q10).But Q9 has to do it's job with only Vbe of Q10. At that Vds the fet doesn't behave as a decent constant source.
Also there is a 2V6 note there, another remain of early thinking?
Mona
It's a series regulator with a constant current load.
I have a problem with the circuit in post 1.
The reference is formed by the drop on R15+R16 from the Q9 CCS + Vbe(Q10).But Q9 has to do it's job with only Vbe of Q10. At that Vds the fet doesn't behave as a decent constant source.
Also there is a 2V6 note there, another remain of early thinking?
Mona
Exact. It is explained in the book, but I wanted to mean that the idea is old. Constant current or variable inversely with the demand, is the same idea, but the first was easier to build in such days.
A fair remark. Question: If I add an additional diode in series with the emitter of Q10, that would make two Vbe drops. Would that be enough of Vds for the fet Q9 CCS for it to behave decently? Is there any better way of providing Q9 with some decent Vds?
(Yes, just discard that "2V6").
True. The gyrator concept is a step to be considered. I have a mental problem with introducing any resistor (R4 as of post #1) in front of the base of the series driver transistor Q2, and this resistor associated with introduced capacitance (C13+C14 as of post #1). The problem being that such an approach could potentially increase the minimum dropout that would then be necessary for the "series" regulating section.
Which, as you already mentioned, is not really a "true" series voltage regulating section (due to the fact that a certain modulated quantity of the driver transistor's Q2 base current is being stolen and routed to ground).
Since you mention it, now I see that there is a similar problem with Q11. In order to pump up the Vds of Q11 to at least 2x Vbe, that would require additional diodes in front of R3 and R17 ... Correct?
But then, my dream of regulating the output voltage, way down low, to say 3.3V, ceases to be achievable. Hmmmm ....
Come to think about it - a very good idea. Will just put in a resistor, say 10kom as a load for Q10 and be done with it. Or maybe even abandon that - just let Q10 pump in some current directly into Q6 and Q8, with the current split as set with the R12 pot.
There is amplification aplenty within the loop(s), so no need to jump to extremes on this.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.