DIYHiFi.org

I opened this topic to be clarified (and maybe other members)

- The Pros and Cons of each type of regulators...
- How each one works...
- Other technical material...

I’ve been sold on shunts for some time. The main advantages that I’ve seen so far;

A shunt can source and sink current, a series can not.

The input to a shunt is a constant current source feeding a control loop. The PSRR is determined by the ratio of impedances of the ccs and the control loop. The higher the ratio the better the isolation.

Shorting the output of a shunt will not destroy the regulator. Current is limited by the ccs.

Generally, you can think of most (not all) voltage regulators as servo controlled variable voltage dividers. The most common exception I can think of to that is one using the inherent characteristics of a semiconductor giving a constant voltage drop, like a Zener regulator (or OD3 style tube equivalent).

In the case of the series regulator, the variable element is in series with the raw voltage source, and it's effective resistance is controlled so that the voltage on the load is held constant. The elements of the voltage divider are the regulator pass device and the load itself.

For the shunt version, the variable element is in parallel with the load, and it's effective resistance is controlled so that the voltage on the load is held constant. You need some kind of current source, even if only a resistance, in series with the raw voltage source for this to work. The elements of the voltage divider are the current source and the regulator shunt device in parallel with the load itself.

Generally, series regulators are more efficient than shunt regulators, since all the DC current ends up in the load. For a shunt regulator, a lot of the current is dissipated by the shunt element.

But, that's not the whole story.

You could imagine that a series regulator might not be as good in isolating the load from the raw voltage source, since the variable element is controlled somehow in a closed loop system, which inevitably loses effectiveness as you go higher in frequency. With a shunt regulator you can make the isolation peformance of the current source pretty good with frequency, especially if you use a resistor for that function.

The other consideration is where the AC currents flow (how's that for redundant redundancy?). In a perfect shunt regulation system, all the AC passes through the shunt element since the current source has a very high impedance at all frequencies. That might be a very good thing. It's even conceivable that you could minimize or even eliminate bypass capacitors in such a system. You can't do this with a series regulator, since the variable element used in that system has a relatively high impedance to ground by nature.

For designs, Google is your best friend.

.............from experience a properly done shunt always sounds better and removes for the most part, the transformer from the equation.

Jam

I've found you can improve the series regs by loading up on output current. That is, put enough dummy load on it such that it never leaves class A operation. In effect, you then get the same push/pull capability of a shunt. You also maintain a lower and more consistent output impedance.

jh

Hmmm... You find that the series regulators are actually turning off on occasion?

Mr. Walt Jung has an article on shunt regulator you can download from his site. It sumarizes advantages and shortcomings. BTW, it was written 37 years ago.
http://waltjung.org/PDFs/Dont_Shun_the_ ... ulator.pdf

It may be good to say that today's usual shunt topology which includes CCS to feed the shunt/circuit was probably firstly advocated by Allen Wright. He also lists most of important features of the shunt regs in his Tube Preamp Cookbook.

Pedja

I've found you can improve the series regs by loading up on output current.


The more output current the lower the output impedance due to greater transconductance in the output device. Series regulators can absorb load current variation just like shunt regulators provided the DC load current ( or bias current) of the circuit under regulation is greater than than the AC load current variations. For anything but the output stage it is.

The advantage of a shunt circuit is in its line rejection. If I was building simple regulator with a know load current I would probably go with a shunt. When you are designing a regulator that will be used by someone who doesn't know what the load current for regulator will be ( almost every audio DIYer) you are inviting disastor with a shunt regulator.

A series regulator with a bootstrapped supply for the error amp brings most of line rejection advantages of a decent shunt regulator design. You can design either type to be good or bad regulators. Attention to Line rejection, noise, stabilibility, and low (preferably resistive) output impedance. If have seen regulator topologies that have both series and shunt parts depending on what part of the circuit you are looking at. It is as easy to build a bad shunt regulator as bad series regulator.

yes, for common tube dude....


in few ole books ,CCS as preceding stage to shunt reg is mentioned,but pretty short; anyway-in these ancient times CCS was somewhat problematic and not cheap part to make with toobz......


btw




have you one?

Absolutely! This is one of their big drawbacks. The situation is even worse with LDO types.

Look at it this way, say you have a big load current which suddenly is reduced greatly. There will be some delay in the response of the regulator (assuming it is an active type) such that the output voltage shoots high. Ok, maybe only millivolts. But the situation is enough to turn off the regulator. At this moment, the servo has gone open loop and you get some integrator windup, and finally a recovery response. It can get ugly.

Now imagine an inductive load, perhaps a motor. The back EMF can easily turn off an unexpectant regulator. Like Fred says, no one regulator works in all situations. You need to understand the specific load you are designing for. There are also other limitations (such as voltage overhead or limited available current) that may force you down a specific path.

jh

I keep wondering what these loads (for audio) to provide such a kick back are. a convertional regulator shunt or series usually has an output cap that will absorb a lot of back EMF. All this stuff about error amps going open loop are extremely rare cases. For audio, regulators that are stable with typical loads loads that are resistive or capacitive your will find the error amp ( not the pass transistor) swinging a few millivolts at far below a milliamp output current.

Your (pre)amplifier circuit is working harder because it is swinging signal voltage and current. I sometimes wonder if people really understand any thing about voltage regulators. A bunch of you want their amplifiers single ended and their regulators push pull instead of the other way around like it should be.

I would reccomend people do some reading:
http://waltjung.org/Regs.html


There are very good IC push pull "regulators" that will handle several amps and have excellent perfomance but no one ever seems to design with them but me. I can't imagine why someone would use a three terminal regulator when something so much better is available for a reasonable price.

What you mean "but me"?

I do... by Allen's kindness.

There is a part where he explicitly talks about how and why he started using CCS here i.e. how he came across "super shunt" topology.

Pedja

Of course, I was speaking generally. If discussion is limited to just audio, then you don't have as much to worry about. Except maybe woofers, turntable motors, possibly class-D amplification, regulated ac power regenerators, motorized volume controls, relays, cooling fans, etc.

The extreme cases are highlighted merely to illustrate the problems. I find it the easiest way to explain something.

jh

Yes there is nothing like the inductive kick back from a smoothly running turntable motor with the constant load of a heavy high inertia flywheel running at constant speed or a cooling fan for that matter. Let us not forget heavy load current from moving a one gram relay armature a millimeter or two.

This is the kind of nonsense that keeps me from having a real disscussion on the issue. Nobody whats to provide some actutal measurements of a specfic audio circuit do they?

PS The stuff that happens when you first turn on the system and before you listen the the music play doesn't count does it?