"A rose by any other name would smell as sweet..."

[Rick Law]

I just put in a 5V regulator using the AZ431 Voltage Shunt Regulator I have at hand.  I also have some LM317 Voltage Regulator (TO-220) but my TO-92 packaged AZ431 fits the space better so I use the AZ431.

After I got it working, I got curious.

For the AZ431, I use a current limiting resistor for the 10mA load I need from the source voltage (20V).  That external resistor will “burn up” the excess voltage and the AZ431shunt the extra current I don't use to ground.

For the LM317, it is not a “buck conversion”, so I assume the LM317 will draw the 10mA at source voltage and burn off the excess energy one way or the other and give me the 10mA for the load.

Either way, I am burning up 10mA 15V (20V source, 5V load).  The LM317’s schematic is a good bit more complex than AZ431.  All that extra in the LM317 must do something, but I can’t quite put my finger in it.  So, I am rather puzzled if apart from having the trouble of choosing the right current limiting resistor, what else?

Okay, okay, I know the LM317 can float, thermo overload...  But from a regulation stand point, what is the main different between a “Shunt Voltage Regulator” verses a “Voltage Regulator”?  Just a different name?

Thanks for enlightening me!

Rick

[codeboy2k]

shunts work best when you have a fixed voltage at a fixed current that you want to supply. They aren't good at variable currents, because they become inefficient when the load demand changes and needs a lower current, and the shunt has to shunt away the excess current as wasted heat. If not engineered carefully the shunt can get past its maximums and be damaged.  A shunt regulator always has to be designed to withstand the worst case current demand.  This can make them inefficient when the load is changing because they are always wasting power.  Shunts are useful in that if you know you have a fixed voltage and a fixed current, you can size the source resistor so that it drops most of the voltage, and not the silicon. Resistors are cheaper and hardly ever need to be heat sinked.

A linear regulator will only waste excess heat as the current increases, so (fixed voltage) linear regulators actually work best at lower current demands. As the current increases a linear will have to waste heat, which may require a heatsink. A heatsink that you might not need if you used a resistor and a shunt.

variable voltage and current (i.e. power supplies, excluding switchers) are almost always linear regulators as the main power element, often times with shunt regulators providing internal references at fixed voltages and really low fixed currents. Sometimes shunt elements are used in the feedback loop of these linear regulators to shunt current away from the main power drive circuit -- this provides the current and voltage regulation.

The actual choice of when to use a shunt vs a linear regulator depends on many factors and there is no real hard answer to say one is better than the other.

[Dave]

There are two types of voltage regulators references: Series and shunt.
The reason I crossed out regulators, is that you would never really use a shunt voltage reference to regulate some supply rail that delivers a substantial amount of current. There is simply too much wasted power for regulation. Shunt references are useful for high impedance loads (like opamp inputs), so the current through the reference remains fairly constant at all times.

Series voltage regulators/references are far more efficient, because the excess voltage is dropped on a series pass transistor - when the load requires less current, the circuit inside a series regulator will slightly close the pass transistor and there will be less power drawn from the source. In other words: The load requests less current, the regulator draws less current from the source.
A shunt regulator is a different story. Let's say you have a fairly constant supply voltage (20V in your example). In order to get a constant voltage at the output of the regulator, you would need to pull a constant current through the drop resistor, to have a constant voltage drop. So, if the load suddenly decides it wants less current, in order to maintain regulation, the shunt regulator would need to draw the additional current the load doesn't want anymore. To simplify: The source will always be loaded with the same current, the regulator will have to burn all the excess current.

To sum up: Series regulators will draw as much current as their load needs (plus a tiny bit more for itself) and a shunt regulator will constantly draw the same power, whether the load needs it or not. Series regulators are more efficient for variable loads.

Here is something interesting:
Check out the schematic of the LM317 series regulator. You will notice that inside it actually has a 1.25V shunt voltage reference fed by a constant current. The internal amplifier then uses this reference voltage, compares it with the voltage on the external divider and tries it's best to keep the two equal by controlling the series pass transistor.

[c4757p]

One point, along the lines of the distinction between series and shunt regulators, is that the 431 can be converted to a series regulator. All it does is try to draw current into the cathode to make the REF pin 2.5V. If the voltage going to REF comes from the output of a pass transistor, it will end up controlling the pass transistor rather than directly regulating the cathode voltage.

[Rick Law]

Thanks, codeboy2k, Dave, and c4547p,

I understand the wording difference now.  It is not just a Rose by another name, there is a substantive difference.

Thanks all for your knowledge.

Rick

[Rick Law]

One point, along the lines of the distinction between series and shunt regulators, is that the 431 can be converted to a series regulator....

Hello, c4757p, as expected, it works.

Since I was being so dense earlier, I decided to make the most out of this experience.

With only 3 transistors at hand, one happens to be a 2222a.  So I disassembled the TL431 regulator circuit and assembled the circuit you depicted.  I use a 2k+2k for voltage divider, 100ohm 1W for current limiter between source and circuit.
@    1ma load it draw 15ma (Vout=5.013 from 20V source)
@  10ma load it draw 23ma (Vout=5.013)
@  20ma load it draw 33ma (Vout=5.013)
@  40ma load it draw 50ma (Vout=5.013)
@  80ma load it draw 86ma (Vout=5.011)
@120ma load it draw 122ma (Vout=5.009)
@130ma load it draw 131ma (Vout=4.844)
Letting the TL431 controls another transistor neatly turned it into more linear than fixed current draw.  With a smaller resistor instead of the 100ohm, I can pull in more current while output stays stable at just over 5V.  Looks like it takes around 10ma to drive this circuit, so it won’t save power in this case - but it sure was a nice learning experience.

In hind sight, I was being dense.  The LM317 is floating without ground.  What comes out of the output pin must equal to what goes into the input pin minus the small amount of reference current.  So, input current varies with the load current.  Obvious.

The TL431 has a ground, the current is controlled by the limiter; so, whatever the load doesn’t use needs to go somewhere.  Ah “shunting to ground” in action.  Obvious.

Only a dummy can’t see that.  I kept locked into thinking – is there a feature/function hidden in the LM that caused the name difference.  No – it is as the name implied for the TL431, shunt regulator - the excess current shunted to ground.

Good thing about a free country is, everyone is free to be a dummy.  Bad thing is, I have been exercising that right too often lately.

Thanks for helping!  I an having fun with this stuff,

Rick

[c4757p]

Don't feel bad - it's not really that obvious. I think it's more one of those things that seem stupidly simple just in hindsight, especially if nobody has explained it to you. You seem to understand it just fine now 

100ohm 1W for current limiter between source and circuit.

Except I'm not really sure what you mean by that. Where's the 100 ohm resistor, again?

[c4757p]

One more follow-up on the behavior of the TL431 - because the cathode voltage and the REF input don't have to come from the same place, there are some rather clever uses for it. One of the most common places you'll see TL431 is in isolated DC-DC converters. You just use an optocoupler, and design the input side so that as current into the opto increases, voltage output decreases. Then configure the TL431 to pull its cathode current through the optocoupler, and it will regulate the output across the optical barrier just fine.

I'm sure you can think of other clever uses for it, too. One simple one that I've used it for is as a reasonably precise constant current source, with just two resistors and one NPN transistor. It has an error term - the transistor base current - which can be small if you use a high gain transistor. See if you can figure that one out.

[Rick Law]

Don't feel bad - it's not really that obvious. I think it's more one of those things that seem stupidly simple just in hindsight, especially if nobody has explained it to you. You seem to understand it just fine now 

100ohm 1W for current limiter between source and circuit.

Except I'm not really sure what you mean by that. Where's the 100 ohm resistor, again?

First time around, I soldered the transistor backward - I spotted the reversal while soldering the TL431.

So, after assembly and recheck, I decided to added a 100ohm between the current source's V+ and the circuit to limit the current to the circuit (in case I had another mistake).

[Rick Law]

One more follow-up on the behavior of the TL431 - because the cathode voltage and the REF input don't have to come from the same place, there are some rather clever uses for it. One of the most common places you'll see TL431 is in isolated DC-DC converters. You just use an optocoupler, and design the input side so that as current into the opto increases, voltage output decreases. Then configure the TL431 to pull its cathode current through the optocoupler, and it will regulate the output across the optical barrier just fine.

I'm sure you can think of other clever uses for it, too. One simple one that I've used it for is as a reasonably precise constant current source, with just two resistors and one NPN transistor. It has an error term - the transistor base current - which can be small if you use a high gain transistor. See if you can figure that one out.

Yeah, that little TL431 is rather flexible and stable.  I am glad I saved that to play with.  Both the TL431 and the 2222A transistor came off a failed old 5V power supply I had.

I am thinking about different projects with this little TL431 just to learn.

[TerminalJack505]

I like to think of the TL431 as a kind of Swiss army knife of electronic components.  There are a ton of different things you can do with them.  The datasheets have a bunch of example applications.

Be aware they they don't like driving capacitive loads of certain sizes.  They'll oscillate.  You'll find a graph in the datasheet that shows the part's "Stability Boundary Conditions" (or something similar.)  Be sure to take that into account in your designs.