# Voltage Regulation

• posted

Derived from a capacitor multiplier with voltage attenuation capabilities. It is simply several capacitor multipliers in series with the voltage drop distributed across the bjt's. R4 controls the load voltage while R1=R2=R3 causes the voltage drops across the transistors to be evenly distributed.

Theoretically this allows one to use large voltages with the capacitor voltage decreasing for each stage. The kth capacitor from the bottom has a voltage of k/n*V while each resistor and transistor has V/n. Tests show that this has significantly better ripple reduction than one stage alone for the cases I've tested. A factor of almost 100x.

What seems nice about this circuit is the reduced voltage requirements of the transistors which also allows for higher beta. The power dissipation for each is also reduced but the same as using one transistor. Overall this seems to be a much better circuit than using one stage and using linear regulation alone.

The real question is how this would work out in practice?

• posted

You don't have to do waste voltage like that--just split the base resistor into N sections and use N capacitors (actually N-1 capacitors, since you want one at the end for short circuit protection of the transistor). You can get 140 dB in one stage in the tens to hundreds of kilohertz.

```--
Dr Philip C D Hobbs
Principal```
• posted

f

I don't know what you are talking about. I don't believe you understand the circuit. The transistor is used for voltage regulation and capacitance multiplication. It has to waste voltage so that one can also attenuate the voltage as it is meant for variable supply. Even without it one still needs the transistor for capacitor multiplication.

The capacitor multiplication is used to lower the capacitance requirements and hence whatever capacitance you are using to get your

140dB I can do the same using a much lower capacitance which essentially gets multiplied by the beta of the bjt's. With your method I do not see any need to use resistors and extra capacitors and one could just use one large filter capacitor. The problem is then how do you get the voltage attenuation/variable supply?
• posted

"James Rollins"

** It wouldn't.

At the moment of applying Vcc, Q1 cops the entire voltage as C1 is uncharged.

Pffffttt....

..... Phil

• posted

Your point? C1 will charge up... duh! Every schoolboy knows this.

• posted

r
s
s

Your right that Q1 in the circuit will start with Vcc across it but it is an easy fix and because in general in a real circuit the supply voltage will not instantly be at Vcc but ramp up to Vcc it is not necessarily an issue. It is an easy fix by adding a soft starting feature. In fact the circuit I plan on using this with will have such a feature since the supply will charge up a bank of large capacitors which will supply another circuit. Alternatively one could use a higher VCEO for the the first Q1. Similarly for Q2 being higher than Q3. It allows one to use progressively lower voltage rated stages and increase the regulation by another factor which I guess is beta^n for n identical and ideal stages. Hence the capacitance is C*beta^n which is significant improvement over one stage.

By implementing some safety circuitry and soft start the problem you described should be eliminated.

• posted

This used to be a ripple regulator, before the term was misapplied to hysteritic switching regulators. In any event a capacitive multiplier is used where regulation is not inherently intended, simply to reduce ripple actively. The normal circuit simply establishes bias for the pass element with sufficient headroom to absorb the ripple.

Linear regulation requires a reference and controlled gain. Your circuit has neither, so it will only regulate as effectively as the input voltage is regulated. The use of multiple series elementsdefeats the simplicity of the original without providing the benefit of regulation.

Check out p12 +13 of the LM10 data sheet as an example of the basic concept of a floating adjustible regulator.

The capacitor being multiplied is C1 in these drawings.

RL

• posted

Perhaps I was unclear. (I'm assuming that your circuit is intended for a practical use, not just to learn SPICE or keep transistors warm.)

Your circuit wastes a lot of power to solve a nonexistent problem, namely feedthrough due to the CB and CE capacitances of the transistors. There's no point in doing filtering at the collector of the capacitance multiplier, since this feedthrough is never the issue--it's always the tradeoff of voltage drop vs RC time constant. If you take your circuit, delete all but the bottom transistor, and short its collector to the positive supply, you'll get equivalent performance with a far smaller voltage drop. The bottom resistor can usually be very big--it needs to pull just enough current so that the voltage drop is a bit more than the worst case ripple.

I do this sort of thing all the time, usually with Darlingtons, e.g. MPSA14s, so that I don't have to use such big caps. The CB and CE capacitances (a few pF each) get swamped out by the bypass caps on the base and emitter, yielding easily 120 dB and sometimes 140 dB rejection at tens to hundreds of kilohertz in a single stage. I didn't invent it--it's a standard technique in low noise circuitry, especially in places like photodiode preamps, where noise on the bias supply looks exactly like TIA voltage noise and so has to be down around 1 nV/sqrt(Hz).

If you really want to use something like this as a voltage regulator, you're way better off using a DC-DC converter feeding a single cap multiplier stage. If the switching frequency is in the sweet spot of the cap multiplier, and you keep the jiggly power supply plane and stray flux away from your front end circuitry, you'll never know the switcher was there.

If you're really interested in reducing the voltage rating of the caps, then you must be intending to waste most of the supply voltage at all times--because otherwise, when the output is at its upper limit, all the caps would see nearly the same voltage.

So I'm puzzled as to why you'd want to do this in a real circuit.

Cheers

Phil Hobbs

```--
Dr Philip C D Hobbs
Principal```
• posted

Q1 might find its self as a Colpitts common base oscillator. This assumes that you don't exceed its SOA at start up.

Have you considered:

--+--- ------+------+------+---+- ! \\ e/ ! ! ! ! ! ----- ! \\ \\ =3D=3D=3D ! ! ! / / ! -/\\/-+ ! \\ \\ ! ! \\! ! ! ! =3D=3D=3D !-----+ +--- ! e/! ! ! ! ! \\! ! --------+ !----+ ! e/! ! \\ ! \\ / \\ / \\ / \\ ! \\ ! GND GND GND It needs only one large transistor.

• posted

r
s
,

s of

.

t -

"> The bottom resistor can usually be very

I love this circuit fragment (with one transistor), but have never used the resistor at the bottom. (R4 in the original post.) What is it's purpose? Is it only needed if the supply voltage is small? (a few volts or so.)

BTW I find that adding a second RC pole helps, but adding a third is not worth the extra bother.

George Herold

• posted

A BJT saturates when the collector goes more than, say, 200-300 mV below the base. (It depends on the device--small signal devices are better than power devices.)

For 60-Hz high voltage supplies, you can sometimes have 10 volts or more of ripple, so the cap multiplier will saturate hard on every cycle. If you use the bottom resistor to drop the base voltage by 11 volts, that

10V ripple won't cause the transistor to saturate.

Cheers

Phil Hobbs

```--
Dr Philip C D Hobbs
Principal```
• posted

or

ts

g
,

s of

Gain is a misnomer. Amplifiers to not amplify but attenuate or modulate. In any case the circuit I gave does have "gain". it is simply R4/(R1 + R2 + R3 + R4).

The reference is R4 in this case. The capacitors hold the gate current steady. Just because there is no zener does not mean there isn't a reference. Of course there are better methods and ultimately I would be using a precision reference because I would be replacing R4 with more circuitry that allows for programmability.

In fact the precision regulator in the LM10 datasheet is very similar.

As far as regulation is conserved I believe you are confused. Linear regulators do no supply power but only attenuate it. Hence why all linear regulators need some headroom. In fact the circuit I gave is not much different than your standard linear regulator(See AOE for the basics of regulation) except a zener is not used for a reference.

I believe you are not looking at the circuit closely enough because it is a simple regulating circuit that accomplishes all the goals for basic regulation. What it doesn't have of course are the more advanced features found in modern regulators such as current limiting, over voltage, etc. The circuit given of course is not meant to be the final result but the basis of a practical circuit.

How does the regulator handle the high voltage? The precision regulator on page 12 still has to handle Vcc across it unless it cannot adjust the full supply. Hence I would need a regulator that can handle approximately 1000V. This is quite easy to see since gate of Q3 is ~ 3 diode drops above the output voltage which needs to swing approximately the full supply and hence the op amp would need to handle approximately the full supply. Unfortunately this circuit won't work. Simulation also

• posted

r
s
,

s of

I do not understand what you mean "wastes" a lot of power. The circuit wastes not much more than using just one transistor. The capacitors and resistor network waste virtually no power. Only the transistors waste power. Yet they have too since this is a linear regulating circuit. It wastes no more power than one transistor except for the additional voltage drops and potentially extra gate current... yet in reality this is not necessarily true since HV BJT's tend to have very low hFE than lower voltage BJT's.

The bottom resistor is used to program the output voltage. This is not just regulating but stepping down the voltage. You are not understanding the purpose of the circuit. You think the extra transistors are useless but they are there to reduce the voltage drop across just using one so that lower rated transistors can be used. Also I am not trying to achieve a low dropout regulator but a programmable one. A few volts below Vcc is not important.

The wasted power is maximum when The transistors drop half the supply voltage which is required since the circuit is linear. A switching circuit would be more optimal but because of the low power requirements this requires a huge inductor in a buck configuration.

This circuit, at least in simulation, gives about 100x more ripple rejection than one stage alone and the power dissipation is almost the same as a single stage and in fact might be better.

• posted

*sigh*
```--
Dr Philip C D Hobbs
Principal```
• posted

Observe the source (googlegroups), apply filter, sip a cold one ;-)

(Reminds me, Monday evening I found the solution to jaw pain after a root canal... one bottle of Claret ;-)

...Jim Thompson

```--
| James E.Thompson, P.E.                           |    mens     |
• posted

I keep hoping...such a cockeyed optimist I am, to be sure. I really like helping people learn stuff, but it's a great deal easier with some than with others.

How does Google maintain such a monopoly on these folks? (I mean, I should probably be grateful as you say, but I really don't get it.) They all want everyone to listen to their latest half-baked idea, but won't listen in return, and won't accept any response other than oohs and ahhs--anybody who doesn't bow in homage is obviously incompetent, regardless of evidence to the contrary. Accepting the fact that their brainstorm is an inferior derivative of a long-known technique is fatal to their self-image, of course, and so must be resisted at all costs. That must be very unpleasant.

My local school has a program where everyone gets to invent the city of the future--and all get oohs and ahhs regardless of whether their idea makes any sense whatever. We need another Sputnik scare, right away.

Cheers

Phil Hobbs

```--
Dr Philip C D Hobbs
Principal```
• posted

I haven't figured that out, either. For some reason googlegroups seems to draw mostly nutcases :-(

Unfortunately I think we're going get far more than a Sputnik (that happened in the Fall of my senior year in high school).

What I'm afraid of is an atomic hit by one of the nutcase nations we now kiss-up-to rather than tell them, "Knock it off, or else".

Pretty hard to deliver "else" when your president is a pansy :-(

The only fortunate thing is that the most likely target cities are full of leftist weenies who voted for the pansy ;-)

...Jim Thompson

```--
| James E.Thompson, P.E.                           |    mens     |
• posted

You know how many of the Jewish laws regarding what's acceptable to eat (Kashrut) just so happen to line up with what was reasonably *safe* to eat given the technology available a couple thousand years ago (e.g., lack of refrigeration and an understanding of microbiology)? I think something similar could be said about the admonishments regarding being overly prideful... ignoring such admonishments is just setting oneself up for even greater unpleasantness as you describe.

I've mentioned before that there is an odd facet of younger peoples' culture today that saying "I don't know" is apparently considered worse than just making up an answer and hoping that no one can decisively demonstrate your error.

Your kids apparently didn't play much no-score soccer, eh? :-)

Hey, now that you're independent, and given that you like to teach... you might think of putting up a "premium" web site similar to Doug Smith's here:

. At \$50/year, it's dirt cheap for companies and even most U.S.-based individual engineers, yet I suspect he makes enough from it to make it more than worth his time adding content (also note that he has a *lot* of 100% free information on his regular web site). You could even start off just going through your book, similar to how it would be used in a college-level course. This could tie in nicely with the upcoming release of the 2nd edition...

Unless they're extraterrestrial, I don't think there'll be one... all countries today have the technology to kill one another quite effectively; it's just politics, the threat of retaliation, and on rare occasion some clear-headed thinking that keeps us -- relatively -- safe.

---Joel

• posted

Nice. Like my town, for instance. :(

You really need to go to Compassionate Conservative school for a refresher. Or find a kindergarten where they serve claret.

Cheers,

Phil Hobbs

• posted

"Jim Thompson" wrote in message news: snipped-for-privacy@4ax.com...

Was it visible from... Massachusetts, was it (where you were living at the time)?

Well, since you set up the context here with our being nuked by some nutcase, I suppose you're right... those guys don't understand that even if Obama wouldn't personally retaliate in kind, he'd rapidly lose control to those who would. I fully expect Jim Yanik's going to be heading on up to D.C. to personally wrestle the red button out of Obama's hands as soon as he hears that the nutcases have launched... :-)

In the end, everyone loses -- plenty of needless death.

Hey, did you read that book, "The Idea Factory: Learning to Think at MIT" that someone here (John L.?) recommended? I read it on a trip back to Wisconsin this past week and enjoyed it, and I figure you might, even though it is set in the early '80s (well after you graduated!).

---Joel

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.