"Black" regulator

Several things. The load current and the 2 nF feedback capacitor are two biggies.

If you mean C1 and C2, they divide the ac feedback voltage which is necessary for this circuit to oscillate. It is a switch-mode power supply and achieves high conversion efficiency via that mode of operation.

Hard turn-on means that the base-emitter voltage is high enough to drop the C-E resistance to its lowest value ... very important in switching applications. Also, not all transistors can achieve the low C-E resistance needed for high efficiency. MOSFETs have mostly taken over in this arena as they can achieve very low drain to source resistance when switched fully on.

Just for the heck of it, I entered the circuit into a PSPICE simulator and, as usual, got crappy results. Simulation of circuits of this ilk is still an iffy science. Especially if one wants to evaluate efficiency and that sort of stuff.

I have a few questions about the Black regulator described in the above article.

-- What controls the switching frequency?

-- What determines the relationship between the two capacitors?

-- What transistor datasheet characteristics would indicate its applicability in that circuit? The article states that "hard turn-on" and "hard turn-off" are important. (Oddly, these aren't spec'ed as such.) Sure enough, playing around in SPICE, a 2N2222/2N2904 pair switches poorly, and works only under very light loads.

Also, playing around in SPICE, the power dissipated in the load is off by almost a magnitude from the RMS power entering the circuit. For a 48V battery voltage source V1, RMS(I(V1))*RMS(V(V1)) gives about 16A, but the dissipated power in the dummy-load resistor is about 2.5A. The dissipated power in all other components seem suitably small, in the micro-, nano-, and milliamp ranges. I'm at a loss to find where the losses are. The PG() and PS() results are similarly non-sensical; less power enters the circuit than the load dissipates.

Sigh... the load current...

There are 3 caps. The feedback cap should be 1/3 of the larger one, and the smoothing cap on the output shouldn't be much larger than 47uF, or the thing won't oscillate.

The transistor needs to be able to pass the maximum current the inductor passes, which is dependent on the inductance, the load, the smoothing cap, and the frequency.

the load is off

My god, man, this thing is meant to be used with up to abou 100mA. Use a real SMPS controller chip. See the linear or maxim site. If you have ltspice (a free spice from

The losses in a SMPS are usually in the pass transistor, the inductor (which has non-zero resistance and dissipates energy, despite what SPICE says) and the diode from ground. The pass transistor will dissipate huge amounts of energy unless it is turned completely off, or turned completely on. When it is partially on, the current is high, but the voltage across it is also high, leading to P = V*I being high. If you graph power in the transistor, you'll see what is happening. Your problem is probably due to the wimpy drive, causing the transistor to only partially turn on...

The way it is supposed to work is that the zener establishes a switching reference voltage at the NPN base. When Vout on the emitter discharges to a minimal Vbe~0.6V below NPN base due to the load, the NPN begins conducting, its collector current drives PNP base causing it to conduct, and the PNP collector current drives into the inductor, free wheeling diode, 2nF junction. It is assumed that this turn-on is fast enough so that most of that PNP current couples through the 2n into the junction of the zener, 6.8n, and NPN base. This is the bad part, that feedback current divides between 6.8n high frequency shunt and the low impedance of the zener, leaving just a fraction of the low frequency component of current to enter NPN base and regenerate the turn-on action. So it makes no sense to have things arranged this way. Continuing with the fantasy for the moment, what little feedback couples into NPN base, with gain greater than one BTW, causes more PNP conduction and more base drive until the PNP is saturated, and the full Vin is applied to the inductor, current continues through the 2n to sustain this conduction in some unknown way, and the current in the inductor builds up at (Vin-Vout)/L amperes per second. This is supposed to continue until Vout builds enough to back bias the NPN base, which is still receiving base current from God knows where, causing a negative regeneration of the NPN-PNP pair non-regenerative thingamajig, assisted by the diode clamp and inductor feedback pulling so much reverse current from base junction that the zener even loses bias, causing its voltage to fall so as to maximize the chance of Vout blowing through the NPN emitter-base junction in reverse mode. Maybe Roger Hamlett can explain things better, but the circuit is a hopeless kluge which in my estimation is probably working off parasitics due to sloppy construction practices along with misapplied component technology.

I agree with this, maybe an inductor in series with the zener, but then you need to account for that stored energy later in the cycle. I suppose the circuit works as-is because the output voltage doesn't change very fast so Vbe in series with Cout can have a lower dynamic impedance than the zener.

from base junction that the zener even loses bias, causing its voltage

Yeah, except for the peak is limited to the capacitive voltage divider ratio, and if it does pass ground level, the zener forward-biases and clamps the pulse.

Doesn't work for Vout > 12V or so, but you could add a zener across the EB junction to spare the transistor.

LOL. You don't understand elegant circuits very well, do you?

Tim

-- Deep Fryer: a very philosophical monk. Website:

nah, I managed it just fine. its a crap circuit though. apply an overload and see how well it copes.

Cheers Terry

A short circuit takes it out of the Standard operating region, thus this condition is invalid and can be ignored.

Check the next page with the three-transistor model. It has current limiting.

Tim

-- Deep Fryer: a very philosophical monk. Website:

apply a short to the output, and tell me again how "elegant" this POS is....

Cheers Terry

Below is a LTSpice drawing for an even worse regulator. It has only one transistor. The total parts count is lower than the "black" regulator.

Version 4 SHEET 1 880 680 WIRE -208 144 -208 112 WIRE -208 240 -208 224 WIRE -80 112 -208 112 WIRE -80 144 -80 112 WIRE -80 288 -80 224 WIRE -80 320 -80 288 WIRE -80 400 -80 384 WIRE 0 288 -80 288 WIRE 0 320 0 288 WIRE 0 400 0 384 WIRE 48 288 0 288 WIRE 176 288 128 288 WIRE 240 112 -80 112 WIRE 240 208 240 192 WIRE 240 240 240 208 WIRE 240 400 240 336 WIRE 288 208 240 208 WIRE 384 208 352 208 WIRE 384 320 384 288 WIRE 432 208 384 208 WIRE 528 208 496 208 WIRE 528 368 528 208 WIRE 528 400 240 400 WIRE 528 400 528 368 WIRE 528 480 528 464 WIRE 672 368 528 368 WIRE 672 384 672 368 WIRE 672 480 672 464 FLAG 384 320 0 FLAG 528 480 0 FLAG -80 400 0 FLAG 672 480 0 FLAG -208 240 0 FLAG 0 400 0 SYMBOL npn 176 240 R0 SYMATTR InstName Q1 SYMATTR Value 2N2222 SYMBOL ind2 224 96 R0 WINDOW 0 47 -15 Left 0 WINDOW 3 42 13 Left 0 SYMATTR InstName L1 SYMATTR Value 220µ SYMATTR Type ind SYMATTR SpiceLine Rser=0.1 SYMBOL ind2 144 272 R90 WINDOW 0 4 56 VBottom 0 WINDOW 3 32 56 VTop 0 SYMATTR InstName L2 SYMATTR Value 0.5µ SYMATTR Type ind SYMBOL zener -64 384 R180 WINDOW 0 24 72 Left 0 WINDOW 3 24 0 Left 0 SYMATTR InstName D1 SYMATTR Value BZX84C6V2L SYMATTR Description Diode SYMATTR Type diode SYMBOL res -96 128 R0 SYMATTR InstName R1 SYMATTR Value 680 SYMBOL cap 512 400 R0 SYMATTR InstName C2 SYMATTR Value 33µ SYMBOL ind 368 192 R0 SYMATTR InstName L3 SYMATTR Value 220µ SYMATTR SpiceLine Rser=0.1 SYMBOL schottky 432 224 R270 WINDOW 0 32 32 VTop 0 WINDOW 3 65 105 VBottom 0 SYMATTR InstName D2 SYMATTR Value 1N5818 SYMATTR Description Diode SYMATTR Type diode SYMBOL cap 352 192 R90 WINDOW 0 0 32 VBottom 0 WINDOW 3 32 32 VTop 0 SYMATTR InstName C3 SYMATTR Value 0.1µ SYMBOL res 656 368 R0 SYMATTR InstName R2 SYMATTR Value 50 SYMBOL voltage -208 128 R0 WINDOW 123 0 0 Left 0 WINDOW 39 0 0 Left 0 SYMATTR InstName V1 SYMATTR Value 18 SYMBOL cap -16 320 R0 SYMATTR InstName C1 SYMATTR Value 10µ TEXT 32 232 Left 0 !K1 L1 L2 0.7 TEXT -254 506 Left 0 !.tran 0.1 startup TEXT 24 144 Left 0 ;A few turns of\\nhookup wire\\non L1

then simulate that, and laugh (I did).

Cheers Terry

PS ratshit gatedrive = ratshit power supply. QED.

If it's so simple, you can make an ASCII drawing so the rest of us can look at it and comment.

Ken Smith said

Just curious. How does one import this into LTSpice?

(1) Put the post into a file called "somethingorother.asc"

(2) Use your text editor to snip off the stuff before and after the LTSpice stuff.

(3) Tell LTSpice to open it.

Here it is:

-----+------------------ ! ! \\ ) / ) \\ ) L1 / 680R ) 220u ! ! 0.1u 1N5818 ! +----!!----+-->!--- ! ! ! ! ! ! ) ! ! ! )220u ! ! K1 L1 L2 0.7 ! ) ! ! L2 !/ ! ! +----+----)))----! GND ! ! ! 0.5u !\\ e ! /-/ --- --------+---------+---------- Vout ^ --- 10u 2N2222 ! ! ! --- 33u GND GND --- ! GND

L2 is a few turns of hook-up wire on L1.

Ken Smith said

Thanks!

I'm finding LTSPICE on hell of an app.

I've used Saber at work for years (ugh). Recently I was convinced to fire up PSPICE for a test drive. I immediately like it better and spent a few months coming up to speed. Then I stumbled across LTSPICE. Nice!

I continue to be amazed at it's capability/cost/ease-of-use tradeoff.

In article , Homer.Simpson wrote: [...]

It works quite well under "wine". It is very fast on my 64 Bit system.

In article , Winfield Hill wrote: [....]

I said it was worse didn't I. :)

Yeah, but I didnt really believe you :)

Cheers Terry