I was trying to reproduce part of this circuit in LT Spice and it seems to conflict with the circuit's stated purpose and I don't know which is correct.
The schematic of the circuit in question is here.

As best I can tell, the voltage at the base of Q2 should be just a bit less than the rectified voltage, meaning the voltage at its emitter should be a diode drop below that. So well good bit above 20V.
On the other hand, the voltage at the base of Q1 will be the results of the voltage divider created by VR1 across the 20Vs created by diodes ZD1 and ZD2. So 0-20V, and thus the emitter of Q1 will be a diode drop below that...

But the emitters are connected together, so who wins? LTSpice tells me the voltage is fixed at over 20 while the circuit is supposed to be an adjustable 20V supply.

And as a separate follow-up question, why couldn't you just remove Q1 and run the voltage divider straight to the base of Q2 to get your 0-(20-diode drop)V?

 

Please post the .asc file.

why couldn't you just remove Q1 and run the voltage divider straight to the base of Q1 to get your 0-(20-diode drop)V?

You can if you don't draw much current from the output.
The two transistors are configured in a Darlington configuration, which greatly increases the current gain, and thus the current draw from the pot versus the load current.

 

Circuit for future reference:


The circuit looks questionable to me. For starters, the fuse is in the wrong place. It should be placed on line before S1. I suspect the author intended for Q1 and Q2 to be wired as a darlington, but didn't do it correctly.

R1 is over specified; could be 1/2W.

 

The two transistors are configured in a Darlington configuration

Maybe that was the intention, but the schematic is wrong.
As the circuit stands, Q1 base could be driven some 20V below its emitter!
What is the purpose of D5?

 

Not sure but from here, Q1 and Q2 looks wired as a schmitt trigger

Q1 will mostly be off because it needs about 19V to turn it on

Q2 looks like a capacitance multiplier

Wrong?

 

I wasn't sure on posting the image directly, so I just posted the link. Sorry, I'll do that next time.

So the sense I'm getting is that it's probably intended to be Darlington Pair but it's mis-wired. That would make WAY more sense to me than what I see there. And really, the Darlington Pair is exactly what I'd want in my own application. In my case I have a fixed 19.5V source and I merely want to create a 0-19.5V variable supply from it. I wasn't considering the current gain needs, thanks for the reminder.

 

Maybe that was the intention, but the schematic is wrong.

Yes, I saw what I expected, not what was there.

 

In my case I have a fixed 19.5V source and I merely want to create a 0-19.5V variable supply from it.

That isn't a very good power supply. The base emitter junction drops are going to depend on current, so the voltage will vary with load.

Do you really need the output adjustable to 0V? If around 1.25V is sufficient, you'll get much better performance from an LM317.

I wasn't considering the current gain needs, thanks for the reminder.

That's generally why darlingtons are used. The beta for 2N3055 isn't very high.

 

I suspect that Q2 should be a PNP transistor - 2955 perhaps, with the collector connected to the output and the emitter connected to D5.

 

That website has many errors. It is managed by an older "student" who says on it, "I am not good enough at explaining Electronics in English. And sometimes the circuit does not work. I am sorry."
Many electronics forums have new people asking why the circuits they find on that site do not work.

The circuit error here causes Q1 to be destroyed when it is turned off because its maximum allowed base-emitter reverse voltage is exceeded causing the base-emitter junction to have avalanche breakdown.

 

That circuit looks more like power control than a regulator, even though it has some elements that make it look like a regulator, it won't work because emitter follower circuits provide no line regulation. (but a modicum of load regulation)

 

That isn't a very good power supply. The base emitter junction drops are going to depend on current, so the voltage will vary with load.

I actually have a bag of LM317s. My 19.5V supply supports about 3.5A and I'd like to take advantage of that. I was actually avoiding the LM317s because I figured they'd probably lose a lot of power to heat. But I could use the 3055 to run the current around the 317. The 1.25V minimum wouldn't be a problem at all.
Thanks, I'll probably take that approach instead.

 

Yes that design is knows as "series pass" but you will lose about 5 volts in the process.

 

This is a recurring issue in this and other similar forums.

Someone copies a circuit found in a website and builds it. Then comes to the forum asking for help, only to find out that the circuit has some fatal flaws.

 

I was actually avoiding the LM317s because I figured they'd probably lose a lot of power to heat.

You will lose that power for any linear regulator no matter what is uses to control the voltage, whether it's a 2N3055, or an LM317.
The power loss is always the input voltage minus the output voltage times the load current.

If you want more than 1.5A, just use several LM317's in parallel (there's an example circuit shown on most LM317's data sheet).

 

I was actually avoiding the LM317s because I figured they'd probably lose a lot of power to heat.

Both regulators are linear and the same amount of power would be lost to heat. If you want better efficiency, use a switching regulator.

But I could use the 3055 to run the current around the 317.

To use an NPN pass transistor, you also need a PNP.
From a Nat Semi Voltage Regulator Handbook:

 

the next uses diode in conjunction with output darlington pair

________________________________
+
heres an old Russian Reg. Schem (although i dont quite get it -- it seems to work) http://tinyurl.com/y6rcwu32 ...
... found the source Pg.31 VIN=21V , IL=300mA , VOUT=2...12V , the "leftmost" transistor protects against the SC as it switches off the darlington if the output falls too low , here's the SC current test http://tinyurl.com/yxwlmyzz