Building a 13.8 volt linear power supply

[ImageJPEG]

I'm into ham radio so I'm wanting to make my own, high power, 13.8 volt PSU. I've got a transformer from an old microwave that I plan on re-winding to output 20v.

I'm currently receiving some of the items I've ordered off of eBay (haven't ordered everything yet though).

I have no real world experience with developing circuits and have a very limited knowledge on them.

Eventually I'm wanting to set it up for being capable of 50 amps (overkill, I know but whatever. I've got 10 2N3055s and my schematic only uses 4).

Let me know what you guys think. Let me know what I've done wrong, I'm sure there's something lol.

https://easyeda.com/normal/Regulator_Schematic-1cab03ddd3164d38adc33d162bef3d1d

[macboy]

I'm into ham radio so I'm wanting to make my own, high power, 13.8 volt PSU. I've got a transformer from an old microwave that I plan on re-winding to output 20v.

I'm currently receiving some of the items I've ordered off of eBay (haven't ordered everything yet though).

I have no real world experience with developing circuits and have a very limited knowledge on them.

Eventually I'm wanting to set it up for being capable of 50 amps (overkill, I know but whatever. I've got 10 2N3055s and my schematic only uses 4).

Let me know what you guys think. Let me know what I've done wrong, I'm sure there's something lol.

https://easyeda.com/normal/Regulator_Schematic-1cab03ddd3164d38adc33d162bef3d1d

Put the 0.1 Ohm resistors at the emitters of the output transistors, not at the collectors. Then they will balance the load current between the transistors rather than just heating up the world.

Your simple emitter follower design should work, but will have relatively poor regulation - the output voltage will sag a bit at full load. That effect will be less with more output transistors but can't be eliminated in this simple design. 

[Benta]

Not a good design.
Regulation will be bad.
R1 should be lower (tens of threads on this subject on this forum).
The resistors should be on the emitters.
No over-current/short-circuit protection.
50,000 uF is too small for 50 A.
and, and, and...

There are lots of tested designs out there.

[ImageJPEG]

I wasn't entirely sure about the capacitor value. Since after the rectifier the voltage should be around 27v. I figured I could afford a 10v drop with the caps and still be above the dropout voltage of the LM317. I just used an online calculator to assist me in determining the capacitor value.

I knew I had to have resistors somewhere on the pass transistors, just the wrong spot, lol. Thanks.

I figured for now, my over current protection would be the fuses.

What will cause the voltage to sag under high load?

I updated the schematic.

[Benta]

What will cause the voltage to sag under high load?

The simple answer is, that you have no feedback from the output. You have voltage drop over the power transistors, over the resistors etc., all depending on load current and the batch of transistors you are using.

[ImageJPEG]

So maybe some kind of variable resistor on R2 controlled by a voltage sensor on the 13.8 rail?

[Benta]

No. Rethink the project and look at how other power supplies are made. There's nothing you can do to make this circuit better.

[Ian.M]

Also look at the 2N3055 SOA graph, specifically the DC line.    You are dropping a peak of about 13V Vce across the output transistors, certainly for longer than 1ms so the most you can count on per transistor is about 7A Ic. 
==> you'll need 8x 2N3055 to supply 50A without blowing.

[Zero999]

Here's the correct way to increase the current capacity of the LM317.

When the current through the LM317 exceeds about 60mA, Tr1 starts to conduct, diverting the current round the LM317.

This circuit has its shortcomings: mainly being the lack of short circuit protection, but this can be added fairly easily if needs be.

Obviously, more transistors will be needed for higher currents.

[Ian.M]

If you use a Sziklai pair for the transistor, you can use the 2N3055 transistors you have.  Then, to form the pair, you only need a high gain PNP capable of about 3A as a driver for them.  With suitable emitter resistors, in addition to balancing the current between the pass transistors, the limit of tha max voltage drop across R3 at the LM317's max current can be used to limit the pass transistor current, protecting them against short-circuits.   The only significant disadvantages are it adds 0.7V + the emitter resistor drop to the regulator's minimum voltage overhead, and can have stability problems with difficult loads.   See https://www.eevblog.com/forum/beginners/variable-regulator-pass-transistor/

[Zero999]

If you use a Sziklai pair for the transistor, you can use the 2N3055 transistors you have.  Then, to form the pair, you only need a high gain PNP capable of about 3A as a driver for them.  With suitable emitter resistors, in addition to balancing the current between the pass transistors, the limit of tha max voltage drop across R3 at the LM317's max current can be used to limit the pass transistor current, protecting them against short-circuits.   The only significant disadvantages are it adds 0.7V + the emitter resistor drop to the regulator's minimum voltage overhead, and can have stability problems with difficult loads.   See https://www.eevblog.com/forum/beginners/variable-regulator-pass-transistor/

I presume you mean like this?

The emitter resistors don't add any voltage drop. Only Tr1's base emitter voltage drop adds to the drop-out and is no worse than the original circuit's monster emitter follower's base-emitter drop.

That's much better. It should be fairly easy to add current limiting with minimal additional complexity.

[Ian.M]

Yes, that works, and I was mistaken about the emitter resistor voltage drop.  However if you want to add over-current protection the simple way, you need a resistor in series with the compound PNP Sziklai transistor's effective emitter so the driver Vbe decreases in proportion to the load current.

[CJay]

As always, PW Marchwood PSU is a good, reliable, RF quiet design and it's pretty simple to implement.

You're going to have to dissipate 300W of power if you load this up to 50 amps so a substantial heatsink is necessary, fan cooled even so figuring out the thermals is a decent part of the design process.

Good luck and post your results

[Zero999]

Yes, that works, and I was mistaken about the emitter resistor voltage drop.  However if you want to add over-current protection the simple way, you need a resistor in series with the compound PNP Sziklai transistor's effective emitter so the driver Vbe decreases in proportion to the load current.

Yes, that would work but the resistor would carry all the current, so how about just putting it in series with Tr1's emitter? R4 can then be changed to a much lower value and the emitter resistors will become the current sensing element.

The main downside to this hack is, it depends on the current limit for the LM317 which is unpredictable. It can range from 1.5A to as much as 3A, if I remember rightly.