Wednesday, December 26, 2007

Variable DC regulator V1

I'm getting really sick of those little "wall wart" power bricks for everything. My collection is getting silly, particularly as the multi-voltage ones seem to never quite have all the voltages I want.

My recent purchase of an Asus EEE PC is a good case in point, I imported it from Hong Kong so it has a power brick with a dodgy mains adapter plug on the back of it. This device needs 9.5V DC to charge - none of my existing adapters will provide this - hence this week's mini project.

It's a classic LM317 variable DC regulator in a box with a knob. The circuit is straight from the data sheet here. In my version I have 100R for r1 and the potentiometer is 1K. On the input I have a bridge rectifier so I can plug either polarity in to it without fear. Construction is on a tag board, probably a heatsink and some ventilation holes will be needed but the LM317 is well protected against over temperature and current problems.

I'm calling this project version 1 as it turns out it can't supply the 2.3A needed to charge the EEE PC, still useful for other things, but a version 2 is going to be needed for my original objective. Looks like I need an LM150 or LM138 for that.


Anonymous said...

You can add a PNP pass transistor to the regulator to boost its current capability. Two resistors are used to set its bypassing point, and current limiting point.

The base of the pass transistor is connected to the input of the regulator, the collector to the output, and the emitter through a resistor to the unregulated source. The base/input of the regulator is also connected to the source through the other (larger) resistor. This resistor in series with the input of the regulator is calculated to drop the base-emitter drop of the pass transistor at the turn-on point you desire, at which load current starts getting shunted around the regulator. As the collector of the pass transistor connects straight to the output of the regulator it automagically keeps the output voltage regulated.

The other resistor in the emitter of the pass transistor (quite small) is sized to drop approximately the base-emitter drop at the current you wish to limit at. Effectively it brings the emitter voltage towards the base and cuts off the bias, shutting down the transistor and preventing further current being drawn at the expense of regulation. You could omit it, but it would defeat the limiting protection of the regulator and the first short circuit would likely do damage.

The main things to watch out for are sufficient Beta in the pass transistor at the currents in question, and the power dissipation in the resistors and pass transistor. You can also run out of headroom with the extra voltage drops. The bias current is easily worked out with from the transistor spec, and at a minimum the regulator must be able to pass it plus the current through the emitter-base biasing resistor.

You can also parallel regulators with some diodes, but I've never been a real fan of this approach, especially with the LM317.

Without doubt google will find webpages which explain both approaches better than I can without drawing a diagram. :)

Peter Marks said...

Thanks Alan. I remember building such a regulator using 2N3055s when I was young but I think they are NPN.

Anonymous said...

Yep, I built a supply for my CB years ago like that too, with a 2N3055. It worked well, and I still have it. The the 2N3055 is a pretty ancient device and it has poor gain especially with large collector currents, I had a second smaller PNP transistor in there in as a Sziklai pair to supply the bias current needed.

The TIP range has some devices with better gain at high currents. Which means you can do it with a single device.

IIRC the back of the DSE catalogue had the circuit and maths for designing such circuits. If not, the 78xx datasheets from National definitely had it.