Power supply for my laboratory bench

[ObliVion]

I have designed a power supply to replace, my 317/337 supply, since it was modified so many times, that I could not trust it any more. I had some multi-turn 10k pots, and wanted the voltage of both rails, to be adjustable by one of these. Therefore the negative rail is made to track the positive.

The circuit has been tested and seems to work quite ok. But I am always up for suggestions, although not along the lines of "Rip the thing apart, and use this chip from [insert brand], since this actually works.

I have some questioned though, first what are good values for the caps, at the output, I am not certain, if they should even be there. I read the TNT regulator articles, and remember, that large caps at the output didn't make things better, for the 317. I wonder if this goes for this circuit to.

The same goes for the C8 and C10, in the simulator and in reality 100uF seems a good value, for ripple reduction, but are there any side effects of having them in the circuit.

The main reason for posting this here, is to give something back, and the fact that I have not seen that many discrete designs of laboratory suitable variable supplies on this forum, and it is a good tool.

Some more details on the circuit is available here:
If it works, mess with it | Power Supply

Thanks for all the endless nights of reading through stuff here .

/Martin

[Mooly]

It's nice to see discrete design... and as always it's how it performs in real life.

Just some thoughts

It's always good to test PSU's with a dynamic load such as a FET switching a known load resistor across the output at differeing frequencies. I did such a test here, with 'scope shots a little further down.
Low cost regulator in between Jung and Flea

When it comes to caps on the output of a lab PSU, usually smaller is better with regard to protecting the circuit being powered... 100uf is getting on the large side there's a lot of energy there... 10uf or 1uf is better.

Make the current limit adjustable down to zero... not easy without adding quite a bit of extra circuitry I admit.

R4 and R7 ? appear to be able to short out the output if both set at zero ohms.
Maybe add protection diodes across each output rail.

It's good though... just test it carefully under all real world conditions.

[ObliVion]

Thank you, nice issues raised, I can deal with a couple right now. You are right even though R7, is a trimmer, I should add a resistor in series to be on the safe side. About the adjustable current, the old case, that the PSU must fit in, does not have enough space for another heat sink. If it did, I think I would simply add a couple of adjustable current sources at the output. I'm seriously considering a new case, to make this possible, it is a really nice feature.

Martin

[ObliVion]

I threw together a load according to your description. Preliminary tests, showed that spice does not burn your fingers like real life .
At this point I have the prototype wired up, dead bug style, and a PCB ready for etching, but I'm holding back for good suggestions, like Mooley has already given me. I am going to test the switching load during the next week.

This weekend I will be off-line , on a little island near Funen called Skareo, drinking bear, and enjoing the sun .

Martin

[multisync]

I am curious about the operation of Q6 bc547b, for it to conduct the base has to be more positive then its emitter which is at ground. Are you expecting that the base of the negative series pass transistor tip2995 to go positive above ground? Or is there a little schematic error?

[ObliVion]

If the negative current limiter kicks in, the voltage divider R9/R13, will go positive, and because of Q6, the positive side, will turn down the voltage somewhat, as well.
You could argue that this is not necessary, since there is a current limiter on the positive side as well. I just added it for completeness, as the negative side will track the positive rail, as the positive current limit kicks in.

Martin

[ObliVion]

Multisync you are right, there is an error in the schematic. The base of Q6 is supposed to be connected to the base of Q8.
I will update the schematic, when I have done more testing,

[multisync]

better

[TechGuy]

FWIW: take a look at using buck regulators, as a LM2576-Adj Regulator that can be adjusted with a pot and supply up to 3 amps of power. A pair or more can be easily used to supply multiple outputs. The datasheet should include a circuit for negative voltage output. Its far, easier to design and build a regulated power supply using a buck controller, than using discrete components. Why spend hours upon hours trying to squeak a linear design, when you can design a regulated DC-DC PS in a matter of minutes.

If you need a low noise output, use a PI filter (documented in the datasheet) or use a coupled inductor filter (http://www.hamill.co.uk/pdfs/ciabfbb_.pdf). Simple and low part count!

BTW: National and Linear tech have other buck regulators that can handle even much high output currents. I believe there are devices in the 10 to 20 amp range, but require external switching transistors, the LM2576 has an integrated switching transistor.

[ObliVion]

Thank you TechGuy, at one point, I was toying with an SMPS design, using a micro-controller, to do most of the stuff, I guess I like doing it the hard way. My PCB drill uses a LM2678, and it was really easy to implement. One of the goals of this design, was that I would like to use what I had in my drawers, that sort of forced me to do a linear supply. Second, I understand their workings better, and would have tried a discrete SMPS design, to get the feel of it anyway...

Another thing, if you look at the schematics, there are many similarities, to a discrete power amplifier, and I am toying with some designs in that department to, I guess that made it easier, for me to do it this way.

Martin