iceNINE Tech

- "Quickie" pre-regulator bench power supply -

"Standards are wonderful things; there's so many to choose from!"

I do not like that game.

I realized a while back that I was spending too much time working with electronics infrastructure issues. Now, working on infrastructure is one thing; developing new things is wonderful fun. But fighting with equipment to make it adapt to what I'm doing this week, isn't. Every development system, kit, prototyping rig, really anything I had that used an external DC power supply, used a different DC power supply. Five power strips under the bench was a bit much. Expensive to keep, too. I could hear the wall-warts buzzing at night. If a few things got unplugged, I could spend 15 minutes hunting for the right adapter.

This has all changed now. I built a universal power supply designed to have anything that needed voltage that would later be regulated down to 5V (or below) plugged into it; it gave me a few amps of ~7.2VDC. I then took a poll of the gear that would run off it, found that the 2.1mm center-positive coax jack was most popular, and made the rest of the gear conform to that spec. Now everything runs off the same supply, and I'm down to 3 power strips. I have a bunch of 6-cell battery packs with these 2.1mm jacks on them, and also made up a bunch of 2.1 -> banana cables (both ways).

It really only took 2-3 hours, and it turned out I had almost everything I needed on hand, scavenged from old equipment and the like. Well, perhaps I cheated a bit and got off to a quick start - the core of the system is an off-the-shelf power supply brick made for OEM instrument use. I just packaged it up cleverly (if I do say so myself).

All high-voltage connections are insulated with heat shrink, and the Lambda unit is bolted down. Exercise care around 120AC!

The 'brick' supply in question is the Lambda LCS-A-2 [pdf], made in 1985. It is rated at 0-7V (it will actually go to ~7.5) at 2A. In 1985 it cost close to $300, I got it for about 2% of that at a swap meet a few years back. The march of innovation, I suppose. It was in a bag (a strong bag) of random rackmount gear, and I was going to use it right away until I looked more closely, and discovered that it was not particularly user-friendly. The AC entered, and the DC left, via screw terminals. The voltage adjustment was done with a screwdriver. Not a banana jack in sight. Hmm - under the bench it went.

I figured that I could use it for this, if I just packaged it up right. A little front-panel goodness, and we'd be all set. Down to the local electronics store - hmm, it's a little big for the cheaper plastic enclosures; 5" x 6" x 10". Why do most of these things cost more than what they enclose? I spent some time wandering, and inspiration struck; the medium Akro-Mills plastic bins are just the right size! A little odd-looking, yes, but this isn't headed to a client, after all.

Again, a search under the bench: a little IEC power inlet here, a few five-way binding posts there, and we're almost there. You know, I think I got that switch and LED off of an old 286? When was that? The meter was the last thing to be added. The power supply is stronger that the things often connected to it; an accidental fuse (ie, a short circuit) on a prototype would give before it would. Some indication of current consumption would clearly help. More investigation under the bench unearthed an old 0-1V analog panel meter.

(Insert standard circuit development saga here.) Suffice to say: Maxim Semi high-side current monitor (MAX4374, thank you samples program), very small resistor, op-amp buffer. The principle of high-side current measurment is simple: insert an X milli-ohm sense resistor between the supply and the load, and amplify the voltage across it. The tricky part is the high common-mode voltage encountered; remarkably close to the supply voltage, in fact, and not something most op-amps are happy with. Maxim Semi makes several ICs that are very good at this exact function, and outline more good information in App Note 746 (pdf). Driving an old, low-impedance analog panel meter from these flashy new low-power ICs also take a little doing.

(someday, a schematic here - the circuit...sort of evolved and is...poorly documented)

The meter, though labelled in Volts, actually indicates 0-1 amps, or, 0-2 amps if the internal switch is thrown (for when I'm running a lot of gear). Good enough for me. I'm aware of the inaccuracies that derive from using a low-impedance element in the circuit like this (the meter draws nearly 60mA when indicating full-scale!), but they don't bother me. I needed a rough indication only; the Fluke 4.5-digit meter is always nearby.

The moral of this story: with a little scavenging habit, it's easy to have everything on hand to build basic, useful gear whan it is needed. When I decided to build this supply, I only needed to spend a total of $2.39 'extra' on it - for the plastic bin! And now

Last updated September 3, 2004
Original content copyright © Christian Weagle unless otherwise indicated.