So my old 24V, 1A supply just wasn't cutting the mustard as much as it used to. just too many projects need just a little more kick, not as much for voltage, but current, atleast 3A would be nice. AND TWO OUTPUTS! Used and commercial supplies available were not appealing.

(Warning, long post, have coffee in hand)

Perusing ebay my eye caught these little regulator modules, voltage and curent limited, reportadly 5A (china amps tho, which are, as everyone knows, smaller than north american amps by about 3/5? ). A close look at these revealed that they used 10k (10 turn) pots for their voltage and current settings, which could be easily enough changed out for panel mount controls.

previous knowledge of small panel mount ammeters and voltmeters made them an easy choice (I like analog as you can see odd load behaviour) \u200b

To power the whole works took a bit of digging, turning over my shop proved unfruitfull, all the supplies I have are just too bulky for the rack-mount-ish power supply form I wanted to end up with, with enough searching a suitable supply turned up on ebay with an adjustable output.

So, this looked easy enough, the most sceptical point being the capability of the regulator modules, some tests would prove them either way.

SCOPE CREEP

What fun would projects be without scope creep? Analog meters are ok, but its really nice to be sure to the decimal of what your getting sometimes, I have to say that I can't count the times I want to know the instantanious POWER that a load is drawing.

And ya know, setting the supply digitally would be nice too....

This COULD be easy, aster all, there ARE 10k digital pots available, I could use encoders on the front, tie it all togethor with an ATMEGA32 and a usb-serial adapter. hmmmm I could make the power supply provide a dumb terminal interface, no software required for connecting, and no clutter of the front of the supply.... Those ACS712 modules looked like a great way to measure current.

PARTS FROM THE MAILBOX

The great and might mailbox started to cough up what I ordered on ebay, murphys law, the regulators did not arrive first..

When the regualtors finally did arrive I put them to the test, obviously, they would need some descent heat sinking for the life I was going to drop them into, other than that, the main concers were their ability to acually handle the 30V they said they would, their ability to handle atleast 3A continious, and their ability to control a short circuit without the lid of a semiconductor flying off.

FEASABILITY TESTING

Buying from china is such an adventure....

I hooked power to the regualtor module as provided, gave it about 12V and dialed things up and down, as the two pots weren't marked, it took some playing to work out who was who.

The current control was quite impressive, smooth, fine control, I was even able to throw on a resistor free led and run in current mode. Voltage regulation was stable, all good, now for the killer tests.

With safety goggles on I dialed it thru increasingly higher currents giving it output shorts, it survived everyone, seemingly without complaint, wow. Continious tests showed they do NOT like 5A continious, lots of heat, but reasonable amounts at 3A.

The old supply can go up to about 26V, I noticed as I was dialing up for voltage limit testing that the leds on it (3 of them, one for power? one for voltage mode, one for overcurrent?) got brighter, I attributed this to poor led biasing and ignored, untill...

The smell of something hot caught my nose, I switched off the power and felt around, leds were warm, op-amp was warm, nothing excessive, power back on, wait a min...

My eye caught a swisp of smoke, from... somewhere... and then definitly. They used a lm317, sot3, for control voltage, it seems that, the one on this board was faulty, it was input-output shorted, huh.... in the process of it frying, it killed one of the smt leds.... would the rest be ok?

Its not hard to take a TO-92 lm317, and cook the leads over to make a smt version, I replaced the burned up led in the same fasion. back to the bench for tests, and, to my surprise, everything was ok... the unit still worked fine. With the help of a 12V battery, I boosted the input to 30V and did all the tortune tests, everything looked good, the current could not be overdialed, it took output short-circuit fine, the only thing it looked like it would need was proper heatsinking.

I decided that if the baord was sandwhiched via a copper spacer to a large heatsink, it would get enough heatsinking, mechanically, only one trace would need to be given carefull attention to avoid a short. There was one more problem with these modules, the 10k voltage feedback was on the + side of the voltage divider, meaning one side of the pot went to 30V, the serial 10k digital pots can only take up to 5V, this would mean digital voltage control is out the window, oh well, I could still monitor voltage and current digitally.

LAST OF THE PARTS ARRIVE

During the 22day ebay propigation delay, I'd been working on my chassis design, I found some knobs, power switch, power lamp (retrofitted with an led), a nice filtered power connector, and fuse holder (and took to ordering some breakers from ebay)

Primary assembly took place on the weekend, by 6am I'd finished the last of the connections, it was time to give it a try.

Initial power up looked good, knobs turned, guages waved, now for a power test. Being 6am, I wanted to get done, so I went for bust, picking one of the outputs, and with its current set to about half, shorted the output.

I should mention that the lights on the regulators had been doing interesting things that I was way too tired to think about.

As I shorted the output, the current guage pinned, I scratched my head, turned the current down and did it again, this time smoke came out of one of the regulator modules. GRRR I went to get some tea...

Thinking about it, I could only work out one possable cause. I'd not thought of this before, but the regualtors used a resistor on the ground rail for current sensing, If at about the 5am mark, I'd crossed the + or - lines between the modules, then there would be no current control, as the wrong module would be getting the returned current.

With tea in hand I picked over my wiring, and sure enough, between 5am and 'the hell with labeling the wires, I wont screw this up' I'd crossed the - leads on the output posts.

DESIGN DISASTER?

This is a problem, the modules could not have their circuits crossed over, this is a problem, at SOME point, its gonna happen. I'd have to design my own buck regulator.... which might not be so bad... this means that I could design the new regulator with 5V control over the voltage, meaning I could have my digital control back!

A note regarding my tests of the ACS712 modules, it turns out, most of the ones sold as "5A" are really 30A chips, this makes for less signal resolution, grr. More over, it turns out, that even the 5A chips suffer a LOT of noise, were talking noise that is atleast 6db over the signal! grrrr.... ( useless if not filtered to death)

Design of a buck regulator begins...