Dare to fail
In the early stages of designing the Tripper Trapper, I decided to use 5V addressable LED strip for two reasons: firstly, the mad resolution of 1-LED pixels which would look nice up-close, and secondly because the newer, faster, flicker-free four-wire APA-102C LEDs that I wanted to use didn’t (and still don’t) have a 12V variant.
So I set about finding some point-of-load (POL) DC-DC converters to convert the lead-acid battery voltage (between 10.5V and 14.4V 13.8V) to the 5V needed for the LEDs. I found some quality converters which were rated for a maximum of 14V at 10A continuous output, but would require the fabrication of a PCB with either sockets for these or solder pads for these. Being lazy[1], I kept looking and found some prepackaged BEC modules which HobbyKing were selling for $14.
Bargain, right?
Turns out, these aren’t the best-designed things. The input wires are thicker than the output wires, even though this is a buck converter which will output more amps than it draws. The heatsink is wrapped in the same plastic used to insulate it electrically, with only medium-sized holes at the input/output ends. I have my doubts that they’re capable of the “20A continuous” rating they’ve been designated, but I only needed 2.5A continuous, and figured that even if they can only handle a third of that rating (say, 7A), then I can still comfortably derate them by a further 50% (say, 3.5A) for use in the awful, awful conditions faced on the playa.
Once I had the BECs, the LEDs, the controller, and a program representative of what will be running on it, I decided to test their consumption to make sure my power budgeting was adequate. My budget had estimated each POL would draw 1.08A, and my bench supply showed the current dancing between 0.92A and 1.12A. Spot-on, so far. Because of my suspicion of these cheap units, I decided to meter the output voltage under load, to make sure they weren’t going to die of voltage-related injuries on the signal line, which is when my DMM showed me a disappointing readout.
4.6V.
I knew it. I shouldn’t have cheaped out, and now I have six flaky, over-specced, crappy buck converters which can’t even handle 1/10th of their rating. I was livid, because now I had the options of either racing through a PCB design and spending even more money on better POL converters, or finding out some other awful problem like a huge voltage sag in the desert heat which would render them useless.
I will do science to it
until it tells me the truth.
After the initial wave of panic passed, I stopped. I knew better than to jump to conclusions. I tried another unit, hoping that maybe it was just one dud out of the batch. No such luck. It was time to find the truth. It was time to test all the things. It was time for… Science!!!1
I tested voltage high, and voltage low. Current come, and current go. Probing all the points, and, lo! Wait a minute here. There’s 5.1V at the output of the BEC. But there’s only 4.6V at the LEDs. That means I’m dropping half a volt over 40cm of wire. 40cm of cheap alligator clip wire. I removed the alligator clips, twisted the BEC outputs onto the LED inputs, and sure enough, 5.1V at the LEDs. There was half a volt of VDROP across nothing more than my test leads.
This exercise should come as little surprise to anybody with programming experience, where no matter what the problem or language, it is always your fault. The trick is to remember this even when you’re sure that the thing you’re testing shouldn’t be doing what it’s doing. It will save you from wasting your time, your money, and most importantly your sanity.
[1] as in, putting off the opportunity to learn how to design a PCB properly.