I have an inexpensive DC motor as part of a project I am doing for work; since it is a proprietary device, I cannot give any more information than is provided here. Suffice to say, it is a portable device with a Li-Ion battery pack and a custom PCB I designed with a MAX1758 Li-Ion charger IC and a baseline Microchip MCU (coded in PIC Assembly – yay!) that controls both an electronic lamp ballast and a small brushed DC motor. To help prevent noise/transients from affecting different parts of the circuit, everything gets its own dedicated power supply with some switching DC regulator ICs built into the PCB; unfortunately, that wasn’t enough!

My first problem was that the MCU would reset the second I tried to turn on the lamp ballast, which is a type of device that I deal with on a regular basis at work for driving many different kinds of mercury lamps – it is known for having an enormous inrush current, up to 5 times the steady-state current draw for the first 100ms of operation. This is due to the nature of the ballast, which requires a 300-700+ VAC “strike” in order to arc the gases inside the lamp; internally, this is achieved with a reactive circuit which resonates until a voltage sufficient enough to arc the gases occurs across the lamp cathodes. This inrush current creates a reactive voltage transient that can affect other parts of the circuit, and a quick and easy way to kill that transient is with a diode so that a transient of sufficient voltage to exceed the diode’s reverse breakdown voltage will be shunted through it and dissipated as heat. I had a few 1N4007 diodes on hand, but that’s actually a poor choice since it has a relatively high Vbreakdown – a better choice is the diode species known as a TVS diode. Of course, as mentioned on that Wikipedia page, there are other devices to choose from and the ability to know what to use where is something that can only come from experience. In this case, the 1N4007 helped enough to solve the problem for now, but a redesign will be required to achieve better long-term performance.

The next problem to come up was that the MCU would randomly reset after an indeterminate period of time of lamp/ballast operation. I took out our new oscilloscope and scoped out the voltage across the terminals of the 1.8VDC motor during operation:

Uh-Oh! +31V and -19V transients!?!? Well, that’s the nature of the beast and also the reason your computer fans use brushless DC motors. A quick search online turned up this robot-enthusiast-oriented wiki page that offered up some suggestions; with just the two capacitor method using some spare 0.1uF film caps I had around, the situation was cleaned up and I now faced a much less dire situation:

Problem solved! The three capacitor method even helped a bit more, but I do not have a screen cap for that. I do have some candid before/after pictures, however!

I finally built a simple bubble etching tank; I’ve seen people with them all over the place in the past, but I just haven’t had the desire to put one together until today. In the past, I’ve been etching my PCBs by just filling a pyrex tray with etchant, dropping the masked board in and manually agitating with a foam brush.

This new piece of equipment has these advantages:

1. Less etchant required; the vertical tank takes up less volume and thus it is more space efficient.
2. Air agitation; the bubbles oxidize the etchant, which allows it to more effectively etch away copper. It also creates turbulence in the liquid that helps etch very small areas.
3. Less manual interaction; the acid in the etchant is toxic stuff, and I really need to be more careful than I have been in the past.

The tank is made from some pieces of spare acrylic sheet I had, which are put together with acrylic cement and the outsides of the joints are reinforced with hot glue (just in case). I had an old aquarium air pump around and carefully hot-glued the end of the tube across the bottom of the tank through some holes drilled into the edges; I then carefully poked holes into the tube using some very small-diameter drill bits. Of course, I tested it with water before trying it with the cupric chloride etchant seen in the photos. I found that there were a lot of droplets splashing out of the tank, so I made a small lid with a spare IC storage tube. The whole construction took a total of two hours.

UPDATE: I put together a small, adjustable acrylic rig to top the tank and hold the part being etched in place.

The green top is a piece of junk acrylic sheet that I found and drilled a series of holes into. The sticks are 1/8″ squared acrylic rods pieced into an edge and covered in double-sided tape (for some extra tackiness); the ends that meet the green top piece are sanded down to round pegs that compression-fit into the holes, which allows the spacing to be adjusted for any size board that may fit in the tank. Small holes are drilled in the pegs so that small screws can be set as an added safety measure should the weight of the board cause the peg to slip, but I’ve found that it isn’t really necessary (maybe for larger boards). In fact, the actual problem is that even the thicker (3.175mm) boards I am etching for my microwave antenna array project are able to fall off the holding rods and down to the bottom of the tank if I am not careful.

So far, the whole setup seems to work well for both developing photoresist (why didn’t I think of that initially?!) and etching boards. The only complaint I have right now is the uneven bubble cover; if the board is not moved around occasionally, the parts in the path of the heavier bubble streams etch faster than the less-covered areas. Etch times for cupric chloride etchant seem to be about 15 minutes for 1/2-oz copper and up to 90 minutes for 2-oz copper (ugh).

I found a wonderful little tutorial on a better way of applying the dry photoresist film: http://members.optusnet.com.au/eseychell/index.html

Following it, I got much better results; I little wrinkling on both sides, but I was able to (mostly) work around it. I should be able to avoid that next time. The 6pt fonts all showed up well; a few missing letters (not enough pressure when I applied the film). Two of the three sections of 4pt font are vaguely legible, and the 3pt fonts are garbled similarly. None of the 2pt fonts made it, though (it was a stretch that they would, anyway).

Three of the four QFN-16 adapter boards are PERFECT (yay!), and the other once can be salvaged. One of the DFN-8 boards are almost perfect, the other two might have too many problems with the 3[mm] traces, for which the precision is extremely important, so they may not be useable (I only needed one, anyway).

Back of board after developing the photoresist. Green electrical tape is there to protect the parts ruined by the wrinkling of the film when I applyed it. The exposed section is for the back of the QFN-16 adapter boards.

Now for the front of the board. A few places of missing text, but everything looks decent, overall.

The boards after etching and cutting (box cutter, ruler, and hand-breaking; jeweler’s saw is too slow, and dremel is too messy)

The end result; two working boards!
These two use the MASW007107 RF switches, and are test boards for my Senior Design project for school; I will find out how well they work next week! They should work on signals up to 8GHz, but I can’t test that without a Vector Network Analyzer.