I decided to tackle my first circuit that includes an IC. I picked an old standby, the Myke Predko 2-wire LCD backpack, where he uses a Hex Buffer (74LS174), wires it as a shift register, and then uses a diode/resistor AND gate to toggle the Enable pin when you get to the 6th bit. It takes some extra code on the µC side to deal with putting together the nybbles and control codes, but hey, one extra pin over your general 3-wire SPI style interface (like using a 74HC595). You need one of these every time you want to drive an LCD, and I find that I don’t use LCDs very much because it’s a pain to wire up, and was pretty difficult to build on a Protoshield, too.
So I put it together in Eagle, I even figured out a way to connect the bits in the shift register a little differently to make it easier to lay out, but I still had to do a lot of gymnastics to get the thing into a 1-sided board (I hate jumper wires; it’s a thing), and had some very narrow traces. There are a couple of places where you have to run in between header pins.
I took the art to OfficeMax, expecting to buy transparency film and completely failing to do so, when I realized that they have a full print shop in the store, and they’re slghtly cheaper than Kinko’s, *and* (most important), they know how to print a 100% sized image. Kinko’s requires scaling the art to 1.03, which gives me the creeps when dealing with mechanically-spaced parts; I later decided that this had to do with the margin gutter, and that 103% was probably the exact correct spacing, but seriously — just print what I give you. This is not a document. I’m not anywhere near the margin. Anyway, OfficeMax sells transparency paper for 97¢ each (50 sheets for $49), or will copy onto a transparency for you for 77¢. Um. Someone needs to explain the difference between “copier” transparency and “laser” transparency, because the “copier” stuff is half the price. shrug. But I digress.
Not realizing that I needed 100% art instead of 103%, I ended up having to make 2 trips to the copier store.
I’d gone out earlier in the week and picked up a $7.50 gallon bottle of 30% Muriatic Acid (sold as cement cleaner) at the hardware store — in a test “etch” (really just stripping all the copper off of a couple of baords to test it out — I’m trying to figure out how to do the Cupric Chloride thing), it performed really well, so I decided to try it out on a real board.
So a couple of things that I noticed in retrospect:
1) I need to diffuse my light source better — there was a definite “hot spot” in the center of the board. I also worry that I’m not using the right type of light source, because my resist often seems weak, not real dark.
2) When you’re putting the exposed board into the developer, do not poke it with your finger to get it to go under the liquid — that smudges the traces.
3) I think my Hydrogen Peroxide is flaky — it definitely took longer to etch than I was expecting.
4) Unless I can repair the traces up front, don’t populate a board that has breaks in the traces.
At the end of the day, I had 2 copies of the board, but neither one was perfect. One had a lot of shorted traces (it was closer to the edge and got “vignetted”), and the other had gotten a lot of the fine detail eaten away during the etch.
I tried populating the board anyway, and ended up with a brick — I don’t know how to repair a broken board (I tried my perfboard-practiced jumper wire technique to no avail).
At the same time this was going on, I was having a Bad Day with the CNC machine, too, and I’m struggling to get the astronomy mount back up and running, so I was pretty fried by the time I had my first official failure to produce a good etched board.
I got mad and decided to spend the evening Eagle-ing up a ’595 version of the circuit, and it went together much easier. Big fat traces this time, too. I put in an order for some PTH and SMT ’595s (and some ’174s, too, and some MCP23008, which apparently do I2C) from Digikey, and went to bed.
In a word, hmph.