Jedi Soldering Skills

I’m currently challenging myself by creating a microcontroller-based project a few orders of magnitude more complex than anything I’ve ever done before. Although this is probably on par with projects you might see being created by senior electrical engineering seniors, keep in mind that I have no formal training in engineering, and that my MS is in molecular biology. I just started learning about circuitry / microcontrollers a few months ago, and challenge myself to learn more by continually attacking greater and greater challenges. Here’s what I began working on last night:
layout

This if the first entry describing the creation of my non-prototype microcontroller-powered prime number generator. I made a proof of concept device a few weeks ago which calculates prime numbers (up to 2^25, about 33.5 million) and displays the results in binary form using 25 LEDs assembled in a 5×5 matrix. I added an extra column of 5 LEDs for a final matrix size of 6×5, illuminated by multiplexing through 11 IO pins of an ATMEL ATTiny2313. My new project will do the same thing, except it can calculate prime numbers up to 2^30 (over 1 billion!). Instead of only displaying 1 number, it will display 3 numbers (last prime, test number, and the divisor) using 90 LEDs. The picture above is of the main circuit board before I began soldering. The empty sockets will house a combination of 8-bit shift registers, binary-to-digital converters, and 7-segment display drivers all powered by an ATMEL ATMega8 microcontroller crystal-clocked at 10.042 MHz (arbitrary, but stable). Here’s what the underside looked like before I began soldering:
underside

I anticipate that this project will develop into a soldering nightmare. The board is nearly too small as it is, and I don’t have good wire for soldering. (I’m actually using the small wires from an old phone cord right now.) I included a potentiometer, 2 buttons, and 3 switches to aid with various settings (brightness, menus, etc). 3 Rows of LEDs (60 pins each) requires 12 shift registers (16 pins each) plus the 28-pin microcontroller makes about 400 solder points (YIKES!), so I anticipate the underside of this project will quickly grow to become a daunting mess of wires. Last night I finished the connections necessary to program the microcontroller, and for the microcontroller to control a single 8-bit shift register, allowing the first 8 LEDs of the first number to be controlled. Here’s what the soldering looked like. Remember, the dense clump of connections only controls 8 LEDs, so multiply this by more than 10 and that’s what I’ll have to do JUST to power the display.
wires

After programming with a straight-through DAPA style parallel-port programmer, I was able to shift data out to the single HC595 I had wired. I spent hours banging my head against the wall because nothing I did in the software would make the LEDs illuminate. I thought I was sending signals to the shift register wrong, or that I soldered something wrong. I finally concluded that somehow (probably when I was troubleshooting by applying 5v of power directly to the pins of the shift registers) I managed to burn out all 8 LEDs of the first light bar. I had to de-solder ALL of the connections you see in that picture, replace the bar with a new one (thank goodness I had an extra), and re-solder everything. I have a feeling that by the end of this project, I’ll be an expert at soldering. Here’s the program running controlling the first 8 bits only:
running

Supposedly the hard part is done for the display. The software was written in such a way that it will automatically begin lighting-up more LEDs as I wire them. The small node for the first shift register and 8 LEDs will be identical to the other 11, and as I solder them one by one I’ll get closer to my end goal. It will probably be many hours of soldering. In retrospect, I wish I purchased a bigger perfboard. Actually, in retrospect I wish I made a PCB!!

Update: After a few more hours (of soldering, troubleshooting, desoldering, and rewiring, and resoldering) I have my second 8-bit segment working. Note all the yellow (newly-added) wires. Multiply this by 10, and that’s what I have left to wire for the display alone!
img_2336
img_2326