Michael did the tedious work of laying out this version of the IR Repeater with VeeCAD on a perf board. I finally got this version built and am excited to be testing it out. Michael has also done the layout for a printed circuit board so I hope to send away for that while the perf board version gets exercised.
The picture above shows the perf board in use. Instead of a micro USB cable I'm using a standard USB B cable for power. The picture below shows the completed perf board version.
I made a lot of mistakes in building it so it took me quite a long time to get this done. Most of the mistakes were due to getting myself confused from flipping the board over differently than I was thinking and soldering things incorrectly. I'm lucky I didn't burn out anything soldering / desoldering over and over. I moved a few things around based on late component changes that I made: PR1 and JP2 are completely different shapes in this than in the original specs I provided. Those changes also caused me to make some mistakes.
This is the underside of the completed board including a major mistake. I had the ground hooked up incorrectly on the RJ45 header: pin 5 instead of pin 4 - oops. I didn't catch this one until I started testing. I didn't hurt anything - the LEDs were just always on so it was pretty easy to spot quickly. Below is the corrected version.
There is one other major mistake that I made that I couldn't correct with desoldering: I was mistaken about what pin 1 on RJ45 was so I have the RJ45 backwards. Pin 1, at least on this SF breakout, maps to brown rather than white-orange. I tried to desolder and flip the breakout from the header but I couldn't get it off. I didn't want to redo the header connections so I ended up making a temporary RJ45 cable that just had all the wires backwards on the one side.
It isn't ideal that I ran wires on both sides of the board but I found that easiest. I really appreciate Michael's work laying this out - its awesome to have a compact version running with near-final connectors and a layout closer to what will be on the printed circuit.