I needed to output some debugging from a micro controller, usually this is easy. Just plug in a MAX232 and connect up a USB to Serial cable and you're away. But in this case I was using 3.3v making the MAX232 useless. I didn't have any MAX3232's in my boxes and I didn't want to fiddle around with resistors etc to make a rudimentary level converter. I needed something, and I knew it was something I would use time and time again in the future so I set out to make a LVTTL USB to Serial cable.

Looking at the various USB to Serial cables I had, I came across one that I could clearly see had a FT232BM. Looking up the data sheet for the FT232BM I found in the app notes that this chip can be configured to be 5V bus powered, 5V self-powered, 5V bus powered with 3.3V I/O and 5V self-powered with 3.3V I/O.

I haven't been posting much lately because I've been busy helping out a friend with some handy little boxes of tricks.

The first one is a simple toggle button controller that uses an ATtiny13 to form a push-on push-off control button with LED and a dry-contact reset to turn the toggle off. The PCB was designed with gEDA PCB and the NC GCode was created with it's GCode export function. It generates passable GCode, but you do need to verify that the tool can make it through all the small clearance gaps otherwise you end up with bridged traces (especially around pads).

gEDA PCB GCode export also doesn't produce a very good outline code to route the board out. It for some reason includes all the drill holes etc etc. So I manually edit the file and remove all the unneeded stuff. I also have to manually add "tool diameter offset" with the G40 and G42 commands otherwise it will run the tool exactly on the outline and not compensate for the tools width. With the G42 command EMC2 will offset the tool width and cut the board out at the correct size. It's a bit of fiddling but the results are great otherwise

Also since I dialled out the backlash the machine is back to being accurate once again. My circular pads are circles again! Also the drill alignment is much much better.

Anyway, the second PCB (box still pending) is a two host to one device USB switcher. It's for a audio-visual automation system that requires to switch a device between two computer outlets. It takes a dry-contact to make the switch. The LEDs where added just to show the selection, otherwise they're not needed.

The board uses telephone grade 5V relays so the contacts should be more than adequate for USB. The device seems to work just fine with the box in line so so far so good.

Anyway just a quick update for those interested.

I managed to get my design over to the CNC mill today. I ended up using pcb2gcode to convert the gerber files. There was a bit of frig'n around but nothing to bad. I didn't change the spindle on the machine to my better low run-out Wolfgang. Instead I left the default Proxxon that has a bit of noticeable run-out when working at these sizes, that's why a few tracks are a bit thinner than I'd like (also the machine has a bit of undiagnosed backlash I haven't dealt with yet).

I've never been happy with the layout in my little CNC controller box. It was a little cramped hard to work on, plus I made a backplate for it but it was far to small and the connectors wouldn't fix. So, for the longest time I just had the cables and sockets hanging out the back.

So the other night I pulled the whole thing apart to see what I could do. I found a piece of plate aluminium that was the same size as the base of the box. I decided to mount all the internal components on this plate and mount that on six stand-offs in the base of the case making the whole assembly easy to lift out and work on. I also removed the rear fan and nibbled out the rear panel so that I can now fit the three Speakcons, one Powercon, four XLRs and a DB25 on a nice rear plate (I've yet to machine, chicken and egg problem) without having everything all bunched up together.