First usable PCB!
I replaced the photointerruptor with a microswitch to eliminate uncertainty where the edge of the PCB left the sensor. I added no debouncing initially, as I couldn't be bothered.
The results were disappointing - about the same as with the optical sensor. So that pretty much persuaded me that the optical sensor is as reliable as the microswitch. Adding debouncing could possibly help. I reckon the board goes at about 0.2ms-1. With a switch bounce time of, say, 5ms that gives 1mm of potential dodginess. Hmm, possibly quite significant!
Then I noticed the left guide fence wasn't perfectly square with the input rollers. A long way back I moved the left guide fence further left, and before that I think alignment was better. That was when I printed a perfect 0.1mm grid twice.
So I removed a screw, squared up the fence and re-drilled it. Alignment improved dramatically.
Here's a printed board. I was trying to avoid printing twice, so I sanded the board first. The image quality was still not good enough, so I printed again. Here is the finished print - two runs, fused after each, ready for etching:
Here it is after etching, modelled by Rob:
After stripping, ready for solder:
So where to go from here? Hmm. Two major issues remain:
1) At the moment I can only align repeatably if the imaging process starts AFTER the board enters the imaging rollers. This is because there is some variance creeping in where the driven rollers mash the board into the image rollers. This means the top 40mm of board is wasted each time.
2) When the trailing edge leaves the sprung drive rollers they ping together, giving the board a little kick forward, resulting in a smear. So either I don't print on the last portion (yet more wastage) or I devise a means of lifting the sprung roller as the end of the board approaches.
Before I approach either of those, I will first add drive to the top rubber roller. I think this will make loading the board easier (at the moment it requires quite a push to enter the rollers), paving the way for automated loading.
I'm also quite tempted to investigate a colour laser printer as some are designed for highly repeatable imaging. It would also allow me to print coloured solder masks and silkscreens. The HP 2100 that I'm currently working on is only a home/small business machine, so not as well made as a more expensive one. The 4550 looks like a possible candidate as it is old (read well made, with nice, modular construction), colour, and possibly quite cheap on ebay.
Going to see how far I can get with the current arrangement though, as a new printer is a major investment of time.
Submitted by jeff on Thu, 03/15/2007 - 16:02. categories [ ]