Week 5 Now I know my PCBs

Week 5 Now I know my PCBs

Creating files

Opened modsproject.org. Opened a predefined progam unded Roland and SRM-20 mill titles mill 2D PCB.

Downloaded linetest.png from Fab Academy website. Uploaded it to the modsproject website to use as the reference cut.

Created three different files for the milling machine, in order to test different milling bits. There were three different bits that we needed to test: - 0,2mm - 0,5mm bit with 90dgree angle. - 0,1mm bit with 60 degree angle - An end bit (blunt tip) with diameter of 0,5mm

The settings in mods project for each bit were as such:

Bit	Tip angle	Tip diameter	Offset #	Stepover	Cut width	Cut depth
0,1mm	90	0	4	0,5	0,1	0,6
0,2mm	60	0	4	0,5	0,2	1
end	0	0,5	4	0,5	0,5	0,6?

In addition to these bits, we used one more to separate the board from the material.

The measurements of these different bits were used in modsproject to generate a file. In order to get the file on our machine, we needed to toggle the settings for WebUSB off and Save file on.

Milling

For milling the test pieces we used Roland SRM-20 precision milling machine.

First we attached the copper plated board to the bed of the milling machine by...

... adding tape to the copper plated board and...

... tape to the bed. Then we added hot glue to the taped side of the copper plate, and attached it to the board. We did it this way because we did not have double sided tape at our disposal, so we made our own make-shift version.

Started by milling one board with 0,1mm bit. The mill bit was focused to the surface by moving the z-coordinate to about 1-2 mm above the surface, and then loosening the bit head screw and gently lowering so it touched the surface. Then the milling process was started with the computer.

After the milling first couple of shapes, the machine was paused and the line width was measured with a microscope. Based on the mill bits tolerances, we expected a line width of 0,1mm, but the measurements came up with a line width on 0,15mm. To fix this issue, we moved the mill bit up 0,03mm. That should have fixed the issue, as the formula change of width based on height change for 60 degree bit is (1/sqrt(3))*heightChange *2. With our measurements it should have changed by 0,034mm, a bit less than needed, but still a measurable change.

After we did that, we started milling again on a different spot, paused again, and measured. The line width had not changed, so we chose to just go with the wrong settings and mill it. This caused additional stress to the mill bit and bigger lines than described in the milling instruction file, but that was deemed not to be a big issue.

After that we changed the bit to the 90 degree angle one and recalibrated its z-axis like before. Changed the milling position to 40mm right along the x-axis and chose a new milling process file to match the bit, and started milling. This time we stopped the process again and measured the line width, were happy with it and continued the process.

Changed the bit the bit the the flat ended one, and once again recalibrated z-axis. Moved to x-axis starting position 40mm right from where it was, and started milling with a new file. Measured the line width once more, just for kicks, as there was nothing to be done with a flat headed end bit, and continued milling.

At the end we had three milled pilot pieces. Of course the end results do not tell us much about the capabilities of the bits, and the bits follow the process file even if the mill bit is different than the one specified in the process file. The one thing that can be deduced is the precision, wobble and additional forces created by the milling process and the used bits. For all of those we were quite happy with the results, as most of the bottom thin lines survived the milling process, only the thinnest couple were torn with the 0,2mm 90 degree bit and 0,5mm end bit.

House of PCB

We got introduced to the website called pcbshopper.com where I could do price comparisons between different pcb manufacturers based on different metrics on borad requirements. There metrics included things like board size, copper weight, and material, to name a few.

We started the process of manufacturing PCBs with the design from the Quentorres board that was used this week to test soldering and simple programming. We chose not to order assembly, only the board that the pieces could be soldered to.

We chose the cheapest option that pcbshopper offered that was based in Europe. The decision was made for these reasons:

• The price difference was not that much for the amount that I was going to order.
• Ordering from EU does not cause hassle with customs
• Supporting local infrastructure that sells for individuals is always nice

The cheapest one listed was AISLER. They needed a file in either ODB++ or one of the few that they supported. As we wanted to just print the Quentorres board with them, we looked through all the files supplied in the QT repository, and compared with the supported file types of AISLER. The first such file was .kicad_pcb so we went with that one.

After I uploaded the file, AISLER showed me the preview of the board, which seemed okay. After that we clicked checkout and proceeded to select "beautiful boards" (to print just the board) and "Simple (1.6mm HASL)" as board configuration and not other addons.

After that we needed to register or login to actually place the order. Did not do that. The price had increased between pcbshopper and AISLER website by around 2€. Not that bad of a difference.