One of the major goals I had when I started working on milling PCBs was to be able to reliably mill boards with surface mount footprints - a number of parts I want to use are only available in that format and I am hoping to reduce the size of the boards I produce. This weekend I finally got some time to test out the process. The results weren't as good as I liked but I am happy with the progress.
To start with I wanted some fairly simple boards to test the various footprints I will be using. I made up three designs:
- A simple LDO regulator board using the SOT-23 version of the MCP1702 and a collection of 0805 discretes.
- A surface mount version of my LED breakout board for breadboarding - this uses SOT-23 transistors and 0805 discretes as well.
- A breakout board for the XMC1100 ARM Cortex-M0 processor - these are TSSOP-16 parts and probably the finest pitch I would attempt with my current process.
The milling worked well so that was a good start. I've recently switched to 0.1mm engraving tips with a 30 degree blade angle - sharper angles (the 10 degree and 20 degree versions of the tips) tended to snap rather easily. I am still having some issues with the tip of the blade snapping occasionally, in future runs I will play with reducing the cutting feed rate some more to try and reduce this.
Some of the boards showed some traces of copper in the isolation cuts, because I paint the boards prior to milling (one of my milling tips) I soaked the board in a mild etchant solution to make sure the isolation was complete. A downside of doing this is that the isolation cuts are widened slightly by the process - with a 0.1mm tip I am getting isolation cuts that are between 0.15mm and 0.20mm wide, any larger than that would make the use of surface mount parts very difficult.
The image above shows my first surface mount soldering attempt. The spacing on the 0805 and SOT-23 parts are not that difficult to manage (be careful though - I lost two capacitors due to clumsy handling, once they fall on the carpet they are extremely difficult to find). Getting a good solder joint turned out to be more difficult that I anticipated though.
I tend to be very heavy handed with solder which is really not recommended for these parts - you need the barest minimum to attach them to the board. Switching from my normal 1mm diameter solder to 0.7mm helped a lot, going to an even smaller diameter will help even more I think.
My first approach was to tin the pads, hold the component in place with tweezers and reheat the solder to allow it to flow around the part. You really need to use solder flux when you are doing this - without it the solder simply will not reflow smoothly. As it was it was difficult to get the folder to flow smoothly on both sides of the part leading to some uneven placement.
In future I think I should be tinning a single pad and use that to anchor the part in place. The remaining pads could then be soldered directly with the part firmly fixed in place.
A lot of the ARM processors I want to use are available in TSSOP format so I skipped the SOIC (1.27mm pitch) footprint to jump straight to the more difficult TSSOP (0.65mm pitch). The board milled fine (there are some variations in the pad width for some reason, possibly because I put it in etchant) and the isolation was good.
I didn't really trust myself to keep my hand steady enough to solder this with an iron so I tried using solder paste instead. I tried very hard to use the minimal amount necessary but as you can see in the image above it was way too much. I tried to remove as much as I could with solder wick but I still had shorts across some of the pins.
My second attempt was a bit better, I used much less paste this time around and it seemed to be closer to the right amount. Both times I used the hot air gun on my soldering station to melt the paste and keeping the part in place while the air was blowing on it proved too difficult and I wound up with a well soldered but misaligned part.
Although it wasn't exactly a successful day I'm happy with the progress and I've learned a lot from it. Milling boards suitable for use with surface mount turns out to be not only possible but not all that difficult to do. I will have to modify some of the footprints to make them a little easier to solder - the lack of a solder mask and the fact that the copper is only separated by a very thin isolation cut make these boards more difficult to solder than professionally made ones.
Having a good iron and thinner solder really helps a lot, flux is essential to help the solder to reflow and keeping the amount of solder you use to a minimum helps as well. Like any fine motor skill the more practice you get the better the results will be - at least now I know I have everything I need to achieve what I want, all I need to do is practice and improve my skills and technique.