During the early stages of planning this project I'd decided to have a booth at our local Maker Faire with a friend Tom, who is also doing an OpenPnP build.
On the left we have Tom's completed and operational build. It's based off of the OpenPnP OpenBuilds reference hardware design, albeit all the custom parts were redesigned. Most parts are laser cut from transparent acrylic, which lets you see the operation of the machine clearly - it looks very cool!
On the right is my in-progress build. :) The setup of the X-axis is completely wrong here, and was just done as a way to temporarily attach the axis for the show (due to the actual mount not being usable as per the last project log).
We had a great time at the show, and got to talk about and show off our builds to many of the over 5000 attendees over the weekend, as we were located very close to the entrance and central. As an added bonus, Jason, the main developer of OpenPnP and our fearless leader, was able to make make the trek to Vancouver and participate on the Sunday - it was very appreciated, and super fun. Apparently Tom's machine was the first machine other than Jason's own that he'd seen in operation in person, which is a huge achievement for both Tom (way to go Tom!) and the OpenPnP project, as it shows it's growing!
After all the swarf machining the Y-axis motor mounts, I decided to hook up the flood coolant system for the CNC mill as the next part I was going to machine was the X-axis motor mount, and that would involve a huge amount of material removal, so chip evacuation was a big concern.
Thankfully this huge mess was expected, and the CNC is fully enclosed! Don't worry, that's just a reflection of the computer's monitor; it's not actually in the enclosure. :)
Oh hey, look - I found an angle where I can kinda see the work piece even.
The aftermath of a several hour machining operation was pretty messy.
Unfortunately the part was a failure due to me entering the incorrect tool diameter into the CAM for the toolpath, causing all the features on the part to be removed. Oops. Another lesson learnt.
The Y-axis sleds are what that attaches the Y-axis carriages to the X-axis.
I started out with some aluminium plate that had been faced down to 8mm thick.
Then I did a 2D profile of the part, leaving tabs to keep the parts secured in the stock.
Oops, or not. What happened here is that the Z-axis on the CNC fell substantially, I believe due to excessive vibrations from the very long end-mill and relatively deep cuts I was taking. This was the first real part I'd run on the CNC mill from g-code (as opposed to just moving it around with a pendant), and first time I'd used this long end-mill, and I had my feeds/speeds all wrong. Another lesson learnt. :)
Luckily there were enough tabs remaining to hold the parts to the stock, just not as securely as I'd have liked, so there was some chatter on later operations.
I addressed the feeds/speeds, touched off again, and was back in business.
The part was then flipped so that the bottom side could be machined.
The relatively poor finish on the bottom of the part is due to the missing support tabs (the photo looks worse than they actually are, as they're covered in cutting fluid and swarf).
Finally the pockets for the threaded inserts to sit in.
That was it for the machining of the sleds. The tabs were cut on a bandsaw, and cleaned up on a belt sander.
Threaded inserts were then pressed in with an arbour press.
For whatever reason I thought I'd save a few dollars and buy t-nuts for my 8020 brand aluminium extrusion from China. It turns out that the t-nuts I bought don't fit as 8020 uses a different slot profile (~5mm wide) than other 2020 extrusions (~6mm wide), so I had to either throw them out and buy proper t-nuts, or machine them down to fit. In hindsight I should've just bought the correct ones, but instead I decided it'd be a simple task.
The issue was how to hold the t-nuts while keeping the 't' area exposed so it could be machined?
Step 1: Machine a jig...
Step 2: For the jig...
Step 3: To hold the t-nuts...
Step 4: So they can finally be machined. All 200 of them.
Lesson learnt. That yak has never been so clean cut in its life.
My goal is to be able to eventually use this machine to place components for prototypes of a wearable device. Being a wearable device, this means 0402's (plausible), 0.4mm pitch QFN's (pushing it), and 0.35mm pitch BGA's (if you have any luck to spare, I'll take it). Speed isn't much of a concern for me with the build as it's only for prototyping - but it'd sure be fun to have a zippy machine, so I'm putting that as a nice-to-have.
These requirements are beyond the performance expected out of current OpenPnP (http://openpnp.org) machines, and as such I'm designing a new machine and expect to have to add some features to the OpenPnP software as I go. The machine will be aderivative of Anthony Webb's DIY Pick and Place that he's documented (well!) here: https://hackaday.io/project/9319-diy-pick-and-place
Where I'm stepping away from Anthony's build is primarily around the design of the X-axis, which will be rotated so the 2040 aluminium extrusion is mounted vertically to minimize deflection. I've redesigned the Y-axis sleds (that hold the X-axis to the Y-axis carriages) and the X-axis's motor and idler mounts (they will now be mounted on the sleds) to allow for the X-axis to sit low enough that the top surface of the X- and Y-axis rails are inline, eliminating a moment as the machine moves. As a consequence of this change, the X-axis sled for the head will also be redesigned as the linear rail's carriage is now under the X-axis rather than beside it. The sleds and all motor and idler mounts will be machined out of aluminium for rigidity.
The downside to this change will be that there will be some a dead zone to either end of the X-axis. This may be reduced by the design of the X-axis sled and head, but it will likely be more than Anthony's machine. This shouldn't be a big deal however as it still has quite a large work area, and I don't need a huge number of feeders or trays.