The infinite joy of hardware development

In this post I just want to give some updates to all testers that have already subscribed and are waiting for their machines.
So what's happening during this "radio silence"?



Well, first of all there's a lot of testing involved:I have already assembled a couple of machines that should be used locally for sorting. That was an invaluable feedback, since I have already found and applied some improvements.

First example: a better way to connect the aluminium bars. At the beginning I planned to use angle brackets, since they are easy and sturdy. But then I realized that there's no need for that: we're not going to sustain high forces, connectors are only needed to keep things together.

Doing some quick math, a 3 way angle connector (on the left) will save you the following: 2 angle brackets, 4 M4 screws and 4 M4 T-NUTs (shown on the right) for each angle. Given that there are 8 angles that should be joined together, this is a really speed up. And don't forget the pain to double check that everything is squared and parallel. In this way you'll only have to insert, tight and you're ready to go.

Of course such change had a little cost: the frame is now a little longer in one direction, so that's something I had to take care of and update the design accordingly.

In the same way, some screws had to be moved since they were clashing against the 3 way angle connector.

 

Then there's the case of camera images. As regular standard, I always run some quick tests when supplier deliver their goods. With the last batch I could not recognize anything (panic!). Take a look at the following two pictures, do you notice anything different?

camera tests

At the beginning I couldn't find anything different (except for the left one being slightly out of focus), but then I realized. In particular light conditions, the camera sensor applies a yellow "veil" over the image, causing the recognition engine to fail. Nothing terrible, but I had to implement an algorithm to automatically fix color balance if needed.

In the meantime I'm working on docs with detailed images of every step, currently I'm very close to 100 images (and counting). Assembling things could be something daunting for somepeople, so I'm doing my best to ease your task.

Finally the PCBs... these are becoming a sort of constant joke with my friends. First some background info: this PCB is required to convert 12v to 5v and power the servo for the vacuum. To avoid you messing around with cables, jumpers and other stuff, I designed a simply voltage converter. Nothing fancy, it's just Electronics 101.
When it was time to assemble it, I decided to outsource it since I don't have the required tools. For the first version of the machine, I chose a Chinese manufactor.

And he messed up big time. Long story short, he applied the wrong chip, the board was not working, had to ship it back, replace them and start again.

For this version, I decided to play safe and choose an European manufactor. And guess what? He got the boards wrong, too! (New  version is on top)

pcb boards

As you can see, the female pins are on the top, while they should be on the bottom. So now here I am, desoldering all those pins and attaching them to the right side.

Never a dull day!