2023¶
23-09-2023 - Electronics¶
Welp… didn’t graduate last year since I couldn’t get the electronics to work back then and had other matters that needed catching up to after full-time FabAcademying it for 6 months. This year I started over again with an ESP32-C3 module to remove a few more potential points of failure.
I go over the changes to my motherboard in Input Assignment. I had to debug the whole system from scratch, aparently I2C doesn’t really like long distances and electronic circuits are a lot more finicky than I very wrongly expected… Since Input Assignment was getting way too big I moved debugging into it’s own page. There I cover a lot of my research. And I have to extend a special thanks to Prof. Neil for helping me prioritize what to debug. The only thing I didn’t get to try yet were Eye Patterns. They might require a fancier oscilloscope than what I currently have access to too.
Funny enough now that I had my own Logic Analyzer at home I figured out my original Mega328p board almost surely worked. I just soldered the first header flipped 180° and after that I soldered all of them wrong based on the first one. It’s why only i2c didn’t work but everything worked fine on an Arduino Uno but not on my boards since I wasn’t using my cables and was making sure I connected everything right.
Pulling Wires¶
The whole structure I had in my mind last year turned out to be a bit problematic, both from not having time to print things properly but also I probably over complicated the wire structure. Trying to print the flat coil springs was also probably a mistake. Fun exercise but burned way too much precious time and attention, one day I might try again with a smaller nozzle… Last year I didn’t want to break a measuring tape just for the spring but after trying to find a supplier a few more times since I just gave up and got some cheap ones of IKEA. 1.5€ each and since they’re surprisingly well made I can put them back together at the end if I want. Which is nice.
Lovely ruberized exterior is easy to take appart thanks to 3 screws.
Spring plus everything else.
The spring is inside it’s own little container which means it’s incredibly easy to reuse for other projects. If you ever need a couple flat coil springs don’t be like me and go straight to Ikea.
Not sure what that extra bit does. I removed it so I can have a thicker 3d printed shaft holding the spring.
With that out of the way I started thinking on the structure I had from last year. The wire wanted to play by it’s own rules so I had to rethink how I could control it’s position better and have enough tension to turn the cylinder with the magnet attached to it.
First step was to go back to my CAD model and review which parts worked and which didn’t.
The steel skeleton I welded and the bottom flexure for the sensor worked perfectly. Also maybe reusing the other 2x sensor holders so I’ll keep the model around.
Gotta throw in a quick and dirty spring:
For keeping the wire from moving around Quentin suggested I try a Capstan like they were using on the Urumbots.
The more I thought about it the more sense it made so I started to look into capstan and V groove pulleys and everything.
As I was looking at how people use capstans in machines and industry (mostly whinches and cable forming machinery) I had the idea to try and search exactly how other people measure cable/rope linear movement.
Each successive image search got me better and better results.
Starting at Measuring Wire Pull I then moved to Measuring Cable Pull Distance.
I started noticing some interesting results… Like a few generic sensors for sale named as a Linear Distance Measuring Sensor:
I must go deeper. Searching Linear Distance Measuring Cable Pull lead me to the same generic sensors but now 1 was called Linear Draw Wire Encoder and that got my attention.
A final search for Draw Wire Encoders hit the Jackpot!
Loads of search results with a many reputable looking companies offering a diverse set of options for measuring distance by tugging on a cable…
A lot of them have different configurations for more or less precision in different use cases.
Looking for example at what might be the cheapest line from TE, an SP1-50 with a listed price around $300, it’s specs suggest precision of 0.25% at full stroke (50 inch). Considering even TE’s models with 550 inch stroke say 0.1% precision at full stroke I’m probably over thinking the mechanical rig for the wire sensor. Even if I can get more precision later a simpler design like formal brands use might not be as bad as I had assumed.
This clip from Micro-Epsilon shows a few examples:
Siko has some sensors for hydraulic cilinders that are quite barebones and have most of the mechanics exposed like the SGH10:
or SGH25:
The SGH10L that was designed for lateral instalation also has an interesting wire path:
Change of Plans¶
I decided to order some small V Groove Bearings to try and mimic the construction of the simpler sensor assembly. I might still try to model something around the other regular 26mm bearings with a printed V Groove but it might not be worth the time.
It was easy to find cheap 4mm ID x 13mm OD, 6mm Thickness V Groove bearings:
But while I was looking for a supplier I found an even smaller 4mm ID x 10.5mm OD, 6mm Thickness but not too cheap and outside the EU so I’d very likely have to deal with Customs.
The ones I ordered are suppose to be 12mm OD so we’ll see. They were around 20€ for 20x with shipping so can’t complain.