03/29/2018 at 03:04 •
On a few occasions I have had the need to secure something to a flat bit of laser cut wood. In all cases, I created precise shapes that allow me to rigidly and accurately grip the object while not depending on the exact dimensions of the cuts themselves. The method requires final alignment to be performed during assembly, but does allow for post-assembly re-alignment and doesn't require any adhesives or fasteners. A few examples from a recent #NeuroBytes patellar reflex model iteration:
The top image is a simple T-joint, used to connect two right angle bits of 3.2mm aircraft plywood. The middle piece goes through the base in two spots and has cutouts for wedges, which are inserted and tightened during assembly to pull the joint tight. The angle isn't exactly 90 degrees due to the aforementioned kerf taper, but it works well enough for this application (a mounting plate for a demonstration model).
This method is still very much a work in progress. I think a lot of optimization could be done to reduce the material requirement and cut length; for example, the width ratios of the tenons and slots could be standardized (probably based on a woodworking joinery book's advice), and the curved shape of the connector plate itself could probably be generated parametrically once a few constraints are figured out. The wedges are also probably beefier than needed and the angles could be optimized. But it's a good start -- feels good to assemble something without glue or fasteners!
02/09/2018 at 21:07 •
[EDIT: I didn't have the form configured right and folks couldn't access the questions. That has been fixed, so try the link again!]
We're headed back for #NeuroBytes EMC testing next week (unintentional radiator) and the chamber requires a 2-hour minimum reservation. If you have a project that you are considering commercializing, I'll run a pre-test scan for you FOR FREE! Some requirements:
- I prefer to only work with open source projects, and I want to share the results publicly. EMC testing is expensive and mysterious (at least for hobbyists), and building a public record of known-good / tested boards would be a great thing for the community.
- Time is short. We are looking at Friday, February 16th, at 8AM in Minneapolis, MN. So you need to be ready to ship your project to me NOW.
- I will need you to provide a self-addressed and postmarked envelope for return shipment.
- Tests sweep from 30 MHz to 1 GHz. Standard unintentional radiator stuff. That means no radios -- ESP32, ESP8266, Bluetooth, Wifi, etc. Sorry folks.
- Your project needs to set up and tear down quickly. As in, I plug it in (or insert batteries) and put it in the chamber, and we go. 5 minutes max.
- This is just a pre-test -- you'll get a report but it won't give you certification.
If you're interested, fill out this Google Form and we'll start the conversation. Again, time is of the essence here.
02/03/2017 at 16:26 •
Yes, it is broken. Step One is to accept this fact:
Step Two is to bitch about the problem on Hack Chat (you have to scroll back a few weeks to see the post). Half an hour later, be gracious when @davedarko throws together a quick PCB design and sends you the @oshpark purchase link. Get stoked and order the boards immediately (including one free!):
Step Three: wait two weeks, find an envelope in the mail, snap apart a header, solder 64 thru-hole joints, and be excited that you saved money:
Yes, it fits -- the board straddles half a breadboard and gives you plenty of space to pick up all 32 pins. @Onion.io: please consider changing to 0.1"/2.54mm headers like the rest of the world for your next spin, it wouldn't change the board dimensions by much and would make your product substantially more user-friendly!
Thanks again, Dave!