My tweakers.net interview on the joystick project!
Categories for Projects
For my simpit i designed several backlit panels that i wanted to engrave using the lasercutter at our hackerspace. To print with this lasercutter, i need to export my drawings, save it on a network share and then run VisiCut on a laptop running some Linux distribution. Even though all the fonts and files are there, the export of the DXF loses all the fine details of the font and line thickness. This is probably because i’m a total n00b with AutoCad and VisiCut. Searching on the ‘net the only solution was to vectorise the image on a Window edition of AutoCad (because that comes with the awesome text tools that aren’t available on the Mac edition of AutoCad). I have no Windows machine so that is an absolute no-go.
However I found a workaround to get the 100% spot-on engraving option of my design.
The trick is to print the design to PDF in AutoCad and save the PDF as bitmap to get the proper DPI:
- Select the layers you want to engrave, either through a WBLOCK or direct from your main design
- print it as PDF. Make sure you print 1:1 and that you paper size is large enough.
- Then, on the Mac, open the .PDF with Preview,
- Crop the image
- Save As PNG. This will allow you to set the DPI of the target image (be blunt and just set it to 1000dpi or something).
Now, to laser-engrave our design you need to make sure that the scale is exactly right. VisiCut is a little pesky about opening bitmaps, but what I do is I open the DXF of the design, and then IMPORT an additional layer with the pre-rasterized bitmap I just saved. With a little adjustment, you can overlay the PNG on the vector file. When the scale is absolutely spot on you can delete the vector file.
it’s just an empty panel for now, but i have all the parts! Time to work on the softwaref!
This panel is part of a long-running project to build a simpit for Elite Dangerous. It is the keypad for the navigation computer that will replace the standard QWERTY keyboard, adding some nice features and a immersive design.
I made the panels and switches by painting transparant acrylic sheets with Humbrol modelling paint (comes in a small spray can). This paint is specially made for plastics, has very good adhesion and doesn’t melt when lasercutting. I have to paint several layers for it to become opaque (hold it against the light and check that no light bleeds through).
After the paint is done I can engrave the panels so that the white below the paint comes through.
The comes the engraving/cutting. All the panels we lasercut with the 40W lasercutter we have at my local hackerspace; i must have spent several hours sitting next to it by now! At first I wanted to make a proper PCB for the switches. But as this is a one-off prototype and PCB’s for this size are pretty expensive i tried to do it by hand. After designing the layout in AutoCAD 2016 I added several layers in my diagram:
1) switch solder panel (2mm,my alternative for the PCB)
2) support panel (actually wasn’t needed but it serves as a good standoff panel right now).
3) button panel (5mm thick creme transparent white with black paint)
4) button retaining panel (2mm white)
5) front panel (3mm thick creme transparent white with dark grey paint)
Because the buttons are 5mm and the front panel is only 3mm, the buttons stick out by 2mm which gives are nice glowing effect when they are lit.
Layer #4 is something that came up last-minute since my previous solution didn’t work. These are little plates that are glued to the backside of the buttons so they do not fall out. The buttons themselves are 13×13 (wider ones are 21×13) and the backplate is 15×15 to give them a 1mm border.
This all comes together in a stack. The distance from the pseudopcb to the buttons is fixed with a few stand-offs, but exactly high enough so that the spring tension of the switches pushes the buttons against the front plate without actually making contact. These tact switches have only 0.5mm travel.
Load of wiring because of the LEDS and the Switches
Today I completed another small project. I’ve got this ‘smart meter’ in my house. The smart bit is that automatically sends off the meter data to the network provider. But I can not see the actual power usage from inside my apartment.
My meter has an infrared LED interface (right next to the meter above 1000 imp / kWh), which gives a pulse for every kWh consumed. By measuring the time between pulses and to count I’m on a current consumption in watts (3.6 / pulse duration to be exact).
The graph shows that it works well. I still can not explain all the peaks, but the high consumption (800W) between 10:30 and 16:00 is my electrical water boiler. The other major consumers are the refrigerator, the coffee and tea (for short peaks).
Measuring is knowing!
IR sensor diagram (led1 not used for now).
Closeup of the meter
I’ve been invited by the guys from Tweakers.net to do a video interview on my chair build! Woohoo.
In line with my previous board i’ve designed a custom breakout board for the HT16K33. Much inspired (cough) by Adafruit’s break out board, this design allows for oriented connectors and a I2C passthrough chainability. As an added benefit i have added mounting holes with the same dimension as my controller PCB so they can site next to eachother.
The HT16K33 is a very nice led driver chip that allows 128 individually addressable leds on a single I2C ID. These leds aren’t individually adjustable in brightness like the MCP7219 but for me that isn’t a requirement.
The board is primarily used to drive the throttle and joystick led effects. Currently it is a little mess with all the wires sticking out.
I just sent it off to the PCB maker, I’m guessing i’ll find a design flaw within the next 24 hours.
UPDATE: The PCB’s have arrived. You can tell they came from the other side of the planet because everything is upside down (bading tssss)
Redesigning the joystick and throttle made them feel much more solid. In fact, they became that much ‘more solid’ that the old rudder control actually started to feel flimsy. So I tackled that one too.
Not much is changed from the previous setup. Most of the changes we basic ‘improvement’ steps. Here they are:
- Main bearings changed from M4 to M5 width. Bigger bolts and bigger bearings means stronger bolts and bearings.
- Ditched the old MDF panels and swapped them for black solid 5mm acryl
- Hall effect sensor parts reprinted with PETG for additional strength and quality.
- TODO: Recut the (white) mounting plate in black.
So, black it is. The A1302 Hall Effect sensor was already in the old setup so I just had to reassemble everything and perform a small calibration to get it working in the simpit.
Here’s version 2 of my joystick mechanical bits. Improvements:
- Metal control stick
- Hall Effect sensors
- Ball bearings
- PETG filament
- Adjustable throttle resistance
The metal pipe is what caused the new iteration. Not only is it much more solid, but because the wall of the pipe is much thinner without a change to the outside diameter means more wires can go through.
The change to A1302 Linear Hall Effect sensor was a logic move after the use of the same sensor in the foot controller proved very reliable.
Also I move from 3Dprinting with PLA to PETG. My Vertex K8400 give much stronger and consistent parts now.
With the use of ball bearings everything feels solid and absolutely free of any kind of play.