In this section I will show some ideas / techniques which I use. Some are based on ideas from others, some are my own. There are several tutorials on the net dealing with home made parts proving that's possible to do with some hard work and dedication. My plan is to let the CNC do the hard work, I do the thinking. You need some drawings - both 2D and 3D dependent of what to make and let the CNC do the rest with some help in between.
Per-Erik - March 3, 2008
Korry
You have certainly seen these from Korry. The company is a major supplier to the aircraft industry concerning switches, indicators and knobs. You can't afford to buy from Korry - very expensive. You could try Ebay, but you never know the quality or you can buy replicas from different suppliers (FDS, Simparts - ref. resource page)
Korry switches
In a 767 cockpit it seems to be two types of Korry switches:
Type 433 - Size 1". Used on the MIP and Overhead Type 389 - Size 5/8 ". Used on the MCP.Here is how I make the Korry 389 as an example by utilizing the CNC unit.
First I make a Corel drawing of the pushbutton part of the switch - using the original measurement from Korry.. It's a view from the top:Export the layer with color information using HPGL format.
Open the HPGL file in the SheetCam program. After selecting the correct milling-tool, tool-diameter and depth for each part of the Korry, the CAM software shows what's going to happen in the next phase
After approx. 8 minutes in Mach3, we have the end result - my first but not the last Korry-389, this one made in MDF. I have not decided on final type of material, just started with MDF. I don't think it's strong enough (very thin walls), so I'm looking for something like black acrylic material - time will show.
It's a good start and the precision is at the numbers - 16,3mm square and the same for the LED chamber, exactly as in the drawing - exciting. But I'm not ready for mass production yet. Some fine-tuning is still necessary!.
Update - July 2008
One step further but still using MDF as Korry-block. The Korry design is now 14mm deep, 16,3mm square and at the bottom there is an edge of 19mm. This edge together with a spring on top of the tactical switch keeps the Korry in place. I got a travel of approximate 3.3 mm which I believe is quite correct, but I'm not sure.
A principle drawing of how this is put together. The most left drawing showing how the of the Korry is milled inside and it is somewhat modified compared to the drawing in the beginning of this section.
Just under the label I have put in two small 2mm Lexan pieces just to diffuse the light from the LED's. The labels are made of ordinary HP Premium Transparency and consist of two pieces. One label is mirrored the other is not. Mounted with the printed sides in the middle. In that way the label front have the print on the back and the letters are automatic protected against "Hamburger-fingers". Another effect of the double layer - completely blocking of the backlight in the "black" area.
Another picture showing the Korry from underneath. The inside is painted white just to spread the light even more.
The Korry attached to the IOBoard / PC and running "controlador" - very well known test program for those of you running OpenCockpit hardware.The color is more "white" on the photo than in real life which is more like amber - Color code RGB - 255,191,0 on both labels (on top of each other) so it should be even darker than ordinary amber. Transparency combined with white LED's can be difficult, but at the same time you have full freedom to make whatever color by filtering (white contains all color). One have to test a lot to find the correct color match. I think I'm pretty close.
The color on the Korry itself is of course black applied by an ordinary Black Marker pen.
Indicators
 |
Type 434. All over MIP and OH. |
More to come
Knobs
Usually builders utilizes silicon rubber moulds and two component resin to produce knobs, but you need an original. Ok - why not make all original by the CNC. Drawings in 3D in lwo, 3ds...- converted to STL and then to Gcode ready to run on the CNC.
Update - July 2008
My first attempt to mill a knob - the one selected is Korry-445.As you can se the material is MDF and that's OK as long as we don't need backlight knob. In that case I will be using a acrylic block. The picture to the right include also Korry-240.
The process: Made a 3D model of the knob in Lightwave. Could actually be whatever 3D tool outthere as long as you can export to STL. I'm unsure if Google Sketchup is able to do this.
Meshcam is the next tool in the chain reading the 3D-object and setting up material, tools & path's. The output from Meshcam is Gcode ready for Mach3. The rest is a 30min CNC workout.The endmill used for the finishing path was actually to big so the flutes is not deep enough. Next run will be more correct.
Tested the new Index wheel. Using Lightwave 3D to make the model of one of the Korry knobs. Then do an export in STL format to MeshCam for generating of the Gcode. The other knobs to the right are from the same process but in stead of 3D, I'm using the build in tools in MACH3 - "Cutting spline and gears". The Disconnect bar is again 3D+MeshCam.
Have a look at the video in the Video section.
Encoder
Gray coded encoders are often a challenge - at least when you would like to merge two of them to make a dual encoder for tuning the radios. Also in a combination with a rotary switch and pushbutton. When using IOboards from Opencockpit you should select quadrature 1/4 cycle like shown in the picture. In that case you just 
use two inputs on the IOboard.
I have found a few which should work but not tested except CTS:
| Producer | Part number | Dealer |
| Bourns | ECW0J-B24-BC0006L | No dealer information |
| CTS | CTS288 | Opencockpit |
| Electroswitch | 690-701-04-16 | Mouser |
| Misc. Mouse | ?? | I've found several PC mouse (Logitech, HP) which uses small mechanical 2-bit encoder. |
Panel-making
Updated spring 2010: Frontpanel is 6mm PMMA-CL and Backpanel is 3mm in the same material.The principle is though the same.
To show my method I select the VOR radio in the glare shield of 767.
First of all - we have to get some drawings. Visit www.simpit.de - a great site for 767 builders looking for approx. dimensions.
After several hours of tweaking in Corel and some help/info from other builders, I have an early version of the panel: In the download section you will find Corel drawings which will be updated from time to time. I'm trying to put all in - from text/figures engraving to cutout, component and wiring/PCB separated in layers. In that way it's easy to make changes and catch conflicting & wrong placement of elements.
I start with the Front panel itself and make an export from Corel Draw in HPGL format just with the cutout information - engraving will be done later as the last step. Bring it over to SheetCam for setup of cutting path, speed, depth etc.
I select a 2mm end mill tool and an "inside" cut profile with the outmost cut with "outside" cut profile as the last one. Cutting depth is 6mm and 1,5mm on each pass - give a total of 4 passes and a cutting time at approx. 8 min. When satisfied with the setup hit P button to produce the Gcode for the next stage - Mach3.
Mach3 is THE controller program for CNC at hobby level. Using the the parallel port to communicate with the driver circuits for the steppers. Read more on Mach3 and HobbyCNC homepages. I will just briefly show how I use the tools.
30/3-08: A milestone date - the first real test of CNC panel cutting.
No more simulation with pen and paper. Still some tweaking to do but I'm pretty close to an acceptable quality.
Update - July 2008
Here are some pictures from the first panel - more like a prototype. Testing a lot of different techniques.VOR - the left instrument in the Glare has become my target for testing different options - both concerning the panel itself and how to connect all parts.I ended up with making a specific
PCB for VOR and collect all the wires in one 26 pin ribbon connector. Schematics in the download section. On the picture you se the dual Frq-encoder - quite easy to implement.
All my encoder comes with an integrated switch and at least on the VOR I have wired two of those (CRS-switch & FRQ-low switch) to the Master controller. Could be nice to have in the future for other specific actions - like MIP test.
Nico; the MIP test and Cold & Dark - both working. I had to switch on the battery in the OH to get things to "light up" - excellent!!!! - So easy, so logical, so realistic - couldn't be better. BTW - used your vor1 script - just changed the I/O addresses for switches, LED & displays. Thanks again for your effort!!
The first back light test. 12 x white LED's should be enough to get the panel to "shine". The 4 mm Acrylic is still not "sanded" or frosted which it has to be to get the light out of the acrylic - if not it just exits out at the end of the panel (se at the top of panel to the left) This is how far I've come 25 july 2008. Start working on the front panel and applied four layers of "Boeing Brown". Next will be engraving text and symbols and then the final test. Updated:
January 20, 2011 22:12
I'm not using the above method anymore - replaced the frosted acrylic front sandwitch with one single 6mm PMMA-CL milled from back to make room for backlight
Sept. 2008 - Some update.
Prototype panel with Dzus and mounting screws in place. All three knobs (not high quality and still wrong color, but anyway....).
A dark shot to the right just to get the details - still fighting some light leakage but getting closer and closer to the "target".
First Bezel attempt (Altimeter): Milled front and back. Made room for a 2mm Plexi "window-glass"
PCB (Printed Circuit Board) production using the CNC.
This chapter is updated by September 2009. Removed the old crap about Eagle and replaced by the following:
All my PCB's starts out from a schematic and ends up as a double sided PCB. To illustrate my method, I will use the PCB for VOR panel as an example. First some background about applications used in this process:
- Diptrace - Inexpensive, easy to use schematic/layout program - output DXF.
- SheetCam - Tooling setup and path generation - input DXF, output Gcode.
- Mach3 - CNC controlling program. Take the Gcode and make some real movements.
The schematics for this example you will find here and the layout I ended up with looks like this:
So how do we proceed?
This is as you can see a double sided PCB (Blue - top side, Red bottom side), so you need a PCB raw material with cobber on both sides. During layout you have to make traces on the "correct" side so you are able to get the component soldered - remember you don't have cobber through holes. Look at the 34 pin connector to the right. The connector is on the bottom side and all traces on the top (blue) - very important!!. The same goes for all parts where you are able to solder just on one side - think a second time about how to mount/solder - you need to get to that pin with your soldering tip! You may need to swap side just for this purpose. In the layout you see several of these "empty pads" for side change. In these goes a wire soldered on both sides to make the connection. Another detail; you see some thin blue lines in the drawing. This is GND (0V) connections and those traces and PAD's are not milled out - only make the hole. I'm using both sides cobber as a Ground plane and just solder the component on both PCB sides.
All my LED's and LED-displays are put in sockets like this:
And then it's actually possible to solder on the component side using a very thin soldering tip. Next; I try to standardize on 1mm traces and whenever practical increasing to 1,5mm or even more. Traces between Pads are reduced to 0,6 or 0,8 mm. The Pad's need to be as big as possible. Again - due to the fact there is no cobber through holes the Pads are not so strong. If you need to de-solder, be very careful.
So when satisfied with the layout, we make one export from Diptrace with all vital information - both sides traces, holes, text and board cutout in DXF format. Bring this to SheetCam and you will see that you've different layers for each feature like this:
I'm using SheetCam not only for PCB, but all my 2.5D work. Here I defining all my tools (endmills & V-cutters) and path setup. For PCB milling I normally use a V-cutter like some of these - more specific the 60 degrees to the right.
The other side of the PCB (not shown in the image above) is run in the same way, only mirrored (MIrror X or Y if you prefer) and placed in the exact same position - just flipped. To be able to align the PCB correct you will need at least two symmetrical pilot holes which is a part of the layout (symmetrical around the axis you flip). In my case I used two of the mounting holes. I made two threaded holes in the scrap board (underneath the PCB) and secured the position while milling.
All milling is based on contour path. This means we just mill an isolation track around all traces. Two runs - one in each direction - One "Up-cut" and one "Down cut". All holes are taken out first with the correct drill bit, then the traces on both sides and finally board cutout.
It's a direct connection between SheetCam and Mach3. When you save your G-code (P) it pops up in Mach3 ready to be run. From then on it is an ordinary milling job so all precautions and setup is put in the layout and SheetCam.
Out of the CNC you probably get a nice PCB. I'm finalizing the board by a light brush with steel wool to get rid of the last strays of cobber and a visual inspection - very important.
You will always find areas where you may need to do some manual correction. More specific, where two milling paths meets with a small angle or between two parallel traces close to each other. Cobber is a soft material and may leave some very thin strays which you need to get rid of. The steel wool removes most of them, but......
In the end, adding a Plastic coating like PRF 202. You can do this before or after soldering.
To the left you see a few PCBs (my own layout of OpenCockpits Display-II board). One is ready and verified working.
To the right is the PCB for VOR-panel.
PIC programming
I have decided to do some custom build of Open Cockpit products. Then you have to be able to program the PIC circuits yourself. I think the first one out will be the Display II board with my own layout using the published schematic and PIC firmware. With the programmer in house I'm ready.
$16 inclusive the RS232 Cable (from China). In Norway I would have to pay maybe 4 times or more, so I order two at the same time (2 x $16 + $16 shipping)!!!!. Software to use can either be IC-PROG, WinPic800 or PonyProg2000 - all free of course. I have so far looked at IC-PROG.
Update July 2008
Programmed two PIC-16876 with success using WinPIC800.
Fetsund - Norway |
©2008 Per Erik Hoddø - All rights reserved.
|
email: peh at pogostick.net |