Accomplishments week ending August 4, 2013


I produced a light cube with the clear resin and transparent green dye. It was very very slow to cure. The result was good, however, except that my current mold has titling etched on the face. Once I have my new blank faced molds, I can cast the light cubes I need.

I made the new mold of a pristine blank button with no text etched into its face; with this I can produce button and light cube instances with any text I need. It will be surface text only, not etched into the face and then painted as IBM fabricates their parts.
Blank faced button used as model for silicone mold

Silicone mold ready to cast buttons and light cubes
I have thought of ways to etch the text into the face of the button or light cube. The resin used for casting produces noxious fumes if burnt so laser cutting the lettering won't work. It isn't suitable to use with a CNC mill and is way too small for the CNC water jet. I guess it might cut with the CNC wood router systems but the jigs for holding the buttons and lightcubes would be a little complicated. Furthermore, getting the sharp corners on the text is really hard with round cutting tools on a router.

My guess is that IBM made castings of the text as an insert they placed into an injection mold. This is not practical for one-off runs but certainly works well for volume runs. I guess I could try to create a casting insert but it seems that getting good alignment for the casting of the button would be a nightmare.


I prepared the lamps by converting the incandescent bulbs from an IBM housing to use my LEDs. I smashed the bulb, removed the glass and the leads, then put in the LED. My logic board for driving the LED display panel and these lights has a common +5V supply to all lamps, with the lamp I want lit having its other contact grounded. For convenience, I chose the barrel of the lamp to be the common +5V side and the button on the bottom to be the ground.

These converted bulbs fit into the IBM bayonet sockets in the holders for the light cubes. The holders use spade connectors, which I soldered onto my wires.  These holders were extracted from recycled 3420 tape drive panels, for those readers who don't remember the source of my parts, physically the same as the holders used in 360 and 1130 systems. I wired these up to a connector, intended to mate with a corresponding connector I will put on the wires coming down inside one of the stands of the display pedestal, since all the electronics to drive any light are installed up there.

I finished wiring the cabling for the lights that sit to the left of the keyboard, including the portion that mates with the LED driver PCB sitting up in the display pedestal box.

I have begun wiring the pushbutton and toggle switches that sit on the right side of the keyboard. I picked up the connectors to use for this cable and the input PCB it will connect with, then finished that wiring.
Lamps and pushbutton switches installed and wired
Soon I will need to build a spot inside the machine frame where all my printed circuit cards can be installed and start hooking these switches up.

Keyboard mechanism and interface board
Before I can finish this task, I have to finish the toggle switches. That involves several steps:

 - fabricating holders for the small switches I found with the waterjet cutter
 - cutting a slot in the plastic toggle switch handles to mount them onto the switches
 - mounting the toggle switches and handles on the button holder
 - adding the wiring for the toggle switches to the remaining switch wiring

Switch mechanism for toggle handle POWER and KEYBOARD switches
The black plate with the keyboard, lights and switches is to be mounted on the frame once the wiring is in place. I can begin operating the system with all the buttons, switches, lights, console printer and keyboard, an exciting milestone ahead.
During testing of lights (LEDs), plate not fully mounted atop keyboard


I dropped the plan to use a 3D printer to build up the separator panel that keeps each LED isolated from the others, so its lighting does not blend into the adjacent openings on the light panel. It is incredibly slow and with only four hour reservations on the printer, I chew up more than 30 minutes each time just getting the printer warmed up, adjusted and ready to print.

Instead, I chose to use the CNC laser cutter to etch the openings out of solid acrylic sheets. My test case using 1/4" clear acrylic worked well, although I only had a short interval and couldn't finish the entire block of LED openings.

I bought a 1/2" thick solid black acrylic sheet at TAP Plastics nearby and etched out the openings in this. I chose to create three separator plates from it, each for a grouping of lights on the panel. I then attached these plates on top of the phenolic board that holds the LEDs, forming a more solid panel to attach inside the pedestal box.

The three LED masks cut exactly as planned and were easily epoxied to the front of the phenolic board on which the LEDs are mounted. Small acrylic pads facing forward were glued to the bottom and top, to keep the mask back from the pedestal lip far enough to allow the light display panel to snap in place. Other pads were glued to the top and bottom faces to vertically center the LED mask and its attached board inside the box.

I made a replacement decal for the front of the mode switch plate, repainted within the black circle to cover scratching and old decal remnants, and applied the new decal. The completed plate was given a cover coat of clear enamel to protect the lettering.

I used some 5 minute epoxy to glue the circular pointer to the mode switch knob, positioning the two parts in an ad-hoc jig made of some boards with drilled holes. A tiny bit of cleanup and this was mounted and ready to go.

I began to glue on acrylic pads to the metal plates and to the light display panel, to attach them to the front of the pedestal box. As the flange around the front face is 1/32" metal, I took some 1/4" acrylic stock and shaved a 1/32 inset along an edge. I cut that stock up to make various pads, which would be glued to the metal and light panel so the 1/32 inset is closest to the panel/plate. This becomes a friction fit to the flange, positioning the plate and panel where I want it.

I originally envisioned using these at the bottom and outside edges of the plate, then adding a thin flexible bit of plastic for the top, which could be slipped into place. However, I didn't count on how well the bottom mounts would work all by themselves. I put two on each plate and two on the light panel. In fact the light panel only had mounts at the midpoint and right initially. However, these were tight enough to hold all the parts in place exactly how I wanted them. Thus, I don't need to add the side and top mounting plastic.

I did epoxy a few small blocks inside the box to act as locating "pins", so that the LED mask assembly inside will sit at the correct "left-right" position to line up well with the light display panel. I proceeded carefully to get a decent fit but not too tight, making it easy to install and remove the LED mask assemblies if they need repair.

Display pedestal box coming together, some wiring work still to do


I had another major battle with Autodesk Inventor in which I finally beat it into submission, getting exactly the knob design I wanted. With the 3D shape completed, I moved on to designing and validating the mold that would be used to injection mold the knob. 

However, before I can finalize that design, I need to know more about the exact plastic I will use for the molding. The raw material comes as small pellets, but it is very hard to buy small quantities. Generally one buys shipping containers worth at a time, since injection molding is usually a volume manufacturing activity. 

The few places I found that sell small batches define that as a 55 pound sack, which costs a couple of hundred dollars, which must come from a larger quantity they have in-stock. Common colors are usually available in stock, but IBM classic blue would require me to order a shipping container worth. 

People who do smaller projects often buy a bigger quantity and then sell off the excess on ebay, but the number of such sales per week is not high enough to give me a good assortment. I found one seller offering dark blue in a five pound bag. This may force me into alternative fabrication methods. 

One option would be 3D printing - at an investment of about $20 for the right color plastic - but I am not confident about the smoothness of the domed part. Further, it may take quite a while to print the part before I would know if it were good enough.

Another option is to cast the part from a mold, but I would need to form the prototype first with a highly polished dome surface and all the details I need. I might use a CNC milling machine to cut the model out of a block of solid aluminum, polish up the surface, then make a mold from it. Another possibility is a CNC router forming it out of a tight grained wood, then painting the wood to a smooth finish. I can use dyes and pigments to get the color of the knob right when casting.


I bought some aluminum plate to add to the desktop, where it held papers and manuals from sliding off the sloping surface of the 1130 tabletop. I still have to paint the rim around the keyboard opening with a flat blac paint. The final step will be to fit the hinges and put it into position, all registered off the location of the main black plate with the keyboard, switches and lights.

Metal plate approximately where it will be permanently mounted on the tabletop

No comments:

Post a Comment