I had some business travel and other obligations that limited the spare time I had, particularly for physical construction tasks that have to be done at home or the shop. I was in Seattle early in the week, and after I was done with my talk and meetings, there was some spare time where I thought I might head over to the Living Computer Museum to say hi and check in on the 360/20 restoration team. Alas, they are not open on Tuesdays. If I knew for sure I would have had the time, I perhaps could have arranged something anyway but this time it was impossible to know in advance.


In order to achieve consistency in color for the light cubes and buttons, I am mixing the pigment for each color into a separate large quantity of the clear resin, then only added the catalyst to small portions of the mixture in a separate container, enough to cast one button at a time. It would have been much harder to try to match the color if I mixed multiple batches with pigment. This system requires me to build all the buttons of a color in sequence, before switching to the next color for another set of buttons. The light cubes use transparent dyes rather than opaque pigments, so these are additional batches in each of the colors. 

I did a trial run on the method for a red button, which flagged a number of issues I have to correct before I go into 'production'. The flexible mold needs a very good support to keep it from flexing and distorting the shape of the cast object. The mixture needs more catalyst as it cures entirely too slowly. Taking an object out of the mold too early can create flaws in the surface as that is still a bit wet. The red opaque pigment does come out very even but the color is a bit too dark compared to the IBM red buttons. Finally, the front face of the button shows some rippling. 

I had the mold hanging in the mouth of a plastic canister, but that was not good enough. I will figure out a good support for this mold or its replacement. The amount of catalyst will be increased by about 1/3 for the next try. The color probably just needs a wee bit of white pigment added, just to lighten it but not to make it seem pink. I will try this change as well on the next attempt. 

 I have to be patient and not release the button from the mold until it has set fairly firmly otherwise I can create flaws on the surface of the object where they are most visible. This may also be a result of the mold containing those flaws in its shape - the top support may not have been thick enough to keep it firm while the mold was setting.  If the mold seems too flawed, I will make another one. That mold will likely be made as a cubic shape rather than the heaped-up free form flexible mold I used originally. The cubic shape will avoid the distortion, at the expense of some additional mold material.

The next attempt I didn't put in the white pigment, but did increase the catalyst and found a better way to support the mold. Still, the result was not great, particularly on the face which had wet blotchy areas. I therefore decided to make a better mold, to decrease the amount of pigment because that is listed as an inhibitor to curing, and to increase the catalyst quantity.

I made the new mold as a full rectangular solid, not the piled up "glove" type mold. Rather than optimize on the quantity of mold material, I went for the block shape that would be self-leveling, more solid and less likely to deform. Next, I stirred in a smaller amount of red pigment and increased catalyst by another third.

Before the resin cured, it looked to light a shade and a bit translucent instead of the solid colored opaque appearance I wanted. That, however, might be misleading so I had to wait until it cured. Removing the cast button from the mold, I found two things. First, the button had a perfect shape and texture on the face - that part is solid progress. Second, the color was indeed too translucent. Looks like yet another test run is needed before I can start the button production process.  


I converted some of the sheet metal screws holding together parts of the sheet metal to pop rivets, but only in locations that are hidden under the tabletop or inside the machine. These help keep things aligned for my seam welding. I had a bit of snipping and shaping of sheet metal to ensure that the finished and powder coated sections could be put back into place. At this point, I have only removed the two front sections that sit under the formica desktop. Once I try the welding, grinding and powder coating, i can judge whether the process is adequate for the more visible rear sections.


I purchased an HP disk cartridge on ebay, a unit that is physically compatible because it uses the same platter size and surface, the same cartridge covering, typically these vary in small details such as sector marking and the logical organization of the blocks on the disk. The number of tracks, the track spacing, the bit density and other characteristics are absolutely common among all the drives that were licensed from the original IBM 2310 drive (ramkit). I negotiated a fairly low price, including shipping, based on the pictures which showed that these had been stored in some unsuitable place rather than a clean data center style environment.

The cartridge outside was filthy, which made me suspect the risk of internal damage was higher than for other used cartridges, but at the right price I was willing to give it a try. Secondarily, it was built for HP drives, not DEC RK05 or IBM 2310 drives. Each maker's drive seems to use a unique method of detecting sector positions - whether it is notches on a rim on the spindle or holes drilled into the hub or some other method - and the number varies from maker to maker who chose different sizes and numbers of sectors for the logical organization.

I expected that I could modify the cartridge or adapt the drive to sense whatever scheme HP used, just as I worked out a method to sense the two different DEC schemes and cause them to appear to be IBM 2315 cartridges to the 1130 logic. In looking at the cartridge I purchased, I saw that they used the drilled holes scheme, allowing me to cut my own sector notches on the rim. That is, if the cartridge surface had been usable.
HP disk cartridge opened to begin inspection and cleaning of platter surface
I cleaned off the outside first, to remove all the dust and crud, since these might migrate into the pack once I cleaned the internal surface of the platter. When I opened the platter, I was initially pleased since there were no scrapes or other signs of head crash. However, when I looked closer, I could see some kind of corrosion and pitting on the surface.

Pitting and corrosion, both on surface and on aluminum hub, probably moisture damage
 I used 99% isopropyl alcohol and Kimwipes to carefully try to clean the platter surface. Alas, whatever process occurred, most likely involving significant moisture on the surface of the platter, structurally changed the oxide surface to form those pits. In some places the oxide had come off completely, showing bare substrate metal. The pack is unusable - these holes and raised dots and pits would crash the heads if I attempted to load this into a drive.
Missing oxide at bottom as well as glued, embedded dirt throughout

No comments:

Post a Comment