L40sx and PS/2e M4-1 Space Saver keyboards [Warning: pics]

[ThinkDan]

Following some discussion over keyboards - including the fate-tempting phrase "I wasn't expecting you to have a..." - another forum member kindly sent me an M3 keyboard from a scrapped L40sx in order to try to fix a failing M4-1 keyboard for a 9533 PS/2e. My M4-1 had been put away some 15 years ago in a fully working state, and now was registering widespread key failures, which pointed to membrane faults:



If you're not familiar with these, the M3 was used on the L40sx and CL57, and begat the M4 and finally the M6 as used in the 700-755 models. Not sure what happened to the M5. The M4 external keyboard came in two versions, the M4-1 having a TrackPoint II embedded. Of course, Admiral Shark has as much detail as you could want on these here - well worth a read. M4-1 shown above the L40sx keyboard:



The L40SX keyboard fits directly into the M4 housing... like it literally slips in and screws down in the right places. Ribbon cables present in exactly the same place, with the same number of connectors/tracks. Of course, that's what the M4 was designed for. I overlaid each L40/M3 membrane over the equivalent M4 membrane, and they were functionally the same. (This strongly suggests to me that the M4 keyboard family are housings for the M3 keyboard assembly, i.e. not a distinct generation of keyboard in and of itself.)



Inside the M4 casing is a simple circuit board to manage the PS/2 keyboard and TrackPoint interfaces, and the status indicators:



Turning either keyboard assembly over, the metal backplate slides to the left once the central retaining clip has been depressed, revealing the membranes:



The Model M membrane tracks are well documented, but the M3 / M4 differ in layout, and in number of tracks presented at the two ribbon connectors (9 and 19). Tracing each track from the connector to the key contacts, it became clear that the M4 upper membrane had three tracks failed within the ribbon part - inaccessible to repair. The L40sx upper membrane checked out fine, and was a direct swap, restoring full function:



Now the only issue is finding the PS/2e...