IBM Model F XT Industrial with Oak Switches

[fohat]

Cindy has sent me a non-functioning F XT with Oak switches branded "Johnson Controls" (a nice big square shiny metal badge) in hopes that I can fix it for her.

I have never seen one of these before and I don't know anything about them.

HID_listen identifies the XT and generally returns R06 for almost any key press, and the letters do not register.

Here is the weird thing: if I *really* press hard (like pounds of force) and hold a key down, it will often eventually register (as well as return a meaningful value in HID_listen).

Any suggestions on how to "re-sensitize" the switches, or whatever else there is to try? The keyboard is truly beautiful.

[snuci]

There's not a lot of them around and I'm not sure how to take them apart right down to the membranes but it sounds like that's what is required. Could be damaged by water, perhaps. Do you have any pictures? I am assuming it's exactly like mine but with a different badge? wiki/IBM_Industrial_XT If so, I don't think that back plate comes off non-destructively.

If all else fails, I would be interested in buying it for parts (I still need a space bar!) but I would do my best to try to fix it first.

[fohat]

There's not a lot of them around and I'm not sure how to take them apart right down to the membranes but it sounds like that's what is required. Could be damaged by water, perhaps. Do you have any pictures? I am assuming it's exactly like mine but with a different badge? wiki/IBM_Industrial_XT If so, I don't think that back plate comes off non-destructively.

If all else fails, I would be interested in buying it for parts (I still need a space bar!) but I would do my best to try to fix it first.

Yes, it looks exactly like the one in the wiki except for the badge.

I don't know that I am up for a complete disassembly and some sort of bolt mod without knowing what I was doing.

More tinkering and tweaking is in order for now.

[snuci]

You should be able to try the membrane matrix through the membrane connectors. You should be able to hold down a key, then try all combinations on the two connectors to see if there is continuity in one connection, no? It's hard to do it blind (without seeing the membranes) and is time consuming but it should technically work. If you get nothing, it's the membranes.

[HuBandiT]

Get a logic analyzer or a scope (latter preferred), and check whether maybe the keyboard does output meaningful codes for normal keypresses, just in a way that is slightly out of spec (timing or voltage) with whatever decoder/converter you are trying to use, and therefore only rarely registers correctly.

[DMA]

I would bet on the membrane deterioration - microcracks in a conductive layer.
Should be easy to fix with conductive pen - but man there seem to be TONS of rivets.

visible parts - membrane connectors and their leads - should be easy to check for integrity, but I doubt a bit that it's those things.