A month or two ago, I purchased a 1993 Lexmark Model M 52G9700 from an
Ebay seller. When I plugged it in, it became obvious it did not work
properly. The "X" key did not work. When I pressed the "Y" key, I
got "6y". When I pressed the "T" key, I got "5t". When I pressed the
"[" key, I got "-[", and so forth. It was plain broke, son. The
keyboard was purchased as-is, so I put it aside, occasionally stealing
springs and keys from it as necessary to repair other various Model Ms
I own. More recently, I decided to use it to verify a hypothesis.
I own about 25 Model Ms in various states of rivet depletion. My
least-riveted keyboard has 31 missing rivets (a 1986); the
most-riveted one has zero missing rivets (a 2008). 10 missing rivets
is about the average across the entire collection. To do most of my
daily typing these days I use a keyboard with zero missing rivets. As
a result, I've become more attuned to the impact that rivet loss has
on typing feel. I think most of the keyboards I own with significant
rivet loss in concentrated areas would probably type better if those
areas were somehow re-riveted. And besides, even if they didn't type
any better if their rivets were fixed, missing rivets just bother me
on some OCD level beyond any rational explanation.
Performing Sandy's nut-bolt mod (or Ripster's "American version" of
the same) would definitively fix the keyboards that might need fixing.
But performing a nut-bolt mod will take at least a couple of hours for
each keyboard. Given that I have maybe ten of them that I believe
could use such a mod, that'd take a long time.
In the meantime, Ebay seller Brian O'Neill (his Ebay handle is
"rarekeyboardstore") tends to put quick-setting epoxy on any missing
rivet holes of the keyboards he refurbishes and sells. The
implication is that this is a reasonable fix for a board with some
missing rivets, and, as a corollary, means that a nut-bolt mod isn't
necessarily required. Most of the folks I've read on Geekhack whom
know and have an opinion about this practice of Brian's seem to
believe that putting epoxy on the rivet holes doesn't actually do
anything; the epoxy acts only as a placebo and Model Ms with
concentrated rivet loss basically must have at least a partial
bolt-nut mod done to them to make them "tight" again. But nobody knew
for sure.
The hypothesis that needed proving: epoxying the places where rivets
used to be is a waste of time because the epoxy does nothing at all.
Was it true? Only one way to find out.
To test the hypothesis, I decided to derivet the broken keyboard, and,
while the steel plate was held to its frame with clamps, put epoxy on
top of every resulting rivet hole, such that the epoxy would be left
as the only thing holding the pieces together. I'd then check:
- Does the epoxy do anything at all? Will it even hold the weight of
the plate to the frame against the force of gravity? Does it seem
like the epoxy has a good chance of holding up over time?
- Assuming the epoxy does anything at all, is there any obvious
visible difference between a keyboard held together entirely with
rivets and a keyboard held together entirely with epoxy?
- Will it type? When held together entirely with epoxy, will the
springs buckle? Are there any "lazy" keys which don't buckle
until noticeably late in their downtravel?
- If the springs buckle and it types at all, does it type well? How
does using it after an epoxy job compare to using it before the
epoxy job?
- Regardless of the quality of a completely epoxy-riveted board, once
epoxied, is it possible to reverse the modification? Can you
"de-epoxy" an epoxy-riveted keyboard in such a way that you can
later perform a nut-bolt mod to it?
For this task, I bought from Harbor Freight some Super Glue Instant
Set Epoxy made by "Pacer Technology". It claimed to set in 90 seconds
(handle item in 10 minutes, full cure in 24 hours). Although it would
be weaker than longer-setting epoxy, presumably a quick setting epoxy
would not have as much time to ooze into places it shouldn't be oozing
into, like the space between the sheets of mylar that make up the
membrane of the keyboard.
Here's the board before I disemboweled it. Ironically, this 1993 board was missing very few rivets.
I hauled out a chisel and hammer and deriveted the board.
A problem with this board was obvious once I took the membrane out.
Sticky liquid had gotten into the membrane and between the sheets of
the membrane. You can see the staining of the liquid on the plate and
the membrane.
You can also see the gunk left on the rubber mat under the membrane sheets.
While it wasn't necessary to prove the hypothesis, while I had the
keyboard apart, I cleaned the gunk off the rubber mat and cleaned the
membrane sheets with alcohol.
I roughed up the surface of the steel plate above each rivet hole so
that the epoxy would have a bit of surface area which to grip while it
cured. Then I clamped the parts of the board back together tightly.
The epoxy looms ominously.
I used a Q-Tip to apply handmixed epoxy to each rivet hole on the
clamped assembly in the places that were clamped most tightly.
Then I allowed the board to dry overnight. The next day, I moved the
clamps slightly, so that pressure was applied more closely to the
rivet holes that I had not already epoxied; I then applied some more
epoxy over the remaining rivet holes.
Once all the epoxy set, I was able to handle the keyboard without the
clamps.
It held the weight of the steel plate while I held it up only by
barrels on the frame. I couldn't pry it apart with my hands even with
a good amount of force. It looked like the plate became pretty flush
with the frame on all sides. (That flappy bit isn't the frame, it's
the rubber mat.)
I attached the keys and made sure it worked.
Side angle.
Does the ghetto epoxy riveting mod "work" on a very basic level? Yes.
Epoxy alone has enough sticking power to hold the steel plate to the
frame in such a way that each of the springs in the each key buckles
satisfactorily. The epoxy bond between the plate and the frame cannot
be trivially forced apart. Visually, it seemed that the frame was
satisfactorily flush with the plate. The keytops were not on
noticeably different planes.
It works, but how about the typing performance? Slightly below
average. All the keys that had springs (I stole some from this board
previously) buckled properly after the epoxy riveting. There were no
lazy keys. The touch and force required to depress a key was slightly
less than average for an M. It was spongier than the feel of it
before I deriveted it. It sounded springier. But neither the feel
nor the sound was throw-it-away terrible; I would personally feel
comfortable using this totally epoxy-riveted keyboard or giving it to
a friend. Someone who didn't own 24 other Model Ms would almost
certainly never notice.
Will it help to apply epoxy to a partially deriveted board? Probably.
The results of this test would seem to imply that if you have a board
that is missing a large number of rivets in a concentrated area,
applying epoxy has a likelihood of making it type better, if only
because the frame will no longer be able to flex away from the plate
during normal typing. This will cause the buckling point of the keys
in that area to become more predictable.
Is epoxying rivet holes a reasonable alternative to doing a nut-bolt
mod? It depends. The performance of a completely epoxied board is
slightly below average. A more dramatic improvement in typing
performance will be obtained from a nut-bolt mod. A nut-bolt mod is
also aesthetically more pleasing, and it makes disassembling the
keyboard in the future much easier. Because the nut-bolt might allows
you to tune the pressure controlling the clamping of the parts, it's
probably a better way to make a keyboard queiter. But, if you're not
terribly particular, and you just want to cheaply get something done
that has a shot at improving unpredictable typing performance of
keyboards with concentrated deriveting, doing a quick epoxy job is
likely better than not doing it.
Will the epoxy hold up over time? Impossible to know. There are lots
of ways it could fail. It may contract and expand with humidity and
temperature over time. It could become brittle and flake away. It
almost certainly will cease bonding if heated to high temperatures
(above 100C). Only time can tell.
And yes, cleaning the membrane of liquid fixed most of this particular
keyboard' problems. The "Z" key still doesn't work 100%. I'll try to
fix that after I derivet it again.
Soon I'll try to "derivet" the epoxied board; I'll follow up here when
I do.