Key bounce problem

[N Cook]

On industrial kit, matrixed 25 switches panel (but no CE key so for any
repeat bounce you have to re-enter parameters from the start) in a dirty
environment, used day in day out.
Buttons through pierced aluminium top cover engage with click switches in a
matrix
through a plastic sheet barrier. The protruding small nibs under the buttons
eventually punch through the plastic sheet, allowing grime in.
Suggested solutions / any other opinions or ranking of the following.
Clean up/ replace the existing panel then
cover the whole panel and surround with some sort of semi-moulded but
flexible covering as found over the till buttons in fast-food joints/ bars
etc.
Add plastic rings around each nib to decrease the punch through force.
Extend the switch lines to an external properly sealed/robust switch panel
and block off the original, problem here is shift function for doubling the
key functions would require legends specific to the shift function for
operator use as well as the basic 0 to 9 etc
Add small caps to the control lines which have 10K pullups to Vcc, going to
the PA terminals of Rockwell R6520A in/out interface chip
Any other possibles

 

[Charles]

On industrial kit, matrixed 25 switches panel (but no CE key so for any
repeat bounce you have to re-enter parameters from the start) in a dirty
environment, used day in day out.
Buttons through pierced aluminium top cover engage with click switches in
a
matrix
through a plastic sheet barrier. The protruding small nibs under the
buttons
eventually punch through the plastic sheet, allowing grime in.
Suggested solutions / any other opinions or ranking of the following.
Clean up/ replace the existing panel then
cover the whole panel and surround with some sort of semi-moulded but
flexible covering as found over the till buttons in fast-food joints/ bars
etc.
Add plastic rings around each nib to decrease the punch through force.
Extend the switch lines to an external properly sealed/robust switch panel
and block off the original, problem here is shift function for doubling
the
key functions would require legends specific to the shift function for
operator use as well as the basic 0 to 9 etc
Add small caps to the control lines which have 10K pullups to Vcc, going
to
the PA terminals of Rockwell R6520A in/out interface chip
Any other possibles

Gee, one would think that the switches would be de-bounced via software. Oh
well, your capacitor idea would be my first attempt. Choose a time constant
of 100 ms or so.

 

[petrus bitbyter]

On industrial kit, matrixed 25 switches panel (but no CE key so for any
repeat bounce you have to re-enter parameters from the start) in a dirty
environment, used day in day out.
Buttons through pierced aluminium top cover engage with click switches in
a
matrix
through a plastic sheet barrier. The protruding small nibs under the
buttons
eventually punch through the plastic sheet, allowing grime in.
Suggested solutions / any other opinions or ranking of the following.
Clean up/ replace the existing panel then
cover the whole panel and surround with some sort of semi-moulded but
flexible covering as found over the till buttons in fast-food joints/ bars
etc.
Add plastic rings around each nib to decrease the punch through force.
Extend the switch lines to an external properly sealed/robust switch panel
and block off the original, problem here is shift function for doubling
the
key functions would require legends specific to the shift function for
operator use as well as the basic 0 to 9 etc
Add small caps to the control lines which have 10K pullups to Vcc, going
to
the PA terminals of Rockwell R6520A in/out interface chip
Any other possibles

Saw similar panel in a garage covered with a ordinary clear plastic bag that
was replaced daily.

Is it sure that the grime causes the problem? Most of the time worn out
switches cause this kind of bounce. (But of course, dirty contacts may
behave like worn out ones.)

Besides, bounce can be intercepted to a wide extent in firmware. But that's
in the hands of the manufacturer unless you are prepared to do a reverse
engineering job.

Did you contact the manufacturer about the problem? (Not a salesman. He will
advice to buy a new panel or even a new machine.)

If al other options fail, you can cut the wires from the panel to the 6520
and place your your interface in between. Guess you need only one micro and
maybe some discretes. You can expand with an input line display and a
backspace switch.

petrus bitbyter

 

[N Cook]

Saw similar panel in a garage covered with a ordinary clear plastic bag that
was replaced daily.

Is it sure that the grime causes the problem? Most of the time worn out
switches cause this kind of bounce. (But of course, dirty contacts may
behave like worn out ones.)

Besides, bounce can be intercepted to a wide extent in firmware. But that's
in the hands of the manufacturer unless you are prepared to do a reverse
engineering job.

Did you contact the manufacturer about the problem? (Not a salesman. He will
advice to buy a new panel or even a new machine.)

If al other options fail, you can cut the wires from the panel to the 6520
and place your your interface in between. Guess you need only one micro and
maybe some discretes. You can expand with an input line display and a
backspace switch.

petrus bitbyter

I tried caps to Vcc or 0V but if on all 5 column lines say then they cross
interfere. One cap of about 1uF on one row or column intercepting the worst
one made a difference, but not consistent. Sometimes it would suppress
bounce registers, sometimes fail to input any parameter and sometimes allow
a bounce entry. The worst aspect of this kit is that to change parameters
you have to enter a "password" and key sequence totally blind. There is
nothing like the 1 asterisk echoed to one password character on a pc, maybe
the wrong character but at least you know how many characters. I would have
arranged the firmware , so for each character entered, then 1 to 7 segments
of a 7-segment display lit as an echo.
I've tried cleaning out each click switch well, quite a lot of fine crud in
the most used ones , with the perforated top cover membrane. Replaced the
cover membrane with some thicker plastic sheet, glued to the welled
structure. Another one of these I've put some graphite+silicone grease on
the contacts under the click domes and replaced the membrane with thicker to
see if that reduces the short duration on/off contacts causing the bounce.
I will have to research vacuum forming clear soft plastic sheet to dimples
for covering the keypad.

 

[N Cook]

I had a go at making a small scale (as its only calculator size keypad)
vacuum forming machine. From a de-solder vacuum pump , 2 oz tobacco tin,
polythene tube, rubber sheet and a hot air gun. From what i remember of
radiant heated sheet, vacuum forming machines, they could do with an adjunct
of hand assisted hot air because of cold spots or tight radii.
Tried doming out in the button positions some expanded ali sheet as a mould
, did a superb job of giving a diamond patterned surface of the plain sheet
and blisters formed but were not large enough or deep enough. Will have to
try step-repeat drilling out some metal sheet with larger holes than the
original front panel and some backing of expanded ali as an internal stop
face.

 

[petrus bitbyter]

I had a go at making a small scale (as its only calculator size keypad)
vacuum forming machine. From a de-solder vacuum pump , 2 oz tobacco tin,
polythene tube, rubber sheet and a hot air gun. From what i remember of
radiant heated sheet, vacuum forming machines, they could do with an
adjunct
of hand assisted hot air because of cold spots or tight radii.
Tried doming out in the button positions some expanded ali sheet as a
mould
, did a superb job of giving a diamond patterned surface of the plain
sheet
and blisters formed but were not large enough or deep enough. Will have to
try step-repeat drilling out some metal sheet with larger holes than the
original front panel and some backing of expanded ali as an internal stop
face.

Apparently, you're reinventing the mechanical part of the switches

If you want to try capacitors in the matrix, you'd better place them across
the switches. Nevertheless it may help against short, fast bounces. I do'n
expect very much effect as your bounces are apparently slow ones.

What's the age of the panel? There must have been times it did not bounce or
at least not that bad.

The more I read about it, the more I'm convinced that the switches are worn
out. Contacts (metal or carbon) may be damaged, domes or sheets are weakened
(or even perforated like yours) and so on. So I think you'll have to replace
the switches now or in the near future. Of course, once you do so you will
take every measure you can to extend the lifetime of the renewed panel.

petrus bitbyter

 

[Franc Zabkar]

I've put some graphite+silicone grease on the contacts under the click domes

I have a bunch of click domes of all different shapes and sizes that
were sampled to me many years ago.

If you can supply dimensions and a photo, I will send them to you for
the cost of postage from Australia.

- Franc Zabkar

 

[N Cook]

domes

I have a bunch of click domes of all different shapes and sizes that
were sampled to me many years ago.

If you can supply dimensions and a photo, I will send them to you for
the cost of postage from Australia.

- Franc Zabkar

Thanks for that but there is a spare, used, keypad that its possible to rob
the little used domes from.
It is a dome problem for if one key produces 2 or more entries per touch
and/or release then swapping that dome to another well transfers the problem
to that well.
The long term solution would be to extend the 10 wires for the 5x5 matrix to
an external covered keypad that uses off-the-shelf switches that can easily
be replaced.
A dab of graphited silicone grease certainly improves the bad domes and
moving them to the little if ever used key positions will solve the problem
for now.
I may try SMD caps on each click-switch as there is only space for SMDs on
the back of the original pads.
The trouble is there are 6 of these 10 to 15 yearold testers costing about
3,000 GBP/6,000 USD each to replace. But the modern replacements are too
slow, they would slow down the production line if they had to buy new, as it
is this final 100% check that limits the production speed.