In search of 4B3A Microswitch Keyswitches (for a restoration, not a keyboard collector!)

Ian Finder ian.finder at
Fri Jul 10 16:28:57 CDT 2020

A few responses-

On Fri, Jul 10, 2020 at 12:29 PM Brent Hilpert <bhilpert at> wrote:

> Reading that datasheet, it appears that to reduce power consumption those
> sensors employ pulsed sleep/wake operation, presenting up to a 1/8 second
> delay in response time. Probably not very good for use in a keyboard, and
> something to watch out for if looking for a modern replacement sensor.

Great catch, Brent. I'll keep looking at other options. I must have skimmed
the data sheet a bit too hard.

On Fri, Jul 10, 2020 at 01:19 PM Santo Nucifora wrote:
> I can't help as I have a full keyboard that I'd like to keep intact but
this is a home made...

I have many of the keyboards with these 3-pin switches. As mentioned
previously, what the sensors need to be electrically is the ***A sensor,
which is harder to find by a considerable margin. Sink level, vs sink
pulse, vs scan. In Al's words-  non-multiplexed contact closure type.

The machine is a Xerox system.

On Fri, Jul 10, 2020 at 02:01 PM Al Kossow wrote:
> It would be interesting to understand the failure mode.
> I wonder if they passivated the die correctly

I agree. What is interesting to me is this- If I look at other hall effect
keyboards, and even other examples of the same sensor:

They appear to be encased in a hard, black epoxy of some kind. Some of the
other keyboards I have with the same SD switches- but alas the wrong sensor
output type- appear to exhibit this as well, and have not failed.

My failed sensors have something with the gelatinous consistency and
transparency of RTV silicone. It did not appear to have become conductive,
but who knows. There were three distinct failure modes:

* 1) Output stuck at Vcc (+5v) - By the spec sheet, this should never
happen. Vo should be a fraction of a volt when not actuated, dropping to 0v
when actuated.

* 2) Output stuck at 0v - Self explanatory.

* 3) Output appears to be correct-ish Vo, but does not actuate to 0v with
magnet. This was the rarest failure.

Based on this, my best guess is that there are two failure modes:

1) RTV silicone decomposes over time, leaching something into the die that
destroys the IC's passivization layer per Al's comment

2) The RTV silicone (or similar) is not robust enough to shield the
delicate sensor die to ceramic substrate connections from vibration and
force-based trauma as it decomposes.

My expectation is that most of these switches have the black epoxy style of
package for a reason- this is the correct design, hopefully should last a
long time, and the ones that use this transparent goop will all inevitably

I have never seen any other semiconductor die encased in what appears to be

This is, incidentally, more about keyboards than I have ever cared to know.
The fact that this is an entire hobby for some people is all the more
shocking to me. I'd far rather be debugging some logic...

Cheers- I

More information about the cctalk mailing list