HP8903B keyboard

[Dickie]

I have an HP8903B audio analyser with some very stiff pushbuttons. Some of the buttons work fine with a nice soft click as they are pressed. But most of them are very stiff and need a firm push to operate. Before I took it apart I had guessed it might be a rubber membrane that had gone hard. It transpires that they have (what I assume to be) a real piece of spring to achieve the click. I have no experience with these switches at all. So is this usual? Can it be improved? Or, since they all function as switches, should I leave well alone? You thoughts and experiences would be welcome!

[Radio Wrangler]

They are called "Bill West" switches after their designer. They have a little piece of gold plated beryllium-copper whose fingers press onto gold areas on the PCB. The rd plastic bit has two pegs which go through the board and are heat-staked at the other side.

The visible bit of stainless strip doubles as a clicker and the return spring.

They're in just about every HP instrument from 1980 to the early 1990s

NEVER ever junk any HP stuff with good ones of these switches, they'll be needed to keep other stuff going!

Details are here:

http://hpmemoryproject.org/timeline/...rt_06_chapt_03


David

[Dickie]

Thanks David. Does that mean they are not fixable? And given that they are heat staked into the pcb, trying to replace them with good ones will require some creative thinking.

[kevinaston1]

See the following video;
https://www.youtube.com/watch?v=OO7gZSm9GrI

The switches are still available (but expensive), part number 5060-9436. The service manual is available on teh web; I cannot post it here as it is 17.4MB.


Regards.


Kevin

[Dickie]

Thanks Kevin, an interesting video. Unfortunately its a bit flawed. To get the spring out vertically you need to remove the keycap. It doesn't say how to do this without wrecking either the switch or keycap. And you can't get the spring out horizontally because of the neighbouring switch body.
I managed to remove and refit the spring the other way up on a switch at one end of a row. It did seem to improve the click action.

[Radio Wrangler]

cut off heat stake dome on back of board with a scalpel and the switch comes off.

fit another switch and apply hot-melt glue dab to the bit of red plastic in the hole in the board.

So they are swappable if you find good ones.

David

[mhennessy]

The worst thing about that video was those resistors. How come they don't all face the same way

A lot of HP manuals have a section about them. For example, the 8656A does. It's a big download, but if you get this and scroll to page 214 of the PDF, you'll see the details: https://www.sonoma.edu/esee/manuals/...0107_ug_sg.pdf

It seems the key cap simply pulls off. As we know, this might be easier-said-than-done in practice.

I was always intrigued by the mention of 8 possible ways to replace it until recently when I worked on a unit which mounted the switches at 45 degrees on the PCB.

I've not needed to repair these switches yet, but can certainly report that they all feel and sound rather different from machine to machine.

Just taken a look at David's link (thanks!). Here's a modified version that takes you straight to the section: http://hpmemoryproject.org/timeline/...rt_06_chapt_12 - some useful and entertaining reading there, so that's the next hour gone!

Mark

[TonyDuell]

I've come across these switches in HP instruments. The keycaps do indeed pull off, but you should hold the red housing against the PCB otherwise you might break the heat stakes.

At one time HP would not supply the switch as a spare part becuase they felt it could only be correctly heat-staked to the PCB on their tooling. You had to send the board back to them and they would replace it. AFAIK they would do this for any PCB with such a switch, the instrument did not have to be one they still supported. If that's still the case it might explain the high price quoted for the switch.

I had one that had lost the metal bow spring that gives it the snap action. I ended up just making one from a strip of spring steel (probably a bit of old clockspring) rather than pay HP to replace the switch for me.

[Radio Wrangler]

Sounds like a job for 3Dprinterman and a bag full of little keyboard switches. Could be a sellable item.

David

[Dickie]

Originally Posted by mhennessy

It seems the key cap simply pulls off. As we know, this might be easier-said-than-done in practice.

Exactly so. It would be a bit of a high risk strategy pulling at a keycap until something gaveway. As these switches are close to unobtainable it's something I wasn't prepared to consider. However, after a couple of hours work I now have a keyboard that feels like it ought to.
The distortion on each spring is barely discernible, and occurs on the "vertical" part of the spring. For each switch I simply inverted and rotated the spring so that the slight kink was in a sympathetic direction, so to speak. I didn't think trying to flatten the spring was likely to produce a long-term fix. For switches I could not get at I removed them from the pcb first by softening the plastic staking with a #7 Weller tip (370C) and pulling them out. This left enough material to hold them in the pcb after re-insertion and resoftening. This isn't a long-term fix so I will be adding a dab of hot-melt glue as per David's suggestion.

Anyway, thanks to all for your help and ideas. Hopefully it will keep my 8903B going for years to come.

[majoconz]

I have a 3586A selective level meter with a lot of these switches - 90% of them work fine with a good solid "click" - the ones that don't click still work electrically - I'm tempted to rip into them and try to repair them but having had enough trouble with this beast the other little man in my head tells me otherwise....