Friday, April 10, 2009

Another interesting problem at work today.

We make electronically controlled beds for hospitals, they have achieved a certain notoriety hereabouts for occasionally exploding (I say occasionally because they rarely do it twice. That would indeed be excessive. Once is hardly worth mentioning, I feel).

A recent batch of 25 turned out to have four failures, when moved up and down they would not remain flat, basically the actuator at the foot end of the bed was not being controlled correctly - it was not moving as far as the actuator at the head end, only about 60% of the distance. The four were so consistent (all showing nearly the same degree of tilt) that we decided it pretty well had to be a software issue, hard as that was to explain.

But it wasn't - let me digress back a few years to the start of the project:

When the first bed hardware (frame and actuators) arrived for us to design the control electronics it came with a example mattress. These medical mattresses are not as you might expect - designed with disease control in mind they are waterproof and made of materials that can survive being steam cleaned. Basically they're dense and covered with some polymer that resembles rubber. And they are supplied, as most things are, in a plastic cover.

I must stress that in the eternal war of man against mattress these may be regarded as special forces, highly trained and motivated opponents. If you are asking yourself "what war?" then you are either damned lucky or have not yet made love in an uncooperative bed. But that's another saga.

So, picture the scene - we have a dirty-great mattress standing up on end against a wall with a polythene cover that needs to be removed. Along comes an unsuspecting Leon, I guess he winds up doing this because he just happened to be tall enough to reach the top of the mattress. As the only other person involved who is tall enough to have done this I must now give thanks to the gods of laziness who were smiling upon me on this occasion, because Leon walked over to the mattress as I was getting to my feet, grabbed the polythene bag near the top of the mattress and tugged it off with a manly flourish.

It was probably the last manly flourish or tugging off he managed for some time, because as he did this the friction between the polythene and the mattress built up a massive static charge which then discharged between the centre of the mattress and the part of Leon that happened to be closest to it as he lent backwards... The flash was visible in a brightly lit room, and the crack, while not exactly thunder, certainly wasn't a tiny little "click!" either... Leon clutched at his groin, managed a strangled squeak and collapsed. The mattress then celebrated its victory over humanity by slowly collapsing on top of him, which was followed by a general collapse of everyone who'd witnessed this event...

"You... Bastards..." Leon managed to say, after some considerable time had elapsed. "Get this... this... THING... off me"

Well, actually he probably said it a few times before we noticed, we were far too busy trying to breathe. Excessive laughter can be dangerous...

Anyway, the upshot of this is that removing the bags from these bloody mattresses is deadly. You don't do it fast. Given the chance you don't do it at all, you give the mattress in its bag to some poor unsuspecting suckers and let them find out for themselves, as we did.

"Do you know what happens if you remove the bag quickly?" I asked the supplier.
"Don't" they said.

"You don't know?"
"No - don't remove the bag quickly"

[pause]

"Thanks. I'll tell Leon that."

Knowing this at the start I designed the electronics with more of an eye to static protection than is normal, and it was successful because we have built a fair number of these beds without further problems (though other equipment the beds take a dislike to has been destroyed by it, they don't always choose the kinetic energy weapon option).

So, what about these four failing beds? Well, yep - static damage. They'd been built, calibrated (to define the range of movement allowed, etc), and then the mattresses had been fitted, only this time they'd done four of them differently - someone had put the mattress on the bed before pulling the bag off and also done this in such a way (yet to be determined) that the discharge was through the same actuator of all four beds. This then overpowered and destroyed a filter component (ironically something that was only there to protect the rest of the electronics from this very static) and did so in almost the same way on all four beds. It went from looking like a ~10nF capacitance to resembling a ~25K resistance, with all four having more or less the same resistance when dead... This added resistance then loaded the actuator's position sense pot, destroying the linearity, so the bed control loop thought the beds were flat when they weren't.

This input protection laughs at human-body model static discharges, you understand. Eats them for breakfast, so gods alone know how much energy is involved with these combat mattresses.

I now suspect this is the real reason why nurses are forever warning people not to sit on the beds during hospital visits. They're trying to do us all a favour, they know the evil forces that lurk beneath that innocent looking exterior. Be afraid. Be very afraid...

Interesting, eh? What? Well I didn't say it was very interesting... Oh, well, suit yourselves...

3 comments:

  1. This comment has been removed by a blog administrator.

    ReplyDelete
  2. To vaporise yourself with this kind of industrial application , knowing the proper dielectric constant of mattresses in an automation is a wisely action to avoid a large occurrence in the bits.

    ReplyDelete
  3. And in the most significant bits at that...

    ReplyDelete