原创 Smoking ban in the emergency room

Typically, medical devices are not supposed to smoke. You don't have to be the United States' Surgeon General to realise this.

Early in my career, I was a newly hired engineer at a large medical device company. The first week my manager, Bernie, gave me full responsibility for supporting an existing product. I'll call it the Smoko-2.

In my first week, I was invited to a "tear-down" session on the Smoko-2. The meeting started out great—the lead mechanical engineer praised the design, which used just one exposed screw to hold together the case and circuit boards. Things are looking rosy. People are smiling. Even the young ladies who assembled the units on the production line are smiling. Bernie himself isn't sweating and fidgeting as much as usual. Life was good.

Then things turn ugly. One of the more experienced women on the repair line says, "Oh yeah, but we do get back some Smokos with melted and smoldering power cords." You could see the enthusiasm leak out of the room. I look around and the other engineers and managers look as bewildered as I am. Nice how they spring this on us at a very public meeting! There wasn't much sparkle or eye contact after that bombshell and the meeting quickly winds down. Bernie's forehead is glistening. He's playing the imaginary drums with two pencils. The discussion peters out to random monosyllables and the group quickly disperses, with some of us on the engineering team slinking away quietly as if hoping we're invisible.

As the new engineer, I get the hairy eyeball from Bernie, and I don't need any further impetus. I go to pull the main Smoko-2.pdf drawing, and I see the problem right away. "Hmm, how did this ever happen?" I ask myself. The design required 9V at up to 2 amps be sent through a connector with 15 pins rated only at 1.5 amps each. So no problem, right? Since only 9 pins are needed for data, the designer allocated the remaining pins as follows: three to carry plus, three to carry minus. That is absolutely okay, in the theoretical plane. Plenty of current capability, 4.5 amps theoretical, 2 amps actual, a huge safety margin you think, sitting on your theoretical cloud.

But in reality, this decision is a disaster waiting to happen. The connector mostly relies on gravity, the weight of the device on its charging stand, to make firm contact. And it expects the connector pins to be clean and perfectly straight and the device and base to be exactly perpendicular. None of these situations happen in the real world. Emergency rooms can be hectic, devices are not always carefully set down perpendicular into their chargers, and bits of crud and cleaning fluids can get into things. All it takes is a speck of a foreign object to interfere, and then we have 2 amps trying to go through not three pins, but maybe just two or even one. Push 2 amps through a 1.5 amp contact, maybe add a little smudge of medical salve or a speck of bandage cotton, and we have Smoke City, Utah. Bad show. At $880 for each Smoko-2, customers would expect it to not smolder.

In an ideal world we would just redesign the whole thing—but that would be a huge deal—there is a real shortage of off-the-shelf connectors with nine or so data lines plus two heftier power lines. We might have to get a custom connector designed, new cases and charging stands and circuit boards and user guides made up, plus FDA clinical trials, easily a quarter-million dollar and year-long project. And we'd have to recall all the devices in use, a many-million-dollar hit. And the nice people on the production line will not be smiling at me in the future as they'd have to increase their build rate 20-fold to replace all the units out in the field. And poor me, I've been here two weeks and I have to tell my manager that "my" device needs a couple of million dollars in bandaging.

I was downcast for several days, even considering going back to my first job, shoveling coal into a greenhouse boiler. While dirtier, it was nice warm work, and even if the worst happened and the boiler exploded, it wouldn't hurt as much as having to run through the view-graphs in front of the managers, especially the bar chart showing $2M of unplanned expenditures on my project. When a boiler explosion starts sounding like a good thing, you know you're in a pretty dark place.
Fortunately, our dog Beau came to the rescue. He needs frequent walks to empty his output queue, if you know what I mean. During one of these long walks when I had plenty of time to think, an idea popped into my mind. No, not a perfect solution, but keeping with the medical genre, a band-aid, inexpensive, and just good enough to work.
As luck would have it, there was just enough room inside the power connector to put a simple CMOS Schmitt-trigger to sense power abnormalities and a flip-flop to shut down the power. Another Schmitt-trigger could be coerced to act like an astable pulse generator, trying to turn on the flip-flop every few seconds to retry the power-on situation. Not a perfect solution, but one good enough to prevent major smoldering and a multi-million dollar recall.

Bernie gave his go-ahead, I got to keep my job, C. Everett Koop stopped appearing in my dreams, I got a good job review six months later, and lived happily ever after.

Well, until the next debacle.

This story was submitted by George Gonzalez for Frankenstein's Fix, a design contest hosted by EE Times (US).

George Gonzalez is by day, officially, a Software Guru, using his ancient degree in Computer Science, plus 35 years of experience, stirring up commercially useful mixtures of C, Delphi, Python, and assembly language. While his father and brother are both accomplished EE's, George just attacks hardware with some general principles learned at the school of hard knocks (and with safety glasses). At home George fawns over and repairs old tube radios from the 30s thru the 60s. At work, when there is no software to do, he is occasionally allowed to use his instincts to keep somewhat less ancient (designed in 1995) products in production.