原创 Tortuous transient errors

Several days ago, when I started my truck after a night when the temperatures had dived to well below freezing, I noticed a warning light glaring at me balefully from the console.

Of course this couldn’t say something useful like "Check Engine." Instead, the message was presented in the form of an inscrutable icon, thereby forcing me to root through the 200+ manual skulking at the bottom of the glove compartment.

When I finally tracked these little rascals down (first the manual, then the icon), I discovered that the icon's raison d'être was to cryptically inform me that I needed to "Check Gauges." Who writes these manuals anyway? I mean, we're not talking about a vehicle that originated in some obscure part of the world -- this is a Dodge Ram truck, for goodness sake. I'm assuming it was built in the USA, so I have no idea why we appear to have commissioned a non-English speaker to write our documentation for us, but we digress...

I'm sure it goes without saying that I treated this warning of impending doom coolly and calmly as is my wont, with only a teensy-weensy whimper wending its way through my lips.

I remember hearing that helicopter pilots in the Korean War would resolve this sort of problem by throwing the errant warning light out of the window. Now, this may be a viable solution when you are scooting around thousands of feet up in the air with no one else around you, but the traffic here in Huntsville has gotten pretty hectic recently. This is only exacerbated by the fact that no one else around here seems to have the faintest clue what the indicators on their vehicles are for, but -- once again -- we digress...

I decided not to do anything hasty, so I continued on to work. When I started my truck again that evening, the warning icon was no longer illuminated. "O frabjous day!" I thought to myself, but my joy was short-lived because the little scamp returned to taunt me on the way into my office yesterday.

Thus it was that I stopped off at the Express Oil Change & Service Center, which is about 1 mile or so away from my office. When the young lad approached my truck and asked me what was wrong, I pointed accusingly at the misbehaving warning light and he said "Ah, the Check Engine light, eh?" (I felt it would be churlish of me to correct him by informing him in haughty tones that this was, in fact, officially known as the "Check Gauges" indicator.)

So I left my truck with the sterling chaps at EOC&SC and one of their merry band ran me into work. If the truth be told, I quickly forgot all about this, so I would have had something of a surprise when I exited the building in the evening and discovered an empty space where my truck should be. Fortunately, the guys at EOC&SC called me in the late afternoon to tell me what they had discovered, which could be summed up in a nutshell as "nothing useful."

Apparently, they started off by connecting my truck to a diagnostic unit, which reported that one of my cylinders had been misfiring. Two obvious suspects were the spark plugs or the coil (or distributor or whatever -- I no longer recall exactly what they said). My understanding is that they swapped the leads to two of the cylinders (I didn’t know they could do that) and then drove the truck around for a while. This was supposed to inform them whether they should exchange the spark plugs or replace the coil.

Sad to relate, however, the warning light never came back on. When I returned to pick up my truck, I took out my wallet in dread anticipation of horrors to come. Instead, the guy behind the counter presented me with my keys and said "There's no charge." Somewhat foolishly, I responded "But I thought there was a diagnostic fee." He replied "But we didn’t diagnose anything." What could I say? You can’t argue with logic like that!

I do have to say that I find this attitude to be somewhat heartwarming in an age where a plumber will happily charge you a $50 service fee just to wipe his muddy feet on your front doormat. The upshot was that we both agreed that transient faults are the worse ones to have; that if I am really lucky the indicator will never return to haunt me; but that the next best thing would be for the indicator to go hard on, thereby allowing them to actually diagnose -- and correct -- the problem.

I know that when I'm diagnosing an issue in an embedded system, I much prefer a hard fault to a transient one. In the case of a hard error, it's almost like solving a mystery by using your intuitive skills to track the issue down to the hardware or the software, and to then keep on burrowing down until you stalk the little scallywag to its lair. When it comes to transient errors, however, it can be extremely frustrating to see them flicker in and out of existence. It's especially irritating when the act of applying your diagnostic tools causes the fault to evaporate, and for the little scoundrel to subsequently reappear as soon as you disengage your diagnostic devices.

I could tell you some stories... but I'd rather hear yours. Have you ever run across a truly tortuous transient error that caused you to gnash your teeth, rend your garb, and pull your hair out in frustration? If so, please share it in the comments below.