This reminds me of a GM minivan that my youngest brother-in-law drove, back in the '80's. He'd gotten it from his father, who was a career GM automotive engineer - and complained that, occasionally & randomly, it would not start. It seemed like the minivan's whole electrical system was dead...
Brother-in-law was known to be "not so good" with cars - so his automotive engineer father didn't take the complaints seriously.
Complaints and emotions escalated, until brother-in-law convinced his dad to swap vehicles for a month, so (he hoped) his dad could experience the problem for himself.
After the problem manifested in the parking lot of the GM Technical Center, and the whole crowd of GM engineers surrounding the vehicle couldn't figure out why the heck the electrical system seemed to be dead, my brother-in-law felt pretty vindicated.
I have a 2009 Mercedes SLK that had the same symptom. It was absolutely fine 99.9% of the time, but one time in a thousand it wouldn't even try to start. No clicks, no indication that it even knew I was in there turning the key. Just dead. And then seven or eight hours later... it would start fine, as if nothing ever happened. Couldn't correlate it to anything.
Had it towed to a service center a few times when this happened. Every time, by the time they got around to trying it (a few hours later)... it would start fine, with nothing to diagnose.
Then, it was 99% of the time it was fine. I was with a group of folks car-camping off road to fly a human powered airplane for a couple days, and... no start. Finally started -- with no sign of any problems -- around noon the next day, and I high-tailed it out of there a day before the rest of the group, because getting stuck there /after/ the rest of the group would have started pushing my comfort level. So at this point it's actually interfering with my life.
I've tried all the usual stochastic troubleshooting (swapping out fuses, light to moderate percussive maintenance, alternate keys) and nothing. Finally it fails to start in my driveway, and I get it towed to an independent mechanic. It's short tow, and it fails to start when it gets there! So now he's seen the problem, and is as puzzled as I am. Of course, when he tries again the next morning, it starts fine.
He proposes two possible fixes: replacing some ECU module, or replacing the fuse box itself (under the theory that it's the connector or connection into the bottom of the fuse box that is having some moisture ingress or intermittent connection). Of course whatever we choose, I won't know if it was right or not until the next time I'm stuck. The ECU is multiple thousands of dollars, and the fuse box is < $200 with labor, so I make the easy choice.
This was six or seven years ago, and that car is still my main and only car. Hasn't had a single mechanical issue since swapping out that fuse box. A good independent mechanic and a good guess!
Finally figured it out after months when I moved a wire and it started, then move the wire close to another wire and it would no longer start. These were wires that would typically be close, so my guess is one of the wires was now generating enough noise that it was bleeding over into some other system and causing it to fail. As the car bounced around the proximity of the wires changed and lead to the random behavior.
EMF can do some seemingly crazy things. I built a kiln controller and the initial version would sometimes randomly lockup, reboot in the middle of an operation, or do other seemingly "crazy" things. Sometimes even the hardware watchdog would stop functioning.
Turns out contactors pulling in and out a 5000w load generates some strong EMF and sometimes that EMF is enough to cause random glitches to the CPU or other hardware.
Switching to high power solid state relays completely solved the problem while keeping the system compact. The actual silicon transistor was so big you could have drawn the mask by hand and it was attached to a heat sink half the size of an adult fist. I was initially worried about reliability but (knock on wood) 8 years later the system is still working without issue.
Yeah. I wrote the code for a controller that managed the temperature of a lubricant. The heater was a propane-fired burner that was mounted on the same skid. Every so often we'd get a random reboot. Finally, when it was very quiet, I heard a soft 'tic' just before I noticed that the CPU had rebooted again. EMI from the spark gap that ignited the propane was coupling back into the I/O lines and would occasionally reset the CPU.
One of those things where if it had happened 100% of the time we'd have figured it out quickly. But it was so infrequent that no one thought of that as a cause.
We had one access point (unifi) in our datacenter which was consistently failing.
fail, RMA, fail, RMA, probably replaced that thing 5 times. It was also incredibly unreliable.
Meanwhile all the other access points (probably 10 of them in total) had 0 issues.
Eventually realized that the cable for that AP was running perpendicularly over the conduits which fed power into our suite, so, about 1MW of power. Relocated the cable so it was farther from the conduit and it never had an issue again. Makes you wonder about the effect of working in such environments.
Most likely this solid state relay has a builtin flyback diode that was not in the circuit of the contactor. The diode should be as close as possible to the load to reduce the EMF.
I know an older guy at church, whose kids all graduated college - except for one.
His "failed" son is the top mechanic at a Mercedes dealership. He does some supervision, training, etc. But the reason the dealership is paying him $200K/year is his skill at figuring out and fixing problems like that, for the dealership's most desirable and profitable customers.
(That I've heard, none of the mechanic's "successful" siblings are making that kind of money.)
I know a story of a certain large engineering firm, dating back to the second World War. They had a senior engineer who habitually came to work drunk and slept through meetings. Every once in a while, they'd wake him up, and he'd save them a couple of million dollars. He had a gift for finding clever solutions.
He probably would have had a better life if he'd gotten his act together. But knowing how to fix subtle issues, or how to design good processes, can be a ridiculously valuable skill.
That definitely sounds like a loose connection, not an ECU problem. Usually it's a bad clamp on the battery terminal. Often it's the ground clamp in particular.
The easiest diagnosis is to rotate the cables on the terminal several times to rub off oxide build up, then leave them in an orientation so that the natural tension of the cable forces the clamp into good contact.
A "dead" (low voltage) battery will still cause some indicators/lights to come on when you crank, while a bad connection usually acts like zero volts.
I was about to say the same thing. My car (nowhere near as nice as a Merc) had the same issue. There would be oxide deposit on the contact points and the minuscule amount of rust and crack on the clamp would render it useless on a bumpy road randomly (1 in 10 times approx). It was scary the first few times but when I figured out the symptoms, it took me less than a minute to get it to start.
Today's car mechanics are a different breed. The ability to diagnose and triangulate problems without advanced tools made car mechanics quite intuitive and resourceful in the past. I take my car to one such mechanic. I always leave with a bill lower than any other estimate I get and he always finds and fixes the problem.
Wow, it's crazy to see this comment because I drive the same type of car and recently experienced a similar problem.
The car was dead for long enough that I could get it towed to the mechanic and they educed it was a problem with the EIS (electronic ignition system).
The EIS computer was sent off to mercedes for diagnosis, they reported that the computer itself was fine but it was an issue with power. The mechanics traced it back to a bad connection of a wire somewhere.
It would have required removing and replacing the box itself... the fuse box has a connector going into the back side, and then a separate harness inside the box that splits out the individual fuse sockets; that connector seems to have been the problem. At various points I swapped/reseated every fuse in there trying to fix things, never occurred to me to remove the box itself.
I have seen similar behavior on vehicles that have a dying alternator. The issue is that one vane of the rotor (or maybe stator?) has shorted and no longer functions. If when the car is shut off and the rotor stops in a particular position the alternator won’t work. Give it some time and with heating/cooling moving things around just a bit plus the act of repeatedly trying to turn it on potentially making it move just a bit, and it works fine.
Alternators aren't needed for starting. Perhaps a similar issue with the starter motor? But even then you'd hear the thunk of the starter solenoid. Any audible clue definitely makes root-causing simpler.
I have a crank sensor that is going out, and though it still works most of the time if the crankshaft stops in just the right position it becomes an absolute bear to start, and runs in limp mode when it does.
This reminds me of an old F250 my dad had. He towed it from Colorado to Missouri and about halfway he noticed his gas mileage got worse. But, you know, coasting down mountains until you get to plains would also have a similar outcome, so he didn't think about it.
Once he got to Missouri he noticed the transfer case (I think) had somehow bounced out of neutral and was partially engaged, which chewed many teeth off the flywheel. To start it we had to get a pipe wrench and manually rotate the flywheel to a spot with enough teeth to catch the starter and turn over. That got harder and harder (as it kept chewing off more teeth) so eventually we had to push start it and pop the clutch. Thankfully it was a manual so it still worked!
My dad had an old diesel Benz - the generator (that old, didn't have an alternator) went out and for a couple of months we just made sure to park it on a hill and drive during the day. Electricity is overrated!
Seen things like this before. Often a bad connection somewhere, possibly involving material that expands/contracts significantly in response to a change in temperature (turning the module into an inadvertent temperature controlled switch).
I don't know if Alec Proudfoot counts as a normy, but DaSH PA flew a bunch of times, with many different pilots. That particular event, which was an attempt to try some records, was a bust; there had been rain recently and the dry lake beds weren't. Landing gear was a recurring problem (light weight and robust don't go together), and even some last-minute attempts to build a runway out of 4x8 plywood sheets was unsuccessful.
I have some very, very fond memories of flight days at Moffett, though. I had gotten to work on the Moffett runway for work previously (so cool out at the bay end of the runway at night, totally silent), and getting back there to help run ground ops (including assembling DaSH PA before sunrise so first flights would hit the calmest possible air) was just lovely.
Heisenbugs are manifestations of whole-system design failings, where projects are not engineered to facilitate troubleshooting, subsystems are strongly coupled, and everything is just barely held together with baling wire and bubblegum.
That GM vehicles from this infamous era would suffer from maddening, mysterious electrical glitches makes perfect sense.
A formative early experience of mine was learning valgrind to track down a Heisenbug for a C project I was working on (which turned out to be an invalid read in a dependency). I’m indeed thinking of this anecdote when generalizing about whole-systems failings, since troubleshooting memory errors is so difficult.
I think there’s an analogy to be drawn when designing large systems on top of unsound foundations.
please expand to include more explanation for why you think this is a memory managed lanugage problem. I used C and C++ professionally for years and ran into all sorts of issues like this. Interaction between subsystems doesn't care at all what language you're using inside each component, they care about design patterns and architecture.
GP is correct that I generally prefer memory managed languages, I just think it’s right to emphasize that this preference is informed by experience. I’ve spent large amounts of my career writing C code, and now when I have a choice I’d prefer Rust for systems projects.
The higher-level point is that Heisenbugs are an emergent phenomenon of complex systems when fundamentals are lacking.
* C systems are lacking because the language is very old and we’ve learned that we need additional infrastructure to avoid memory errors.
* 1980s GM systems were lacking because of a management culture which didn’t value reliability, leading to inevitable issues in e.g. poor grounding and electrical isolation.
It’s my belief that many contemporary tech companies have management cultures similar to 1980s GM, and subsequently waste tremendous resources when troubleshooting complex systems which are not designed to facilitate troubleshooting. That’s why the original article resonates strongly with me.
GM cars are notorious for (sometimes) developing strange problems that would have you think they are possessed with the devil. (Despite post-1990 GM cars being near peers to Japanese cars for reliability overall)
I learned a lot of things from my father. Unfortunately a lot of them were what not to do. Don’t buy cheap tools that you’ll have to replace three times in the lifespan of one that costs 50% more. And don’t buy GM.
Mechanically they may be reliable, but 90’s GM forgot how to make paint stick to metal and had to pay to repaint a massive number of vehicles that simply pealed if parked outside for too long. How?
And there is absolutely no forgiveness in my soul for the Chevy Citation. I joked when I moved to Seattle that the main problem is since there is no salt, there are still Citations on the road and that is unnatural. Their place in the natural order is the junk yard.
After owning my 1988 Chevrolet Beretta for about a year, the paint came off in one big sheet one morning when I was brushing a light dusting of snow off with my wool toque.
The first call to GM revealed that it was the acid rain (acid rain in the 1980s was what global warming is today -- the cause of all evil). Exposing my car to rain voided the warranty.
The second call to GM revealed that ultraviolet light destroyed the bond between the paint and the primer. Exposing my car to sunlight voided the warranty.
I spent hours researching this issue through the trade mags and published court filings. Plenty of legal findings about implied warranties of fitness for purpose. Evidently GM had an unpublished policy that it would pay the cost of a repaint to dealers for this situation, and the dealer was expected to provide the work for free.
Of course, the greasy grin of the dealer as he quoted full price while knowing he would collect the same from GM was enough to make me drive the car with no paint for the next 10 years so everyone could see, and recommend nobody purchase any GM product ever again.
I'll say this though: that primer sure prevented rust.
I wonder if the irony of the situation is that some chemical engineer invented a new primer that sticks to metal extremely well and it turned out that it may stick to metal like glue but it doesn’t stick to paint as well as the old stuff.
But based on what peeled, I’d say thermal expansion or UV damage were involved. The former could still be the primer’s “fault”.
>I joked when I moved to Seattle that the main problem is since there is no salt, there are still Citations on the road and that is unnatural. Their place in the natural order is the junk yard.
This was the fate of many British Leyland cars, even the ones that people genuinely liked such as as the classic Mini and the MGB were practically hygroscopic.
I will say I was somewhat delighted by the number of good looking mustangs and british roadsters there were. And so many Beetles. My roadster had a little too much bondo for my liking.
I didn't realize the Countryman was bigger than a midsized sedan until I parked next to one the other day. Walking back to my car something seemed off. Wait, that's a Mini??
None of the 'Mini's' are Mini. They're just BMW that bought the brand and builds cars loosely inspired on the original, with sizes ranging from 1.25 times the original all the way up to definitely non-Mini crossovers. Our company had one of the smaller ones as a lease car and even though it drove ok'ish it definitely didn't feel or drive like a small car. It was a way for BMW to appeal more to women (with that demographic BMW never really caught on) who loved the original Mini to bits. And then wealthy women want a larger but still trendy car.
BMW is hyper aggressive about the brand, including going after each and every use of the brand for the original cars, even when used in the context of spare parts.
I think about eight, ten years ago they stopped being mini when collision safety laws changed. They had to raise the hood for pedestrian safety (hitting legs and spinning heads into the windshield is not allowed) and raise the door height for side impact safety. The whole car got normal sized because of it and then they just kinda said fuck it.
Of course there are: "a mention of a praiseworthy act or achievement in an official report, especially that of a member of the armed forces in wartime" Don't focus on the North American usage of "a traffic citation". Citation is almost a contranym, which is a word that has at least two meanings that are opposites of each other, i.e. bolt, bound, buckle, cleave, clip, consult, ...
I think it's funny how GM came out with the "Cavalier" to compete with Honda's "Civic". Or that matter, there was a Chevy Cobalt (e.g. a "Kobold" is a demon that causes mine accidents) or an AMC Gremlin.
Sure, the name is derived from "kobold", but that's like saying you should never call anything good "terrific", because it derives from the root "terror". Etymology isn't destiny.
> The statement refers to a popular anecdote in international business and marketing about a supposed blunder made by American automaker Chevrolet with the car model, "Nova."
> According to the story, when Chevrolet tried to market the Nova in Spanish-speaking countries, the car reportedly did not sell well because in Spanish, "no va" translates to "doesn't go". This led people to joke that a car named "doesn't go" wouldn’t be a popular choice.
> However, it's important to note that this is largely a myth. In reality, the Chevrolet Nova was relatively successful in Spanish-speaking markets. "Nova" as a word is understood to mean "new star" in Spanish, and it's unlikely Spanish speakers would naturally break up the term into "no" and "va", just like English speakers wouldn't naturally break up "notable" into "no" and "table".
> But the story remains popular as a cautionary tale of the consequences of not considering linguistic and cultural differences when naming products for international markets.
Technically Chevrolet did but the entire thought process for that vehicle was questionable so the name is IMO a harbinger. This is not a place of honor.
A few other GM vehicles have this issue, Chevy in particular. A well known example is the market failure of calling a car Nova (No-Va) in South America.
You know, I've heard that rebuttal, but I've been told this anecdote of the car's notoriety by family members from Columbia and more recently from a friend from Argentina. So perhaps "no va" and "Nova" are pronounced differently, and perhaps the car did sell well, but the Spanish-speaking peoples most certainly did find the term "no va" in the car's name.
You should know what they say about the Mitsubishi Pajero, too!
Sure, we have these in my language as well. For instance, the car Kia should be written like "vomit" - so we have a convoluted spelling and say the name slightly wrong to distance the word from "vomit". But it's still clear to everybody how the name should be pronounced and spelled.
Doesn't seem to affect sales, though, probably like the Chevy Nova. Those Kia are everywhere.
> Mechanically they may be reliable, but 90’s GM forgot how to make paint stick to metal and had to pay to repaint a massive number of vehicles that simply pealed if parked outside for too long. How?
Could be an older component stopped being available. Like when Apple switched to environmentally friendly lead free solder, but then the NVidia laptop GPUs got so hot they unsoldered themselves.
> 90’s GM forgot how to make paint stick to metal and had to pay to repaint a massive number of vehicles that simply pealed if parked outside for too long. How?
Dodge Ram Trucks had exactly the same problem. I also have no idea how two different companies with 100 years of history simply forgot how to paint cars.
i learned that these things typically are grounding issues. it usually came down to a single source of ground being a loose connection which is why it was intermittent. as a personal anecdote to this, as a first car as a teenager, i drove a GM/Chevy S-10 that one day started to have issues where all of the gauges on the dash would just go crazy and the lights would go on and off, and then suddenly just start working again. after taking it to the shop my dad recommended, a mechanic walked out to greet me. after i told him the symptoms, he stepped back to look at the truck model, asked me to confirm the year model. he promptly opened the driver side door, reached under the dash, located a specific screw, hand tightened it as a test, and everything worked. he came back with a screw driver to properly tighten in before sending me on my way free of charge. he told me that specific model was notorious for the screw holding the ground wire to come loose. it would cost him more in time to write up a sales slip to charge me.
I think independent mechanics tend to opt for earning goodwill in this situation rather than charging fair value for a quick and easy fix. Objectively, he provided more value than someone who might have spent several hours of troubleshooting time, but customers don’t necessarily think logically about it. So it ends up being better for the mechanic to earn that goodwill rather than appear “greedy”.
Some do the freebies because that’s who they are, and the free publicity is just a secondary effect. Some do the freebie because they know the publicity as the primary reason. I feel like this experience was the former.
> Despite post-1990 GM cars being near peers to Japanese cars for reliability overall
American carmakers really needed that kick in the ass from Japan. Around 1990 was when my parents went from being protectionist, "buy American" to never buying another domestic car again in their lives. They were angry, angry at the reliability difference and angry knowing that domestic carmakers could have done better but instead relied on people like them to buy the flag.
I had a 1989 Chrysler LeBaron that developed insane electrical issues. Windshield wipers would randomly turn on, the radio would change stations by itself. It really did seem possessed.
When I was a kid my (not wealthy) family had a Lincoln Towncar that was probably purchased used and fixed up and it ended up with some freaky electrical problems like you describe- most notably (because it freaked me out as a small child) I remember the automatic door locks would start locking/unlocking themselves rapidly, and they were big chunky metal switches that pop up and down and made an awful sound when this happened
I dunno. When my son needed a car to get to work we found a 1996 Buick Park Avenue in almost perfect condition at 100,000 miles. He didn't like the look of modern cars, didn't want a touchscreen, besides a 2010 Japanese car can cost almost new prices and they wanted $9k for a Chevy Sonic with 180,000 miles (didn't know they went that far!)
Granted the traction control and anti-lock brakes occasionally fail to boot up, but it seems to be a pretty good car, but it was close to the top of the line. Gets 27 mpg which is not bad for a big ass car. I like how it has a lot of the feel of a 1970s boat but it has airbags, OBD II and most of the good features of a modern car... And we didn't need to get a loan to afford it. Driving home though I was looking in the mirror and seeing it dwarfed by today's XXL trucks and SUVs.
I mean...it's possible they've improved a lot compared to where they were 30+ years ago, but to call GM's cars "near peers to Japanese cars for reliability" just doesn't hold up.
On Consumer Reports' list of car brands by reliability[0], none of GM's brands even crack the top 10. GMC and Chevy are 20 and 21, respectively, out of 25 brands. (The top 5 include, unsurprisingly, Toyota, Lexus, and Honda—your classic reliable Japanese brands.)
Plenty of cars that intermittently won't turn on in this thread, but I once had a car that intermittently wouldn't turn off... or at least the headlights wouldn't. I assume a relay somewhere was overheating and that was making it stay closed, but I never debugged it, and nor could my cousin who was an auto mechanic. We never tried too hard, though: I would just take the fuse out if it happened and put it back the next morning.
True story: I had that happen to an airplane. I was a student on a long solo flight and on the last two legs of that flight, the engine wouldn't shut off using the usual method of pulling the throttle back to idle and turning off the magneto.
It's been long enough that I don't remember what I did to get it shut off (maybe I toggled a circuit breaker?), but when I got back to base, I made a note in the airplane log about the problem and also left a note for my instructor, who was out that day.
When I came in for my next lesson, instructor mentioned that the next person to use the airplane, also a student on a solo cross-country, got stranded 100 miles away because when the engine wouldn't shut off, he panicked and pulled the throttle hard enough to rip the cable through the firewall. Airplane had to be put on a flatbed to get it back home.
That's probably what I did. Thinking back, the first time it happened, it just took about 10-15 seconds to shut off. The next time, when I was back at my home airport, it wouldn't stop running at all and I think I turned the fuel cutoff to get it to stop.
The radiator fan in my Jeep did that in Sudan when it was 48C. Better to have it stuck on that stuck off in those temps!
I just tapped the relay with a wrench and it un-stuck and turned off.
Funny enough that was almost 5 years ago, and it hasn't done it once since in more than 80,000 miles of driving. Not in -35C, not in +45C. Odd little relay.
I'm going to armchair and guess that your brother needed a new battery.
My (GM) car gets really funky startup behavior when the battery gets old. It will often turn the starter fine, but the electronics can get stuck in weird states until I disconnect the battery (essentially a hard reset).
The dealers wanted a ton of money to diagnose the issue, even though I suspected it must have been the infotainment system draining the battery. I just ended up replacing the infotainment system with a cheap CarPlay compatible one and the problem went away.
Problem is I cannot get money from the class action settlement since the original infotainment system is already out and I fixed it myself.
But I cannot prove it without reinstalling the original infotainment system. That is many, many hours of work for little gain.
What I do not understand is how the settlement did not require Subaru to compensate all Subaru owners for the faulty design. Why was there proof required, other than the purchase of their car with the faulty components?
I worked tech support that included a device for software developers that had Ethernet connectivity. I learned very quickly to say "Try reseating the Ethernet cable" because if I said "Is the Ethernet plugged in" about 1/3 of the people would respond very negatively (e.g. "I'm not a fucking moron") but by having to reseat the cable, they would sometimes discover it's not plugged in.
The classic way to get the consumer to power cycle the hardware is to ask them which color ring is at the base of the DC plug. It doesn't really matter what color it is, what matters is that it was removed and the hardware definitely restarted.
I once has this happen with my father's new standing desk. He, PhD with an Electrical Engineering degree, couldn't get it working. Turned out the C13 power cable was only 90% inserted, and felt firmly in place but did not have enough connector to power the desk.
A friend who is an automotive engineer shared once that most colleages were not so great with cars. Engineering new cars is one thing; fixing them is another. It's like asking a programmer to do system administration - two different jobs.
Brother-in-law was known to be "not so good" with cars - so his automotive engineer father didn't take the complaints seriously.
Complaints and emotions escalated, until brother-in-law convinced his dad to swap vehicles for a month, so (he hoped) his dad could experience the problem for himself.
After the problem manifested in the parking lot of the GM Technical Center, and the whole crowd of GM engineers surrounding the vehicle couldn't figure out why the heck the electrical system seemed to be dead, my brother-in-law felt pretty vindicated.