Family members works in parts quality (PQ) at a major automotive manufacturer.
Once they crash the car and build the car to conform to the regulatory framework - the manufacturer of the parts in the car can't change. You must use the same Nippon semiconductors etc.
Actually there is quite a lot of change going on. German car manufacturers have standardised on the so called delta-qualification-matrix [1] which basically says what kind of tests should be done to before implementing an change (which can be anything from nothing if e.g. the label print of a semiconductor changes up to a full retest of the whole ECU).
But its probably that there is just no alternatives on the market, since Ties Ones have extensive supply chain management with second and third sources for the majority of semiconductors.
This makes sense. One would at least plan what do do in case components suddenly become unavailable due to unexpected events.
E.g. when the AKM plant recently burned down this already was havoc for a lot of audio applications. I'd guess that at least 15% of all cars in the >35kUS$ range use one or more AKM ICs. Unless car manufacturers (or their suppliers) have enough stock of AKM ADCs and DACs (or assembled ECUs), they have to either stop delivering they're more expensive cars - or adapt their hard- & software to use other ICs (or get questionable stock at inflated prices, which might be an expensive option if you're building millions of cars, and expensive again if those IC have a higher failure rate due to bad storage/being recycled/...). With the tendency to do as much as possible as jit, I doubt they can bridge the duration until the production ramped up again.
I wish cars came without stereo systems, just a DIN/2x DIN slot and you slip in your preferred system. But that kind of thinking is too modular for regulators and manufacturers.
Probably depends on the use case. On a budget car I can see where you're coming from and I agree. However, in the more expensive range I was thinking about, not so much. These can have pretty good audio systems (some even come with DIRAC these days). The "infotainment" display also usually integrates things like cameras and allowing you to control various settings*.
On a 80.000US$ car you don't swap the radio and let some shop do magic adjustments - you just get the 2000US$ "high end" audio package if you care for that.
*) Regarding settings, I'm a opponent to "control every little thing with the touch screen" and don't buy cars with this safety hazard. However, I can adjust the a lot of "offline" settings in mine. Toggle adaptive head lights, lane departure warning sensitivity, registered keys, and a many more features one does NOT change during driving.
You can change them, it just comes with a lot of design, testing, V&V and Manufacturing work to satisfy the needs of a robust QMS system for a regulated environment.
Yes, that is true. I work in a factory who makes (mechanic) car parts. For example, if we buy a new machine, we have to send some pieces to the car company and they do some tests again. If think it is more, because who will pay the recall of all cars, if the new part is not in spec.
This can happen even without regulation, simply because the company doesn't want the risk of using a less-than-fully-tested design.
I used to work at a kiosk manufacturer that would do the complete hardware design, fabricate it, put it on a truck and drive it around the city for a week straight, and then make sure nothing was broken. This was to ensure that it wouldn't break during shipping to the customer. After that, not even a zip tie was allowed to be changed.
In our case, our development process was such that the software engineers didn't get ahold of it until afterwards. One time we built a machine with an internal cable that was too long and so its baud rate was too slow to perform to the customer's spec. Rather than incur the cost of re-testing with a different cable, the company spent two years trying to get the customer to accept a slower machine. Fun times.
I don't see why they'd be exempt, generally. Obviously they only need to certify the changes that make it to production but, yeah, that still seems like more than other automakers.
I'm guessing one advantage Tesla has over traditional automakers is their entire company has been built around the existing regulatory environment. I'd be willing to bet their recertification process is quite streamlined compared to companies that have iteratively patched it on to their tradition vehicle design processes. Just a guess though; I have no background knowledge on this topic.
The traditional automakers are the ones making the regulation, trough lobbying (in the good and bad sense of the world). There isn't a regulation passed on the car industry that the car industry isn't part of writing.
I highly doubt that the traditional carmakers are slowed down by regulations.
Would be cool to find someone who has actual insights to ask this.
There are lots of mechanisms to fast-track changes. If Ford changes the headlight assembly on the next year Mustang, they don't go and crash the car again. They (or a supplier) fill out the certification form showing that their testing indicates the new headlight assembly abides by all the regulations imposed on vehicles, and submit it to the relevant state and federal bodies.
You can generally tell what revisions are straight-forward to certify and which are not by noting common year-to-year changes in a model.
Frankly it's a brilliant innovation, assuming that's the case. Applying new regulations quicker is definitely a moat in a heavily regulated environment, as long as it's not half-arsed
Not unique to them... An entirely reproducible power glitch in my early honda hybrid causes the power steering to steer at full power to the left... End result is the wheel spins with so much force it broke my thumbs... Lucky I was only doing 5mph at the time!
It's 100% a design fault, and applies equally to any car of this model (I tested it on a friends, same result!). There are simple reproduction steps which can totally happen by accident, yet there has been no recall or service bulletin issued.
I ended up pulling the power steering controller apart and adding my own safety circuit to the microcontroller output so my car is now the only car of this model which won't break your thumbs or steer you into a ditch!
The design fault is having a failure mode that leads to full power to the left, not the fact that there was a failure.
I've noticed that automakers are about 60 years behind aircraft makers in design methodologies that prevent things like hardovers due to single component failures.
I'm not exaggerating about the 60 years thing. I worked in Boeing engineering 40 years ago and the "no single component failure causing a crash" was the rule and literally the law.
(A lesson Boeing had to relearn with the 737MAX MCAS system.)
Nuclear reactors (Fukishima) and drilling rigs (Deepwater Horizon) are far behind, too.
This isn't even this level... There is a certain sequence of actions to make it do it... and it isn't hard to do by accident - it basically involves turning on every electrical thing (as you would do on a cold dark night), having a quite flat 12v battery (as would happen if you left the car unused for a few months), then turning the steering wheel while revving the engine (as happens while you're pulling out of a parking space).
So park the car for a few months, then go to unpark it on a cold night... Not at all rare!
Crazy in a general context of cars but not sure they issued recalls? That seems normal wrt Teslas. IIRC the root cause was water intrusion into power steering ECU on MS P100D, could be rain or condensation.
So even the small electronic components are locked in after the vehicle passes crash testing? That seems weird, I wouldn't have thought that the IC you use on the indicator modules (say) would affect the crashworthiness, especially if you're replacing like with like.
These things aren't microservices for a dating app that is going to fold in two years. The core of working in any safety-critical field is strict empiricism. You have to accept how little you know and not believe anything you haven't seen with your own eyes (and even then, it's wise to distrust what you've seen). To believe that you have predicted all of the thousands of scenarios that will arise out of "just one small change" and forego testing is a good way to surprise somebody an early trip to the pearly gates.
Crash testing will be part of a wider testing plan. The logic there is that you have shown all the tests to pass using that particular hardware+software configuration. Only that configuration (and any future ones you put through testing, verification and validation) is what you're allowed to ship to customers.
(I work in a regulated environment, but not automotive, and we run into the same barriers, often for even minor changes. We work agile, but changes will get released in planned batches every few months, rather than shipping instantly. For automotive, I'm guessing their update cadence would be every model year.)
Yet we allow people to replace almost any part on the car themeselves, with varying grades of quality made by aftermarket or OE manufacturers. Brakes, suspension components, body panels, electical components, etc. purchased from Amazon and made somewhere in China. Once it's in the customer's hands, it seems a lot of those rules go out the window.
As someone who has worked on cars and motorcycles seems ok to me. Most failed aftermarket modifications will be like random DNA changes- no affect, a purely cosmetic affect, or a dead machine. And risks are mostly to the driver so there is an incentive not to mess up.
It's not difficult for me to get the difference: manufacturers' actions can kill thousands of people by subjecting them to additional risk without them knowing. Individuals have a right to subject themselves to risk. The food industry is regulated to avoid food contamination or food poisoning - but there are no laws against eating mouldy food in your fridge if you want to; that is on you.
Obviously, if the aftermarket product puts other people at risk, then the consumer who did the modification would be legally liable
E.g. there are these additinonal brake lights one could place at the top of the rear window. For a long time it was illegal to use them, even if you just glued them on the inside
It would seem not all car manufacturers are so strict...[1]. Some have no issues to use tape and some hardware store components in a pinch to keep production running. Theres no way they put that through crash testing...
I'm not in the car industry, but I am in the electronics industry and I have definitely felt the chip shortage. Some of my parts just aren't available, and lead times are 3-4 months. Seems to be the result of production stopping at the beginning of the pandemic and they still haven't caught up.
I was on a call with our Future Electronics reps yesterday and they more or less said the same. Every CM is booked to capacity as well. They said it's a combination of stopping production for almost 3 months and then an explosion of latent demand for consumer devices + WFH demand.
Btw, another interesting ramification of both the trade war with China and the travel restrictions is that a lot of low/mid scale electronic manufacturing is being moved to Mexico.
> Btw, another interesting ramification of both the trade war with China and the travel restrictions is that a lot of low/mid scale electronic manufacturing is being moved to Mexico.
That's a first I have seen manufacturers moving to a country with similar wage levels. Last I heard, wasn't a lot of manufacturing shifting to Vietnam, Malaysia and to some extent, India?
It's true for a lot of commodities too. The prices for Steel have increased to over 45% of the price levels seen in January 2020. Capacity isn't coming up fast even enough to meet the demand, and manufacturers won't invest in new capacity because they know the demand is transient.
Yeah, definitely noticed this. For example, I designed something (low volume) at the start of 2020 with a precision gyroscope that was easily available with over a hundred in stock each at Digikey, Mouser, Avnet etc. and when it came to starting building them around October/November, nobody had any parts of any of the variants in stock and the lead time said 22 weeks...
Although looking now for that part at least the shortage seems to be easing - there are about a dozen showing up as available across a few places on Octopart and the lead time seems to have dropped back to eight weeks.
When car manufacturers open up after Covid shutdown, they started seeing massive demands. We couldn't keep up with their demands since July. They learned their lesson from Covid and not taking any chances this year.
For what it's worth, my somewhat educated guess from someone who covers Autos slightly:
2019 was expected to be a "peak" year in the automotive cycle, so everyone in the supply chain adjusted accordingly by modeling in lower demand following that peak. I'd wager with COVID-19 the trough came significantly faster (and deeper) than anticipated, and as we gear up for an equally fast (expected) recovery, inventory levels aren't adequate.
Are there still shortages in some places? In my (EU) country at some points the shelves were half full, but there was never anything out of stock. I'm somewhat surprised they didn't sell to other countries, we can drive a truck of tomatoes across Europe without problems, why not toilet paper?
Where I live in the US, the shortage only lasted a couple weeks, where you couldn't get certain brands. There was only a few days where the shelves were totally empty. I work at a grocery store so I saw the whole thing start to finish.
At least in the USA, due to COVID there has been lower than normal new car inventory at dealerships for the second half of 2020. This drove new car prices up which in turn drove used car prices up.
At the end of 2020 it seemed like the inventory problems for new cars might be coming to a close in early 2021, but if this shortage of semiconductors is real then maybe early 2021 still won't see new and used car prices come back down for a bit longer.
The whole “dealer inventory” thing seems so strange as an outsider. Is it specific to the US? Or North America? In Europe (or at least the European countries I’ve been in) and Japan, you go to the dealer, they will have a few showroom and test cars, and then you sit down, spec your car, haggle the price (this is changing in Europe with third parties who will negotiate for you), and then wait for your car to come out of the factory
From time to time there will be a showroom car on sale that matches, especially if you’re near the end of that generation
In the US (and I assume also Canada) it's almost the exact opposite. Custom order cars are possible, but rare. The vast majority of new cars sold are built with a factory-chosen set of options and sold from dealer inventory. Transferring a car from another nearby dealer if needed is also fairly common, if they had the particular color or option group the buyer wanted but the local dealer didn't.
Which made car shopping quite a hassle for me because I wanted something that isn't very common (non-sport trim with highest spec options and manual gear) but which wouldn't have been any problem at all for the manufacturer (since they list those options on their site for those models).
But why do you think the system is setup this way? Car dealerships in the US having more money (or financing access) to keep stocks than other parts of the world? US customers wanting instant gratification for their purchase? (although, likely, anyone likes that over delayed car delivery, it's just that once you get used to instant car delivery is hard walk away from it)
Did you actually try to order what you wanted? I've been ordering my cars from the dealer almost exclusively for the past 25 years. Never had a problem. Sure, they'd rather sell a car off the lot, but I've never had one tell me they wouldn't order the car for me.
I did get rejected by Audi when I wanted to order a car in a special paint color. They advertised "any color you want, just $2500 extra". When I tried to take them up on it, the reply was, "We don't know you, your dealer isn't a top performer, and our custom paint shop is booked up over a year anyway. So kindly bugger off."
I suspect you’re paying a large premium. The last car I bought included a $10K discount off MSRP or a smaller discount with 0% financing, but only for in-stock inventory. That was before haggling, as those were pass-through dealer incentives.
Presumably, no one pays anywhere near MSRP for that model unless they custom order from the factory.
I’m sure it varies from manufacturer to manufacturer, but still.
Spitballing here: more land in the US means lower cost to stock cars in dealerships. Also makes it easier to do an impulse purchase with all the easy financing available, compared to the European scenario of waiting for delivery (and possibly changing one's mind).
Car dealerships often sit on prime retail locations. So that vast inventory will often sit in a location close to but not conveniently adjacent to the actual dealer.
"From an automotive standpoint, dealership inventory remains lean – currently sitting at a nine-year low and down 22% for year-ago levels – which has already led to some reversal in the very generous summer incentives", which indicates higher prices. I couldn't find pricing data.
This new shortage + current shortages + low interest rates = high prices on used/new. I was thinking of buying something last year, but now I think I'll wait until '22 or '23. (or if I can WFH, never!)
Yeah, I was also hoping to buy a vehicle in 2020 but ended up holding off due to not driving nearly as much now and due to the prices going up. I was hoping prices would come back down in 2021 but we shall see...
I'm complete confused about the economy in the United States. You look at the news and all you hear about is the enormous amount of financial pain families are going through at this time. But then you hear about cars and trucks flying off of dealership lots.
I guess I'll be holding onto my 2011 Honda Odyssey until it catches fire on the road.
The average income (including stimulus checks) in the US went up, while at the same time lots of people losing their jobs. People who didn't lose their jobs (or have their hours curtailed), are doing significantly better because they aren't spending as much money and got stimulus checks.
Meanwhile those who did lose their jobs are screwed. These people were probably also poorer before COVID hit, so fewer of them would have bought a new car absent COVID anyways.
This is also why the stock market can't seem to crash; bond yields are super low, and there is more money being saved (and thus looking for investments) than ever.
That's been the problem in the US for the past 45 years or so. We have a lot of wealth, it's just not particularly evenly distributed. Every single financial shock in my lifetime has made the problem worse.
I've heard its a K-shaped recovery. Some parts recovered quickly but others didn't. For example service and tourism didn't. But tech and construction is less impacted.
Used car pricing went sideways after the cash for clunkers program, and the market still hasn't re-normalized, at least here in California. I'm surprised at how long the effects are lasting.
Sorry, but this is bullshit pedaled by people who've never given an ounce of thought to the economics of car buying.
C4C effected 700k cars from the 90s, all of which were worth less than $3,000. There were, at the time, roughly 250 million registered vehicles on the road. There's absolutely no chance that this had any effect what-so-ever on the market for used cars. 700k is literally a rounding error when compared to 250 million.
What hurt the used car market was the fact that new car sales fell from roughly 15 million a year in 2008, to 9 million a year a year in 2009. It took until 2015 before cars sales reached that of 2008.
All told 40 million fewer cars were produced between 2009 and 2015 compared with how many would have been produced if not for the GFC.
Those 40 million cars that were never produced are what caused used car prices to spike. Not the 700k Ford Windstars and Explorers lost to C4C. Anyone who tells you different is a fool.
I couldn't read this article, but another spelled it out:
"Industry officials say semiconductor companies diverted production to consumer electronics during the worst of the COVID-19 slowdown in auto sales last spring. Global automakers were forced to close plants to prevent the spread of the virus. When automakers recovered, there weren't enough chips."
Maybe they need to pay more then? or it's not actually an issue. I wonder what the margin on a car and associated components is.
“The problem is, we are lower in the chain than companies like Apple and HP,” said one executive. “The automotive industry doesn’t pay that much for its semiconductors.”
I have witnessed automotive companies looking at shaving fractions of pennies off of BoMs for ECUs. Obviously cost savings is only somewhat related to margins, but that's a datapoint.
It would not surprise me to find out that Apple makes more money per iPhone manufactured than Toyota does per Corrolla, but I could definitely be wrong about that.
If I remember correctly VW makes 400 Euro per car they sell in Germany (pre covid). I believe its just another example of Parkinsons Law (https://en.wikipedia.org/wiki/Parkinson%27s_law) and hope the legacy oems go bust rather sooner than later to free up the massive ressources they allocated inefficiently all the time.
How is that Parkinson's law and not a success story of market competition? Isn't it that the highly competitive car market forces legacy manufacturers to accept lower margins? If anything, we should welcome greater competition to drive down the margins of electric car makers.
Yes, it's true that the margins were low, which is a strong indicator for competition. However, I think the weakness in electrification and software makes it clear that there's little true innovation. Legacy automotive turned into a stable market, and participants were happy to fight over pennies. Eventually US manufacturers realized they couldn't compete and needed to change, but even then decided to retreat into trucks and SUVs rather than truly innovate.
It seems like anyone that had over 50% of the market share of large SUVs and made 30% profit on each one would probably double down and ride that gravy train to the end?
I see ID.3s everywhere. They are more common than any other electric car. They also tends to blend in with other VW vehicles.
>However, I think the weakness in electrification and software makes it clear that there's little true innovation. Legacy automotive turned into a stable market, and participants were happy to fight over pennies
If you sell millions of cars you can hire a whole team of penny pinchers who will then save your business 10s of millions dollars. Every penny pays for a $100k salary.
If anything the semi conductor shortage is causing downstream issues in VW's ability to produce more ID.3s.
Budget car makers are more like gaming console manufacturers and SaaS subscription services every year. They make almost no money when they sell you the car but they make all of it back from overpriced optional extras, the maintenance services you have to take at the overpriced authorized dealer and through their financing services.
I live in Europe and keep pondering moving to a bigger city and downsize just to give up the need of a car since the maintenance and the rest of the costs are going up faster than my pay. I can see a lot of low-middle class people getting screwed by the rising costs of car ownership long term.
which is weird, because $1k - $2k retail prices per some electronic modules, like ECU, are not unusual (couple friends had issues with their car where the only solution is the replacement of the ECU). That is the price range of pretty decent laptops, which includes not only chips and such, but also display and batteries.
Don't assume that the retail price of a replacement part is the BOM cost used by the manufacturer. It's likely integer multiples (if not an entire order of magnitude) higher.
e.g., the replacement cost of a control board in my furnace is around $800, IIRC. I could probably build it for about $30 in parts (ignoring the software in the microcontroller). However, to someone who's evaluating buying a new furnace vs. repair, that $800 is a hell of a lot cheaper than a new furnace.
When you buy a replacement you are paying for their entire replacement parts chain. Try buying a replacement part for a 10year old computer, if you can it will be more than a new computer.
That makes sense, but then why the same market dynamic does not work between retail sales and car parts makers? If the end customer is ready to pay 10x for the part, yet the maker of that part struggles to keep costs down - something weird is going on.
The control board for my furnace is ~$150 (gas/condensing). $800 sounds like that includes labor of installation since they are somewhat annoying to replace as you practically need to take half the furnace apart to do so.
That's exactly it. Nobody else can duplicate the part, and it turns a brick into something valuable. People will pay up to $WhatTheNonBrickIsWorth to make that happen.
Stupid question: does this have anything to do with the recent cryptocurrency boom? This always seems to happen right around the time people start scooping up GPUs and such.
Based on other sources of this story it basically goes like this -
1) Covid hits and sales and demand for new cars drop.
2) Automotive Industry cancels a load of silicon orders.
3) Covid hits and there is a surge in demand for replacement/upgraded IT equipment.
4) Silicon Fabs adapt and take up the slack / Re-Tool
5) Automotive experiences a sudden surge
6) Silicon Fabs go - Uhmm yeah, you are going to have to wait.
[Hit the FT Paywall so am not commenting directly on the linked article]
Good distinction. The difference between eating all of dinner and still being hungry, versus eating all the food in the house and still being hungry. Hand waving metaphor, but more visceral.
I was thinking why not make cars like cars, not smartphones. I hate when a car after a number of years is still perfect except for the extra bells and whistles.
Well there's a lot of safety features that require semiconductor components. Back up cameras, collision warning, lane assist, brake assist, tire pressure warning, etc. Humans are bad drivers.
we dont need half of that stuff. backup cam is fine (because of the stupid b pillars now) , tire pressure warning is probably cheap enough and not going to be hard to fab at more open foundaries but lane assist and brake assist. naw, fucking drive your car . plus all that self driving stuff makes it easier for cars to be totalled and bring repairs up because of all the sensors.
The pillars are there to save lives during a crash.
The lane assist and brake assist - along with adaptive cruise control - are great systems that also increase safety, if used properly. The difference in fatigue alone that these systems contrubute to is (anecdotally from my experience) night and day.
I believe that the improved safety is worth the increasing complexity of maintenance.
Taking this further. Statistically I might be better off with full autonomous as soon as available, but I wouldn't feel as bad if I crashed myself vs dying while letting the ML data do it's thing. If harm came to others, I'd still be the one who chose to buy/use it and only technically free. At some point I will change my opinion but it will have to be many many times better than humans.
car manufacturing mysteriously hits unsolveable shortage at the very same time as industry-wide regulatory changes regarding carbon emissions are now inevitable ?
Once they crash the car and build the car to conform to the regulatory framework - the manufacturer of the parts in the car can't change. You must use the same Nippon semiconductors etc.
That's the simple explanation I was told.