Elon Musk’s Machines Monologue
Originally published on TeslaMondo.
The last few minutes of the shareholder event, beginning at 1:22, during which Musk describes his ongoing re-examination of his own factory, warrant a verbatim transcription here on TeslaMondo. This is an alpha engineer at work, folks. It’s an awesome thing to behold. It might be dry, and it might lack an exciting visual, but it represents Musk’s dogged first-principles reasoning — the same reasoning that resulted in reverse rocketry and Tesla’s multiple novelties — “caught on camera.” He’s realizing that the factory’s design and production processes embody a little too much reasoning by analogy. They could be re-worked and dramatically improved. This will result in a better, stronger, and faster Tesla.
Maybe you watched the Six Million Dollar Man as a child. Remember the opening? For a couple of seconds, you saw Steve Austin running at full gallop in real time, not slow motion as in the actual episodes. This was pre-CGI, of course, so we weren’t jaded to such spectacles. He ran faster than anyone could possibly run, and it looked real. It inspired lots of children to go outside and run around the yard.
Well, this is every bit as inspiring. If you have any appreciation for engineering, this makes you want to burst open the door, go outside and … and … re-think something, dammit.
Musk’s thoughts may be a little hard to follow at times, but that’s okay. Nobody could keep up with Steve Austin either.
“We’ve realized that the true problem, the true difficulty, and where the greatest potential is, is building the machine that makes the machine. In other words, building the factory. And really thinking of the factory like a product, not a hodgepodge of things where the machines are bought from a catalog.
“Just like we do with the car. We don’t try to create a car by ordering a bunch of things on the catalog. We design the car the way it should be and then we make them — either we or working with suppliers — make all of those individual components. There’s almost nothing in a Model S that’s in any other car. And I think the same approach is the right approach to take when building the machine-maker. The factory.
“I actually think the potential for improvement in the machine that makes the machine is a factor of ten greater than the potential on the car side. I think maybe more than a factor of ten. I’ve really come to appreciate that over the last two or three months in particular when I’ve just been on the production floor all the time and seeing things, running production personally at a detailed level. I don’t even have a desk or an office anymore. I’m just basically standing on the production floor and occasionally meeting in a conference room.
“And it’s like, wow. I do my favorite thing, which is apply physics-first principles. When you think of a production facility on a fundamental level, for a given size of factory, the output is going to be volume times density times velocity. So let’s look at our factory and say okay, what is the density of useful to non-useful volume? It’s crazy low. It’s like two or three percent. If you look at it volumetrically, not just on a planar level, but volumetrically, it’s literally two or three percent when you say car to non-car volumetric ratio. Like wow, that seems like a lot of room for improvement.
“And then you say velocity. What is a reasonable expectation for the exit velocity of vehicles from the factory? And at first you may think that, say these advanced car factories from around the world, and they’re very good at making cars, and they may make a car every 25 seconds. That sounds fast, but actually, if you say, well, the length of the car plus a buffer space is approximately five meters. And so it’s taking 25 seconds to move five meters. Okay, that’s .2 meters per second. Basically, you’re not much faster than a tortoise at that point.
“So it’s like, that doesn’t seem fast. A slow to medium walk would be one meter per second. And a fast walk would be 1.5 meters per second. And the best car factories in the world are doing .2. Seems like we should be able to have cars exit at at least walking speed. This doesn’t seem so crazy.
“And then, the density improvement. There may be as much as an order of magnitude in density possible as well, going from maybe two or three percent to twenty or thirty percent of the volumetric density being optimal. And you also think of it like the design of a modern system on a chip or a computer. If you look at, say, the complexity of the board, and you see how close together the line tracings are and how focused things are on clock speed and data transfer from RAM to solid state disk or the internal CPU cache. It’s like wow, there’s crazy potential for improvement here. I think at least an order of magnitude in potential for improvement on production. And so with significantly less engineering effort, we can make dramatic improvements to the machine that makes the machine.
“I think probably a lot of people will not believe us about this, but I’m absolutely confident that this can be accomplished. We’re basically designing our factory the way you’d design an advanced computer. And in fact, use engineers that are used to that, and have them work on this. And I find once you explain this to a first-rate engineer, the light bulb goes on and they’re like, wow, as JB was saying, they spend huge amounts of effort trying to get a fraction of a percent of improvement on the product itself, but actually, that same amount of effort is an order of magnitude if you focus on building the machine that builds the machine. And it’s just that a lot of engineers don’t realize this is possible. They think there’s a wall. They’re basically operating according to these invisible walls. We’re in the process of going through and explaining that those walls don’t exist, and I think it’s going to be pretty amazing.”
Have a tip for CleanTechnica? Want to advertise? Want to suggest a guest for our CleanTech Talk podcast? Contact us here.
CleanTechnica Holiday Wish Book
Our Latest EVObsession Video
CleanTechnica uses affiliate links. See our policy here.
I hope Elon continues to tinker with things for a long time to come … 🙂
And doesn’t shoot away to Mars….
looking forward to modern car manufacturing.
All you need to do is buy a Toyota Corolla,
When I was an auto techician a lifetime ago, I had tremendous respect for Toyota. We just didn’t see the kind of issues other cars routinely had.
I don’t have any recent repair shop experience, but my dad recently sold a 4 year old Toyota because he got tired of paying for out of warranty repairs on such a new car.
That’s only a single data point, but after hearing about what he had to have repaired, it did make me wonder if Toyota has been slipping.
Explain “need to”. Who is going to make people buy a piece of junk like that for 10 times the price? Or perhaps you were not paying attention?
Thinking like a physicist is what is completey out of reach for many stock pseudo-analysts.
The first thing I think about when Mr. Musk speaks about making the machine which makes the machine is a Swiss watch. A multitude of components working with exact precision creating a specific consist outcome.
Getting cost down through greater efficiency is great. Fabulous.
But I’d be more impressed to hear Musk talk about how he’s going to improve quality by an order of magnitude.
Improve reliability so the car lasts 2x as long and you can make a case that the car is worth 2x as much. (same cost per mile). And less environmental degredation (fewer raw materials per mile).
I find it interesting when people assume one of the most innovative companies of all times might not be able to eliminate “squeaks and leaks” in their manufacturing process.
I don’t make any assumptions at all. But as a Model S owner who owns several other EVs I have plenty of experience to comment on the need for Tesla to focus on quality before volume.
Quality is not seasoning you sprinkle on after the car is baked. It starts with the ingredients and involves every step of the process.
Accelerating Model 3 production schedule makes me less likely to exercise my 2 reservations, not more likely.
Fundamental engineering triangle:
Speed : Quality : Cost
“Quality is not seasoning you sprinkle on after the car is baked. It starts with the ingredients and involves every step if the process.”
Certainly, but if one looks at the learning process Tesla has gone through from the original roadster idea through the Model X it’s clear to me that these folks learn stuff and apply it. The Model S may be the best car ever produced, short of “squeaks and leaks”. The knowledge of how to build without minor quality problems like those is common in the car industry. Easy enough to hire a few specialists and add that knowledge into the mixing bowl before pouring the cake batter into the pan.
> Easy enough to hire a few specialists…
I have tremendous respect for you Bob.
But your assertion begs the question why didn’t they do that three years ago?
Just to be clear with everyone reading this: I like my Model S very much. I still want Tesla to focus on quality.
That’s not just from personal experience, that’s also fron talking to hundreds of Tesla owners personally (I am the Vice President of a Tesla Owners Club) and reading thousands of owner reports online over the past 3 years.
I do believe Tesla will get it right. But I hope that is before cranking out a few hundred thousand Model 3s.
New company learning.
They may have, in fact, brought in the people they thought they needed. Or thought they didn’t need additional help. Obviously they didn’t perform at the A level in terms of squeaks and leaks.
Look at other problems they’ve had. They had two transmission manufacturers let them down. They turned to companies in the business but that wasn’t enough. Now Tesla seems to have solved its drivetrain issues and almost certainly knows how to prevent problems in future models.
Tesla certainly is capable of making mistakes. But I’d be surprised to see them repeat a mistake.
—
Your take. Outside of early drivetrain problems (mostly restricted to 2013 models?), squeaks and leaks, and sticky automatic door handles are there any other higher frequency problems that owners are reporting?
Not the sort of “one owner found their right side mirror screws were not adequately tightened” sort of low level stuff.
I can sympathize with supplier problems. But that’s not a remedy.
I’ll be writing about it at length later this year, so current and prospective owners can be fully informed when making a decision.
You’ve mentioned the most frequent serious issues; drivetrain, battery. Door handles less serious but potentially very expensive out of warranty.
Rear wheel alignment/tire wear has been another issue I’ve heard about from several owners. Noise from cabin temp control and battery cooling fans.
Reports of an $8,000 brake job and $3,000 Model X windshield should give anyone pause who isn’t used to exotic car repair pricing (I am, but less enthusiastic about that level of investment in my daily driver at this stage in my life).
$8k for brakes and $3k for a windshield have been discussed here. Apparently those prices are not out of line for cars in this price range.
Have there been drivetrain issues for Ss built in the last couple of years? What I’ve read is that there are still some 2013 models needing upgrades but I’ve read nothing about newer models.
Battery problems. Some problems with the 12 volt battery? Solved?
“Rear wheel alignment/tire wear has been another issue I’ve heard about from several owners.”
Fixable? Still occurring?
“Noise from cabin temp control and battery cooling fans.”
One would think that a minor issue of adding more sound deadening. Correct?
The issue, for me, is not how issue free the 2012 and 2013 models were but whether the problems has being eliminated with more recently built models. (Remember the two fires from punctured battery packs?)
Musk apparently said that the final design for the 3 should be locked down in a few more weeks. I would think that would give Tesla time to test the 3 can take care of most issues before release. Plus delivering the first units to employees should catch any bugs that initial testing might have missed.
My objective is not to nitpick through every issue. That the X had the kind of issues it did is frankly worrisome.
Like I said, I believe they’ll solve the quality issues. Until they do, I’ll call it like I see it.
I’m not suggesting any problems be ignored. I’m wondering what any current problems might be.
The need for a shield for the battery pack? It’s on the checklist.
The drivetrain solution? It’s on the checklist.
Squeaks and leaks? I assume those are on the checklist.
I think Elon & the Tesla team learned from the X to completely change up (to “design for manufacturing”) for the 3, if that wasn’t already planned.
JB & Elon have expressed something akin to regret or apology several times regarding the X, and said they were just trying to build a car that worked with the S. They’ve come a long way since the Roadster, which I think was a main point of the layout of the shareholder meeting.
Granted, caution is a fine thing to have. But I’m a lot less worried about reliability and such for the 3 *because of* the X problems.
Each of the Model S problems taught Tesla a lesson, or should have.
Lug nuts: watch for dissimilar metals and check for whether things will swell or shrink
12V battery: be very careful about the load profile on this battery and specify it more carefully than Tesla did. (They should use li-ion instead of lead-acid, bluntly… lead-acid can’t handle the load pattern)
Drivetrain grease problem: always use automated and precise lubrication
Drivetrain shim problem: always be very precise with stuff involving gears
Side mirror metal rusting problem: always apply a protective coating to any metal which is even incidentally exposed to water
UMC: always leave a safety margin on your electrical wire diameters, and be very careful about plug contact design to avoid loose connections (it’s harder than it looks)
Door handles: expect people to jerk the door handles violently and allow for it (among other lessons)
Noise: put vibration insulation materialss around any reciprocating equiment, pumps, compressors, or fans
Windshield fogging and freezing: design and test cabin airflow in a WET cold environment like Detroit or Toronto, not in Norway or California
Bad temperature sensor behavior : design airflow over temperature sensor, and exposure of temperature sensor to direct sunlight, and proximity to heat-generating equipment, very carefully
Et Cetera.
The question is whether they learned all the lessons they should have. I see no way of knowing this until the next car comes out.
“design and test cabin airflow in a WET cold environment like Detroit or Toronto, not in Norway or California”
I’ve long felt that the Big Three largely lost their dominance of the US car market to Japan because they were located in Detroit. In the upper Midwest. Had they been located along one of the coasts they may have gained a better feel for how markets were changing.
Holland, Michigan! Worst winter weather in the continental US. I went to college there.
Holland, Michigan. Where “Dutch doors” refers to the hidden backdoors on bars where members of the Dutch Reformed church can sneak in for a couple of drinks without being seen by members of their congregation. ;o)
(I did some graduate work at WMU.)
I wouldn’t know. I’m not Dutch Reformed, so I didn’t need to look for secret doors. If I wanted to go to a bar, I could walk in through the front door.
It appears, from the information which has come out, that the alignment machine used at the Tesla factory prior to mid-2014 — which they used to align all their service center alignment machines — was misaligned. (!!!!) It took an owner getting an outside alignment, and forcing the Service Center to compare the results to their own machine, before Tesla figured out what the problem was and fixed the alignments on their alignment machines.
Since then Tesla’s alignments seem to have been fine. Owners were not compensated for the excess tire wear caused by Tesla’s misaligned alignment machines.
Wow, where did you see that? Never saw that thread.
The drivetrain design had multiple problems. The first revision was “A”. They’re currently on “Q”. So far it looks like revision “Q” has solved all the problems.
The 12 volt battery problems are constant and continuing and have not really been fixed. The 12v batteries need to be replaced quite frequently.
The “vampire drain” is very high, which indicates bad software design.
The Universal Mobile Connector was poorly designed internally and almost everyone who uses it regularly has had several replaced. Tesla seem to have completely replaced the design a few weeks ago.
I have been hoping and assuming that each of these design failures has caused Tesla to *learn a lesson* and that they will be able to avoid this design mistake on Model 3. We will, of course, see whether this is the case or not.
All of your bloviating is accounted for in the tight customer feedback loop, the three generation approach to the Model 3, the scaling factor of the business to remove financial compromises and hiring a bunch of people most of whom are at the top of their game in their respective fields – most likely all of them capable of giving you intellectual vertigo on any subject even remotely related to engineering.
Lastly – Delivering the first several thousand Model 3s to in-house customers (Tesla and SpaceX employees) is a masterwork of QC / QA genius. Like a pool of the most highly qualified engineers get to give customer feedback under NDA. Beat that!
It was actually a really good thing to go looking for trouble with the Model X (Gen 2.1). That was a supplier and organizational stress test.
F* the NDA.
“More worrying, Tesla seems to have forgotten some quality lessons between Model S and Model X.”
That would be definitely bad. Can you give examples?
I don’t own an X or have any firsthand knowledge. But there are many written reports on the Tesla Motors forum and Tesla Motors Club forums, with photos. Just the widely reported door problems – not just the FWD but the “valet” front door problems – are kind of astonishing. How, especially after the Model S door handle problems, do you ship a $130,000 car with doors that don’t work properly? I take Elon Musk at his word when he says “Hubris”. I’ll accept that. And a long-delayed car that simply had to be shipped before other bad things happened.
Plenty of reports about all kinds of things, I don’t intend to hash them all out here, not really my point. My point is that Quality is Job 1. Not a radical redesign of the manufacturing process so they can get cars through the factory at faster than a walking pace.
Quality is not necessarily achieved by slowing down and adding more complex attention to detail.
Speed and quality go hand in hand with consistency of compionent yield – getting it right first time every time and also if you take the computer on a chip analogy – multifunctional component aggregation.
Can talk ’till the cows come home – bottom line Musk is absolutely right with zero compromises.
I hope so too, but if they don’t I will not lose sleep. 300,000 will be a small percentage of global sales.
I also feel confident they will detect and fix any genuine flaws before producing all the cars reserved by employees. That was a master stroke right there, putting the staff ‘first’.
That triangle is flawed. It is the invisible wall that Elon is talking about. Mind trap, mantra, thinking inside the box – call it what you will. Genuine innovation shrinks the entire triangle.
While “T:Q:C” may sound true, it only hold if you don’t consider input of intelligence. Lets take a simple example, get the lug nuts tight on a wheel.
Guy doing it be hand. Has to remember to do star pattern, and get tired as the day goes by. So quality/speed varies over the day.
— add smarts —
Give him a torque wrench, can now see instead of guessing by feel. Quality and likely speed go up.
— Add smarts —
Powered torque, with 3 setting for 3 pass star pattern. Faster/better
— Add smarts —
Power smart wrench with 5 heads. Quality speed goes up.
–
Yes much simpler than a whole factory, but you have to be willing to look for changing the way things are done. And most people can’t see a basic concept shift until after it happens and can see someone else doing it. Ask for how to improve and many only can think of little tweaks. But the work changes with those reaching much farther. It is why it is so important to force (if needed) you engineer to get outside their office. Engineers/Designers need to wall the factor floor and also live with their products. You need to reward those people that come up with the changing idea, no matter who come up with the idea Bellboy, engineer, factory floor sweeper, CEO, who ever.
Tesla’s reliability problems have been, almost exclusively, *design* problems. Since Model S #1000 came off the line, very few of the problems have been manufacturing defects. Almost all of them have been design errors.
Just so you know.
Hopefully they’ve learned from their mistakes with Model S and Model X and will avoid the design errors of the past (ranging from the wrong sort of metal for the lug nuts to a missing coating on the interior metal for the side mirrors to an off-by-less-than-a-millimeter shim in the transmission to a bad spec for the 12v battery…)
Can you point me to data?
I like having facts at hand.
Oh, god, I can’t even list all of the design errors off the top of my head. I’ve mentioned a bunch of them before. I suppose I could go through my extensive paperwork from the Service Center visits, but I don’t feel like doing so.
Basically, I’ve had a lot of service. It’s *all* been TSBs where they replaced a part with a new, redesigned part. I’ve never seen a manufacturing defect.
Important point: Tesla owns a tool & die company.
How do you make those manufacturing machines, the ones you put in the factory? Well, Musk wants to design the “right” machines, and then manufacture those machines. How do you build those machines? You have to do it using a different set techniques; they can’t be mass produced because there is no market for mass production. What are these techniques? These are the tool & die techniques.
There’s another factor I hope doesn’t get overlooked: recyclability.
Just read that Apple’s Macbooks get harder to recycle with each new, thinner, model. Things glued together instead of screwed was one big example. Virtually impossible to separate different materials for recycling when junked.
Greater durability reduces impact more strongly than recyclability. Post-consumer recycling is economically marginal for most items, and sometimes environmentally marginal as well.
Both of these tend to be utterly dominated by energy use in normal lifespan.
A typical passenger car in the U.S. is driven about 200,000 miles in its working life, Average U.S. light duty vehicle fuel efficiency in 2014 was 21.4 mpg, so it would burn about 10000 gallons of gas (about 28 metric tons). Multiply by pi to get CO2, you get about 89 tons CO2. The gasoline (post combustion: CO2, H2O, and bad stuff) is really, really, not recyclable. I think it is worthwhile to work to improve recycling, but it is far, far down the list.
Good points. Should do same calculation for EV. Think will make recycling bigger factor. And thinking long term – you have to do something with junked cars. One obvious solution is transportation minimizing number of cars (seems we’re heading that way).
But with 7 + billion people, how does anything get us to truly sustainable? I mean, we’ve been “civilized” for 6 to 10 thousand years. We’ve had cars 100 years. 5000 years from now….
The calculation is a lot harder for EVs, because it is strongly dependent on where electricity comes from. The grid is greening most places, so EVs produce lower emissions over time.
I realize that I missed the second biggest waste stream from ICE cars, the waste created in extracting oil and making gasoline. Full life-cycle analysis is time-consuming and difficult. If you do recycle something, that does NOT mean that you don’t create waste in the process.
By avoiding Apple products you can just replace the parts and keep most of your investments over the decades. Buying the best computer products and only disposing of them when they actually go bad saves lots of money.
Used to work for desktops but not laptops. I doubt anyone now makes a laptop with anything replaceable except screen and OS. Macbooks do tend to last and last and keep taking OS upgrades.
Dell used to be a good choice. Their old top of the line M6800 allowed 4 disks and every option of CPU and RAM up to 32GB. Now however Dell’s top of the line is far from top of the line and mandates card reader or fingerprint scanners on you. (Who wants to carry a card to access your laptop or give away your biometrics to the web in the name of “security”?) Anyways you can get some good deals on their old top of the line on ebay. As you can see the m6800 was made to be fiddled with and modified by the end user. Not their new crap though.
Change drives the wheel of life 🙂
Having worked in a company that was a mix of consulting and software.I saw over and over where one group (say auto) saw walls where another group (aircraft did) and that you could cross pollinate ideas. It was even a new idea, just something that had not been done in the market segment, or was something that everyone there “knew” you could do. So just by bringing a different view point to the problem you could look like a hero.