Episode 1: "You Asked For It" Tour -- Day 1: Slate Factory Tour

[FlyHappy]

yeah, that was fantastic commentary by @kvermeer , for sure.
he/she could have stopped right at pointing out the previously unseen rocker panels, and I would have been impressed

 

[kvermeer]

I learn something new nearly every day on this forum.

Today it was that "Dual check safety" cartesian position checks do a good job of keeping robots from crashing into each other.

Who knew?!

It's really designed for keeping robots from getting out of the cell, the big ones could punch through that welded wire and barely notice it was there.

The fences on those enclosures aren't there to keep the robots from getting out, they're there to keep the people from getting in!

DCS tells the robot where that "in" is allowed to be. You can just repurpose it to do other useful things.

 

[19SL80s]

New video up!



As a play-by-play, I'm going to note some interesting stuff that I see as an industrial controls engineer which might not matter to you in your truck but I'm familiar with and you might not be:

The shot at 0:05 has a view under the rocker panels, which I hadn't seen before:

I know Slate has running boards in the Maker, but I'm wondering what it would take to make some custom ones.

At 0:06 (yes, moving slow, but they cut through scenes so fast) you can see the back of the headlamp assembly:

Wonder if there will be aftermarket halo lights available that fit the same enclosure. I also wonder what they're learning from the word "Chassis" scrawled on the whiteboard behind the hood latch that's so important they had to make a smiley face and write "SAVE" in the corner, and I want one of those crew neck sweatshirts on the merch store. I am surprised they're wearing crew necks in Warsaw, it's currently a heat index over 100 so that might be recycled B-roll.

I'm surprised that they've got a uniform for "CAMERA CREW" on the wall at the factory:

Whose cameras are those? Presumably the person holding the camera here is in blue, and we've seen 30 seconds of factory shots in the past, but are we going to see more in the future?

That robot moving the body-in-white is being driven by the programmer with the teach pendant at the bottom:

He's running it in test mode, holding the deadman and shift-forward button, not yet in full auto. It's still got the lifting eyes and even a sling installed after being dropped in, that paper taped to J5 is a calibration and mastering report from the factory...it's still being commissioned. It's a Fanuc M2000iA: https://www.fanucamerica.com/download/asset/41013?download which is an absolute beast of a robot, I've got hundreds of hours on those little M710iC and arc mate 120iD weld robots but the heavy lift ones are on another level.

On that note, there's a general-purpose welding robot, love those hollow-wrist units, and there's a whole assembly and spot welding cell:

Hopefully the thing is foreshortened quite a bit, because keeping those all coordinated is a nightmare. Fanuc DCS or "Dual check safety" cartesian position checks do a good job of keeping robots from crashing into each other, but it's the programmer's job to ensure that they take turns and never have a reason to make use of those checks. I've not used BizLink cable management before, I'll have to have them quote for our next cell. That shot shows them running a "dry cycle", moving the robots through their paths without any product in place and just pretending that sensors made or weld commands were completed. Can't quite read the label on those contractor vests to see who their Fanuc ASI is, I'd be curious to learn if they're doing those in-house or outsourcing some of that work.

I get why the host is wearing Meta glasses - they're probably great for filming, way less obtrusive than a GoPro on your hard hat - but I hate that little blinking white light by his right eye all the same:

When the camera defocused during that shot, the cabinets in the background showed that they're using Bosch-Rexroth resistance welding controllers for the spot welders:

(More at https://www.boschrexroth.com/en/us/c/resistance-welding/). It must be so nice to work in a green-field (brown-field, I guess) development environment where they're not forced to fix some ancient Entron weld controller with mercury vapor current amplifiers that they've had since the 60s!

The top-center thing on this giant EOAT (end-of-arm-tool) labeled "230125-HI6" or something is an SMC EX600 remote IO valve controller:

With that component, they only have to run that teal Ethernet cable, blue air hose, and yellow 24V power up the robot arm through the dark blue BizLink cable management system, and they can put basically as many sensors and valves as they want out on the tool. They've got a whole block of spares. The shiny silver rounded boxes are power clamps: https://www.smcusa.com/products/ckz2n50-steel-body-naams-power-clamp-cylinder-50-bore~161318. SMC is based in Japan but their North American headquarters are right down the road in Indianapolis. I've had to wait for weeks for products from Japan in the past, but if they're buying standard, stocked parts, it should take no time at all to get to Warsaw!

I'm curious what cable harness they're running through the seat back shown at 1:40:

That's not part of an occupant classification system or seatbelt reminder system which uses weight/pressure sensors in the seat base. It's not power lumbar adjust, and I wouldn't expect the buckle switch to go there, perhaps it's for seat-mounted side/thorax airbags?

I'd love it if someone in Scottsdale or wherever the next events are could stick a camera on a selfie stick up near the static truck's DeDion tube and grab a close-up of that motor label:

JJE SD180 compatible?

Love to see the roll cage and fastback/open-air trim going together:

I know that's been giving them trouble.

Looks like there might be quite a bit of space under the front seats:

which is a nice spot to tuck seldom-used stuff... or to misplace often-used stuff!

it's fascinating that they're just laying some concrete for a test track as they drive the vehicles off the line.

I've been involved in some projects with 4-poster end-of-line shake test machines, we shake stuff on the line in a quiet room where you can ID the squeak. But just laying some concrete and painting some yellow lines is a lot easier! I wonder if they'll have microphones/accelerometers/dataloggers on the trucks as they drive that track, operator subjective results are notoriously inconsistent.

That's been my third-favorite video from Slate so far, just behind the Munro engineering philosophy review vids. Awesome stuff, I'll keep the playlist bookmarked!

Wow! Thanks for the rundown. You really know your stuff. I used to work for Honda R&D as an IT guy, so I never got to see the factory floor. There were some cool test vehicles though and we all got to check out the new NSX before that released in 2016. I’m more of an affordable Honda Fit kind of guy and drove my Gen 2 Sport for 16 years and 276,000 miles. That thing was like a truck with 57.3 cubic feet of space with the rear seats folded down.

I know a lot can change in 90 days, but in your opinion do you think they’re ready to start deliveries in the next 3-4 months? Seems like a lot of testing validation is still going on. I can’t wait to take delivery in the Summer of 2027.

 

[KevinRS]

It sounds like they intend to test and revise right up to the end. If they had everything locked in, it would be time to announce that they were taking orders now for October deliveries. So they have a couple of months left for testing and revising before they really have to lock in final revisions.

 

[kvermeer]

Wow! Thanks for the rundown. You really know your stuff. I used to work for Honda R&D as an IT guy, so I never got to see the factory floor. There were some cool test vehicles though and we all got to check out the new NSX before that released in 2016. I’m more of an affordable Honda Fit kind of guy and drove my Gen 2 Sport for 16 years and 276,000 miles. That thing was like a truck with 57.3 cubic feet of space with the rear seats folded down.

I know a lot can change in 90 days, but in your opinion do you think they’re ready to start deliveries in the next 3-4 months? Seems like a lot of testing validation is still going on. I can’t wait to take delivery in the Summer of 2027.

Start deliveries? Sure. Build 450 trucks per day? No, the ramp will be slow. The downtime will be high for the first few dozen/few hundred/few thousand cars off the line where they find and fix problems.

Re:build has talked a lot about the fusion of manufacturing and design, the design drives the manufacturing and the manufacturing drives the design, so you have to finalize both at the same time. I think they've gone from 0 to 95% done incredibly fast, but that last 5% won't be totally done with the line running at full rate until ~Q3 2027. From what they've shown publicly they're well positioned to hit their targets, I'm unsurprised by the dry cycles they're showing on video at this stage.

I will say that if the issue was that any single one of those weld/assembly robot cells had a lead time of more than 90 days, they could call me or any one of a thousand other integrators and we could push that one to completion in less than 3 months if we had final production CAD and fully representative test parts. It would be tough to get a thousand integrators to all build to the same cohesive automation equipment spec, but those cells all appear to be in really good shape with top-notch engineering.

If there's anything that will hold them up, it will be unforseen design changes driven by last-minute issues with results from crash tests or integration tests. A change that cascades through the design can really lead to a ton of work if it means you need to update every cell after that part, but they did a ton of simulation and early design work and are programming by CAD-to-path so there shouldn't be too many unknown unknowns left to discover.

 

[Rocks]

The fact that they're willing to take a DV truck on a road trip is either confidence or arrogance. Looks like we'll be along for the ride either way.

 

[19SL80s]

Start deliveries? Sure. Build 450 trucks per day? No, the ramp will be slow. The downtime will be high for the first few dozen/few hundred/few thousand cars off the line where they find and fix problems.

Re:build has talked a lot about the fusion of manufacturing and design, the design drives the manufacturing and the manufacturing drives the design, so you have to finalize both at the same time. I think they've gone from 0 to 95% done incredibly fast, but that last 5% won't be totally done with the line running at full rate until ~Q3 2027. From what they've shown publicly they're well positioned to hit their targets, I'm unsurprised by the dry cycles they're showing on video at this stage.

I will say that if the issue was that any single one of those weld/assembly robot cells had a lead time of more than 90 days, they could call me or any one of a thousand other integrators and we could push that one to completion in less than 3 months if we had final production CAD and fully representative test parts. It would be tough to get a thousand integrators to all build to the same cohesive automation equipment spec, but those cells all appear to be in really good shape with top-notch engineering.

If there's anything that will hold them up, it will be unforseen design changes driven by last-minute issues with results from crash tests or integration tests. A change that cascades through the design can really lead to a ton of work if it means you need to update every cell after that part, but they did a ton of simulation and early design work and are programming by CAD-to-path so there shouldn't be too many unknown unknowns left to discover.

Cool. I figured the full ramp up would take 18 months or so. I also didn’t realize many of the team members are former Munro & Associates employees. They seem to know what they’re doing more than most of these startups. It really is a unique operation and a unique product. I’m here for it.

 

[RumbleRed]

I was reading the comments on the video, and I am seeing a real shift. There used to be a lot of hate in those comments. All of the typical complaints. But the comments on this video were surprisingly positive. I think seeing how the donuts are made has really shifted perception.

 

[EVfun]

I just wanted to thank James for the video link and Kevin for the analysis. 👍

 

[tubes]

If there's anything that will hold them up, it will be unforseen design changes driven by last-minute issues with results from crash tests or integration tests. A change that cascades through the design can really lead to a ton of work if it means you need to update every cell after that part, but they did a ton of simulation and early design work and are programming by CAD-to-path so there shouldn't be too many unknown unknowns left to discover.

Exactly!

I wasn't in mechanical design. I supported electronics design with test software. Our "failed crash test" was a problem with routing in a deep layer on the printed circuit board (PCB). Right before I retired, my test software was finding about a 1% failure fallout at this late DVT phase. They tried to blame my software. It was a bad way to retire with a cloud over my head. A few months after I was done, I ran into a co-worker and he said the simulation had a bad parameter and the routing was wrong and would cause "cross-talk" effects. I wish I heard it from the boss, but I was gratified I didn't leave with a bug in my code. And indeed, it set the project back 3 months.

 

[cadblu]

Q. How may assembly techs does it take to install the Open Air kit?

And we were led to believe this is a DIY in your driveway?

 

[cadblu]

They tried to blame my software. It was a bad way to retire with a cloud over my head. A few months after I was done, I ran into a co-worker and he said the simulation had a bad parameter and the routing was wrong and would cause "cross-talk" effects. I wish I heard it from the boss, but I was gratified I didn't leave with a bug in my code. And indeed, it set the project back 3 months.

Man, I feel your pain. Working through production hell (it seemed like every day) I tried to keep the management team apprised of all the issues, they only half-listened and just expected me to find a way to get it done. Generally, when things go well, management takes all the credit. But when it kind of falls apart, all the finger pointing leads to the Subject Matter Expert.

 

[VigilKnotEagle14]

I think they've gone from 0 to 95% done incredibly fast, but that last 5% won't be totally done with the line running at full rate until ~Q3 2027.

This pattern doesn't sound too dissimilar to the charging of an EV 😉

Quick to 80%, but that final 20%....

 

[Rocks]

Did anyone notice they had a truck baking under the lights in the last building visited?

UV testing on the panels, maybe?

 

[bartflossom]

I learn something new nearly every day on this forum.

Today it was that "Dual check safety" cartesian position checks do a good job of keeping robots from crashing into each other.

Who knew?!