navguy12
Active Member
Electric motors 101 in the 09 Dec 2023 Toronto Star
-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
Investor Engineering Discussions
- Thread starter MC3OZ
- Start date
I'm nerding out here, may have some errors, not meant to be argumentative...
Accuracy is a bigger deal unless dealing with really low precision. Accuracy needs to be better than 10*precision for that precision to be useful for collision avoidance. Excess precision is useful for trend monitoring (closer/ further).
If the issue is side clearance when turning there are no mid car latteral sensors. The mirrors give a reference for width and when extended (or using fender cams) show side gap.
USS are more handy for distance to bumper when maneuvering to allow the widest (or tightest) turn.
Generalized for passage:
Gap = (path width - vehicle width)/2
Generalized for bumper
Gap = Required closest approach
With perfect accuracy: If the per side gap is less than 1 units, you need tenths level of precision. If the gap is less than 0.1 units, you need hundredths level. One additional level is very helpful to establish the trend.
With infinite precision: If per side gap (required minimum) is 1 unit, you need better than 1 unit accuracy (uncertainty better than +/- 1 unit).
Required decimal places of precision = floor(log(1/(gap ))
1 = 0.x
0 = x.0
-1 = x0.0
Maximium allowable uncertainty = (gap) (show > 0 when centered)
Combined performance is insufficent to pass if
(gap - accuracy) < 10^(-precision)
One issue I haven't seen discussed is whether or not the USS array has some "blind spots" where detection of a nearby object in some locations isn't guaranteed.
This is more likely at an approximate 45% angle to the front of the car, rather than directly in front. (i.e. It can be an issue backing out of a parking spot and turning the wheel to eventuality travel a 90% to the direction of parking. The arc of the car may pass though a "blind spot".)
If the speculation (IMO some evidence) for these USS "blind spots" is true, then it is also true that the problem is long standing and Tesla could not easily fix it by adding more USS.
Eventually, vision may work with USS to improve the solution even for cars with USS.
They are not using vision on cars with USS for now, but that could be because they want to merge the information streams. I don't know how important the front facing camera is, but for cars with USS it probably isn't needed.
So the most interesting test case is cars without USS or a front camera.
Knightshade
Well-Known Member
It's known USS can't see things at certain specific angles or distances from itself yes.
....huh?
How would adding more USS, such that the added ones cover the blind spots of the existing ones, not be a fix?
I'm not suggesting it's the idea solution, but also not seeing how it couldn't be a fix.
Using vision and "memory" of things that vanish from USS but are still seen by cameras (only really useful for the rear though given the front blind spots below the hoodline) would be a nice augment though.
That's what USS cars already have. See below.
AFAIK vision is absolutely used on USS cars for parking--- in fact that's the only way it's able to park in spots without a car on either side. The OLD parking code was USS only and you'd have to move very very slowly along a lane and only if you passed an empty spot where USS saw a gap between 2 cars would it offer to autopark.
Since moving to vision I can now have it autopark in a lines-on-the-ground spot and no other cars around.
it still uses USS for distance calculations though-- because it's objectively more accurate so far to do so.
Vision based parking (with USS still active) began rolling out ~August 2021, first to refresh S/X, later to 3/Y (that still had USS).
First look at Tesla's new vision-based Autopark feature
Tesla has started updating its Autopark feature with its new Tesla Vision computer vision system, which now powers Autopilot and...
electrek.co
So ideal world we USS owners get the cool new visualizations- but they continue using the more-accurate USS distance measurements whenever available.
It would be a fix, but Tesla has chosen not to do it...
To me this suggests that a USS solution with "full coverage" is significantly more expensive and not an easy upgrade to do.
I'm a suggesting that the USS limitations, and the cost of overcoming those limitations, was a consideration.
TL;DR cells bought from Munro teardown, nothing new - thick cell casing, no Si in anode, dry coating with teflon of anode. Cathode is called as NMC811 although molar ratio seem 12:1:2... Electrode thickness is quite high so slow charging as expected, I didn't see Cu/Al thickness mentioned. 1rst / last 40 cm don't have connection to tabs.
Great work as usual from Limiting Factor...
The only relevant things he didn't consider is costs and supply chain for PDO.
Tesla might be prepared to trade a small amount of longevity for lower cost and more certain supply.
I see the relevant factors as ;- Range, Longevity, Cost, Raw Materials Supply.
Tesla is thinking of moving to high nickel anyway, if high nickel works well with PDO and PDO is lower cost that is a big win.
Tesla, TSLA & the Investment World: the Perpetual Investors' Roundtable
Wow, are you barking up the wrong tree. Tesla never said 3/Y deliveries would increase at 50% per year, that is overall for the company. Model 2 / Gen3 will be available for sale and starting at the bottom of their sales 'S' curve before the 3/Y plateaus. Tesla's guidance was for the decade, not...
teslamotorsclub.com
The rumoured chemistry at Austin to NMC 955 and later NMC 973 seem compatible the the aims outlined and battery day and also the type of formulation that might be compatible with PDO.
IMO Asymmetric Lamination might be making a thicker graphite Anode to match the improved energy density on the Cathode side and as a means of avoiding silicon for now.
I think asymmetry refers to the two sides of one electrode, not thickness of cathode vs anode.
You are right at least that is what Joe posted.
If it is both sides on the one electrode apart from packing a little bit more into the can, I can't think of a major advantage.
Might have something to do with surface deformation due to flexure. The inner surface must squish whereas the outer surface gets stretched (assuming foil doesn't deform). Too thick may cause ripples resulting in increased distance to separator and other electrode. Smoothing that interface would then allow more layers boosting cell energy (more foil but less voids). This also reduces ion path possibly increasing power handling and maximizes electrolyte utilization while minimizing volume needed per cell...
... maybe
So it could just be mainly improving yield and reliability, with only minor energy density improvements.
However, those improvements might still unlock a thicker graphite anode.
I just have a hunch Tesla will not be doing "Panasonic-style" doping of the graphite anode with silicon any time soon.
As far as we know there is no silicon in V1 or V2, 4680s, there are chemistry changes for V3 so silicon might be possible, but I would be surprised if it is included.
@MP3Mike, pulled the discussion over here.
They can't just 48V the steering, it's a full vehicle update, not just wiring. Yes, they could make gateway nodes that source the legacy power and comms to existing end nodes, but that negates the carry over idea.
Cybertruck has large firewall mounted pass through modules, S/X likely don't have the packaging space. Similarly, S/X don't have room for a rear steering actuator.
@TSLA Pilot
If the modules were appropriate. As @mongo said, there likely isn't room in the firewall for the Cybertruck ECUs.
What 12 volt system in the Cybertruck? I haven't seen anything about a 12 volt system. (They do use 24v amplifiers because 48v ones weren't available yet.)
The confusion may be that in the interview with Sandy, Elon mentioned the seat adjustment as a thing that had to be change, not that it didn't change.
They might not need to do rear wheel steering...
Yes moving Model S/X to 48 would involve some body shop changes, new stampings (or perhaps castings, etc), but overall I think most of the existing body and the existing stampings can be retrained.
I am saying it is possible, not that it is quick or easy..
Lots of effort but lots of reward, it is the logical upgrade to do for Model S/X in particular weight and costs savings will be significant.
Ultimately only the stampings around the firewall and casting (and possibly removing some stampings entirely if the casting gets larger and subsumes some of them) would need to change, probably, as far as those things go. Small chance they revamp some other stampings to change cable or bracket retention but seems unnecessary, just change what attaches.
Most anything else not electrical (i.e. interior surfaces, plastic mouldings, etc) would be unchanged unless they decided to also make cosmetic changes. Going to 48v changes the guts of electronics but doesn't necessitate cosmetic ones.
Probably everywhere outside the firewall can simply have updated guts and identical mounting / enclosures, ignoring the wiring itself, which only gets smaller not bigger.
Still enough work I'd be doubtful they do it before Gen 3 engineering is finalized and started production ramp, but doesn't necessarily need to wait until ramp is done.
I expect we see highland at Fremont and juniper everywhere before s/x 48v too, but I'm less sure where those will line up time wise versus Gen 3. Might be there's a highland 48v variant first if supply chain can ramp faster than CT, in which case juniper would too (before s/x, not necessarily launching as 48v, but one can hope)
Buckminster
Well-Known Member
Similar threads
- Marketplace listing
