-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
Model 3 Motors on the Tesla Parts Catalog
- Thread starter EVMike3
- Start date
AlanSubie4Life
Efficiency Obsessed Member
As mentioned above by @Knightshade, the 3P Stealth is not lowered, but has better efficiency.
Also see EPA datafile showing that highway efficiency of the 3P+ is actually the worst of all of the Performance vehicles, even though its performance relative to the 19" wheels in the city is similar (suggesting that rolling resistance factors are similar between the two).
I'm not sure how much to really read into these datafiles, to be honest, but if you assume there is no vehicle-to-vehicle variation I think it's reasonable to think that the lower ride height and spoiler and big wheels actually make things worse on the highway. It's a bit surprising that there is much difference but that's the suggestion.
AlanSubie4Life
Efficiency Obsessed Member
Yeah, if there were a max power output for the 990 motor (nominally at least it is lower power output), it would be a bit tight, with the power output required for the SR+ with current performance. I went through the calculations a while back, and as I recall it would have left very little room for any sort of future power boost. So, 980 motor it is.
pdx_m3s
Active Member
Can we get a summary of the different motors? There’s so much detail in this thread I’m already mixed up on the difference between 990, 989, etc! I have a May 2019 SR+ by the way.
Edit: from what I can gather, the 980 motor is used on RWD and P configurations and has a higher output compared to the 990 motor used on the AWD configurations. There are various revisions of each with no identifiable difference.
Edit: from what I can gather, the 980 motor is used on RWD and P configurations and has a higher output compared to the 990 motor used on the AWD configurations. There are various revisions of each with no identifiable difference.
Last edited:
Knightshade
Well-Known Member
Edit is right.
990 is used (based on currently known info) only in the LR AWD non-P, and only since sometime early 2019 (with some 980s showing up as late as I think April or June maybe?)
980 is used in the P and all RWD versions of the car.
Based on this info we can conclude:
the 990 is cheaper, less efficient, and less capable of the max power output when compared to the 980.
Because if it wasn't cheaper- there'd be no reason for it to be used in anything- they certainly wouldn't use a same/higher cost motor especially a less efficient one.
And If it was cheaper but equally or MORE capable, they'd use it in everything.
Has it been ruled out that the 990 (MOSFET-LC) is not just a binned lower capacity version of the 980 (MOSFET) drive unit?
pl8dlikafiddle
Member
It is right behind the driver side rear wheel.
If you have the 18” aero wheels, look through the spokes and you’ll see it. I put my phone between the back of the wheel and the car (arm around the front of the wheel/tire), and took photos with the flash on.
If that were the case, costs would be identical. No one knows unless a full tear down takes place between both motors.
I’ve been in manufacturing where lower performing parts were given a different part number but cost the same to produce.
Right, and I've designed with binned parts.
My point is that binning would allow Tesla to have higher output motors without needing to pay more for binned power devices. They pay for the standard distribution of power devices and then cherry pick the better ones to make higher performance drive units.
Basically, pushing back on the cheaper statement:
The reason would be: Making only 980s would drive up the cost due to needing pre-sorted parts from the supplier. As long as the non-binned semiconductor performance distribution curve supports the needed 980 990 split, the overall cost of both stays lower.
Hypothesis:
The average motor cost is lower because Tesla can use higher and lower power versions, thus the reason to have it. However, the 990 itself is not cheaper.
Alternatively, the parts vendor is presorting and the 990 has a cheaper BOM. However, that requires running inverters and drive units as two builds and part numbers (parts catalog only shows onenl inverter #, but that doesn't mean much)
Restating with different words:
Making one motor for everything is cheaper than making two designs - one set of parts, one manufacturing line, lower unit price with higher volume.
Making two different motors costs the most - different parts on different lines, and both are manufactured in lower quantities.
Binning splits the difference - you get the lower cost of only manufacturing one part, but based on tests/capability, you split them into two different model numbers - higher-performance, lower-performance. It costs slightly more than a single part (because of the cost of doing the tests and managing two parts in inventory), but not as much more as manufacturing two different motors.
If the "capability" testing has strong overlap with QA testing, then the additional cost of binning can be quite low.
Also, with a binned design, both motor versions may be compatible with any model 3 - so if replacing in the future, you have the option of stepping up or stepping down. Or, 20 years from now when stock is limited, you can use whichever part is available!
Making one motor for everything is cheaper than making two designs - one set of parts, one manufacturing line, lower unit price with higher volume.
Making two different motors costs the most - different parts on different lines, and both are manufactured in lower quantities.
Binning splits the difference - you get the lower cost of only manufacturing one part, but based on tests/capability, you split them into two different model numbers - higher-performance, lower-performance. It costs slightly more than a single part (because of the cost of doing the tests and managing two parts in inventory), but not as much more as manufacturing two different motors.
If the "capability" testing has strong overlap with QA testing, then the additional cost of binning can be quite low.
Also, with a binned design, both motor versions may be compatible with any model 3 - so if replacing in the future, you have the option of stepping up or stepping down. Or, 20 years from now when stock is limited, you can use whichever part is available!
Knightshade
Well-Known Member
But we know that's wrong.
Because the 990 is less efficient than the 980, all else being equal but the motor.
If they're the same motor, but "binned" for the capable-of-more-power ones being stamped 980, that would not be the case.
Further, if this was NOT a physically different motor, just the result of the magical no-evidence-for-it-ever-happening-ever "binning"- why did it take them like a year and half from first production to start actually doing it?
Because the 990 is less efficient than the 980, all else being equal but the motor.
If they're the same motor, but "binned" for the capable-of-more-power ones being stamped 980, that would not be the case.
Further, if this was NOT a physically different motor, just the result of the magical no-evidence-for-it-ever-happening-ever "binning"- why did it take them like a year and half from first production to start actually doing it?
Efficiency and power handling go hand in hand.
The drive unit is comprised of both the motor and the inverter. The inverter contains MOSFETs that have variation in terms of resistance and gate charge/ capacitance. Some will be lower resistance and thus both more efficient and capable of higher power at the same heat dissipation level.
An inverter with 21% worse MOSFET resistance would be built exactly the same, but the peak current would be 10% less for the same power dissipation and overall efficiency of the inverter stage would be 10% less.
How do we know that binning is wrong?
A delay in implementing binning does not mean they do not bin. Nor does the delay in differentiated part numbering prove they were not tracking sone other way (what were the blank motors?)
Only having one inverter part number indicates they may bin.
The part number being laser etched into the case indicates they may bin.
Elon tweeting that they bin indicates they may bin.
Is there any hard data like a tear down showing they use less MOSFETs in the 990 to indicate they are not binning?
Another option is pre sorted MOSFETs and they track assemblies by PCB barcodes. Still identical designs, just sorted at a different stage.
Knightshade
Well-Known Member
The fact ordering a replacement 980 doesn't require a VIN tells us that.
Got a P with a "magic super special" 980? Got an AWD with a "regular" 980? Tesla doesn't care.
Everybody gets the same 980 replacement. They don't even want to know what you had originally. And that was true before the 990 was even in the catalog.
If there really were super special 980s that'd not be the case.
Huh? Isn't 980 the better motor of the two? As such, of course you would replace a 980 with a 980.
There is no better 980 to replace. The new 980 might be better than the old 980, but it is not going to be worse.
The 990 is (potentially) a lower efficiency 980 and you would not replace a 980 with a 990.
pdx_m3s
Active Member
Thanks. I’ll have my wheels off to do springs and spacers in a couple months and will get a shot of the motor. I’m guessing my May-19 SR+ will have 980 rev F. We’ll see.
AlanSubie4Life
Efficiency Obsessed Member
We really need @Ingineer to repeat his tear down, this time with a 990 motor! Must be possible to get a junked one to pull apart at this point...
I’m sticking with my old theory that they just use three devices instead of 4 in each of the six groups used to drive the three bus bars of the motor. As far as I can tell this is possible. For a while I thought those 4 devices in each group might not actually be in parallel (which would make this a silly suggestion), but I think effectively they are...
I only say this because it explains the 840/630 references that have shown up periodically as well.
It would also explain a clear efficiency difference due to higher switch resistance for a given output power, etc. I have a hard time explaining such a marked efficiency difference (which, keep in mind, shows up when the car (and the FETs) are operating FAR below peak power, due to the very leisurely test conditions).
Last edited:
Knightshade
Well-Known Member
Yes. A 980. Not a special magic binned one.
This disproves the idea there was any "binning" before hand.
A 980 was a 980.
There wasn't some magical "special" 980 the P got and the rest got the rejects...which was the theory- that you yourself just suggested again- where there was some secret hidden "other" way they tracked which the "special" 980s were to insure that went into a P.
Because if there were then ordering a replacement would require a VIN, for them to see if you need the "special" P version or not.
And that was never true- even when the 980 was the only PN orderable for a rear motor.
(it would also make the idea the 990 is a binned 980 even dumber because it would mean Tesla had a special secret totally hidden binning method using JUST one PN that somehow worked fine... and then after 18 months of production decided to abandon that and use a second PN for the same process for no apparent reason)
If all service parts were the higher performing motor then there would no need for a VIN or additional tracking outside of Fremont (beyond checking on upgrading D's to P's at the delivery center.
Anyway, I was just wondering if binning had been fully ruled out. No need to debate the available data.
pdx_m3s
Active Member
My guess is that they binned motors early on in the older builds because it was necessary (consistent quality want there yet). Now that they have built so many of these motors and are on newer revisions, the need for binning is not necessary, as all 980s can now support the demands of the P configuration. They can replace any older 980 (binned/P or not) with a newer one knowing that it will take the power.
