100% drive unit failure rate??

[ohmman]

My SvC originally wanted to do an on-site repair of my DU for the highway hum noise, but after listening to the DU we also heard some electrical buzzing noise and that's when they opted to replace rather than repair. Sounds like they could have repaired the highway hum, but the buzzing was an inverter issue that prompted the full replacement.

Mine was the bearing/milling noise we've been discussing here. Hard for me to figure out how that is something they want to outsource to numerous SvCs but they have better insight into the problem than I do.

 

[Max*]

That's not the point. The point is that Tesla revises and re-engineers components, and when they do, they assign new revision number. A revision is a revision. What you are attempting to say doesn't make much sense to me, sorry.

You could make whatever point you want, battery packs are not on Rev Q (thank you to whomever corrected me upthread). They're on what, F?

DU are on Q. They're patching a problem, might be time to spend the cash upfront (maybe the are, like I said in my post, I hope I'm wrong) and redesign the damn thing so that it stops failing.

 

[AmpedRealtor]

You could make whatever point you want, battery packs are not on Rev Q (thank you to whomever corrected me upthread). They're on what, F?

DU are on Q. They're patching a problem, might be time to spend the cash upfront (maybe the are, like I said in my post, I hope I'm wrong) and redesign the damn thing so that it stops failing.

You have no idea what Tesla is doing behind the scenes, we have multiple revision letters indicating that Tesla is actively revising the drive units, and yet you keep insisting that they need to redesign something that they are already actively modifying. Perhaps you should stop proffering a point of view and making a demand that has no basis in fact, and which quite obviously contradicts the facts.

 

[wdolson]

I never said they were sitting on their hands. From my observations, since we're still on V1 of the driveunit with Rev P, they're patching problems. The engineered solution at this point (well, actually a while back), might be to start from scratch and have a DU V2 Rev A.

Pushing revisions is usually cheaper short term than starting from scratch. But at some point, when you're on Rev P (and sure, not all the rev's are due to problems, they're probably doing simulatenous improvements) and people are still having problems, it might be time...

I still hope I'm wrong, but that's what it looks like from the outside.

Actually Tesla is on version 17 of the driver units. They just use letters instead of numbers.

If you want to see more problems, continue insisting on a redesign. The bigger the changes, the more room there is for new problems. Re-starting from scratch has the advantage of lessons learned, but there are all sorts of bugs that can creep in that weren't in the original design. They are approaching a true fix for the drive unit problem now, starting over from scratch would create more problems than it solved unless the root of the problem was a core design flaw, which is rare and does not sound like it is the case here.

It sounds to me like they have had some trouble isolating the cause of the problem, but they have zeroed in on micro-pitting of the bearings fairly recently. The last couple of revisions are probably different attempts to mitigate the problem. This is a difficult problem to diagnose, it only happens with wear to the motors, so any change they make will take a while to be sure if it worked or not, though they are probably out thrashing some mules to test the changes. Additionally, if it is due to eddy currents flowing through the bearings, eddy currents are notoriously difficult to measure. If you know specific spots where they are flowing, you can put measuring instruments in those spots, but to see how they got to those spots can be difficult to measure.

Electrical current is like water flowing through metal. It prefers a good channel, like a copper wire, but it will flow over flat land too. Water you can see where it is and where it isn't, and it's fairly easy to measure it where it isn't obvious to the eye, but current can't be seen and can only be detected with instruments. One way is to measure the electric field generated by the current flowing through metal. This is not difficult in something like a wire where you know the channel for the current. In something like a metal chassis, especially one with complex curves like the drive unit, it's very tough to isolate how much current is where.

 

[JOEV1]

I take it you mean the gear design change for the klunk problem and not the wirr problem?

Having read most of the threads, the only comment I can make is that the DU has an extremely high Power Density.
Refering to the Power to Weight Ratio one must admid that hardly is there another system which handles so much power:
An electric motor which ist virtually the size of a football delivering 450 hp!!! Amazing.. really!
To me it is not really surprizing that some parts fail due to fatique, over-load or under-design.
Luckily, the DU has very few moving parts. I am sure TESLA will get on top of the issue.

Dual Motors will logically improve the situation as the Load (a 2-Ton Vehicle) is distributed to 2 DU´s

 

[Avenger]

Drive units are made by Fukuta - not Tesla

clearly the milling noise isn't normal and that's why tesla replaces the DU. I'm sure if tesla is asked on earnings call, they'll give an update on it.

Since they've done many revisions of the DU I gotta believe that the reliability overall per revision is generally improving.

Lastly, I'm confident tesla will ultimately get the DU to be ultra reliable. I believe Elon is telling the truth when he said they are moving towards a DU that lasts 1 million miles.

Fukuta is the manufacturer of the AC Induction motor Elon Musk and Tesla Motors secret motor supplier. | Bridging Money, Finance, and Technology

 

[Max*]

You have no idea what Tesla is doing behind the scenes, we have multiple revision letters indicating that Tesla is actively revising the drive units, and yet you keep insisting that they need to redesign something that they are already actively modifying. Perhaps you should stop proffering a point of view and making a demand that has no basis in fact, and which quite obviously contradicts the facts.

We have the same facts, lol. You can't argue that my opinion is wrong based on the same facts that you're basing your opinion on.

Actually Tesla is on version 17 of the driver units. They just use letters instead of numbers.

If you want to see more problems, continue insisting on a redesign. The bigger the changes, the more room there is for new problems. Re-starting from scratch has the advantage of lessons learned, but there are all sorts of bugs that can creep in that weren't in the original design. They are approaching a true fix for the drive unit problem now, starting over from scratch would create more problems than it solved unless the root of the problem was a core design flaw, which is rare and does not sound like it is the case here.

It sounds to me like they have had some trouble isolating the cause of the problem, but they have zeroed in on micro-pitting of the bearings fairly recently. The last couple of revisions are probably different attempts to mitigate the problem. This is a difficult problem to diagnose, it only happens with wear to the motors, so any change they make will take a while to be sure if it worked or not, though they are probably out thrashing some mules to test the changes. Additionally, if it is due to eddy currents flowing through the bearings, eddy currents are notoriously difficult to measure. If you know specific spots where they are flowing, you can put measuring instruments in those spots, but to see how they got to those spots can be difficult to measure.

Electrical current is like water flowing through metal. It prefers a good channel, like a copper wire, but it will flow over flat land too. Water you can see where it is and where it isn't, and it's fairly easy to measure it where it isn't obvious to the eye, but current can't be seen and can only be detected with instruments. One way is to measure the electric field generated by the current flowing through metal. This is not difficult in something like a wire where you know the channel for the current. In something like a metal chassis, especially one with complex curves like the drive unit, it's very tough to isolate how much current is where.

I've heard we're close for a while now. I don't know the inner working of the DU, I understand everything you're saying though.

I'll believe it when I see it.

 

[scottm]

People are thinking the letter suffix is a revision, it could just be a build date indicator, or which batch... not necessarily a difference in anything physical inside.... just saying.

 

[apacheguy]

People are thinking the letter suffix is a revision, it could just be a build date indicator, or which batch... not necessarily a difference in anything physical inside.... just saying.

That theory doesn't seem to jive with the way Tesla catalogs other parts.

 

[cwerdna]

Re: patching or whatever Tesla's doing to mitigate/solve the DU noise problems, as I've wondered aloud in the last part of Drive Unit failure symptoms and thresholds for replacement - Page 5, what type of durability and reliability testing Tesla does and for how long? And, if whatever they do passes w/flying colors, why does it seem to not translate to the what the customer receives?

On this note, I started automotive reliability and durability testing, including a bunch of info that's been publicly released from other automakers. Please post anything interesting that's factual about that subject there.

 

[Cyberax]

Then what do you call it? It's not a catastrophic failure that leads you stranded, but it's a design failure. Otherwise it wouldn't need to be replaced in the first place.

And we're still getting catastrophic failures - like sudden propulsion loss on a highway, even with new cars.

- - - Updated - - -

Please, please post the tesla part number of the swapped DU. You have AP and AFAIK we don't know the part # for AP cars.

Here's my DU number - AP car, delivered June 2015. Drive unit failed catastrophically and without any prior warnings, 3 days after a regular service.

Parts Replaced or Added
Part Quantity

ASY,DRIVE,UNIT,COMPLETE,3.0-150,REAR 1 (1037000-00-F)

- - - Updated - - -

This is the point of disagreement.
The sound IS normal, however, Tesla doesn't want that sound their.

I don't think so. The sound is a symptom of an impending drive unit failure. Just like a knocking rod in a regular ICE car. You might get a fair amount of mileage from a failing unit, but eventually it will fail catastrophically.

It also looks like Tesla has poor diagnostics for this problem - the car can't sense that something is out of order until it's too late.

- - - Updated - - -

The problem is they have had their gear/mech guys on the team to fix the problem while the problem is not mech per se
1) either ground the shaft with a brush (which would make it a service item)

That's not right. Tesla uses an asynchronous motor, its shaft does not need to be 'grounded' and it won't help anyway.

2) use expensive non-magnetic ceramic bearings (which they did on the Roadster and have an other set of problems)

You mean non-conductive, not simply non-magnetic? It's unlikely to help.

Edit:

I read the linked PDF. It's quite interesting - turns out that the problem is not caused by magnetic fields, but rather by the way the motor power output is controlled.

Modern motors use PWM (Pulse-Width Modulation) to control the current - essentially switching voltage completely on and off several hundred thousand times a second. The ratio between 'off' and 'on' periods regulates the power output.

The problem is that electrically insulated rotor acts as a plate of an electric capacitor (with stator as the second plate). And if you have a capacitor connected to an AC circuit then there'll be a current flowing through a capacitor. With 'old' asynchronous motors that used 3-phase grid electricity at 60Hz to drive them the induced current was completely negligible, but with modern high-frequency PWM-controlled motors it becomes appreciable.

There are two ways to solve it: either insulate the rotor and stator completely, so that no breakdown can not occur, or provide a low-resistance path between rotor and stator. The aim of both approaches is to make sure that the breakdown doesn't happen in a tiny contact spot where the bearing balls contact the outer ring (through a thin film of grease).

 

[yobigd20]

My SvC originally wanted to do an on-site repair of my DU for the highway hum noise, but after listening to the DU we also heard some electrical buzzing noise and that's when they opted to replace rather than repair. Sounds like they could have repaired the highway hum, but the buzzing was an inverter issue that prompted the full replacement.

I am guessing that the specific failure which requires a shim as a fix is the one they do repair on-site. Perhaps that's the "drone noise" issue. The "milling noise", if it's bearing related, probably necessitates DU swap instead of a shim so they can't do that repair on site. My theory at least.

 

[kort677]

I am guessing that the specific failure which requires a shim as a fix is the one they do repair on-site. Perhaps that's the "drone noise" issue. The "milling noise", if it's bearing related, probably necessitates DU swap instead of a shim so they can't do that repair on site. My theory at least.

the shim fix is for a separate issue, I've had the shim fix, then a year later I got the dreaded cicada noise and had the drive unit repaired. I am getting a n 85d soon and hopefully the newer car won't exhibit these common problems

 

[wdolson]

And we're still getting catastrophic failures - like sudden propulsion loss on a highway, even with new cars.


I don't think so. The sound is a symptom of an impending drive unit failure. Just like a knocking rod in a regular ICE car. You might get a fair amount of mileage from a failing unit, but eventually it will fail catastrophically.

It also looks like Tesla has poor diagnostics for this problem - the car can't sense that something is out of order until it's too late.

The only early symptom is the noise. The car drives the same and the bearings are still doing their job. It would probably be pretty difficult to catch that problem with a sensor.

That's not right. Tesla uses an asynchronous motor, its shaft does not need to be 'grounded' and it won't help anyway.


You mean non-conductive, not simply non-magnetic? It's unlikely to help.

Edit:

I read the linked PDF. It's quite interesting - turns out that the problem is not caused by magnetic fields, but rather by the way the motor power output is controlled.

Modern motors use PWM (Pulse-Width Modulation) to control the current - essentially switching voltage completely on and off several hundred thousand times a second. The ratio between 'off' and 'on' periods regulates the power output.

The problem is that electrically insulated rotor acts as a plate of an electric capacitor (with stator as the second plate). And if you have a capacitor connected to an AC circuit then there'll be a current flowing through a capacitor. With 'old' asynchronous motors that used 3-phase grid electricity at 60Hz to drive them the induced current was completely negligible, but with modern high-frequency PWM-controlled motors it becomes appreciable.

There are two ways to solve it: either insulate the rotor and stator completely, so that no breakdown can not occur, or provide a low-resistance path between rotor and stator. The aim of both approaches is to make sure that the breakdown doesn't happen in a tiny contact spot where the bearing balls contact the outer ring (through a thin film of grease).

PWM motors are also very noisy electrically. Tesla must do some pretty good shielding to prevent it from getting into the passenger cabin, otherwise it would be impossible to listen to AM radio. One simple way to detect EM noise is to take a portable AM radio and you will find where the noise is worst by the radio getting very noisy.

Both inductors and capacitors produce inductance when you run AC through them. Most motors have pretty high inductance, so the capacitance might balance that a bit. However, with the EM noise generated by the PWM motor, you are getting a wide range of harmonic frequencies which makes the inductance equation nuts. I haven't done any Fourier Transforms in close to 30 years, but I remember the topic. I don't know if they drive the motor with a square wave or sine wave. Sine waves are pure frequencies when they are complete waves and not chopped like they do with a PWM motor, but at full power a sine wave would not be noisy at all.

The best solution is probably copper grounding straps around the critical areas to give the current a better path. It probably wouldn't channel all the eddy currents, but it would get most of it. Hard to say for sure, I don't know the motor or how the currents are flowing exactly.

 

[SW2Fiddler]

One simple way to detect EM noise is to take a portable AM radio and you will find where the noise is worst by the radio getting very noisy.

Slightly OT - You can also "see" it by using a portable analog TV, tuned to the lowest UHF frequency, calibrated by turning Brightness down to darkest grey when there is no noise.
(This is the old Redneck Tornado Detector trick - lightning shows as flashes, tornadoes create a "visual hum" that gets brighter ...and that's when you have to hit the cellar.)
Back to scheduled programming... :smile:

 

[vortexz]

I am wondering how a guy who is CEO for SpaceX, a company that designs rockets for NASA can't make a working, reliable DU for a car?how can you make a f**** spaceship that is supposed to go outer space reliable enough to go there without any problems and you cand make a freaking DU for a car?how?

 

[lolachampcar]

Vortexz,
You underestimate the difficulty of what it being done at Tesla. When these guys/gals mention high power density, they are not kidding. The Tesla drive unit is unlike anything ever developed for a car and it is no surprise that it will take some de-weeding.

Much was made of opening up Tesla's patents but that is not where the value resides. The real value is in the scars and associated solutions that come from working through these exact types of problems. We are just lucky enough on this forum to have people who have some insight to similar issues thus allowing us all to paint a mental image of what it going on.

 

[electracity]

Vortexz,
The Tesla drive unit is unlike anything ever developed for a car and it is no surprise that it will take some de-weeding.

Was that necessary? If I recall correctly, Tesla wanted to buy the motor, but ended up making their own.

 

[yobigd20]

I am wondering how a guy who is CEO for SpaceX, a company that designs rockets for NASA can't make a working, reliable DU for a car?how can you make a f**** spaceship that is supposed to go outer space reliable enough to go there without any problems and you cand make a freaking DU for a car?how?

I ran my original issue through an visual audio spectrum analyzer and posted it on YouTube. You can "see" the sound happening (and hear it) too. Let me find it ...

 

[Max*]

I am wondering how a guy who is CEO for SpaceX, a company that designs rockets for NASA can't make a working, reliable DU for a car?how can you make a f**** spaceship that is supposed to go outer space reliable enough to go there without any problems and you cand make a freaking DU for a car?how?

I'm am 100% sure that they are capable of doing it. It's a time/budget/effort required issue. If the car crashed each time the DU started making noises, I bet they would have fixed it/re-engineered it by now.