Car disabling track mode when drifting

[LoL-Gas]

Anyone have any issues with the model 3 throwing the car out of track mode and disabling traction control? Happened the other day while I was doing some drifting/donuts. Car freaked out mid spin, exited track mode, threw about 10 codes, regen braking disabled along with traction control. Had to exit the car and get back in for the car to reset. Never had an issue like this before

 

[Mash]

Sounds like intermittent wheel speed sensor issue. Check those cables.

 

[gearchruncher]

I've had the same thing happen in track mode, but I never had any wire damage or sensor failures.
In my case, what fixed it was changing the tire config. When I ran the track pack config with actual tires that were smaller in diameter that the track pack, I would reliably have this occur on about 25% of my AutoX runs.
I changed to the 18" Aero config, and it hasn't happened a single time in 100 runs since then.

So try a different wheel config if you don't find any damage to the sensors.

 

[Gauss Guzzler]

Mine does the same thing now but didn't in the past. Everything is stock and properly functional but when I light em up it triggers a plethora of errors disabling regen, forward collision warnings, stability control, etc.

 

[Mash]

Still sounds like something gets disconnected temporarily.

 

[gearchruncher]

Still sounds like something gets disconnected temporarily.

While a possibility, I am 100% sure this hasn't happened on my car. I can literally set the wheel type to the track pack, go out and do 5 laps and get 3 of them with an error, change the wheel config to the 18", and then run 20 laps without an error.

I've never not gone more than 3 laps without an error when set to track pack, and never once had an error on 18".

I've been in a car that did cut the ABS wire, and the messages and such are identical. TC, ESC, ABS, power steering, etc all go into fail safe mode and the driveline drops to about 100HP.

 

[LoL-Gas]

It's unfortunate Tesla still limits the car in track mode. Would be nice if they let you turn off all safety features and allow the owner to send it

 

[gearchruncher]

It's unfortunate Tesla still limits the car in track mode. Would be nice if they let you turn off all safety features and allow the owner to send it

The only thing track mode doesn't allow you to disable is traction control and ABS.
As we've seen from the old dyno mode, the car is basically un-drivable without TC. The tires will just immediately spin up to 100+ MPH. A human can't really manage EV torque. If you've driven drift mode much, you can feel what this would be like without any limits.

Plus, in a car with completely independent front and rear drivetrains. with open diffs, what do you expect it to do with front and rear axle power when you press on one pedal?

 

[LoL-Gas]

I expect the car to spin the tires and not go into limp mode and disable mid drift. It's nothing special; disable everything and let the owner send it, even if it breaks.

 

[gearchruncher]

I expect the car to spin the tires and not go into limp mode and disable mid drift. It's nothing special; disable everything and let the owner send it, even if it breaks.

I personally am glad that enabling track mode doesn't immediately invalidate my whole warranty, which would be required for what you want.

I think you don't get what drift mode is also- it is a mode that manages front and rear wheel speeds to make the car basically RWD, and actually encourage a drift. An AWD machine with no front/rear speed control but needing to limit power (the motors can draw more than the battery can give) would be pretty weird.

 

[Gauss Guzzler]

The motors cannot "draw more power than the battery can give", they're not even connected to the battery. They're connected to inverters which deliver exactly as much power as the software specifies for any given instant. And while it's possible for something like a wiring short to cause an instantaneous current spike beyond the prescribed limit, the motor has no ability to trigger such a spike.

But electric motors can misfire. Much like fossil engines they use variable timing advance based on crank position sensors and can "knock" very destructively when the timing is off. But electric motors need much more precise timing than fossil engines -- and at a much higher rate. So it's possible that wheel hop could cause the motor RPM to change so quickly and unexpectedly that the next derivative of the crank position sensor reading produces an inaccurate RPM estimate and resulting misfire.

Maybe this is happening and being detected as a drivetrain fault? Or maybe there's some acceleration threshold that triggers a safety shutdown before the risk becomes too great?

 

[gearchruncher]

The motors cannot "draw more power than the battery can give", they're not even connected to the battery. They're connected to inverters which deliver exactly as much power as the software specifies for any given instant. And while it's possible for something like a wiring short to cause an instantaneous current spike beyond the prescribed limit, the motor has no ability to trigger such a spike.

But electric motors can misfire. Much like fossil engines they use variable timing advance based on crank position sensors and can "knock" very destructively when the timing is off. But electric motors need much more precise timing than fossil engines -- and at a much higher rate. So it's possible that wheel hop could cause the motor RPM to change so quickly and unexpectedly that the next derivative of the crank position sensor reading produces an inaccurate RPM estimate and resulting misfire.

Now this is the kind of pedantic, completely technically wrong take I expect from someone that thinks a "simple voltmeter" can be used to diagnose a "khz bridge" and was 100% sure that higher voltage was used on the parking brake system in different modes, with no proof.

We all know what I meant by motor- this is like saying an engine doesn't use fuel because the fuel pump, ecu, and fuel injectors technically deliver the fuel. The point here is that the max nameplate rating for the front and rear motors in a Tesla generally exceed the rating for the battery. They even lost lawsuits over this. Which means Tesla has to manage the torque to the front and rear axles, which they can't effectively do without wheel speed sensors.

The idea of "destructive knock" in an electric motor is flat out made up. Getting the current wrong in a phase once is not that big a deal, and nothing like knock in an ICE engine. So is the idea that an electric motor needs faster processing than an ICE engine. They are very much in the same ballpark. A Model 3 at 60 MPH is at about 6,000 RPM, and deals with needing to track the rotor to a degree or so. Meanwhile, a modern direct injection engine runs the same speed, and needs even finer tracking for fuel and spark- which are controlled to way less than a degree for modern emissions. Meanwhile modern ICE engines are often dealing with very fast gear changes in modern DCT transmissions.

Also, if you've ever had this happen to you in drift mode, you'd know there's no wheel hop involved. It's all very smooth. But it's due to high steering angle inputs and wheel speed differentials. If it was this "knock" that you suggested, then why would the car still allow you 500HP in a straight line afterwards, but just disable things like cruise control, stability control, traction control and abs, all of which require wheel speed sensors, while the motor's output is left untouched? That's a totally illogical response for losing track of the rotor for a moment and then gaining it back.

This is all just made up hogwash.

 

[Gauss Guzzler]

Exactly, it'd be like claiming that a fossil engine might gulp too much fuel when drifting when in fact it only consumes exactly what the computer delivers thru the injectors. There's nothing a fossil engine can do to pull more fuel from those injectors than the computer is willing to supply, and likewise there's nothing an electric motor can do to pull more power from the battery than the computer is willing to supply. To claim otherwise would be made up hogwash.

When a fossil engine misfires, the loads on that one connecting rod are far beyond the normal maximum, but the crank is in a near-zero torque orientation at that instant such that the geartrain is not necessarily overstressed. But when an electric motor misfires, it misfires "on all cylinders" and the rotor has very low inertia so it can decelerate very quickly, creating significant shocks thru the geartrain lash. We've all heard electric motors play music by way of rapid +/- torque oscillations, imagine what 300HP "music" could do to the motor and/or geartrain. The loads aren't nearly as high as a misfiring cylinder, but they are incredibly fast and can do a lot of damage.

A typical 6 cylinder fossil car might drift with the tires running around 40mph at 6000rpm with a 12:1 gear ratio, or 8 ignitions per second ,per cylinder (6000RPM / 2 revolutions per firing / 6 cylinders / 60 seconds per minute).
A 6 pole Model 3 could run the tires at 162mph with a 9.0363:1 ratio and a motor speed of 18,500 RPM or 5,550 commutations per second (18500RPM * 6 poles * 3 phases / 60 seconds per minute).

So yeah, 5,550 commutations per second is more than 8 squirt & sparks per second.

But in either case, the motor RPM is never known - the computer only gets periodic crank position angle readings and it calculates an estimated RPM from that. So you could imagine how a change in RPM between crank angle readings could result in a slight error in the RPM estimate, and a car that needs to "fire" 5,550 times per second might be more sensitive to this than one that just lazily ignites a slow burning campfire maybe 8 times per second.

 

[mongo]

A typical 6 cylinder fossil car might drift with the tires running around 40mph at 6000rpm with a 12:1 gear ratio, or 8 ignitions per second ,per cylinder (6000RPM / 2 revolutions per firing / 6 cylinders / 60 seconds per minute).
A 6 pole Model 3 could run the tires at 162mph with a 9.0363:1 ratio and a motor speed of 18,500 RPM or 5,550 commutations per second (18500RPM * 6 poles * 3 phases / 60 seconds per minute).

So yeah, 5,550 commutations per second is more than 8 squirt & sparks per second.

But in either case, the motor RPM is never known - the computer only gets periodic crank position angle readings and it calculates an estimated RPM from that. So you could imagine how a change in RPM between crank angle readings could result in a slight error in the RPM estimate, and a car that needs to "fire" 5,550 times per second might be more sensitive to this than one that just lazily ignites a slow burning campfire maybe 8 times per second.

Model 3 PM motors have direct motor phasing (and thus speed) sense. AC induction are SW inferred with wheel speed as ground truth.

 

[gearchruncher]

But when an electric motor misfires, it misfires "on all cylinders" and the rotor has very low inertia so it can decelerate very quickly, creating significant shocks thru the geartrain lash. We've all heard electric motors play music by way of rapid +/- torque oscillations, imagine what 300HP "music" could do to the motor and/or geartrain. The loads aren't nearly as high as a misfiring cylinder, but they are incredibly fast and can do a lot of damage.

A typical 6 cylinder fossil car might drift with the tires running around 40mph at 6000rpm with a 12:1 gear ratio, or 8 ignitions per second ,per cylinder (6000RPM / 2 revolutions per firing / 6 cylinders / 60 seconds per minute).
A 6 pole Model 3 could run the tires at 162mph with a 9.0363:1 ratio and a motor speed of 18,500 RPM or 5,550 commutations per second (18500RPM * 6 poles * 3 phases / 60 seconds per minute).

So you're claiming a "misfire" on an electric motor is applying *exactly* full reverse current at *exactly* the wrong time and thus producing a full 300HP in the wrong direction? Despite you acknowledging that there are 6 poles and 3 phases in the motor, and a few degrees of error won't do this. Meanwhile, a few degrees of error on an ICE does cause knock and excessive rod stress even without knock.

And yeah, let's assume the ICE car drifts at 40 MPH wheel speed, but let's assume the Model 3 does it at 162 MPH, when if you have ever drifted a Model 3 and had it error out, you know it's the exact reverse and the wheel speed is very limited and well controlled.

And you flat out don't get what is needed for timing- You don't just fire a spark plug or fuel injector once per 2 revolutions. You have to fire them down to less than 1 degree of accuracy, and for GDI, you have to have under a tenth of a millisecond control of the injector pulse width. You said that an EV needs much more precise TIMING than an ICE, but that's flat out not true. The ICE needs more accurate angular control than an EV- like you said, the EV only has 6 poles, so 1 degree of error isn't that big a deal. Yet on a GDI ICE, that matters a LOT. So yeah, an EV runs 2X the RPM, but needs less accuracy. It's a wash.

Here's the processor in a Model 3 inverter:

TMS320F28377S data sheet, product information and support | TI.com

TI’s TMS320F28377S is a C2000™ 32-bit MCU with 400 MIPS, 1xCPU, 1xCLA, FPU, TMU, 1024 KB flash, EMIF, 16b ADC. Find parameters, ordering and quality information

www.ti.com

It's a 200MHz, single core, that is about $5 in automotive volumes.

Here's an gas engine CPU from 2015:

MPC5777C|Engine Control MCU

MPC5777C 32-bit MCUs target high-end automotive and industrial powertrain applications for advanced engine control, functional safety, and security.

www.nxp.com

300 MHz, dual core, with a safety critical 3rd core. About $15.

You realize you can buy a $30 hobby electric speed controller that will easily support a 30K+ RPM model aircraft motor? If it needs such amazing timing control and so much more processing than an ICE engine, this wouldn't be posible.

So you could imagine how a change in RPM between crank angle readings could result in a slight error in the RPM estimate, and a car that needs to "fire" 5,550 times per second might be more sensitive to this than one that just lazily ignites a slow burning campfire maybe 8 times per second.

Again, totally misleading. The number of events per second is irrelevant. The accuracy of when they need to occur is all the matters. Those 8 injection and spark events need just as accurate of timing as the 5Khz commutations. You can't just arbitrarily fire 8 times in that second and have a working engine.