Optimizing track mode for snow driving

Hello everyone,

I've just returned from Winter Driving School, now that my MYP has track mode. I thought I would share the settings I came up to optimize track mode for high performance driving on snow (high performance here refers fine control, not outright speed).

Here's a video - see below for discussion of settings & analysis.

Before my MYP, I had Audi's and Subaru's that would let me turn off the stability program. That was much more effective in snow than an MY without track mode. I had hoped they'd eventually give us track mode, and that with it, my MY would be able to get into the same ball park for car control as my prior cars. I am happy to report that that is exactly what happened - but it was not a direct, or short path to that goal. If you just want to know the numbers I'll give the spoilers now. Read on if you're interested in why these ended up being the right numbers or in how we got to them.

Notes:

• These settings were with around 200 lbs of cargo in the trunk
• The track had a solid foundation of ice with several inches of snow that we wore through in several spots.
• These bias and regen settings are likely not optimal for either all-ice or all-snow, but they'd be fine as a starting point
• The stability setting is critical, and would be the same for any conditions

Settings:
Front/rear power bias: 30/70% (anywhere between 35/65 and 20/80 was acceptable)
Stability: -9
Regen: 75% on the skidpad, 30-40% on the slalom

IMPORTANT: Watch out for ABS on ice - with regenerative braking on even partially, if you trigger ABS, the car will keep doing ABS after you take your foot off the brake, b/c it still does ABS for regen-braking. That means you can't threshold-brake to get back your grip. Or at least to do it w/ regen on, you would have to actually press the accelerator to the point where the car wasn't braking anymore, reverse-modulate by lifting to brake more, and pressing to brake less. Crazy-making, that was too next-level single-pedal driving for my brain, I didn't try.

Ok, I'll jump directly to the biggest spoiler from all that: the -9 stability setting. It took a lot to get to this.

-10 does not work well because the MY has open differentials, and uses the brakes to force power from a slipping wheel to the other side of the differential (some call this EDL - Electronic Differential Locking). However, Tesla considers this part of "stability," and when stability is at -10, it's completely off. -9 is the magic number because it enables just enough EDL to reliably get wheels on both sides of an axel to spin together, without cutting out power or doing other tricks to stop you from sliding intentionally.

Here's how we came to all the above...

On the first day, my results were really frustrating - about 1/3 of the time, I could initiate controlled oversteer readily, and keep it controlled for 3-8 seconds. 1/3 of the time it wouldn't come easily, I'd have to slam the accelerator to get oversteer, and then it would be wild/uncontrollable. And 1/3 of the time the car just wouldn't do anything except just keep going slowly in whatever direction I was pointing it, with no understeer or oversteer, no matter how hard I pressed the accelerator.

I tried many combinations of settings that day, ranging from 50/50 to 20/80 bias, 30-80% regen, and -10, -7, and -6 stability. The other tweaks only had a minor impact on how often or how long I could maintain controlled oversteer, or how likely I was to wildly oversteer & spin out. There was no happy medium - always a split between too much, not enough, and always unpredictable.

We had a social event that evening, and I had a great super technical conversation and super constructive conversation with one of the instructors. We had several theories:

1. Maybe even with stability as low as I can put it, the car is still doing some stability control
2. Maybe there is a timer, and it's only willing to let me spin wheels wildly for a limited time
3. Maybe part of the problem is that when the details are just right, and open differential can still spin both wheels (like if they have exactly the same amount of grip and are going the same speed when you overpower them)
4. Maybe the built-in "Drift Mode" setting - which you can not edit, but you can see what settings it is using - has a hidden change that it does that is not available to me in the custom settings

So, we came up with the following experiments to answer the above questions, and he let me go out to the track early on day 2 to conduct them:

1. Get on some ice, point the car straight, and floor it for a long time. See if it eventually cuts power
   1. It did not
2. Same as above, but have observers on both side of the car - see if all four wheels are spinning at the same time
   1. They were not. Sometimes wheels on both sides of an axel would spin, other times it would go back & forth between one side and the other.
3. Turn on Drift mode, and see if I was maintain a continuous drift without it ending except by my choice
   1. I was
4. Duplicate Drift mode's settings in a custom setting group, and see if I was still able to maintain continuous drift (thus proving or disproving whether the built-in Drift mode has a special hidden setting)
   1. I was, and thus, it does not.

Armed with all of the above information, the instructors all concluded that the car still has some kind of traction or stability control going on, even when it's set to -10. I was not convinced. The one thing we didn't have an experiment for was to determine whether the inconsistent results were simply because an open differential has inconsistent behavior.

That is when I thought of trying -9. It was more of a hail-mary, I never believed it was going to work. My thinking was this:

• If an open diff sometimes will still spin both wheels, then maybe all the times I was able to cause oversteer, it is because in those cases, by chance, I got both wheels to spin.
• I had a feeling that the alternating spinning and stopping of wheels that they reported during the straight-line acceleration test was because as I moved forward (slowly b/c of all the slipping), a wheel that was once on ice would eventually find snow, and if the opposite wheel was then on ice, I'd expect the slipping to switch from one wheel to the other
• Maybe, with a lot of luck, if I add the smallest amount of stability programming that I can add, it will bring in just enough EDL to force both wheels to spin whenever one is spinning, but not so much as to prevent me from getting the car to slide.

Well it worked! We did one more test:

• Straight-line floored acceleration with observers on both sides, and front/rear bias at 50/50
  • All four wheels spun continuously until I took my foot off the accelerator!

After that, I tried 25/75 front/rear bias and hit the skid pad again. For the first time, I was able to "drift" in all-wheel-drive mode for as long as I wanted, and it worked every single time. We never did bother to figure out why Drift Mode also enabled this, even though it uses a Stability setting of -10. I think it is a combination of me already being moving and turning before flooring it, and turning off the front motor means the equation for keeping wheels on both sides of the read differential became much simpler.

 

[avramd]

The normal power delivery of MYP is rear wheel drive. It would engage the front wheel as needed depending on the situation. [...] When you adjust the handling bias in track mode, it adjust the power delivery during turning. Going in a straight line it still uses mostly rear wheels in track mode unless you're mashing on the accelerator pedal. I have the S3XY button app that displays front and rear inverter output in real time, and there isn't much difference going in a straight line whether or not I'm using track mode.

[...] For starters, the battery cooling fans come on full blast when track mode is turned on, so you have extra noise and potential wear and tear on the fans if you use it too much. I also think you would need to be driving pretty aggressively to notice a difference, but I'm not much of a performance driver, so perhaps I don't know what I'm talking about.

These comments are surprising. It sounds like you have some legit authority w/rt the app, so I won't presume to contradict you. However my experience from WDS doesn't reconcile with your comments. Did you observe this during slip conditions, or is this just in normal driving?

I'm pretty sure that in TM, the bias setting really just does exactly what we set it to. Every aspect of my WDS experience was that adjusting bias had exactly the effect I expected, even single 5% steps. In particular, when I had the bias nearly centered at 55 rear/45 front, there was a lot of evidence that that was always the split, no matter what I was doing with the steering wheel or the throttle:

• When starting from a dead stop on a slippery surface, pointed straight ahead, all four wheels started slipping together. I didn't get anyone to verify this from the outside by watching, but the car would crab sideways immediately if I gave to too much throttle. "Too much" here is about 10%, and you always apply the throttle gently b/c sudden changes are destabilizing. So, I wasn't mashing the throttle, and I wasn't turning. The car can't know in advance that the wheels are going to slip from 10% throttle when I'm not moving yet, so it doesn't have a chance to decide which situation it is in.
• On the slalom, they taught us to kick the tail end out early and straighten the steering wheel so that the car is pointed at the next turn before you've gotten around the current cone. You slide sideways for about 1-1.5 car widths, with the wheels pointed straight forward. You start accelerating as soon as you've cleared the current cone. If you do it right, the car is still sliding sideways, but not rotating at all. If you have rear-biased power and start to accelerate when already sliding sideways, then the car would start to rotate. This is what it did when I was using 75/25 bias. As I improved, I tried lowering it. With each step, the rotation reduced when I accelerated during those slides. That's how I got to a 55/45 split, and at that point there was basically no rotation, partly I think b/c I had extra weight in the trunk.
• At a more basic level, once I had dialed in -9 on stability control, everything became perfectly predictable. If the car had been ignoring my bias settings in some cases but not others, I would have still experienced a lot of "WTF just happened? I didn't do that..." That didn't happen. I made tons of mistakes, but they were always clearly my mistakes, not inconsistent car behavior. The car's behavior was consistent, there was no evidence of the power bias being different depending on whether the steering wheel was turned or how hard I pressed the accelerator.

I wonder if what you are observing is more of a natural side effect of weight transfer. When you accelerate, weight shifts towards the rear wheels. I wonder if the extra load from that creates more back-pressure on the rear-motor, and that shows up as a higher inverter reading.

Additionally, I had virtually no battery cooling going on. My understanding was that even in track mode, the car only does that when the batteries are hot, not all the time. At WDS things didn't get very hot b/c it doesn't take much power before you're sliding or slipping.

Also, to state the obvious although I'm sure you already had this in mind: Track mode makes a night and day difference in snow driving, which is to say when not pushing the car very hard.

 

[gx9901]

These comments are surprising. It sounds like you have some legit authority w/rt the app, so I won't presume to contradict you. However my experience from WDS doesn't reconcile with your comments. Did you observe this during slip conditions, or is this just in normal driving?

I'm pretty sure that in TM, the bias setting really just does exactly what we set it to. Every aspect of my WDS experience was that adjusting bias had exactly the effect I expected, even single 5% steps. In particular, when I had the bias nearly centered at 55 rear/45 front, there was a lot of evidence that that was always the split, no matter what I was doing with the steering wheel or the throttle:

• When starting from a dead stop on a slippery surface, pointed straight ahead, all four wheels started slipping together. I didn't get anyone to verify this from the outside by watching, but the car would crab sideways immediately if I gave to too much throttle. "Too much" here is about 10%, and you always apply the throttle gently b/c sudden changes are destabilizing. So, I wasn't mashing the throttle, and I wasn't turning. The car can't know in advance that the wheels are going to slip from 10% throttle when I'm not moving yet, so it doesn't have a chance to decide which situation it is in.
• On the slalom, they taught us to kick the tail end out early and straighten the steering wheel so that the car is pointed at the next turn before you've gotten around the current cone. You slide sideways for about 1-1.5 car widths, with the wheels pointed straight forward. You start accelerating as soon as you've cleared the current cone. If you do it right, the car is still sliding sideways, but not rotating at all. If you have rear-biased power and start to accelerate when already sliding sideways, then the car would start to rotate. This is what it did when I was using 75/25 bias. As I improved, I tried lowering it. With each step, the rotation reduced when I accelerated during those slides. That's how I got to a 55/45 split, and at that point there was basically no rotation, partly I think b/c I had extra weight in the trunk.
• At a more basic level, once I had dialed in -9 on stability control, everything became perfectly predictable. If the car had been ignoring my bias settings in some cases but not others, I would have still experienced a lot of "WTF just happened? I didn't do that..." That didn't happen. I made tons of mistakes, but they were always clearly my mistakes, not inconsistent car behavior. The car's behavior was consistent, there was no evidence of the power bias being different depending on whether the steering wheel was turned or how hard I pressed the accelerator.

I wonder if what you are observing is more of a natural side effect of weight transfer. When you accelerate, weight shifts towards the rear wheels. I wonder if the extra load from that creates more back-pressure on the rear-motor, and that shows up as a higher inverter reading.

Additionally, I had virtually no battery cooling going on. My understanding was that even in track mode, the car only does that when the batteries are hot, not all the time. At WDS things didn't get very hot b/c it doesn't take much power before you're sliding or slipping.

Also, to state the obvious although I'm sure you already had this in mind: Track mode makes a night and day difference in snow driving, which is to say when not pushing the car very hard.

I'm sure your observations are correct. I was just commenting on track mode generally, and I was thinking on dry road because the poster seemed to ask about it in a normal dry road setting. When our cars got track mode, I tried to observe the power output on the front and back wheels as shown on the S3XY button app and it did appear to be about the same going in a straight line on dry road, with the rear wheels getting most of the power, even when the handling bias is turned 100/0 toward the front.

As for the fans, every time I turn on track mode I hear them come on pretty loud, and they sound similar to when I'm preconditioning the car's cabin temperature. I guess that loud humming sound might not be fans, but on my car they are clearly audible when track mode is activated.

 

[avramd]

As for the fans, every time I turn on track mode I hear them come on pretty loud, and they sound similar to when I'm preconditioning the car's cabin temperature. I guess that loud humming sound might not be fans, but on my car they are clearly audible when track mode is activated.

I believe the sound you are referring to is the heat pump. I'll pay more attention to it this winter. Perhaps once you're in a cold enough environment then as long the battery is cool, it doesn't turn on?

 

[timurx]

Amazing post! Love to have seen all the testing!
75% Why is regen on at all? I would have thought 0 would be best. I’m not trying to drift, just get to work in bad conditions. My YP fishtails (default settings - sport or chill - non-track-mode) even if I feather the accelerator carefully.
Thank you!

 

[avramd]

Amazing post! Love to have seen all the testing!

Aw shux... Thanks

75% Why is regen on at all? I would have thought 0 would be best. I’m not trying to drift, just get to work in bad conditions. My YP fishtails (default settings - sport or chill - non-track-mode) even if I feather the accelerator carefully.
Thank you!

Well, a few reasons:

1. Have you ever actually tried driving with 0% regen? It is terrifying. Most ICE vehicles engine braking has an effect equivalent to - just guessing - a setting of about 40-50% regen on my MYP. However, that's not a measure of how much farther the car will coast compared to 100%. When you turn off regen completely, the car will coast for MILES. It is freaky AF to take your foot off the accelerator and have NOTHING change. It is so unintuitive that it feels like the car is accelerating. You have to use your breaks every time you turn a corner even at low around town speeds when you were already below the speed limit.
2. My goal for winter driving is to maximize my skill, control over the car, and ability to drive at non-grandma speeds without getting into trouble, and with the ability to get out of any trouble I get into. Turning off regen would require you to switch pedals much much more than you would ever normally need to. That pedal switching is an unnecessary and unhelpful distraction and delay compared to being able to modulate the car's weight transfer just by subtle lifts or presses of the accelerator. This is not Tesla/EV specific - using careful engine braking to modulate weight transfer is fundamental to all high performance driving, from F1 on a dry track all the way down to 5mph on an iced-over skidpad.
3. Setting it at 75% makes lift-throttle weight transfer manageable. At 100%, I found it too touchy - when the track was slick enough, I did not have precise enough control over my foot to get mild enough engine breaking to cause subtle enough weight transfer. Also it would trigger the ABS, and once that's active, you've lost 80% of your grip budget. 50% was little enough that I found myself having to switch to the brake pedal a lot more, and that is a clunkier way to try to influence weight transfer. For me, 75% was the sweet spot where I could get the amount of engine braking I wanted most of the time, rarely too much, rarely too little, and about an even split of those two.

If your primary concern is avoiding fishtailing, I suggest trying track mode with 50-60% regen, and experiment with power balance between 40 and 60% front/rear. I think you will find the car doesn't really want to fishtail with those settings as long as you are lifting off the accelerator gently. I suggest you do not look for a setting where you never fishtail even if you lift off aggressively. That is called lift-throttle oversteer, and you want a car to be able to do that. If you can't get a car to fishtail even by aggressively lifting off the throttle, then the car heavily out of balance, and you are going to end up understeering too easily, and in situations where there didn't need to be any skidding at all.

Beyond that though, I suggest going to a winter driving school or just experimenting in a very large empty open parking lot, and just learn how to control fish-tailing; It's fine that you don't want to drift, but it's better to know how to get out of it than to live in fear of it.

 

[timurx]

Thank you for that amazing response!
I’ll give these a try.
Much appreciated!

 

[avramd]

I love that the MYP got track mode, but I've only really used it on an HPDE track because it's not really practical to use in everyday driving. For starters, the battery cooling fans come on full blast when track mode is turned on, so you have extra noise and potential wear and tear on the fans if you use it too much. I also think you would need to be driving pretty aggressively to notice a difference, but I'm not much of a performance driver, so perhaps I don't know what I'm talking about.

Just thought I'd chime in here after another WDS: My experience was also that the heat pump was always running when I was in track mode. I wouldn't call it "full blast," it seemed the same to me as when I'm running a/c on a hot summer day. I don't know what it's doing since most of the time I was sitting still waiting my turn. It may have been warming the battery though, b/c I never once experienced reduced regen like when I go out to my car on a cold day.

 

[avramd]

And more generally, I figured I'd report in after my third Winter Driving School, and first one in 11 months. This time there isn't much new to report. Only one thing stood out:

It appears to me that Tesla has tweaked the definition of the 0-100% regen slider. In the past when setting it really low like 20% or 10%, I remember feeling like the car was really coasting, like I'd shifted into neutral. This time I felt like there was a lot more braking effect even way down at those low settings. It was mostly my friend using these low settings b/c one of the instructors decided it would be best for her.

Other than that, for me this class was about nuance. Most of my effort went into fine-tuning the timing of all of my maneuvers, and trying to get finer control over my foot so that transitions between accelerating and decelerating would be smoother, and so I could ask for smaller transitions.

I was moving between 45% and 75% regen myself. I was finding that different amounts of regen were most effective depending on the exercise, and on how slick the surface was. For the slaloms I used mostly around 50% regen - just enough so that I could drive single-pedal. Not having to move my foot between pedals made me less clumsy - that's a fairly gross adjustment, so by not having to do it, I got to spend 100% of each transition from accel to decel focusing on smoothness and just the right amount of input. For the skidpad, 75% regen made it a lot easier to induce lift-throttle oversteer.

One thing I did not experiment with at all this time was real-time adjustment of the power distribution slider. I still feel like that would be extremely effective in certain scenarios, but having to reach over to the display and brace my hand to have precise control of my thumb was too distracting a transition. I would have had to keep one hand braced on the screen at all times to avoid those transitions, and then I'd have only one hand on the wheel, which is asymmetric and harder to be smooth or precise with. I wanted to just focus on getting into a groove with the car's behavior, and increasing my regular control ability.

<--- nothing tesla specific for a moment --->
That worked well. I spent most of the class in a really nice groove of what they call "pendulum turns." That's when you time the recovery of your right fish-tail around the current cone so that it swings right into the left fishtail that you'll be using to get around the next cone, and repeat. Another thing I learned is that when you change directions, you can use the car's weight shift to get extra grip. When you're turning left, the right suspension is compressed.

When you switch to turning right, the right suspension springs the car over to it's left side. There's a key moment when the car is loading up the suspension on the other side that there is extra pressure on the tires. Sort of like how if you stand on a scale and then hop, you appear to be lighter for a moment, then heavier, then back to normal. I learned to delay changing direction so that I would get extra grip on the slick spots right in front of each cone from that key moment of suspension loading. It would kinda "jerk" the car for a split second onto a new trajectory. Then on the new trajectory I didn't need as much grip on those slick spots. Time after time it looked like I was going to slide sideways right over the next cone - and them "oomph" the outside tires would dig in for an instant and redirect me. It felt almost like a billiard ball bouncing off a bumper.
<--- end non-tesla specific moment --->

Out of 20 students and instructors, I got the fourth best time on the timed long-slalom/road course. Only one student beat me, and two instructors. While I can take some credit for applying myself, learning, and improving, I think the car has a few of key characteristics that contributed to those results.

1. Single-pedal driving opens up a lot of opportunity for being smoother.
2. Instant acceleration saves fractions of a second here & there that add up
3. Very low body roll means the car's inside tires have more grip more often
4. 50/50 power balance lets me apply more power on turns without causing excessive fish-tails

The top five students finished within 1.2 sec of each other, times ranging from 1:31.6 to 1:32.8. The top two instructors beat us by a whopping 3.6 and 7.2 seconds! The #2 instructor was driving an RS4 with manual transmission, studless Nokian Hakkapeliitta's (same as me), and a WaveTrac limited-slip differential. The #1 instructor was driving a basically bone-stock old A4 1.8t with studded Nokian Hakkapeliitta's.

Now it's time for the bad news: The student who beat me was a Russian woman driving an old FWD Honda Accord. Although it did have studded tires. She was really good - whenever we were on the skidpad together she always caught up to me and had to stop and let me build some distance so that she could push her limits again.