Is it OK to ask questions about what's in captured data in this forum?? Hope I'm not crossing a line...

I was participating in a HPDE session the other day with an instructor sitting beside me. This is a person who's used to more of a momentum car (944) and I'm driving a 991TT.

Some basic traces...



And one with a bit more math...



Right about the point in the lap in the images, I get asked about how the brakes work on my car and we get into a discussion about nannies in the car once we get to debrief at the end of the session.

Looking at the data afterwards, I see some times where the car is applying pressure on just one of the four channels. See the spot at the cursor in the pictures for an example.

My car has Porsche Torque Vectoring (PTV). Plus, I drive with all the TC/ESC gadgets enabled (only one I turn off is the Porsche Active Safe, radar assisted braking gizmo which just doesn't like cars with high closure rate but obviously about to turn off-line out of the way ).

So this here are my questions:

1) Is it safe to assume that what I'm seeing is PTV at work, which would imply that the PSH_BRAKE_PR3 is the left rear corner??

Thinking more about nannies...

2) I don't believe there was any sense of oversteer as I was pretty much settled well on the way to exit cone at this point. But if there was oversteer, I'd guess that traction control nannies would dab the outside, right rear corner at that point on the track to help bring the tail back. Is that a safe assumption about how the TC would reach in??

And the real question out of the debrief discussion...

3) So I'm trying to work on my driving technique; part of the reason that I have this data at all. The instructor suggested that one should strive to avoid nanny intervention of any kind. If the kind of thing I see at the cursor there, small brake pressure clearly not applied by me since it's only one corner, is in fact PTV is that really a nanny thing or just part of the way this car works -- in other words, would it be advisable to alter technique to try and get rid of that sort of trace in the data or not??

[And of course since I posted pics of a whole lap I'd happily take any advice on what to work on. I've pulled a few things out for myself but I'm sure experienced eyes will probably see what needs attention first where I may be looking at merely incidental things...]

 

Very good post. Some of this intervention is just the way the car works. Keep pushing!

 

OK. So.

Disclaimer: day job, I tune this stuff. For the OEMs. Only the American ones, not the Germans. I'm a Principal Engineer for Chassis Controls at Bosch Engineering here in the US. So I know the technology, but not the specific tunes we're looking at here.

Regarding the PTV vs. TCS vs. ESC (Electronic Stability Control)... you should also pull in the individual wheel speed data to ensure your conclusions are correct on assigning brake signals to corners.

If TCS is responsible for one rear wheel brake activation, it'll be braking the inside, unloaded wheel, in response to wheel spin-up. So you should see a wheel flare at the same time. I doubt this is likely the case on-track. You may or may not also see an engine torque reduction with this case.

If PTV is responsible for the activation, it'll be the same (inside) wheel, but it won't be in response to a wheel flare... so you would see the brake activation, but not inside wheel spin. You might also see some interesting things in there if you pull in yaw rate as well as the steering (which you already have), if you can get the yaw rate signal from the CAN bus.

If ESC is responsible for the brake activation, it will be on the outside wheels, but most likely the FRONT, not the rear (or both front and rear). This is not Traction Control. TCS manages acceleration, ESC manages yaw response (for the most part). TCS response to wheel slip on the driven axle(s). ESC responds to, mostly, yaw error (oversteer or understeer, hopefully).

Regarding your last question... yes, your instructor's right... but your car is unfortunately designed to rely on the PTV function to go faster. You'll learn more without TCS/ESC on, sure - heck, even try just driving TCS off, if no Sport/Chrono/whatever mode - but you may not get quite as much rotation as you might like in the low-speed stuff.

Then again, your rear pads might last a little longer...

 

Wow, I'm constantly amazed at the knowledge resident here and peoples' willingness to share and help. Thank you! Can I push my luck and pile on...totally understand if you don't want to get into it more.

So the channels on my car for individual wheel speeds are labeled to suggest which corner is which. I presume that Aim got this right but, given that, I didn't see what I expected from your descriptions so I'm not groking the whole thing quite yet apparently. Here's a pic of just that one corner from the same session/lap as above but zoomed in on X and Y axes.

I don't have yaw rate from the car but the SoloDL delivers a yaw rate measure so maybe that's good enough to see what you were thinking about(?).



I guess now you can see just how ragged my driving is blown up large for all to see

Does that give you enough information to form a definitive conclusion about the brake channel to corner correspondence for that red measure??

I see. Thanks for the advice there. It sounds like you are implying that, to a degree, TCS is a performance enhancer while ESC is more oriented towards catching unsafe conditions. Unfortunately for me it seems that I have only one button to control things from the cockpit which turns off "PSM".

I think some of the other 911's give you separate means to turn off ECS and TCS -- I see that in the 991 GT3 for instance. Although it seems odd to me that there's an ESC OFF and a ECS+TSC off button -- I guess I really don't understand this stuff well because I would have guessed TSC OFF and then ESC+TSC off would make more sense.

 

...and now I'll respond like the typical engineer - MOAR DATA!!!

So we can see you're in a traction event; the rear wheel speeds are both clearly above the rough visual average of the fronts. But they're together, so you're definitely not getting a traction brake hit. A display of engine torques would definitely show if there's any engine torque reduction from traction - but is perhaps not critical for this discussion.

Well, unless it causes what appears to be a massive understeer event that precipitates the PTV event! Which yes, it now pretty clearly appears this is; yeah, I'd feel comfortable calling that PR3 channel the inside rear brake.

It looks like your yaw rate is building well up to almost -19/20 deg/s in the corner, then drops off a lot (to nearly half the value) at 2480 ft., recovering as the brake is applied.

This is where the driver coach can just as easily chime in; we'd want to see lat and steer again along with that yaw rate, but odds are you've just steered excessively, are in understeer, and PTV's trying to help deal with that. Though peeking at the earlier data set, it's hard to be totally certain without a zoomed-in data overlay...

So the corrective action, as a driver, is less steering (obviously, standard understeer management technique), and/or less gas, or better line, or less speed, or different apex timing... etc etc, I don't know this track so can't say exactly what you're doing wrong here, but you do appear to be overdriving the car, trying to hustle it through there faster than it wants (on that line, at least)...

PS - it would also be worth looking in close detail on this corner for multiple laps, see how likely different inputs or approaches might change both the vehicle and system reaction... that might help quantify how much is driver input...

 

Hahaha! Rock on, Vaughn!

By adding lat g, steering and calculating RATE of input, you could really fine tune driver inputs to minimize the intrusion of the nannies.

Agree with overlaying multiple laps. It's amazing the inconsistency almost all drivers exhibit. Sometimes by looking only at the fast lap, hidden gems remain hidden...

 

Steering input rate is helpful, but actually actual steering input magnitude vs. steering limit on the surface is more important for ESC control activation.

Finding that limit is fairly easy; look where lat stops increasing despite increasing steering, and you're there (for a given speed). Typically the yaw rate will also drop, though sometimes the drop is very subtle and harder to see, once you exceed the max useful steering angle...

 

LOL -- OK when Peter said above "keep pushing" I guess he meant: "you keep pushing in that corner" perhaps
Being an engineer, I knew that was coming

Unfortunately the CAN decode doesn't provide engine torque data so that one I can't do.
OK great. I didn't consciously register understeer at the time but of course that just means my butt-dyno isn't up to snuff yet perhaps.

I'm happy to be able to slap a Left Rear tag on PR3 though -- thanks, that is useful to be sure about.
OK, how does this look??



Interesting. That's definitely something to work on then and not what I thought going into this discussion! Great insight, thank you.

In case you are interested, here's a track map with "the school line" marked on it on their web site -- this is Ridge Motorsports Park, Shelton WA for reference.

This corner is T5, I'd guess the third most significant on the track. It follows an uphill left-right kind of S pair of linked turns, comes just after a crest and the terrain is gently sloping down away from you and to the outside of the track as you look at it from the right edge of the track turn in point.

The advice from instructors is actually kind of split...there's a T4 apex cone on the left that some want you to make. Others, including one of the chaps that drew the school line, say pass wide of the T4 apex cone and make the T4 turn-in to T5 turn-in one single radius arc. The objective is to get back to power well before the T5 apex...so that would probably correlate to "ok but you're over-doing it, too much and/or too soon" given the above analysis.
OK, here are 4 laps from the same session. These are sequential actually -- I got mostly clear track ahead for these so the times are somewhat consistent -- first three +/- 1 sec or so and the last one is a shade slower (but that's mostly about having to slow to make a pass coming onto the main straight at the end of the third of these lap).



If I back up and squint at this it seems to me there's a dip in yaw rate preceded by a drop in Lat G in more or less the same place on the track in all these laps. Am I reading that correctly??

If so and it's consistent I guess the homework item is to find a line and inputs in this turn that bring Lat G to a sustained max that doesn't dip until I start unwinding steering past the apex, right??

I'm assuming that accomplishing that would end up being "faster"; safe assumption??
OK, now I'm going to stick my neck out -- so looking at the chart with the 4 traces, maybe coming up to around 2500' there's a consistent drop off in Lat G but the steering amount input is still increasing. Is that enough data to conclude that useful input at the steering wheel is limited to 55-58 degrees in this turn?? I'm not a test pilot and don't play one on TV and I suspect that going out and running test patterns on the track during DE events might not be welcome but I see what you're suggesting.

This is fascinating stuff. Thanks for playing along and indulging me

 

Well, we can define what it'd take in the way of driving inputs to reduce the PTV intervention. That may not, however, equate to going faster.

It's equally possible that this is just a area of the track where the elevation change just induces understeer, and the "correct" technique is to just ride it out, knowing that the car will recover as the road settles.

What I see here is when the yaw rate starts to drop, your steering is actually constant - this is where the PTV intervention starts. So it looks to me that it's actually not so much your inputs as a loss of grip that's causing the initial event.

Then you add more steering, clearly betrayed by the steering rate signal, which doesn't help expect perhaps to induce more PTV output... and then the car recovers after, as you come out of the steering.

It's a shame we don't have ride height data to validate the theory of an elevation change causing a loss of grip...

Anyway, I could see grounds for an argument that the PTV is hurting your ability to learn to sense and react properly to understeer. Damn software engineers...

That said - probably the best place to learn such is in a car without such aids (or in this car with them turned right off) in a snowy parking lot. You appear to be reacting to understeer with more steering - though mildly, maybe an extra 5-10 deg steer... So working harder to dial that reaction back or out would seem desirable.

Also, side note - that GPS yaw rate signal seems heavily filtered. I'd look into dialing that back, if possible - looks like you're losing a lot of useful information there. Steering rate signal also looks over-processed, but I doubt you have a chance to do anything with that in the AIM system with the A-to-D converstion and loop calculation time... just take it with a grain of salt... besides, you can visually infer the necessary information directly from the shape of the steering angle anyway, and your brain can process the data more accurately than the dumb math channel, in this case...

 

Excellent observations!

We DO have vertical acceleration, I believe. Which, if properly calibrated initially and not heavily filtered, could provide insight to the topography...

As to the latter, true dat!

 

Well, measuring the vertical body acceleration might provide some hints, but I personally wouldn't rely on it like I would the ride height data, since the latter's a much better indicator of what's actually happening to the wheels touching the road, vs. the body we're trying to keep off the road.

Maybe sometime I'll show you some FlyingCar data, Peter...

 

Hmm, pretty sure I wasn't flying

More data?? Why certainly! Unfortunately I can't get ride heights from this car with the built in sensors and decode so...

Here's a chart with the Vertical G added -- this is from the SoloDL not the car.



And for giggles here's the RSA "GPS" chart for just that section of the track. I've rotated it and tilted it a bit so you are looking at it from slightly above grade and to the outside of the track. Kind of looking down and through towards the corner exit, if you will. The coloration is again the Vertical G value.




What it's showing is that I'm descending to the datum plane and then the middle part of the corner is below that datum before it climbs up again towards corner exit. The datum plane is where the uncolored edge of the grey transparent vertical plane is drawn. I don't know how to control the scale in the vertical of this -- it relates to topology height somehow I think but it definitely exaggerates the terrain height changes quite a bit.

Interesting bump (?) up for Vertical G right before the traction event -- could that be a piece of the puzzle??

On this day the instructor was asking me to make the T4 apex cone and I think if I look at the video I can see a bit of a bounce in the front end of the car right after that apex. It almost looks like the front corner dips into a low spot and then gets bounced out a bit - I surmise that's what the little yellow/orange section is.

. That's climbing up the hill through T3 and T4 and then left into T5.

On other days if I look at the data in the same track spot when I've been trying for the line that makes T4 entry to T5 exit one big sweep passing wide of the T4 apex...well, gee, I see no PR3 blips. Hmm...possible eureka moment there??

FWIW, subjectively I noticed on this day that trying this tighter line around T4 that I was not really using all of the track out by the T5 entry cone where on the wider line that was easier and felt more comfortable too...by which I mean my recollection is that the car was better pointed through the apex to the exit on the wider line (...which, if I were really post-justifying, might account for me trying to add steering to make the apex through that little understeer moment(?)).

p.s. The Vertical Acc channel is right from the SoloDL -- I don't have any control over the filtering of that so far as I can tell. I believe I did calibrate the SoloDL though...well assuming I did it right; I tried at least at the start of the day. Yaw rate is the same thing -- direct from the SoloDL, don't think I have much control over the processing. The Steering Rate one is a straight RSA math channel [in this case: deriv(PSH_STEER_POS)] so again I don't think I can get better granularity. The car does spit out a PSH_STEER_SPD but this appears even more filtered and compressed: values in a 0-5 range for this section of data and less active than the math channel as well as being unipolar so I didn't add that because it's probably less helpful than the math channel one anyway.