SundayDriver

The Motec does not come with any sensors - you buy and install what you want to use. In my case, for braking I have pressure sensors front and rear on a dual MC setup. This is where these systems can get very expensive - most sensors cost $250-$350 each so when you start doing multiple pressures, 4 shock pots, steering, throttle, etc, etc, is really adds up. I am currently only measuring 1 wheel speed - I'm sure I could learn quite a bit from more speeds.

As faras Yaw, I have not considered that (and not discussed it with my coach). I will bring that up when we talk again in a week or so. Care to share your thoughts on this? I guess I can immediately see how smooth rates would tell you a lot. How do you get good yaw measurement?

SundayDriver

Tim,

I don't see how I can provide support for a position that is someone else's - though I do agree with it. His opinion is based on personal experience and my personal experience supports that view (as does a few others I have spoken to about data).

Now I have no doubt that there are many, many things done in F1 that differ from what the rest of us do. I am sure they are valid for the F1 environmnet, but how can I (as an amateur or even a low level pro team) use G data when I don't know the g limit for the car/corner/condition combo? I mean, F1 teams have 6-8 data engineers and everything imaginable is instrumented. They have systems and software costing more than my whoel racing budget for a decade. I am SURE they know what their car will do on a given corner, but the rest of us don't. What do we do? Go out on a skidpad and measure g's at the limit. Fine - that tells us the limit for flat cornering at whatever throttle balance you had on the skidpad on that set of tires, etc, etc. Now, how do I relate that to the keyhole at Mid-Ohio? I can promise you that if you take your skidpad limit, and apply it to the radius of the keyhole and use that for you minimum corner speed, you will crash. How, as an amateur, do I make that leap and possible get real value from G data?

There are two more very significant issues, in my mind, when it comes to trying to use F1 info.

1) Can you trust what these guys are saying? They have a heavy interesest in not telling what they know. Even retired, I am sure there are strong NDA's in place. How much of any given advice is really useful, how much is a partial story and how much is BS to try to throw off competitors?
2) F1 cars follow very different rules than the rest of us. It is not even the same 'physics', if you will (yes I know that physics is physics, but there are huge things happening with an F1 car). Here is an example, that takes us back to a favorite topic - trail braking...

So we have a F1 car that can do 3.5 g's at the corner entry speed but is down to under 2 g's at min corner speed for a given corner. Now let's assume for a street car, that the characteristics of this corner provide very little time difference between trail braking and not. Even in this type of corner, that does not hold true for the F1 car. Because of the huge downforce, you need to carry as much entry speed as possible and take advantage of that grip. Thus you are going to try to take this corner with an early turn in, and make it a decreasing radius. That way you can do as much turning as possible when you have 3.5 g's available, rather than take a line closer to geometric and corner at 1.8 g's.

But if you take that sort of line (like an early apex, except you actually increase the turning and hit the apex) you have to slow down as you turn. And guess what, that sort of line promotes heavy trail braking. So you haul *** into the corner, turn early, trail brake hard and use all those g's at high speed at the expense of being a little tighter and slower at the apex. But in a street car with consistent cornering, the lower apex speed hurts and you have nothing to gain with the trail braking. (Remember I said this is a corner where the characteristics do not make trail braking faster for whatever reason. I am not trying to argue whether or not you should trailbrake but there are some corners where you do not dare because of the characteristics, and some that clearly would offer no advantage.)

If you think about this huge g capability shift, it makes me wonder if the F1 line and style is anywhere even close to what I want to do in a 'mortal' car. Now with the huge grip chages, it may not make much sense to even look at speeds in a corner for an F1 car. Maybe it works much better with g's.

But I am not driving an F1 car or anything really close to it. I don't have 3 data engineers to determine what the cars capabilities are at each and every corner, I lack the test time to gather that data, AND I LACK THE DRIVING TALENT TO GET THE CONSISTENT LAPS, AT THE LIMIT, GO EVEN TRY TO GET THAT DATA.

So that is an awfully long winded response - sorry, I got on a roll.
How about starting with a couple of questions?
1) Would you agree or disagree that in order to use g data in corners, you have to know the limit of the car in that corner?
2) If you agree with that, then how do I get that data for my car?

Geo

I don't believe for a second that F1 teams use g data for driver development.

Some engineer: "Michael, you're not generating enough g's."

MS: "Good, you go out there and generate more."

Some engineer: "No, seriously Michael, you're under performing."

MS: "Where's Jean? We need a replacement for this guy."

I still maintain it's trivia. So what? It's like saying "I turned a 64.8 at my last autocross." So what? Is that good or bad? Where can I improve?

Ah, that's the good question, "where can I improve?" That is the only thing that matters. The rest is simply the result. I'd bet my next paycheck that F1 teams don't look at g plots, at least not at the track (they probably do for programming their 7 post shaker rigs). They are going to look at all the things your coach did (and more) looking for something specific they think they can improve upon. Once again, they will compare with a teammate or a prior run.

As my buddy who is a DA professional said, teams don't look at FC plots. They expect their driver to be on the edge already. They look for the small things they can improve upon.

Let's look at it another way. If the 7 post shaker rig and various other calculations say MS should pull certain g's and he does, do you think they sit in the paddock and say "Mikey, another good run. Jolly good. Nothing to improve. Keep it up lad."

ColorChange

OK, an actual discussion. Me likey.

OK, the practical way to establish the limit is driver near it and occasionally past it. Yes, F1 can model the hell out of it and know awfully close in advance. We could do this but it is time and money prohibitive. But, what we can do, is drive near the limit many times and after collecting enough data, you will be able to see where you had too much slip (exceeded the limit), and therefore accurately identified the limit on that line, at that point in the track. By doing this enough, you get a real accurate picture of the fc on a given turn. Again, the limit does not change dramatically from lap to lap normally so this is an accurate enough and fully usable approach.

1. OK in general you could be right about throwing up chaff, but not in this case. No one is going to be able to snow another F1 team about a technical issue. Also, Peter is not associated with Ferrari and to spew inaccuracies would harm his reputation immensely. This book is widely regarded as the most revealing and accurate F1 book ever.
2. No, physics is (are?) physics, and maximizing g-sum on the best line is the fastest always. This is as absolute as it gets. You are correct about the F1 car taking a different line then our cars, but both are still maximizing g-sum on their ideal line. I agree with you that the ideal lines are quite different, as primarily controlled by lat g limit and hp to weight ratio. Mark, send me your data in an excel file, speed, rpm, lat g, long g at a minimum (throttle and brake pressure preferred) and I will do some free analysis and share the results with the ‘”class”. You may be closer to the fc than you thought (10 hz should be fine.

1. Yes, but this can be determined empirically (as discussed above), and usually does not change dramatically unless the turn is banked, etc.
2. See explanation above. Drive near the limit and collect the data. It really does work. I have data from my cars to show as it would be easier than your car as we may have to correct for aero effects and I’m not particularly good at that (yet).

924RACR

Yes, those sensors are ugly expensive... since the goal (for driver instruction) is to evaluate the brake apply (and consistency and smoothness) - would not a cheaper method be to simply use a string-pot or the like on the brake pedal? I mean, we don't really care about the actual pressure so much as how well you're putting it down, don't we? Unless you're looking at the efficiency of the ABS system, who cares what pressure's at the wheel? Rather, give me wheel speeds, that's the right way to calibrate the proportioning and driver!

OTOH, once the driver's trained, just the one driver axle wheel speed should be adequate. That's all I'm using now. But I do need to add a string pot for steering angle, definitely. Maybe a throttle position sensor wouldn't be bad, too. Still cheesed off that the AIM system is so hard to get sensor connectors for, though.

Yaw measurement - note that it's yaw rate, not actual vehicle body yaw angle. That requires some rather expensive equipment to measure (and calculate), not conducive to the racing environment. If you're familiar with ESP (electronic stability program) systems, like Porsche Stability Management (PMS ) you'll know that they use a combo sensor, usually mounted close to the CG of the car, for the control. This sensor is a combination lateral G and yaw rate sensor (micromachined). It could be obtained and installed, then the sensor output plugged into the data system.

Of course, no need to pay Porsche dealer prices. I'd recommend getting the Bosch combo sensor for, say, a late-model Olds Intrigue (2000-onward, I think) with PCS - what Olds called ESP. Or go for the same sensor for an '04 Pontiac Grand Prix GTP Comp G with Stabilitrak. I have no idea what they'd cost, but that's what I have and intend to use, and I'm sure it's cheaper from a GM dealer than from the Porsche dealer! You could install it sideways and thereby have a long. g-sensor as well as the yaw rate. There are some sensor calibration issues, with temps, etc., that you'd have to work around, but that could be handled after the fact.

So you could record the actual yaw rate of the vehicle and evaluate it for smoothness, etc... you could even compare it to a theoretical yaw rate based on vehicle speed, steering input, Ackerman model, etc... and then you've pretty much got an ESP system in the making!

Should be pretty cheap and easy to do, when you already have a data system. Hard part is convincing the dealer parts guy to order a part for a car you don't own?

Of course, if you really wanted to have fun, you'd plug that analog input into your AIM dash warning lights and set it to go off when you exceed your given tolerable thresholds... sorta like a "YER F*CKED!!!" warning "OH $HIT" light! LOL!

carreracup21

CC assuming you can accurately identify the limit by retrospective analysis of g data plots, after in fact driving at the limit on the proper line, what then is the point of the DAS in the first place ? I mean this retrospective analysis of the limit seems useless to me, just an academic exercise. You have to drive the car and be able to "feel it" real time. I'm with Sunday on this one as it seems like what he is saying is DAS can help you recognize specific issues with shock set-up which will allow you to fine tune the suspension to get more g and raise the limits. Now that would be a huge advantage for us amatuers because the ability to fine tune a cars suspension set-up with precision is what really seperates the men from the boys in racing. I think it's the kind of thing only truly top drivers can feel. Finding g cornering limits when you are actually driving the car, only requires the use your own internal g meter.... your rear end, inner ear or whatever.

SundayDriver

I don't think a string pot will do much good for brakes unless all you want is on/off. Otherwise the brake force is not really related to travel but to brake pedal pressure. The wheel sensors would certainly provide valuable data. In my case, I am very interested in brake bias so the F/R pressures are needed for that. My bias is actaully a math channel where the pressures are adjusted based on piston sizes and bias represents a ratio of clamping force, not pressure. However, I am not clear how jsut wheel speeds get you much except detection of locking a wheel. But up to that point, what do you learn about braking from wheel speed?

Yaw stuff sound interesting. Maybe a winter project. Hmmmm

SundayDriver

I can assemble some data after the runoffs. Until then, I am a bit too pressed for time to go through that. Sounds like a good off-season excercise.

Where I really disagree here is the ability to establish what the limit is for a car/corner. As amateurs, we are vere limited as to how much time we can get on a track and can not get enough to ensure track conditions are the same between different visits. Even if we could, how do you establish what the limit is for a given corner? My system does not tell me what my slip angles are. The steering angle seems too dynamic to relate that to slip. There is an internal calculation of oversteer but there is huge debate amongst engineers as to how to calculate that and what the number really means.

As far as skidpad data, it just isn't going to work. If you only run tracks that are flat and have 'ordinary' corners, it will tell you something. As I recall, Gingerman fits that description. But let's try Mid Ohio in terms of skidpad data and known corner radius.

T1 - Might work but it has quirks that pretty much demand full throttle at turn in and through the corner. Also concrete and asphalt = changing grip during the corner (which is true in many of the corners).
T2/T3 - The line is not the geometric line - you compormise an apex to get a better angle of entry at T4.
T4/T5 (Keyhole) - Ain't no way to predict that one. Huge drop, some off camber and grip changes. True speed there will always be way less than 'calcualted'.
T6 - Not a corner
T7 - Again, maybe on this one except for surface changes.
T8 - Nope - big crest near the apex and that is what limits the corner speed. You compromise the line to max T7 and T9.
T9 - Probably not as it induces lots of understeer so you trail brake to get the car to turn.
T10/10a - Nope - line is again compromised for angle of entry to T11.
T11 - Another big crest past the apex.
T12 - Not a turn
T13 - Maybe, just like T1
T14 - Nope - Crest at entry so car is off balance as you enter.
T15 - Probably work here.

Most tracks are in between Gingerman and Mid Ohio (one of the least technical and one of the most) but I think your lack of experience with different tracks is biasing your view.

SundayDriver

I'll share a great story about a "YER F*CKED" warning. A certain DSR driver could take T10 at Sears flat out but only if there was no tail wind. A tail wind is a very bad thing in an aero car - it takes away downforce and makes you a little bit faster as well. The data system was set up using a pitot tube and privided a warning if there was any tail wind, based on pitot tube compared to wheel speed. The message was actually set up to recognize the approach of T10 and said either "Go For It" or some other message.

SundayDriver

Unfortunately THE most valuable thing you can do with DAS is not being discussed here. That is getting a faster driver to lay down some data and overlaying that with your data. There is no amount of analysis that can provide knowledge anywhere close to what you can learn from that.

RJay

At the risk of confirming that I'm a simpleton, it seems to me that in the end only one thing really matters: lap time. The tried and true method of rev measuring on corner exit seems far simpler and given that this technique can be used in car realtime to determine the optimal line seems far more useful as well. After all, if G-loading was the most critical indicator of performance wouldn't those digital bars on Schumi's steering wheel indicate G-Force instead of revs?

944TURBOS

Alright I was also arguing with CC about many things and this weekend I worked very closely with AIM and tried to work with G sum and a couple other DAS tools and I agree that G sum really does help in many situations. I am not going to get into it. But i saw lap times drop trying to maximize my G's. And their theories and coaching really works. I am a faster driver from getting coached by them, so I also give in

924RACR

LOL... I like the sounds of that (warning light)...

Wheel speeds can tell you a lot as compared to vehicle speed if you can look at them quick enough; naturally low-res systems like our 10Hz AIM systems are not quick enough, but I seem to recall the higher-resolution systems intended for chassis work do have the resolution. Then it's just a matter of having the sensors. You can watch the wheels fall off into deep slip and lock. From that you can determine an optimal operating point. Then it's up to the driver to hold the wheels at that slip level - which can either be calculated or just visually estimated. Since the rears usually aren't braked at the same degree of slip as the fronts, they can be used as a reference. Obviously you're not going to have a true vehicle speed to reference, but a visual estimation will suffice. More importantly, for the purpose of driver training, what you'll be able to review and critique will be the consistency of wheel slip, how quickly the driver gets to and how well they maintain optimal wheel slip and max decel when needed.

I have to admit, and this comes from my ignorance in not having an adjustable prop valve, I'm surprised that you'd be so interested in having such detailed proportioning data in your plots... I would think that in-car feedback via gauges would be of more use, to actually get the car dialed in while driving - since the proportioning needs would likely change enough from day to day or session to session to require possible regular adjustment. I mean, since you would presumably be going for the ragged edge of too much rear pressure. Wouldn't having that captured on the DAS simply be a check for after the fact, or something to log and keep track of? How does that factor into car or driver setup - especially without four wheel speeds to keep track of rear wheel lock? Or is the one rear sensor on the drive axle adequate? You don't run a spool, do you? Is that proportioning data needed as an input for the shock tuning?

DJ

Vaughn,

Re: "I have to admit, and this comes from my ignorance in not having an adjustable prop valve, I'm surprised that you'd be so interested in having such detailed proportioning data in your plots... I would think that in-car feedback via gauges would be of more use"

Mark can elaborate more when he has time, but I don't think real-time guages would be any good in his car because it brakes so hard, and drops speed so quickly. There'd be no time to look at the guages. Only 1,000 lbs., on super-soft slicks with a ton of downforce (well, actually, probably only about a half ton) makes for a really unbelievable ability to shed speed. Combine that with >2.0 lateral g ability in the corners (and hence higher corner entry speeds, and thus the need to bleed off a lot less speed) and you've got extremely short braking zones. Mark's car is a lot lighter than mine, and stops much better (I've never driven his car as I don't fit, so I'm not absolutely sure) but I know that my car compared to my previous 993 track car on slicks, my braking zones are far less than half the distance than those in the 993. Probably around 1/3. There's no time to look at guages.

924RACR

Ah, yes, good point, forgot about that... Duh! Figures, since I'm not playing with a prop valve!