Data Logging

[djr81]

I do some fairly basic data logging for my car at the track. I use some maths function which I have written in Excel to help analyse the data. I guess my question is can anyone offer some more ideas as to how to draw more information out of the otherwise impenetrable squiggles on the screen?

I have done the usual things:

Rpm versus speed (at 100% TPS) to determine where the engine spends it working life.

% of time versus negative gees to show how crap my braking is.

Terminal velocity for various straights etc.

Done the overlay thing to see where I am dumping time on the slower laps. (see .pdf file)

Graphed the X-Y function to check the effectiveness of corner entry & exit.

Used the logger to determine which of the corner entry options is quickest.

Also spent time looking at squiggles to determine if there is too much understeer in the setup or too much wheelspin.

But I can't help but feel most of the "understanding" is only there because I was driving the car & intuitively knew what was going on anyway.

So can anyone help point me in a direction to further my understanding?

lap_comparison.PDF

[Sydneykid]

First off, what reference books do you have? My data favourite is Competition Car Data Logging by Simon McBeath followed by Data Power by Buddy Fey. If you haven't read both of those, then I strongly suggest that you do. My guess is there are only a handfull of guys on SAU that will have any idea what we are talking about, but here goes;

The first thing I look at is the TPS, I want to see 100% as much as possible, that's maximum acceleration. If it's not 100% then it should be 0%. On the transition from 0% ot 100% I like to see a smooth progession on corner exit. Jerky, on, off throttle indicates that the driver is compensating for something. What? Understeer or oversteer? If you have it logged, look at the steering angle. If not, then wheel speed and lateral G force.................

Hang on......I am going to ask more questions than give answers.................what have you got logged?

Cheers

Gary

[Roy]

I was thinking about changing my oil pressure channel to throttle position. Ignorance is bliss with regards to the oil pressure the poor old RB20 is running at teh track

Only with my turbo its pretty easy to guesstimate when i pick the throttle up by looking at manifold pressure and revs. I did actually think i ouwl be better off using that channel for steering angle, but the sensors cost teh earth Thats not the case with the things i am logging at the moment. And throttle position would be pretty easy/cheap to do

[djr81]

Thanks for the reply, Gary.

The problem I have with most of the books is that they spend an inordinate amount of time banging on about how to install data loggers & alot of basic stuff that because of my work (Mechanical Engineer) I pretty much already understand from uni days.

As for logging, at the moment it is pretty much just the basics:

Lateral, longitudinal gees & RPM in one unit as shown by the graph, above.

I run a camera on the back shelf so I can check for steering input smoothness & general car behaviour.

I have the Ruzic attessa controller so can log (or will be able to when it gets wired up) the 4WD function.

Plus a Power FC so just about any of the engine parameters are accessable via datalogit

Between that lot most of the inputs can be accessed, but not necessarilly in one application. Unfortunately I don't have a GPS system which sucks although a mate has a 5Hz phone logger which I can pinch I suppose.

To use the case of understeer as an example obviously you get square waveforms (for want of a better word) on the lateral plots. Which basically shows you having to hold lateral acceleration for too long on corner exit - in other words not trading off lateral for longitudinal grip as should be done. This shows up on the X-Y plot too. But I guess I am whittering on now.

Questions are two fold:

Do any books move much beyond the basics & focus on the interpretation of the data. As an example check the X_Y plot. Obviously the quicker time (This is a scatter plot of the chart at the top of the thread) shows a fleshing out of the braking plus turn in component & a happier car through the right hand sweeper - something that helped lap times. Pretty basic stuff, do the books go much further than that?

Are their any maths functions you would recommend that may illustrate setup issues beyond what may otherwise be intuitively obvious?

scatter_plot.PDF

Edited May 23, 2008 by djr81

[salad]

Roy: Have you thought about using a linear sensor on the rack? We've found that to be cheaper and still gives decent data, just have to do a bit of maths in the software to convert from rack travel to steering angle.

[Duncan]

what about using the output from the hicas sender at the steering wheel?

[Sydneykid]

Sorry I haven't been back to this topic for a few days, I was at Eastern Creek all weekend. Lots of cars running that needed input.

By looking for chassis set up data you are tackling the hardest to diagnose with the limited data you are logging. It's the last thing I try and work on from data alone. My order for getting more speed out of the driver and car are;

1. Engine rpm, throttle position and LongG for maximising the acceleration. Look for the rpm range where the engine produces the highest postitive LongG and make sure the driver knows to keep the engine in that range whenever he is at 100 WOT.

2. Engine rpm and LongG for gear change rpm optimisation. Make sure the upchanges are done at the rpm that produces the highest LongG. Plus gear change speed is obvious from the rpm and TPS, work at shortening it as much as possible.

3. Transition from acceleration to braking, look at reducing the time spent in limbo between 100% WOT and maximum negative LongG. No dallying, get of the throttle and straight onto the brake. Do some straight line tests and determin what the maximum G under brakes is for that car set up, then aim to achieve that maximum on every stop.

4. Speed, KPH or wheel rpm is the most often used data. If I only logged one parameter it would be KPH. It's obvious why, increasing the KPH lowers the lap times. This is really #1 but I fnd it easier to work with the driver and maximise 1,2 and 3 above and then work on speed. I navigate my way around the track straight by straight, corner by corner looking for places to increase the KPH. Obviously acceleration and braking are fairly well covered by the above, so this part mostly involves increasing the corner speed. The ability of the software to select the potential best lap time from multiple laps is a feature I use a lot. Either automatically or manually cut and past the sectors. I don't really care if the car is oversteering or understeering, I am looking to simply increase the corner speed, generally raise the LatG. I tackle it in one corner at a time, the most important corners (the ones leading onto straights) being first priority.

I sit with the driver and look at as many laps as we have available, why was the speed in corner 4 faster on this lap than that lap. Look at the data and talk to the driver, higher LatG, less steering input, earlier throttle application, later or earlier braking, whatever made this lap better. Move on to the next corner, do the same.

Since you are both the driver and data analyst you have some advantages which should outway the dissadvantages. What I find works for me is to sit in the car immediately after a session has finished and make some quick notes, before I look at the data. I know what sectors the lap is divided into by the logging software, so I make my notes the same way. Or it can simply be be Corner 1, braking, entry, mid corner and exit, corner 2 etc. For novice drivers I have a prompter sheet pre printed, with some boxes to tick and space for comments on each straight and corner.

Then I have 2 semi independant records, the written notes of the driver straight after the session and the data to refer to. On a busy day, I might not get to look at each session individually, so the written notes help at the end of the day or a few days later when I have time to look closely at the data. If I am driver, mechanic, race engineer and data analyst all in one, I tend to find driver improvements are done during the day, as as are important/easy set up changes. Major/ time consuming set up changes are generally left until the car is back in the workshop. If I have a crew, then we may do some major set up changes during the day. But that's pretty rare, due to the previously gathered data I have a pretty good book of what set up works at what track. So we only do minor tuning on the day.

The logging software I use has provision for notes and set up data for each session, I either use that to record what we had in the car that session, tyre pressures, swaybar settings, camber, caster, toe, tyre temperatures if we recorded them etc, everything goes into the log. You can't have too much data.

Cheers

Gary

[djr81]

1. Engine rpm, throttle position and LongG for maximising the acceleration. Look for the rpm range where the engine produces the highest postitive LongG and make sure the driver knows to keep the engine in that range whenever he is at 100 WOT.

Um I usually do this via the engine dyno print out. Basically you plot the rear wheel torque in each gear on the y axis and the road speed on the x axis. It gives you a series of cascades which allows you to pick the gear change points in road speed & then average and transfer them to rpm. In the case of a Gt-R it just shows you need to rev the berries out of the motor in every gear - well mine does anyway.

2. Engine rpm and LongG for gear change rpm optimisation. Make sure the upchanges are done at the rpm that produces the highest LongG. Plus gear change speed is obvious from the rpm and TPS, work at shortening it as much as possible.

Mmm quick gear changes = unhappy synchros. Just the price you pay, I guess.

3. Transition from acceleration to braking, look at reducing the time spent in limbo between 100% WOT and maximum negative LongG. No dallying, get of the throttle and straight onto the brake. Do some straight line tests and determin what the maximum G under brakes is for that car set up, then aim to achieve that maximum on every stop.

Max negative gees is slightly over one. In other words my tyres have more grip than the brakes have stopping power.

4. Speed, KPH or wheel rpm is the most often used data. If I only logged one parameter it would be KPH. It's obvious why, increasing the KPH lowers the lap times. This is really #1 but I fnd it easier to work with the driver and maximise 1,2 and 3 above and then work on speed. I navigate my way around the track straight by straight, corner by corner looking for places to increase the KPH. Obviously acceleration and braking are fairly well covered by the above, so this part mostly involves increasing the corner speed. The ability of the software to select the potential best lap time from multiple laps is a feature I use a lot. Either automatically or manually cut and past the sectors. I don't really care if the car is oversteering or understeering, I am looking to simply increase the corner speed, generally raise the LatG. I tackle it in one corner at a time, the most important corners (the ones leading onto straights) being first priority.

I sit with the driver and look at as many laps as we have available, why was the speed in corner 4 faster on this lap than that lap. Look at the data and talk to the driver, higher LatG, less steering input, earlier throttle application, later or earlier braking, whatever made this lap better. Move on to the next corner, do the same.

Since you are both the driver and data analyst you have some advantages which should outway the dissadvantages. What I find works for me is to sit in the car immediately after a session has finished and make some quick notes, before I look at the data. I know what sectors the lap is divided into by the logging software, so I make my notes the same way. Or it can simply be be Corner 1, braking, entry, mid corner and exit, corner 2 etc. For novice drivers I have a prompter sheet pre printed, with some boxes to tick and space for comments on each straight and corner.

Then I have 2 semi independant records, the written notes of the driver straight after the session and the data to refer to. On a busy day, I might not get to look at each session individually, so the written notes help at the end of the day or a few days later when I have time to look closely at the data. If I am driver, mechanic, race engineer and data analyst all in one, I tend to find driver improvements are done during the day, as as are important/easy set up changes. Major/ time consuming set up changes are generally left until the car is back in the workshop. If I have a crew, then we may do some major set up changes during the day. But that's pretty rare, due to the previously gathered data I have a pretty good book of what set up works at what track. So we only do minor tuning on the day.

The logging software I use has provision for notes and set up data for each session, I either use that to record what we had in the car that session, tyre pressures, swaybar settings, camber, caster, toe, tyre temperatures if we recorded them etc, everything goes into the log. You can't have too much data.

I find myself being chief sponsor, driver, engineer, gofer, brollie dollie, mechanic, crew chief, owner.....and basically not having enough time on a track day to think things through properly. Combined with a dodgy memory from old age & infirmary the tick sheet/prompt sounds like a good idea. Otherwise I am forced to rely on the inference that ending upside down in the sand trap means the latest brilliant idea didn't quite work.

From what you have said I will go away and try & sort out a more efficent way of overlaying segments of laps.

With regard to lateral gees is there a particular part of the corner sequence where most time can be gained. Obviously the corner exit is up there, but I was more thinking of where to go after you have found all the grip, ie finding more grip more often rather than just peaking the thing at 1.25 gees or whatever.

[Sydneykid]

Um I usually do this via the engine dyno print out. Basically you plot the rear wheel torque in each gear on the y axis and the road speed on the x axis. It gives you a series of cascades which allows you to pick the gear change points in road speed & then average and transfer them to rpm. In the case of a Gt-R it just shows you need to rev the berries out of the motor in every gear - well mine does anyway.

Mmm quick gear changes = unhappy synchros. Just the price you pay, I guess.

Max negative gees is slightly over one. In other words my tyres have more grip than the brakes have stopping power.

I find myself being chief sponsor, driver, engineer, gofer, brollie dollie, mechanic, crew chief, owner.....and basically not having enough time on a track day to think things through properly. Combined with a dodgy memory from old age & infirmary the tick sheet/prompt sounds like a good idea. Otherwise I am forced to rely on the inference that ending upside down in the sand trap means the latest brilliant idea didn't quite work.

From what you have said I will go away and try & sort out a more efficent way of overlaying segments of laps.

With regard to lateral gees is there a particular part of the corner sequence where most time can be gained. Obviously the corner exit is up there, but I was more thinking of where to go after you have found all the grip, ie finding more grip more often rather than just peaking the thing at 1.25 gees or whatever.

For reasons that a smarter person than me could workout, I have found that maximising the LongG doesn't align with what the dyno says. It's closer with an N/A engine but with a boosted engine there can be some noticeable variations between gears. I put this down to the turbo response to the differing loads that each gear provides. So these days I definitely rely on the LongG for gear change points, plus what gear is the best to use. Personally I get caught up in the rpm thing, more rpm = more noise so I think the car is going faster, when an upchange may well give higher LongG (more acceleration). That age old question, "is it faster using 2nd gear at 6,500 rpm out of turn 2 or 3rd gear at 5,400 rpm?" Maybe there is more wheelspin in 2nd, so the actual LongG is higher in 3rd even though the horsepower is lower. It takes the guess work out of what gear to use where on the circuit.

Do you subscride to Race Car Engineering? One of the things I like about the current format is the section on practical data logging that is in every issue. For example, in the current issue there is an interesting practical example of using a derived maths channel for examining understeer and oversteer by overlaying the speed corrected steering angle with the LatG. Well worth the monthly investment.

Brakes wise, so you can't activate the ABS even with really big push on the brake pedal? That's unusual, I can't say as I have ever had an R32/33/34 that I can't get that level of braking from. Maybe only once, not lap after lap, but once is all it takes to get an idea of the maximum LongG possible.

Cheers

Gary

[djr81]

I usually get racetech, but unfortunately that has degenerated into what the component suppliers have got new to sell this month. I used to get racecar eng but it disappeared from the shelves at the newsagent ages ago & I have not seen it since. I will have to see what I can find.

Not sure quite what you meen by speed corrected steering angle. I did some maths work on determining the rate of change of yaw angle in the car to ensure the car was turning less as it exited the corner. That may be similar but it would perhaps be more useful plotted against distance or speed rather than time. See the attached.

As for the brakes, well when the tyres are warm I can't get the ABS to activate. Occasionally it will do it over the bumps at Collie, but generally no chance. The most I ever saw was 1.1 gees which was during a holy crap moment when a Cobra infront of me didn't check his mirrors.

Long_circuit_with_yaw_rate.PDF

Edited May 26, 2008 by djr81

[Sydneykid]

I usually get racetech, but unfortunately that has degenerated into what the component suppliers have got new to sell this month. I used to get racecar eng but it disappeared from the shelves at the newsagent ages ago & I have not seen it since. I will have to see what I can find.

Not sure quite what you meen by speed corrected steering angle. I did some maths work on determining the rate of change of yaw angle in the car to ensure the car was turning less as it exited the corner. That may be similar but it would perhaps be more useful plotted against distance or speed rather than time. See the attached.

As for the brakes, well when the tyres are warm I can't get the ABS to activate. Occasionally it will do it over the bumps at Collie, but generally no chance. The most I ever saw was 1.1 gees which was during a holy crap moment when a Cobra infront of me didn't check his mirrors.

Is that the stand out 1.1'ish neg G on the 51.49 lap? Still 1.1 G should be the target, if the braking capacity is there for "holy crap" moment then you can do it every time, braking performance deterioration not withstanding. I notice that there is a lot of braking at ~0.7 G, the capacity is there to pull 1.1 G, what is it that prevents you from using the max every time? I know I do it, my brain says the car doesn't need to slow down that fast. I know the route to faster lap times is to spend longer acccelerating and then brake harder, use the maximum capacity. But the signal from the brain to the right foot somehow gets lost.

I only glanced at the current RE this morning on the way out, I'll have more of a read tonight, after 3 of 4 readings I generaly get what they are on about.

Cheers

Gary

[djr81]

The -1.1 wasn't on the 51.49 lap. It was a whole other seat trim puckering experience. The most I usually see on a "normal" lap is 1.05 or thereabouts.

My bad braking is basically a combination of a lack of talent, bumps and creeping up on the braking markers during the day. I have to try & convince myself I can grab an extra yard or so before hitting the anchors.

Plus the brakes go away because they are too small anyway.

The pedal effort is very high in the Skylines for some reason. Compared to my daily driver Foulcan it is rediculous.