ZPOST
BMW Garage BMW Meets Register Search Today's Posts Mark Forums Read


Go Back   ZPOST > BMW Z4 Technical Talk > Track / Autocross / Dragstrip / Driving Techniques
  TireRack

SUPPORT ZPOST BY DOING YOUR TIRERACK SHOPPING FROM THIS BANNER, THANKS!
Post Reply
 
Thread Tools Search this Thread
      12-19-2011, 12:19 PM   #111
Shipkiller
Colonel
Shipkiller's Avatar
United_States
187
Rep
2,138
Posts

Drives: Me
Join Date: Oct 2007
Location: Virginia Beach

iTrader: (3)

Garage List
Frank,

I left foot brake, and trail brake....
Appreciate 0
      12-19-2011, 07:32 PM   #112
MFGJR
First Lieutenant
27
Rep
341
Posts

Drives: 2007 Z4 M Coupe
Join Date: Sep 2008
Location: Richmond, VA

iTrader: (0)

Quote:
Originally Posted by Shipkiller View Post
Frank,

I left foot brake, and trail brake....
No problemo with trail braking, as that's still just one pedal at a time.

When you left-foot brake, are you actually applying the brakes and throttle at the same time, or doing it to minimize transition time between brakes and throttle but not applying both simultaneously. If the latter, then it'll still work fine. If the former, the math channel will return the net value of throttle and brakes at that moment, which'll be pretty meaningless.
__________________
Appreciate 0
      12-19-2011, 08:04 PM   #113
Shipkiller
Colonel
Shipkiller's Avatar
United_States
187
Rep
2,138
Posts

Drives: Me
Join Date: Oct 2007
Location: Virginia Beach

iTrader: (3)

Garage List
At the beginning, it is just one peddle at a time. At the apex, start to release the brake and apply throttle. Many times I have to do this to transfer weight to the rear (the front is already set) to keep the rear planted...
Appreciate 0
      12-19-2011, 08:07 PM   #114
Matt Romanowski
New Member
United_States
0
Rep
13
Posts

Drives: Porsche
Join Date: Nov 2011
Location: Manchester, NH

iTrader: (0)

Reading quickly, you may be able to nest another if/then function (if RS2 allows it) or a greater than to make it choose only one input.
Appreciate 0
      12-19-2011, 10:08 PM   #115
MFGJR
First Lieutenant
27
Rep
341
Posts

Drives: 2007 Z4 M Coupe
Join Date: Sep 2008
Location: Richmond, VA

iTrader: (0)

Quote:
Originally Posted by Shipkiller View Post
At the beginning, it is just one peddle at a time. At the apex, start to release the brake and apply throttle. Many times I have to do this to transfer weight to the rear (the front is already set) to keep the rear planted...
So you're doing it to smooth the transition from braking to accelerating, and make the resultant weight transfer subtle, yes?

Quote:
Originally Posted by Matt Romanowski View Post
Reading quickly, you may be able to nest another if/then function (if RS2 allows it) or a greater than to make it choose only one input.
Yes, we should be able to nest a couple more if-then-elses in there, keying off both bakes and throttle being greater than 0, but the question is which input would we want to select in this situation? Throttle or brakes? I'd say use the brake data, and then create yet another math channel that's only throttle position, but with the downshift blips suppressed as per above. Then, that new trace could be overlayeded on the same graph as he combined throttle-brakes trace; most of the time, they're exactly the same line (with a little "clutter" at zero). When doing the 'Shipkiller Techniqure", however, you'd be able to see each input separately.

The only problem here is that it wouldn't work well when presenting the 'pedal inputs" in map format using Track Report view. You'd want the combined channel to have, I think, the throttle input instead of brakes there. Of course, you could make yet another combined math channel just for the Track Report view...

Any of this make any sense?
__________________
Appreciate 0
      12-19-2011, 10:24 PM   #116
Matt Romanowski
New Member
United_States
0
Rep
13
Posts

Drives: Porsche
Join Date: Nov 2011
Location: Manchester, NH

iTrader: (0)

Quote:
Originally Posted by MFGJR View Post
Yes, we should be able to nest a couple more if-then-elses in there, keying off both bakes and throttle being greater than 0, but the question is which input would we want to select in this situation? Throttle or brakes? I'd say use the brake data, and then create yet another math channel that's only throttle position, but with the downshift blips suppressed as per above. Then, that new trace could be overlayeded on the same graph as he combined throttle-brakes trace; most of the time, they're exactly the same line (with a little "clutter" at zero). When doing the 'Shipkiller Techniqure", however, you'd be able to see each input separately.

The only problem here is that it wouldn't work well when presenting the 'pedal inputs" in map format using Track Report view. You'd want the combined channel to have, I think, the throttle input instead of brakes there. Of course, you could make yet another combined math channel just for the Track Report view...

Any of this make any sense?
You could do it where to default to which ever input is great as that is doing the most i.e. if the driver is giving it 40% throttle and only 10% brake, then the car is still accellerating.

I guess you have to step back and say what are we really trying to see? If we want to see the major thing the driver is doing, then we just need whichever input is greater. That is an easier problem to solve. If we want to see all inputs, then we have to figure something different out. I'm not sure the best way to do that.
Appreciate 0
      12-19-2011, 11:26 PM   #117
RandysV10
MPOWERD
RandysV10's Avatar
15
Rep
338
Posts

Drives: 2010 M6
Join Date: Nov 2010
Location: Cloud Nine

iTrader: (0)

Garage List
2006 Z4M  [8.00]
2010 BMW M6  [9.00]
Sunday turned out to be a great day. Had tons of track time with no traffic, weather couldn't have been better for December. There were times that I had the whole course to myself. Had issues with my in-car camera mount so had to go with the front bumper mount. Here is some video from my 5th run of the day.

Appreciate 0
      12-20-2011, 07:32 AM   #118
MFGJR
First Lieutenant
27
Rep
341
Posts

Drives: 2007 Z4 M Coupe
Join Date: Sep 2008
Location: Richmond, VA

iTrader: (0)

Quote:
Originally Posted by Matt Romanowski View Post
You could do it where to default to which ever input is great as that is doing the most i.e. if the driver is giving it 40% throttle and only 10% brake, then the car is still accellerating.

I guess you have to step back and say what are we really trying to see? If we want to see the major thing the driver is doing, then we just need whichever input is greater. That is an easier problem to solve. If we want to see all inputs, then we have to figure something different out. I'm not sure the best way to do that.
I like picking the dominant input when both throttle and brakes are applied outside of the downshift scenario, but better yet let's use longitudinal accel as the arbiter. I'm not sure that comparing the gross values of throttle and brakes will actually identify the dominant input--25% brakes probably trumps 30% throttle, for example--but long accel will. If the car is decelerating beyond a nominal threshold (long acel < say, 0.1), then the brake input is dominant. Otherwise, throttle is dominant. We need the threshold before picking brakes to allow for lifting the throttle, in which case I think that what’s going on with the throttle is the information we’re most interested in. I’ll look at some of Randy’s data to find an appropriate threshold.

I’ll work on constructing the formula, which will get a little bit complex, but testing is going to be the issue. Right now, my left foot is programmed in on-off fashion for the clutch, and so I’m a lousy left-foot braker…

Quote:
Originally Posted by RandysV10 View Post
Sunday turned out to be a great day. Had tons of track time with no traffic, weather couldn't have been better for December. There were times that I had the whole course to myself. Had issues with my in-car camera mount so had to go with the front bumper mount. Here is some video from my 5th run of the day.

Found four seconds! Awesome! And I’m jealous of all that open track. I don’t think I saw another car in your video. Are you going to post some new data up on TrailBrake? Having some comparative data—same car/driver/track would be interesting for folks; we can see where the four seconds came from.
__________________

Last edited by MFGJR; 12-20-2011 at 01:26 PM.. Reason: typo
Appreciate 0
      12-20-2011, 07:35 AM   #119
Matt Romanowski
New Member
United_States
0
Rep
13
Posts

Drives: Porsche
Join Date: Nov 2011
Location: Manchester, NH

iTrader: (0)

The standard I've seen for choosing whether someone is on the brakes or not using Long G is .25 G. I would guess something around that would be good here too.
Appreciate 0
      12-20-2011, 08:29 AM   #120
Shipkiller
Colonel
Shipkiller's Avatar
United_States
187
Rep
2,138
Posts

Drives: Me
Join Date: Oct 2007
Location: Virginia Beach

iTrader: (3)

Garage List
Quote:
Originally Posted by MFGJR View Post
So you're doing it to smooth the transition from braking to accelerating, and make the resultant weight transfer subtle, yes?
Partially. Even though I am normally not driving at 10/10's, I still do this for several reasons.

1. It's the faster way around the track
2. Smoother transition, more controlled transfer
3. Carrying more speed into the turn, slowing to the same speed for turn-in, but getting on the power sooner
4. Practice.. practice...practice

You will brake a little later than using the 'one foot' method, but your turning speed is not really any faster. Yea, some but not significant.
As you know this ‘braking a little later’ is relative. Maybe 5-10 feet. If that.


This allows you to be smoother, which results in being faster. I am still a believer in slow in, fast out, but my corner entrance speeds are only about 3mph faster but my exit speeds are now 5-6mph faster

Still learning left foot braking. I find left foot braking in the MC much harder than other cars due to peddle placement.
Appreciate 0
      12-20-2011, 07:31 PM   #121
MFGJR
First Lieutenant
27
Rep
341
Posts

Drives: 2007 Z4 M Coupe
Join Date: Sep 2008
Location: Richmond, VA

iTrader: (0)

Quote:
Originally Posted by Matt Romanowski View Post
You could do it where to default to which ever input is great as that is doing the most i.e. if the driver is giving it 40% throttle and only 10% brake, then the car is still accellerating.

I guess you have to step back and say what are we really trying to see? If we want to see the major thing the driver is doing, then we just need whichever input is greater. That is an easier problem to solve. If we want to see all inputs, then we have to figure something different out. I'm not sure the best way to do that.
Quote:
Originally Posted by MFGJR View Post
I like picking the dominant input when both throttle and brakes are applied outside of the downshift scenario, but better yet let's use longitudinal accel as the arbiter. I'm not sure that comparing the gross values of throttle and brakes will actually identify the dominant input--25% brakes probably trumps 30% throttle, for example--but long accel will. If the car is decelerating beyond a nominal threshold (long acel < say, 0.1), then the brake input is dominant. Otherwise, throttle is dominant. We need the threshold before picking brakes to allow for lifting the throttle, in which case I think that what’s going on with the throttle is the information we’re most interested in. I’ll look at some of Randy’s data to find an appropriate threshold.
Quote:
Originally Posted by Matt Romanowski View Post
The standard I've seen for choosing whether someone is on the brakes or not using Long G is .25 G. I would guess something around that would be good here too.
Here are the rules that I came up with to accommodate heel-toe downshifting and left foot braking. Not quite as simple as what I started with.

If the clutch is in, use brake pressure. If it’s a heel-toe downshift, we want the brake input and this ignores the throttle blip. If it’s a non-heel-toe downshift or an upshift, there’s 0 brake and 0 throttle input—effectively coasting—so the 0 brake input reflects both brake and throttle input.

If the clutch is out, and if the throttle is down, and if the brakes are also down, and if long accel exceeds the braking threshold, use brake pressure. This is the left foot braker giving both throttle and brake inputs, and braking is the dominant input.

If the clutch is out, and if the throttle is down, and if the brakes are also down, and if longitudinal acceleration does not exceed the braking threshold, use throttle position. This is the left foot braker giving both throttle and brake inputs, but the inputs are equal or throttle is dominant. We us the threshold to allow for some manipulation of the throttle as acceleration begins. Note that the threshold is a negative value, so “exceeding” means more negative, i.e., less than the threshold. Looking at Randy’s data, coasting generally resulted in up to -0.16 longitudinal accel, so I’ll use that as the threshold.

If the clutch is out, and if the throttle is down, and if the brakes are not down, use throttle position. Straightforward acceleration.

If the clutch is out, and if the throttle is not down, and if the brakes are down, use brake pressure. Straightforward braking.

If the clutch is out, and if the throttle is not down, and if the brakes are not down, use throttle position. This is coasting, with 0 brake and 0 throttle inputs. The 0 throttle position value will represent both.

When If-THEN-ELSEd, some items above become “THENs” and others ‘ELSEs”. With the statements nested together, it looks like this:

( IF clutch is in , THEN use negative brake pressure , ELSE ( IF throttle is down , THEN ( IF brakes are down , THEN ( IF longitudinal acceleration exceeds the threshold , THEN use brake pressure , ELSE use throttle position ) , ELSE use throttle position ) , ELSE ( IF brakes are down , THEN use brake pressure , ELSE use throttle position ) ) )

Note that wherever we use brake pressure, we’re using the negative of the logged value as we want braking to appear “below the line” on the Measures Graph. In RS2 syntax, and using the channel names from the Z4M ECU protocol, it looks like this:

IF(EQ(Z4M_CLUTCH_SW,1),-Z4M_BRAKE_PR,IF(GT(Z4M_PPS,0),IF(GT(Z4M_BRAKE_PR,0 ),IF(LE(GPS_LonAcc,-0.16),-Z4M_BRAKE_PR,Z4M_PPS),Z4M_PPS),IF(GT(Z4M_BRAKE_PR, 0),-Z4M_BRAKE_PR,Z4M_PPS)))

This exactly replicates my simple “throttle – brakes” approach using Randy’s data, which doesn’t include heel-toeing or left foot braking/simultaneous throttle input. I’ll work out a test protocol involving those techniques, and will see how the formula works with those. Might be after the holidays before I can do that and report back.

My head hurts now. Time for a beer.
__________________
Appreciate 0
      12-20-2011, 07:46 PM   #122
Shipkiller
Colonel
Shipkiller's Avatar
United_States
187
Rep
2,138
Posts

Drives: Me
Join Date: Oct 2007
Location: Virginia Beach

iTrader: (3)

Garage List
Frank,
Just a reminder, the car does a plausibility check when both the throttle and brakes are applied. The more brake pressure is applied, the more the throttle is retarded. I don't know when or what the intervention points are when reached and the ECU intervenes, but I know it will happen.

I don't think it should matter in these cases, but we/you have to be aware of it when reading the data and something is screwy.....
Appreciate 0
      12-20-2011, 07:58 PM   #123
MFGJR
First Lieutenant
27
Rep
341
Posts

Drives: 2007 Z4 M Coupe
Join Date: Sep 2008
Location: Richmond, VA

iTrader: (0)

Quote:
Originally Posted by Shipkiller View Post
Frank,
Just a reminder, the car does a plausibility check when both the throttle and brakes are applied. The more brake pressure is applied, the more the throttle is retarded. I don't know when or what the intervention points are when reached and the ECU intervenes, but I know it will happen.

I don't think it should matter in these cases, but we/you have to be aware of it when reading the data and something is screwy.....
Good thinking, thanks for the reminder.

IIRC duration is a factor in the car's programming, mayybe allowing a couple seconds of simultaneous input. Note that we can still blip the throttle even with mega-pressure on the brakes. Also, I have an extra-wide throttle pedal on my car, and when other guys have auto-Xd it, they frequently hit the gas when braking, and the car allows a good bit of it, even with the clutch out.

I'll keep it in mind when I look at some test data, see if I can figure out how it handles the situation.
__________________

Last edited by MFGJR; 12-20-2011 at 08:18 PM..
Appreciate 0
      12-20-2011, 08:15 PM   #124
Matt Romanowski
New Member
United_States
0
Rep
13
Posts

Drives: Porsche
Join Date: Nov 2011
Location: Manchester, NH

iTrader: (0)

You'll know when the ECU cuts things becuase of LFB. I have a couple of friends who ran a 3 series in the Continental Series and had to shut the car off and on about 4 times per lap at Watkins Glen becuase the ECU went into limp mode. Had the same problem at Trois Rivers too.
Appreciate 0
      01-02-2012, 08:06 PM   #125
MFGJR
First Lieutenant
27
Rep
341
Posts

Drives: 2007 Z4 M Coupe
Join Date: Sep 2008
Location: Richmond, VA

iTrader: (0)

Let’s see…when we left off before the holidays, we were trying to figure out a math channel that would indicate what was going on with the pedal inputs, allowing the production of a track graphic with information more granular than longitudinal accel would show. The single channel would combine the inputs from the accelerator and brake pedals, but complexities were introduced when both pedals are applied, either in a heel-toe downshift or when left foot braking and feeding in gas while still not completely off the brakes.

I went out and logged some data, including a heel-toe downshift, and a series of six instances of left foot braking, each time beginning to add throttle before the brakes were completely released.

I had previously identified that my original, match channel that was simply the difference between throttle and brake pressure inputs would show wacky results on a heel-toe downshift. In the graphs below, the upper pair of lines are the individual throttle (dark green) and brake pressure (red) channels. Note that, to present them this way, I reversed the brake pressure channel so that it’s negative—below the line—and set the scale on both throttle and brake channels to range from -100 to +100.

The lower graph’s orange line is longitudinal acceleration, so we can see where the car is actually being accelerated or decelerated. The bright blue line is the original simple formula, and you can see where it gives misleading info when the throttle is blipped for the downshift. The dark blue line—which covers most of the bright blue line as they have the same values much of the time—does a much better job of depicting when the car is actually being decelerated or accelerated. It uses an if-then-else statement conditioned on the clutch switch to ignore the throttle blip.

Name:  heel toe.png
Views: 450
Size:  157.5 KB

On to left foot braking… We last discussed using additional if-then-elses to select the dominant input, conditioned on which pedals are engaged and on longitudinal accel as the tie breaker. As I feared, the single channel makes big, instantaneous jumps from the brake value to the throttle value as soon as the long accel value is crossed. It seems that the best solution, when both pedals are being applied, is to blend the values. Weighting the brakes 80% and the throttle 20% seems to do a pretty good job of crossing the line from braking to accelerating around the point where longitudinal accel goes from negative to positive. In the below graph, once again the upper set are the throttle and brake traces—note how throttle is being added before complete brake release in each braking/accel event. The lower set’s orange line is longitudinal accel, the bright green line is the flawed formula that makes the instantaneous leaps, and the purple line is the good, blended, 80/20 version. The 80/20 ratio might need to be adjusted to something else at track speeds, however.

Name:  left foot.png
Views: 410
Size:  162.1 KB

So, the final formula using RS2 syntax and Z4M ECU protocol channel names is as follows:

IF(EQ(Z4M_CLUTCH_SW,1),Z4M_BRAKE_PR,IF(GT(Z4M_PPS, 0),IF(LT(Z4M_BRAKE_PR,0),(Z4M_BRAKE_PR*.80+Z4M_PPS *.20),Z4M_PPS),IF(LT(Z4M_BRAKE_PR,0),Z4M_BRAKE_PR, Z4M_PPS)))

Note that earlier, when I was using the default settings on the brake test channel, it was a positive value and so I was using the negative of its value in the new math channel. Now that I’ve reversed the brake channel as discussed above, it’s already a negative value and so it can come straight into the formula and it’s added to the throttle value when blended.

Now, I’m going to have to learn to left foot brake so all this isn’t for naught.
__________________
Appreciate 0
      01-02-2012, 08:21 PM   #126
MFGJR
First Lieutenant
27
Rep
341
Posts

Drives: 2007 Z4 M Coupe
Join Date: Sep 2008
Location: Richmond, VA

iTrader: (0)

Speaking of braking, pretty interesting article on Matt's site: http://www.trailbrake.net/2/post/201...oa-horsey.html
__________________
Appreciate 0
      02-27-2012, 10:28 PM   #127
MFGJR
First Lieutenant
27
Rep
341
Posts

Drives: 2007 Z4 M Coupe
Join Date: Sep 2008
Location: Richmond, VA

iTrader: (0)

Finally got out on VIR with the Solo. Best laps were 2:30s, just OK for C Group... Was able to compare to Randy's data from December, where he has a nice 2:24 best lap.

In the below graphs comparing our respective best laps, I'm red and Randy is blue. Corner numbers are at the top, straights are shown in green. (Note that AiM doesn't let you label corners "a" or "b", so VIR's 5 and 5a are all labeled as 5, 11a and 11b are shown as 11 and 12, 14a and 14b are 13 and 14, and 17a and 17b are combined as 17.)

The top graph shows throttle (above 0) and brake (below 0) application. The middle graph is speed, and the bottom graph is the time differential--my time in excess of Randy's as the baseline. All three are plotted against distance around the circuit on the horizontal axis.

Name:  Frank v Randy.png
Views: 394
Size:  179.6 KB

It's easy to see that I give up the bulk of the time approaching and through the Climbing Esses and South Bend (from turn 6 through 10). Randy's much more confident attacking this section, entering the beginning of the esses at turn 7 at 111 mph vs. my 88. If you look at the top graph, you see me coast off a bunch of speed on the straight before turn 7, where Randy's only gradually rolling off the gas. That coast alone cost me 1.5 seconds. I did have a 98 mph entry on another lap, but then blew the braking for 10 and ruined the lap... no good corner goes unpunished.

Looking at the braking zones, Randy is quite noticeably more efficient in compressing the braking. See how steep his transitions from throttle to brakes are approaching turns 1, 3, 4, 10 and 13. Plus, I'm a wuss and just can't keep the hammer down all the way through and past the front straight's kink, hence the trailing off the gas and another big coast between turns 17 and 1.

I feel I can make up significant time just focusing on these couple areas... gotta get some new (stickier) tires first, though. That, and more seat time.

Ain't data acquisition great?
__________________
Appreciate 0
      02-29-2012, 10:23 PM   #128
sabbanick
Lieutenant
13
Rep
505
Posts

Drives: 08 SBM M coupe
Join Date: Sep 2010
Location: NY

iTrader: (1)

Randy you the man!

Seriously nice breakdown and interpretation of this data. Nice to know HOW to improve lap times... applying what you see in this data will net you a serious reduction in time.

I am still working on getting myself situated with using RS2...biggest issue is getting a track map incorporated with the data like you have above.
Appreciate 0
      03-01-2012, 06:38 AM   #129
MFGJR
First Lieutenant
27
Rep
341
Posts

Drives: 2007 Z4 M Coupe
Join Date: Sep 2008
Location: Richmond, VA

iTrader: (0)

Nick--my maps are here: http://www.trailbrake.net/aim-track-maps.html

One map is individaul corners (see above on numbering), other is segments. Download them and, assuming you did the standard install of RS2, put them in C:\AIM_SPORT\RaceStudio2\TRACKS and you should find them available in Map Manager.
__________________
Appreciate 0
      03-01-2012, 09:33 PM   #130
MFGJR
First Lieutenant
27
Rep
341
Posts

Drives: 2007 Z4 M Coupe
Join Date: Sep 2008
Location: Richmond, VA

iTrader: (0)

Thinking about Randy's 2:24 (actually, he was down to 2:23 this past weekend) compared to my 2:30...

Looked at my split report with the last two Sunday sessions loaded up. Lo and behold I have a theortical best time of 2:24.709 using my best individual segment times from among the 15 laps in those two sessions. Actually, only 5 laps contributed to the theoretical best time, and three of the seven segments came from a single lap. So, maybe I'm closer to a decent lap than I thought--just have to string all the good segments together into the same lap.

Gotta love data acquisition: improved lap times and self-image at the same time!

Name:  splits.png
Views: 319
Size:  102.1 KB
__________________
Appreciate 0
      03-02-2012, 06:25 AM   #131
RandysV10
MPOWERD
RandysV10's Avatar
15
Rep
338
Posts

Drives: 2010 M6
Join Date: Nov 2010
Location: Cloud Nine

iTrader: (0)

Garage List
2006 Z4M  [8.00]
2010 BMW M6  [9.00]
That's awesome Frank, just need so more seat time to bring them together.
Appreciate 0
      03-02-2012, 09:09 AM   #132
ZStig
instagram 997turbotom
279
Rep
7,376
Posts

Drives: Interlagos Z4MR
Join Date: Jun 2008
Location: Delray Beach, FL

iTrader: (4)

my head exploded trying to understand page 6, so I'm running away from here
__________________
Appreciate 0
Post Reply

Bookmarks

Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off



All times are GMT -5. The time now is 03:58 PM.




zpost
Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2024, vBulletin Solutions Inc.
1Addicts.com, BIMMERPOST.com, E90Post.com, F30Post.com, M3Post.com, ZPost.com, 5Post.com, 6Post.com, 7Post.com, XBimmers.com logo and trademark are properties of BIMMERPOST