Members Login
Username 
 
Password 
    Remember Me  
Post Info TOPIC: Dragtools - development starting...


Senior Member

Status: Offline
Posts: 149
Date:
RE: Dragtools - development starting...


Petrik
My bike is showing F1 light with dragtools activated.
I am running Nitrous activation on DSM1 and Shifter on DSM2.
If I de-activate dragtools and reflash F1 light goes off.
Strange thing is bike still limits @ 4000 with clutch in.

I now rebuilt the maps from a stock base map from V.56 , maybe this will help.

Will post result.

__________________


Senior Member

Status: Offline
Posts: 149
Date:

It worked , no more F1 will test bike tomorrow.
Think it was a Version thing with editing a previous version map with later version ?

__________________


Senior Member

Status: Offline
Posts: 186
Date:

I have new ideas about evaluating the RPM(t) function in order to find "t" candidates for activating the traction control: I'd like to evaluate not just the
first derivative RPM'(t) but also the second derivative RPM"(t).
Regions where RPM"(t) is positive (i.e. RPM(t) has upper concavity) are more likely to need attention. To verify this idea is woudl be useful to have a data log for time t,
RPM(t), RPM'(t), RPM""(t) and indication of t where the wheel was slipping. This would help to find a useful correlation between wheel slipping  and shape of RPM(t).

Here the numerical approximation foruma needed:

If we have RPM(t1), RPM(t2), RPM(t3):

RPM'(t1) = (RPM(t2) - RPM(t1)) / (t2 - t1)
RPM'(t2) = (RPM(t3) - RPM(t2)) / (t3 - t2)

RPM"(t1) = (RPM'(t2) - RPM'(t1)) / (t2 - t1)


If the time instants are equally spaced (i.e. t2 - t1 = t3 - t2 =... = dt)
the recurring formula for the second derivative becomes much more simpler:

RPM"(t) = (RPM(t + 2*dt)  -2*RPM(t + dt) + RPM(t)) / (dt*dt)

Seen from a different perspective I'd say that regions where RPM"(t) is negative should be pretty safe and so excluded from any traction control, while
for positive values an upper threshold should be estimated.





-- Edited by ballad2 on Saturday 8th of January 2011 09:41:24 PM

__________________


Guru

Status: Offline
Posts: 2338
Date:

Yes, youre absolutely correct in what youre saying there, but time instances are not equally spaced, this is lower in the interrupt chain and basically executed 22 times per engine revolution where last one is lenght of 2 teeth.

I discussed about the filtering while playing badminton a couple of days ago with my friend who is deep into industrial automation of paper production machines. We come to a conclusion that there are ways to improve the filtering compared to average of 2 full engine cycles.

http://www.dsptutor.freeuk.com/dfilt4.htm

__________________

When asking a question, you can also consider posting it to facebook:

http://www.facebook.com/ecueditorcom



Senior Member

Status: Offline
Posts: 186
Date:

PetriK wrote:

Yes, youre absolutely correct in what youre saying there, but time instances are not equally spaced, this is lower in the interrupt chain and basically executed 22 times per engine revolution where last one is lenght of 2 teeth.

I discussed about the filtering while playing badminton a couple of days ago with my friend who is deep into industrial automation of paper production machines. We come to a conclusion that there are ways to improve the filtering compared to average of 2 full engine cycles.

http://www.dsptutor.freeuk.com/dfilt4.htm

---------------------------------------------------------------------

Well, that's the engineering view of the world,
I'm just a mathematician :)

If the time samples are arbitrarily spaced (variable dt), the second derivative formulas coming from its definition still applies, using 2 consecutive values RPM'(t) and RPM'(t + dt)  to calculate the difference:

RPM"(t) = (RPM'(t + dt) - RPM'(t)) / dt

My next homework could be to add 2 columns into the data logging output for RPM' and RPM" in order to find a valid filtering logic.



 



__________________


Guru

Status: Offline
Posts: 2338
Date:


Then we speak the same language, almost .... fourier and first or second element transofrmation.


__________________

When asking a question, you can also consider posting it to facebook:

http://www.facebook.com/ecueditorcom



Senior Member

Status: Offline
Posts: 186
Date:

ballad2 wrote:

 

PetriK wrote:

Yes, youre absolutely correct in what youre saying there, but time instances are not equally spaced, this is lower in the interrupt chain and basically executed 22 times per engine revolution where last one is lenght of 2 teeth.

I discussed about the filtering while playing badminton a couple of days ago with my friend who is deep into industrial automation of paper production machines. We come to a conclusion that there are ways to improve the filtering compared to average of 2 full engine cycles.

http://www.dsptutor.freeuk.com/dfilt4.htm

---------------------------------------------------------------------

Well, that's the engineering view of the world,
I'm just a mathematician :)

If the time samples are arbitrarily spaced (variable dt), the second derivative formulas coming from its definition still applies, using 2 consecutive values RPM'(t) and RPM'(t + dt)  to calculate the difference:

RPM"(t) = (RPM'(t + dt) - RPM'(t)) / dt

My next homework could be to add 2 columns into the data logging output for RPM' and RPM" in order to find a valid filtering logic.



 

In one of my data logs with PC onboard (raw .csv) I added 2 columns for the first and second RPM derivatives using excel.

Here below a fragment where we have column A for time, B for RPM, C for RPM', D for RPM", other column unchanged.

A3=time,  B3=RPM, C3=(B4-B3)/(100000*(A4-A3)), D3=(C4-C3)/(100000*(A4-A3))

 

It looks like negative RPM" values are a good trigger to filter high RPM' values not related to wheel slipping (see row 21:40.8).





21:40.5 3490 -4028.5 29291.5 57 96 39 65535 175 94 3 In False 255 90 186 166 67 179 1 801 786 809 788 False 0 100
21:40.6 2730 1497.6 22166.8 57 86 41 65535 143 63 3 In False 255 100 186 166 67 180 1 810 814 810 816 False 0 100
21:40.8 2990 5346.0 -30989.1 58 79 41 65535 146 57 3 Out False 255 107 186 166 67 181 0 904 862 881 871 False 0 100
21:41.0 3980 -392.7 2203.4 58 94 45 65535 187 97 3 Out False 255 92 186 167 67 180 1 889 889 893 899 False 0 100
21:41.1 3910 0.0 -1244.7 59 92 41 65535 201 97 3 Out False 255 94 186 167 67 180 1 910 908 910 906 False 0 100


__________________


Senior Member

Status: Offline
Posts: 186
Date:

In order to have meaningful values for RPM' (in RPM/sec) and for RPM" (in RPM/sec2)
the scaling factor in the excel formulas must be 86400 (number of seconds in a 24 hours)
instead of 100000.
That said, if anybody has data logs and can isolate the fragment where the wheel was
spinning, applying the formula and posting the fragment would help to refine the filtering logic.




__________________


Guru

Status: Offline
Posts: 2338
Date:

This would be a good approach to use enginedata . Have you looked at the videos that sportbykeryder posted here earlier. The spikes can be seen on those videos. They seem to have different algos for detecting the spin, within one engine round if I remember correctly. That sensitive data may be difficult to get from a normal engine log.

Our current detection mechanism is average over 100 samples. The number 100 comes from besides being a round number also:
- one engine turn is 24-2 samples
- one full working cycle is 48-4 samples
- To have one full engine working cycle to compare agains another cyle we need around 100 samples to avoid frequency interference if and when more andvanced filtering is applied.

There is often some variation between cylinders so therefore really need to compare one full cycle to another if we look for minor differences.

The easiest to implement detection algorithm tuning is to provide more weighting in the average for the cycles that are at the end of the measurement period. E.g. weight last 10 samples with 25% etc. This way the algo would better anticipate what the next sample should be. Anyway we are limited by amount of variables and recursive loops are very difficult if not impossible to implement - so simple algos are what we are stuck into.

Then there is other type of filtering of RPM, thats the spikes in the digital RPM signal itself - but as we use a digitally filtered RPM signal from ecu alrealdy we dont really need to take that into account.




__________________

When asking a question, you can also consider posting it to facebook:

http://www.facebook.com/ecueditorcom



Senior Member

Status: Offline
Posts: 186
Date:

I see the filtering algorithm in two stages:

1) low level filtering, as you wrote, calculating a suitable weighted moving average (WMA),
in a C routine should be possible to implement even EMA (exponential moving average) over
100 points.

2) once the RPM(t) function has sufficient regularity out of step 1), with a granularity in the order
of 1/10 of second, RPM' and RPM"" could be evaluated in order to trigger the control.

Let's have a look to the orders of magnitude:
Stock B-king 1st gear at 5520 RPM speed is 65.9Km/h = 18.3 m/sec.
Tire circumference 1.985m. => (18.3 m/sec) / 1.985m = 9.2 rev/sec.
5520 RPM / 60 = 92 rev/sec. => 1 wheel rev = 10 engine revs

If we limit to consider the wheel spinning for a single revolution, it corresponds
to 10 engine revolutions that happens in about 1/10 of second.

That's why I think that datalogs, with a granularity of 1/10 of second are suitable for
refining stage 2) of the filtering.



__________________


Guru

Status: Offline
Posts: 2338
Date:

Good stuff and thinking ongoing there.... a few comments.

Datalog software tends to add filtering to the rpm, particularly innovate does so.

The engine management must be able to react within a couple of engine rounds, (without checking the maths) just lets say hopefully less than one wheel round at 8000rpm where the peak torque tends to be and above. If the wheel starts to spin it continues through out the track and its game over for the rider.

Another element which I am worried is hysteresis - worst possible scenario is to detect the wheelspin too late and then overreact by causing a nose dive then getting grip again and nose up. Lost seconds and with a lock up will mean interesting moments.

Have a look at the videos, those are very good. Some say that even a single ignition kill (60ms) will kill the wheelspin so we may need to consider also different algos in reducing the power.



__________________

When asking a question, you can also consider posting it to facebook:

http://www.facebook.com/ecueditorcom



Guru

Status: Offline
Posts: 1247
Date:

May be a bit off topic, but many automotive FI systems can detect a single cylinder misfire by monitoring instantaneous rpm from the crank trigger wheel. They are basically determining rpm many times per engine revolution and compare it to where it should be.

Also, the Davis Technologies system uses a driveshaft trigger. Assuming wheel spin happens in the lower gear ratios, this signal contains multiple engine revulotions per driveshaft revolution. Additionally, it contains multiple driveshaft revolutions per tire revolution.

Usign the crank trigger will net a multiplier of 7.27 of wheel revolutions to crank revilutions with stock gearing and in 2nd gear (this was using gen1 ratios).
Averaging the rpm over ~2 engine revolutions will give an rpm input to the ECU ~ 3.6 times per tire revolution.  Assuming at least two rpm inuts are needed to get a rpm rise rate, that drops to 1.8 times, or slightly more than half a tire revolution before the ECU recognizes a problem. 

If this logic is correct, it may not be high enough resolution to use 100 cycles average as I would think half a revolution prior to correction being added may not be quick enough.



-- Edited by sportbikeryder on Sunday 9th of January 2011 07:47:18 PM

-- Edited by sportbikeryder on Sunday 9th of January 2011 07:48:16 PM

__________________


Senior Member

Status: Offline
Posts: 186
Date:

I had a look to the Davis' patent on traction control. In the abstract it spends 10+
lines to illustrate the concept of RPM second derivative :).
I knew a funny story in which every day in some part of the world an engineer
rediscover the Newton's iterative formula. Well, it was not a joke, it is reality!




__________________


Guru

Status: Offline
Posts: 1344
Date:

great stuff guys, keep up the good work.....thank you..marc

__________________

09 busa.????? now what....still got what it takes.......!

I got what you need...!
www.poweredbyford.com

www.marc@poweredbyford.com

 



Veteran Member

Status: Offline
Posts: 49
Date:

tested these out yesterday the limiter works much better now. as for the traction part it works pretty good to alot of tuning and testing out of different settings will have to be done to find what works best. had it set to come on at 6500rpm, 2 step set the same, 1400rpm/sec with 15 out in 1st and 10 in 2nd. gave it pretty much full throttle and dumped it, still left black lines for couple of meters not huge spin though just a little till it really hooked up. my head still doesnt want to let go and give all control to the electronics yet even so im probably getting off the line quicker. when i get another chance and bit more used to it i will try and log some files. hopefully the rain goes away and the sun comes out but its not looking good here for the rest of the week

__________________


Veteran Member

Status: Offline
Posts: 95
Date:

OK guys......I've looked over the dragtools and downloaded it to my laptop for testing. I'm not a dragracer so I'm not familiar with 2 steps and stuff. I think I understand the concepts but I have some questions.

1. When dragtools are activated and the clutched ignition limiter is set to 7000 RPM the bike will accelerate to 7000 RPM then the two step kicks in and holds the RPM's at 7000 RPM......right?

2. Will this clutched ignition limiter work only in first gear or will it limit the RPM whenever the clutch is in? I ask this because it is a drag tool and I know some drag guys upshift without pulling in the clutch.

I'm going to test this stuff out this weekend. My plan is to take the bike to an open road and have a friend follow in a car with my laptop so I can reflash at the test site.

The way I see it testing the clutched ignition limiter will be easy. Either it's going to hold the RPM's or it won't. Testing wheelspin probably won't be as easy. My bike is prone to wheelies not burnouts so getting the rear wheel loose and finding the sweet spot where the bike doesn't spin may take a few trips. Additionally, I think there will be issues trying to determine if the rear wheel is spinning and what RPM that happens at while I'm on the bike and from a bystanders perspective. We'll see this weekend.

I'd like to be able to report back and say that the bike normally spins in X gear at X RPM's and with the max acceleration set at X RPM's/sec with X ignition retard/100RPM it prevents wheelspin. Again, well see this weekend.

Please let me know if you have a better idea for testing any of the above. Thank you.

Jason

__________________


Guru

Status: Offline
Posts: 1247
Date:

Id be a little cautious trying to get a swb bike to spin Jason, especially since you aren't used to spinning / dragracing.
The 2 step works as you mention. The purpose is to be able to basically hold the throttle wide open (or some do it partially open) any be able to launch with just the clutch. Again, use caution when testing.

The main thing that could potentially have issue are possible "hiccups" at the set rpm whine driving the bike with the clutch out (this seems to be resolved) and another woudl be the relative "smoothness" of the 2 step rev limiter.



__________________


Veteran Member

Status: Offline
Posts: 95
Date:

sportbikeryder wrote:

Id be a little cautious trying to get a swb bike to spin Jason, especially since you aren't used to spinning / dragracing.
The 2 step works as you mention. The purpose is to be able to basically hold the throttle wide open (or some do it partially open) any be able to launch with just the clutch. Again, use caution when testing.

The main thing that could potentially have issue are possible "hiccups" at the set rpm whine driving the bike with the clutch out (this seems to be resolved) and another woudl be the relative "smoothness" of the 2 step rev limiter.



I understand all of that.  Thank you.


 



__________________


Senior Member

Status: Offline
Posts: 149
Date:

Tested Yesterday,
Went out on the road with a mate on Zx14 with all the bells and wistles.
Nailed him off the line Time after time after time after time.
This stuff is sooo- cool.
Tried it from a rolling start too . Bike carries front wheel for about 20 metres when pinning throttle in 1st gear from about 40Kph.
There is a small amount of wheelspin , will fiddle with settings next.

I have never been that good of the line , but now I seem to be faster than my friend who is a very experienced drag racer !


__________________


Veteran Member

Status: Offline
Posts: 95
Date:

LenTurbo wrote:

Tested Yesterday,
Went out on the road with a mate on Zx14 with all the bells and wistles.
Nailed him off the line Time after time after time after time.
This stuff is sooo- cool.
Tried it from a rolling start too . Bike carries front wheel for about 20 metres when pinning throttle in 1st gear from about 40Kph.
There is a small amount of wheelspin , will fiddle with settings next.

I have never been that good of the line , but now I seem to be faster than my friend who is a very experienced drag racer !




That's awesome!

Can you post up what settings you have your drag tools set-up with?

 



__________________


Senior Member

Status: Offline
Posts: 149
Date:

Made some changes to STP maps.
Neutral Stp Maps at 45% Max Through rev range.
This holds Rpm more stable at Launch setting.
Set 1st gear STP to open slower through range.
Copied 4th gear Stp to 2nd and 3rd gear.

Did mod on clutch lever switch so that it will open when clutch is partially engaged.
As standard it alows the Rpm to climb before clutch engagement starts.

ROA settings as per Base map.

Can eMail .bin if anybody wants to see.Dont know how to attach a file to post.



__________________


Guru

Status: Offline
Posts: 1344
Date:

got tire spin today on a cold tire with 1st rpm/sec set at 1600 rpm...?will try a lower setting i a little while, this was leaving on 2 step at about 60% tps and going wot when clutch was thrown...

__________________

09 busa.????? now what....still got what it takes.......!

I got what you need...!
www.poweredbyford.com

www.marc@poweredbyford.com

 



Guru

Status: Offline
Posts: 1344
Date:

i lowered the activation threshold to ensure tirespin will happen suddenly, and i also lowered the rpm/sec to 1000 rpm, and have 10 degrees out.....and boom...when the software activated the changes were so extreme the engine stuttered hard and did not acellerate..now off to test more realistic real world settings....i think the local sherrif may be onto my noise making....lol...so testing intrevals may be drawn out a little further....also been airshifting with 40ms and working fine.....!

__________________

09 busa.????? now what....still got what it takes.......!

I got what you need...!
www.poweredbyford.com

www.marc@poweredbyford.com

 



Guru

Status: Offline
Posts: 1247
Date:

Cool....that hook-up tire is too damn good for this kind of testing!

Good to see the bike reacting to changes and decreasing the torque.

__________________


Guru

Status: Offline
Posts: 2338
Date:


Let me know if you would prefer a different type of algorithm. Now its simply reducing ignition degrees * rpm/sec. That means that if the RPM after cluch is e.g. at 9000 rpm constantly then there is a constant wheelspin. This ROA works only during the sudden increase of rpm - which is easiest to implement.

Next level up would be to calculate a time based maximum rpm for each time point after dumping a clutch. Maybe Ballad2 could be interested to work on the time based algo ?







__________________

When asking a question, you can also consider posting it to facebook:

http://www.facebook.com/ecueditorcom



Senior Member

Status: Offline
Posts: 186
Date:

I'm trying to keep in sync GenII changes with BKing. At this time the BK Dragtool module is coded but still under review.

The clutch limiter functionality is based on already existing BKing limiters code.
so it should work or should be easily fixable.

If somebody wish to give it a try, it is available at:

http://bitbucket.org/ballad2/ballad2_fork/overview





__________________


Guru

Status: Offline
Posts: 2338
Date:

Yep - quite a lot of changes lately, this happens occasionally when working on something new. I think it will get soon more quiet on development front.

Anyhow the dragtools.c module is exactly what we are talking about here, adding a more complex algo in that - but maybe its more relevant when the existing one is avail for bking too.


__________________

When asking a question, you can also consider posting it to facebook:

http://www.facebook.com/ecueditorcom



Guru

Status: Offline
Posts: 1247
Date:

I think there may be much more testing of the ROA type control when the US / notthern Hemisphere drag racing season begins. There are MANY platforms that will be able to easily spin the tire at will that are sitting in garages at this time due to weather.

-- Edited by sportbikeryder on Sunday 16th of January 2011 05:53:13 PM

__________________


Guru

Status: Offline
Posts: 1344
Date:

or you could send me your bikes and i can test it at any of the 4 tracks here in florida that is open and we can develop it faster....john, i will come up to your place to pick yours up first....?...lol...wink

__________________

09 busa.????? now what....still got what it takes.......!

I got what you need...!
www.poweredbyford.com

www.marc@poweredbyford.com

 



Guru

Status: Offline
Posts: 546
Date:

I'm hoping to test the traction control on the salt in march, and maybe before that on a hard dirt airstrip
Last season in the US i was chunking a tire every run over 220mph (current tire regs :( )
now brought 07 gix 1000 ecu and loom that i will fit to both my 750 and gen 1 busa lsr bikes
flash the gen 2 busa codes in for boost fuel etc and see if i can make it all work
It will be interesting as the amount of slip needed on the dragstrip for best times is way different to what will suit the loose surface , i figure i need around 6-8% for best drive

Don't know if you guys realise it yet but you have effectively turned these ecu's into a world class standalone

__________________


Senior Member

Status: Offline
Posts: 498
Date:

Maj750 wrote:

Don't know if you guys realise it yet but you have effectively turned these ecu's into a world class standalone



AMEN To that!!! smile When you Really Sit back and look at it ALL........ IT'S AMAZING, The Kind of thing's we can Now Do with a "FACTORY" ECU! biggrin
Petrik & Crew, As Always...... THANK YOU for what you have Done for the Busa/Bike Community. wink 

 



__________________
08 Busa AKA: }ToXSicK{


Veteran Member

Status: Offline
Posts: 95
Date:

jgennaro wrote:

OK guys......I've looked over the dragtools and downloaded it to my laptop for testing. I'm not a dragracer so I'm not familiar with 2 steps and stuff. I think I understand the concepts but I have some questions.

1. When dragtools are activated and the clutched ignition limiter is set to 7000 RPM the bike will accelerate to 7000 RPM then the two step kicks in and holds the RPM's at 7000 RPM......right?

2. Will this clutched ignition limiter work only in first gear or will it limit the RPM whenever the clutch is in? I ask this because it is a drag tool and I know some drag guys upshift without pulling in the clutch.

I'm going to test this stuff out this weekend. My plan is to take the bike to an open road and have a friend follow in a car with my laptop so I can reflash at the test site.

The way I see it testing the clutched ignition limiter will be easy. Either it's going to hold the RPM's or it won't. Testing wheelspin probably won't be as easy. My bike is prone to wheelies not burnouts so getting the rear wheel loose and finding the sweet spot where the bike doesn't spin may take a few trips. Additionally, I think there will be issues trying to determine if the rear wheel is spinning and what RPM that happens at while I'm on the bike and from a bystanders perspective. We'll see this weekend.

I'd like to be able to report back and say that the bike normally spins in X gear at X RPM's and with the max acceleration set at X RPM's/sec with X ignition retard/100RPM it prevents wheelspin. Again, well see this weekend.

Please let me know if you have a better idea for testing any of the above. Thank you.

Jason



Yesterday I tested out the two step and tried to test the traction control.  The two step was easy.  I set the two step at 4K RPM and it worked perfect.



I flashed it at 5, 6, and 7K RPM also and again it worked perfect.  To test it out further I took the bike for a 1.5 hour trip through the country roads and surburban traffic.  The bike ran great.  

Testing traction control wasn't quite as easy or successful.  I flashed the bike without dragtools and took it to a test site.  I tried to get my bike to break the back tire loose but I couldn't do it.  Even with water on the road it would still wheelie and not break traction.

Since a few people mentioned that setting the traction control calmed down wheelies I figured I'd give it a try.  I looked at my logs from the dyno and in 4th gear I noticed the bike was accelerating a little over 2K RPM's per second.  I flashed 3rd gear and up at 2100, 2000, and 1900 with 5 degrees of timing taken out without success.  The bike continued to wheelie.  I ran out of time so I went home after the last flash.  

When I went back home I took another look at my logs and I did have 2000RPM/sec acceleration from 9-11K RPM; however, at max torque (around 6500RPM) my logs showed acceleration of 1600RPM/sec.  So........I'm thinking to control wheelies with traction control I'll have to flash at 1500RPM/sec or less.  The only issue with that is it may control wheelies but I'll loose out at higher RPM's.

Bottom line is this stuff worked great.  Thank you men for your hard work and I hope this post and test result helps.

Lastly, I dowloaded PetriK's lastest version of ECU Editor before I tested any of this stuff.  I think the version was .33 at the end.

Jason      

 



-- Edited by jgennaro on Monday 17th of January 2011 11:58:49 AM

__________________


Guru

Status: Offline
Posts: 1344
Date:

i have been studying davis traction control, on thing i noticed is that when tire spin is detected, they immeadiately retard timing, but the ramps it back in, not insant...now how does ee2 function..?, i have to look into this, it may work on my low hp bike, but on a big hp bike, a ramp in will keep the tire hooked....more on this later..

__________________

09 busa.????? now what....still got what it takes.......!

I got what you need...!
www.poweredbyford.com

www.marc@poweredbyford.com

 



Guru

Status: Offline
Posts: 1247
Date:

Marc, do you have any bikes, or maybe Josh, to do some higher hp, longer bike testing? Mine will not ba available for a few more months, both due to weather and "completeness"

__________________


Member

Status: Offline
Posts: 12
Date:

I was reading a book on statistical method for engine control systems; it was way outside my abilities or knowledge. It looks like back to the maths books to try and understand the methods used in the book. The book contained a chapter on torque calculation and another on a misfire calculation method.

The methods used would have to ramp the timing back in or you would just create more wheel spin as you hit the tyre again with a sudden increase in torque. I am not sure that the best method is to just retard timing a fixed amount when wheel spin is detected and then bring back timing until you reach max timing or wheel spin occurs again.

If a timing figure of 26 degrees creates wheel spin I retard timing 5 degrees so now it is 21 degrees increment timing back and find wheel spin occurs again but at 23 degrees does the ecu retard timing again 5 degrees which means back to 18 degrees and increment it back till wheel spin occurs at 23 degrees again and keep this continual loop possible occurring.

I will check some documents I have on how another manufacturer implemented torque control this weekend.


__________________


Guru

Status: Offline
Posts: 2338
Date:

There is currently a fairly simple algorithm here.


The ramp up rate is just calculated as rpm/sec using average of 100 measurement points (4 engine rounds is 96 points).

If the actual ramp up rate exceeds the calculated rpm/sec rate then for each 100rpm/sec the ignition is retarded by user defined retard in degrees, i.e.:

if (actual_rpm_sec > defined_rpm_sec) then ignition_advance = ignition_advace - (retard*((actual_rpm_sec - defined_rpm_sec)/100))

I have spoken about this with my friend who is a specialist in how to control processes in manufacturing environment. The conclusion so far is that instead of the average over 100 points we could weight the latest points more than the points in the beginning making the algorithm react faster - in case the 4 full engine rounds is too slow.

Then the power controlling method is another issue. In the current algorithm we bluntly just add ignition the more the wheel spins. A feedback algorithm would anticipate the future rpm/sec ratio and apply the ignition retard. In more advanced methods there should be a feedback loop that actually finds the optimum value, ie. when the retard is applied the amount of retard applied would be adjusted to keep the wheel at optimum acceleration curve. That could be the next step, but instead of writing complex algorithms I believe in testing and rider feedback to make the improvements so that we start to shoot into the right direction.

(I have a toyota prius which has a traction control alogrithm which seems to make the engine loose all the power, which is a pain if there is a lot of snow and I would like to keep the wheels spinning at optimum range. So a good drag racing algorithm should not prevent the wheelsping, but make it easier for the user to control.)



__________________

When asking a question, you can also consider posting it to facebook:

http://www.facebook.com/ecueditorcom



Senior Member

Status: Offline
Posts: 148
Date:

Totally agree with Petrik i am using the antislip function from the first day and i like it cause it doesn't cut the power from the bike, it spins but very controllable even with 1st gear at wot.

I just feel that it must be more quick response when it is initiating after that is perfect, i still can't believe that this comes from the stock ecu.

__________________

LIFE IS SHORT RIDE HARD



Senior Member

Status: Offline
Posts: 148
Date:

Just a thought what about using stp in addition with ignition to control traction? e.g when tyre slipping reduce 10degrees+close 10% stp. is this doable?

__________________

LIFE IS SHORT RIDE HARD



Guru

Status: Offline
Posts: 2338
Date:


STP is doable, but the effect will be very slow. Rather would e.g. put one shiftkill activation cycle there ?


__________________

When asking a question, you can also consider posting it to facebook:

http://www.facebook.com/ecueditorcom



Senior Member

Status: Offline
Posts: 148
Date:

OK lets try that and see how this is going to affect the acceleration.

Any news with the yaw sensor?

__________________

LIFE IS SHORT RIDE HARD



Guru

Status: Offline
Posts: 2338
Date:


Been too tied up (entusiastic) about gixxer stuff and designing a new interface to avoid Headlight off instability - so not been much paying attention to yaw sensor.


__________________

When asking a question, you can also consider posting it to facebook:

http://www.facebook.com/ecueditorcom



Guru

Status: Offline
Posts: 1247
Date:

From the small amount of testing I have been able to perform on my GEN2 conversion, I have had no issues with connecting to the engine datastream without having a headlight (or many other components of the stock bike) installed at all.

I just received most of the parts to make a spare engine for my bike, so I hope to really be able to test the drag tools functions once the bike is finished and the weather breaks (just hope it makes enough power to break the rear tire loose)

__________________


Member

Status: Offline
Posts: 12
Date:

Is missfire detection part of the checks performed by the ecu?

__________________


Guru

Status: Offline
Posts: 2338
Date:

not really

__________________

When asking a question, you can also consider posting it to facebook:

http://www.facebook.com/ecueditorcom



Senior Member

Status: Offline
Posts: 186
Date:

ballad2 wrote:

 

PetriK wrote:

Hi, thanks, not really - at this stage the most difficult thing is to find proper time to get the ideas into a simple implementable format.

The prevailing idea starts to look like something below to the end users... there is no logig or anything yet, this is just a mockup what the end result could look like to give the forum members to give feedback about the thinking.

This is the simpliest form I can get the idea working at least when sketching.


slewrate.jpg



There should be the possibility to input the final transmission ratio, for example speed at 1000 rpm in a specified gear. Knowing this there is a 1:1 relation
between rotational acceleration (RPM/sec) and linear acceleration (meter/sec2).
I think m/sec2 target is more meaninful, for example 400m in 9.5 seconds corresponds to an average acceleration of 8.8 m/sec2. Knowing the target final speed the average acceleration can be estimaded easily:
(254 Km/h / 3.6) / 9.7 = 7.42 m/sec2.
Since people use different gearing, it makes more sense to discuss and compare linear acceleration values, which are more closely correlated to traction results rather than rotational acceleration.

 

 



dragtool.jpg

I'm convinced that a frame of reference where the  crankshaft is still and

the bike rotates around it is not as significant as the frame of reference

where the earth is still and the bike moves around it (Galileo Galilei Docet).

Hoping not to be jailed for this assumption, I coded an improvement to

the dragtool module where linear acceleration is displayed given

front sprocket, rear sprocket and tire size. No functional changes at all.

If useful it is ready to be integrated.

 



__________________


Guru

Status: Offline
Posts: 1247
Date:

Ballad, Are you saying that the coding you did is just a reference calculation that uses the inputs of max rpm/sec and outputs a linear correlation? If so, I think the info can be useful. One thing I woudl reccommedn is the ability to add in generic tire circumference as soem of the dragracing tires are somewhat "custom" even though the size is listed in standard dimensions (i.e. not all 190/50 tires are the same)





__________________


Veteran Member

Status: Offline
Posts: 95
Date:

I have a question about the linear acceleration. Where you have 1st, 2nd, and 3rd gear with the drop down menu does the 3rd gear linear acceleration set-up apply to 3rd, 4th, 5th, and 6th gears or just 3rd?



Jason

__________________


Guru

Status: Offline
Posts: 1247
Date:

On another, general programming note, Is there a particular reason why all of the programming is performed with pulldown menu's rather than a user input option?

Seems it could be much easier to program if the box was left open for typing rather than to have to code in all of the options for every pulldown. Also, this would allow users to fine tune or even try new ranges of items as appropriate.

perhaps a suggested max / min range listed woudl be appropriate itf this was done.


Note, this is not necessarily directed at you Ballad, just an overall observation.

John


__________________


Senior Member

Status: Offline
Posts: 186
Date:

sportbikeryder wrote:

Ballad, Are you saying that the coding you did is just a reference calculation that uses the inputs of max rpm/sec and outputs a linear correlation? If so, I think the info can be useful. One thing I woudl reccommedn is the ability to add in generic tire circumference as soem of the dragracing tires are somewhat "custom" even though the size is listed in standard dimensions (i.e. not all 190/50 tires are the same)




Yes you got it perfectly. Input of Tire Circ in mm  would have been even more easy

for me to code, but I will think how to integrate that option as well.



__________________


Senior Member

Status: Offline
Posts: 186
Date:

jgennaro wrote:

I have a question about the linear acceleration. Where you have 1st, 2nd, and 3rd gear with the drop down menu does the 3rd gear linear acceleration set-up apply to 3rd, 4th, 5th, and 6th gears or just 3rd?



Jason



In my coding 3rd gear acceleration is just a display. I could display
the values for 4, 5, and 6  as well but I did not want to modify the form
too much. Acceleration is more meaningful in low gears, but if
it is interesting I can display all of them.

The functionality is ruled by RPM/sec just as it is now.

 



__________________
«First  <  1 2 3  >  Last»  | Page of 3  sorted by
 
Quick Reply

Please log in to post quick replies.

Tweet this page Post to Digg Post to Del.icio.us


Create your own FREE Forum
Report Abuse
Powered by ActiveBoard