|
julex
|
 |
« Reply #30 on: January 16, 2012, 03:15:24 PM »
|
|
|
Spreasheet works for me brilliantly. First time for a long time my car is driving butter smooth through all rpms at a given pedal request. I also completely linearized requested pedal torque across rpms/pedal position. I hate the stock curve as if you want to have you car to accelerate constantly, you have keep compensating with pedal by pressing it more and more as the rpms raise, unless you wot ofc.
The car is finally holding the torque without hiccups and heavy handed corrections, A+++++ on this.
As to the shaping of the table and corrections... there are tables in ME 7.1 which have specific notes on them that the ECU is not taking the two most adjacent cells into consideration but is actually "drawing" a curve through five relevant cells and picks a calculated value from that line, not an interpolated value from two adjacent cells.
When you think about it, it makes sense. If the ECU was just interpolating, the derived value would follow a very edgy curve following the values directly. If you draw a curve through multiple cells, it smooths the resulting curve considerably.
I think the note was on
|
|
|
|
|
Logged
|
|
|
|
|
TTQS
Guest
|
|
« Reply #31 on: January 16, 2012, 04:22:57 PM »
|
|
|
As to the shaping of the table and corrections... there are tables in ME 7.1 which have specific notes on them that the ECU is not taking the two most adjacent cells into consideration but is actually "drawing" a curve through five relevant cells and picks a calculated value from that line, not an interpolated value from two adjacent cells.
When you think about it, it makes sense. If the ECU was just interpolating, the derived value would follow a very edgy curve following the values directly. If you draw a curve through multiple cells, it smooths the resulting curve considerably.
I always presumed that the interpolation was based on the relationship between the entire row or column's values be it linear, power, logarithmic or whatever, not just a simple linear one through two adjacent cell addresses otherwise, as you say, the result would be a jerky response. TTQS |
|
|
|
|
Logged
|
|
|
|
|
julex
|
|
« Reply #32 on: January 16, 2012, 06:56:08 PM »
|
|
|
I like my new map though, it really smoothed the things out.
|
|
|
|
|
Logged
|
|
|
|
|
jibberjive
|
|
« Reply #33 on: January 19, 2012, 11:57:05 PM »
|
|
|
Nice efforts, it really helps so every person doesn't have to re-invent the wheel every time they learn this stuff.
|
|
|
|
|
Logged
|
|
|
|
|
nyet
|
|
« Reply #34 on: January 20, 2012, 02:28:24 PM »
|
|
|
I always presumed that the interpolation was based on the relationship between the entire row or column's values be it linear, power, logarithmic or whatever, not just a simple linear one through two adjacent cell addresses otherwise, as you say, the result would be a jerky response.
TTQS
I actually recall reading through a random motronic patent regarding the table interpolation method. iirc its a simple low order poly fit done in integer math. There are (iirc) separate routines for 2d and 3d interpolation. I'll bet if we ask nicely setzi might disassemble the interpolation routines for us  I'm especially curious to know what happens when you go off the reservation: I'm a assuming it does a hard cap, and does not extrapolate... but i don't know if the cap is smoothed off. |
|
|
|
|
Logged
|
ME7.1 tuning guideECUx PlotME7Sum checksumTrim heatmap toolPlease do not ask me for tunes. I'm here to help people make their own. Do not PM me technical questions! Please, ask all questions on the forums! Doing so will ensure the next person with the same issue gets the opportunity to learn from your ex
|
|
|
Matt Danger
Full Member
Karma: +17/-1
 Offline
Posts: 116
|
|
« Reply #35 on: January 20, 2012, 03:59:44 PM »
|
|
|
Thanks BerTTos! This template is very useful. My car drives less crappily now.
|
|
|
|
|
Logged
|
|
|
|
Tony@NefMoto
Administrator
Hero Member
Karma: +132/-4
Offline
Posts: 1389
2001.5 Audi S4 Stage 3
|
|
« Reply #36 on: January 23, 2012, 12:02:02 PM »
|
|
|
I actually recall reading through a random motronic patent regarding the table interpolation method. iirc its a simple low order poly fit done in integer math. There are (iirc) separate routines for 2d and 3d interpolation. I'll bet if we ask nicely setzi might disassemble the interpolation routines for us I'm especially curious to know what happens when you go off the reservation: I'm a assuming it does a hard cap, and does not extrapolate... but i don't know if the cap is smoothed off. For 2D map interpolation, the ECU does a linear interpolation of the two adjacent cells bounding the test value. For 3D map interpolation, the ECU does two 2D map interpolations, and then interpolates those two results. The ECU hard caps the map lookups which go beyond the map range. |
|
|
|
|
Logged
|
|
|
|
|
nyet
|
|
« Reply #37 on: January 23, 2012, 12:41:46 PM »
|
|
|
Wow. Lazy motherf'ers
I guess the patent was for show, and not in ME7 and earlier
|
|
|
|
|
Logged
|
ME7.1 tuning guideECUx PlotME7Sum checksumTrim heatmap toolPlease do not ask me for tunes. I'm here to help people make their own. Do not PM me technical questions! Please, ask all questions on the forums! Doing so will ensure the next person with the same issue gets the opportunity to learn from your ex
|
|
|
|
TTQS
Guest
|
|
« Reply #38 on: January 30, 2012, 04:20:49 PM »
|
|
|
For 2D map interpolation, the ECU does a linear interpolation of the two adjacent cells bounding the test value.
For 3D map interpolation, the ECU does two 2D map interpolations, and then interpolates those two results.
The ECU hard caps the map lookups which go beyond the map range.
Wow. Lazy motherf'ers
Second that.  |
|
|
|
|
Logged
|
|
|
|
|
prj
|
|
« Reply #39 on: March 02, 2012, 12:25:38 PM »
|
|
|
That's how it's done in all the older Motronics as well.
Generally Bosch does not part with used traditions easily.
If something works, they use it for a very long time, until there is a compelling reason to switch. They are masters of copy paste.
|
|
|
|
|
Logged
|
|
|
|
|
savages4
|
|
« Reply #40 on: March 04, 2012, 06:54:23 PM »
|
|
|
I tried this calculator on my, and got extremely low values for the first few columns.. Car runs pretty good but I feel like there is way more engine breaking when the a/c is on for some reason... it also just doesn't feel as responsive down low.
|
|
|
|
|
Logged
|
|
|
|
lulu2003
Full Member
Karma: +11/-1
Offline
Posts: 242
|
|
« Reply #41 on: April 06, 2012, 06:03:22 AM »
|
|
|
sorry for being late in this discussion, but how does this make sense: when KFMIOP does not show the correspondig inverted values in the last columns? |
|
|
|
|
Logged
|
|
|
|
berTTos
Full Member
Karma: +24/-0
Offline
Posts: 91
|
|
« Reply #42 on: April 09, 2012, 09:45:19 AM »
|
|
|
sorry for being late in this discussion, but how does this make sense:
when KFMIOP does not show the correspondig inverted values in the last columns?
i played around with the last column to arrive at values that worked best for throttle at or very near to WOT and i got better results with the posted values. |
|
|
|
|
Logged
|
|
|
|
lulu2003
Full Member
Karma: +11/-1
Offline
Posts: 242
|
|
« Reply #43 on: April 10, 2012, 12:22:13 AM »
|
|
|
do you have any explanations why this is working fine for you?
|
|
|
|
|
Logged
|
|
|
|
A6turbofrance
Full Member
Karma: +0/-5
Offline
Posts: 119
|
|
« Reply #44 on: August 10, 2012, 05:21:15 PM »
|
|
|
i have use your italian kfmirl "tuner"  that work fine for me i have change my kfmiop colums value and kfzwop/kfzwop matching to kfmiop colums but i can't use your kfmiop interpolator when i click on generate nothing happens it seem only modify the table in excel ? what software u use for it ? i have open office nobody can have my kfmiop ? or help me sorted with it ? that seem at great improvment... my kfmiop row are : 13.50 / 27.00 / 50.25 / 62.25 / 74.25 / 97.50 / 120.75 / 144.75 / 168.00 / 191.25 / 219.16 thanks |
|
|
|
« Last Edit: August 10, 2012, 06:35:20 PM by A6turbofrance »
|
Logged
|
|
|
|
|
