|
s5fourdoor
|
 |
« Reply #15 on: April 03, 2012, 12:53:55 AM »
|
|
|
also - don't touch the _UM maps or you will end up with a bricked ecu on your first startup (boot loop accompanied by twitching throttle plate sound familiar?). thx to phila-dot for advising me on those (used for calibration).
a 100x this. thanks for clarifying it! |
|
|
|
|
Logged
|
|
|
|
|
s5fourdoor
|
|
« Reply #16 on: April 03, 2012, 12:58:19 AM »
|
|
|
I love this thread. This exact topic has been on my mind, but I wasn't going to try to address it until I got a couple of other things ironed out. I definitely think there is some good optimization that can still be done with this regarding bigger turbos.
Awesome file! Did you ever get around to trying the RS4 interpolation that you did on your previous excel file where you convert RS4 to S4 and scaled the top end? How did it run if so?
get this, i definitely did. the car felt massively intensely sharper than ever with throttle response. however i backed off that path when i encountered the _UM bootloop failure. the biggest change however was moving over the RS4 PID maps. i literally celebrated with a bottle of champagne because the car finally felt like a normal street car after having blown my k03's in week 3 of s4 ownership. i'm going to do an intensive part-throttle logging session fairly soon to sort out my final fuel corrections. after that's done though i'll certainly switch my kfmirl and kfmiop back again. one of my observations from this also was that you shouldn't run low values below 2/3 on kfmiop table and also that your first few columns shouldn't be played with too much... this echo's berttos's sentiments. however the inverse facts that if max rlsol goes from 191 to 248, then 4% torque should goto (4%) / (248 / 191.25 ) = 3.085% because these low values of kfmiop. i'll have quick peak at stuff tomorrow at work, but i think most of my maps on the s4_rs4 tab (tab 3) are ready to be used. be careful with the _UM stuff and anything not KFMIRL / KFMIOP in name... tbh |
|
|
|
« Last Edit: April 03, 2012, 01:10:20 AM by nehalem »
|
Logged
|
|
|
|
NOTORIOUS VR
Administrator
Hero Member
Karma: +58/-7
 Offline
Posts: 1056
|
|
« Reply #17 on: April 03, 2012, 06:48:31 AM »
|
|
|
I've modified the _UM maps many times.. never had an issue unless I was relying on the MTX plug in to do checksum corrections... that never worked for me and always bricked the ECU.
When ever I touch the TQ maps now I check the files in WinOLS (MTX plug in will say the checksums are corrected) as it finds the ones the MTX plug in doesn't see and the file then works as expected.
That said, from what I see and am reading here at this point is to keep the stock KFMIRL/IOP maps from the M-box and tweak them instead of replacing them with something else. And while I agree the RS4 box maps (even just IRL) has more resolution there must be enough there to manipulate for us to get the desired results even with all the TQ monitoring on/enabled no?
|
|
|
|
|
Logged
|
|
|
|
NOTORIOUS VR
Administrator
Hero Member
Karma: +58/-7
Offline
Posts: 1056
|
|
« Reply #18 on: April 03, 2012, 07:04:43 AM »
|
|
|
examine my K04 file that i posted and note the axis changes and the changes to the Level 1 TM tables.
that file is perfectly smooth at all throttle inputs and has not once exhibited a Level 2 intervention in over 20,000 miles. it is also pretty aggressive and is a hoot to drive, even just down the street to get coffee.
Unless I've missed it I don't see where you show/list which tables are TM1 and which are TM2... would you mind listing the table names for that? I'm going to examine your STG3 file right now though... Thanks! |
|
|
|
|
Logged
|
|
|
|
berTTos
Full Member
Karma: +24/-0
Offline
Posts: 91
|
|
« Reply #19 on: April 03, 2012, 07:38:50 AM »
|
|
|
Are you sure about this? I found more interventions with the MAF too HIGH, not too LOW.. a lot of my throttle/wg interventions went away when I slightly underscaled my MAF.
I'm still tracking down the source of my timing interventions... do these also fall into two types if intervention?
you'll know a Level 2 intervention by the EPC light and having to cycle the ignition. I've only noticed it (as it relates to MAF) with an artificially low MAF signal for a moment in an otherwise healthy MAF scale as would be caused by turbulence and i've also noticed it with a very large vacuum leak. i bet you were experiencing Level 1 interventions. what i have observed so far is that Level 1 will attempt to get things under control via WG, throttle plate and ignition and if this fails and the Load is still abnormally high then Level 2 intervenes for safety. example - if you raise KFMIRL and leave KFMIOP stock and leave all Torque management stock - you will likely never encounter a Level 2 intervention. Level 1 interventions will be experienced as a terrible throttle response, surging, light bucking etc under part throttle. now - try loosening or FFing Level 1 maps RLVMXN RLVSMXN KFMIZUFIL KFMIZUOF and your vehicle will feel like it's going to catapult itself into the car in front of you at part throttle. it will be wild and nearly uncontrollable because of the grossly miscalculated load values (it will also be exciting and will give you hope). Level 2 will intervene with a throttle cut, EPC light and limp mode. |
|
|
|
« Last Edit: April 03, 2012, 08:01:04 AM by berTTos »
|
Logged
|
|
|
|
berTTos
Full Member
Karma: +24/-0
Offline
Posts: 91
|
|
« Reply #20 on: April 03, 2012, 07:52:34 AM »
|
|
|
I've modified the _UM maps many times.. never had an issue unless I was relying on the MTX plug in to do checksum corrections... that never worked for me and always bricked the ECU.
When ever I touch the TQ maps now I check the files in WinOLS (MTX plug in will say the checksums are corrected) as it finds the ones the MTX plug in doesn't see and the file then works as expected.
thanks for sharing this! - i was, indeed, using the MTX plugin. That said, from what I see and am reading here at this point is to keep the stock KFMIRL/IOP maps from the M-box and tweak them instead of replacing them with something else. And while I agree the RS4 box maps (even just IRL) has more resolution there must be enough there to manipulate for us to get the desired results even with all the TQ monitoring on/enabled no?
take KFMIRL ad multiply all data points by .25 - .35 to start. you will need to completely recalculate KFMIOP. i also removed the first x axis value (9.7502) and shifted all columns to the left and changed the 18 value to a 13.5 (we don't need that much resolution down low) to give me more resolution in the high load ranges. notice my last 3 columns become 175, 216 and 248. once you have an axis then you can interpolate data points from your KFMIRL values. now, realize that the KFMIOP axis is shared with KFZWOP/2 maps so you will need to address those as well. see my file for an example. |
|
|
|
|
Logged
|
|
|
|
NOTORIOUS VR
Administrator
Hero Member
Karma: +58/-7
Offline
Posts: 1056
|
|
« Reply #21 on: April 03, 2012, 08:03:53 AM »
|
|
|
The more I look at your file and the more we discuss (also from some recently playing around with tables these past few days) this the more I see what is going on and just to see if I am going in the right direction here.
RLVMXN
RLVSMXN
^^^ These two are basically TQ limiting maps are they not? Looking at the FR they basically set the load limit for the IOP table it seems. Although the SMXN map is not really something our car uses is it? We don't have an active intake manifold so that SU switch would probably never flip to the RLVSMXN map I would think?
|
|
|
|
|
Logged
|
|
|
|
|
phila_dot
|
|
« Reply #22 on: April 03, 2012, 08:50:29 AM »
|
|
|
The more I look at your file and the more we discuss (also from some recently playing around with tables these past few days) this the more I see what is going on and just to see if I am going in the right direction here.
RLVMXN
RLVSMXN
^^^ These two are basically TQ limiting maps are they not? Looking at the FR they basically set the load limit for the IOP table it seems. Although the SMXN map is not really something our car uses is it? We don't have an active intake manifold so that SU switch would probably never flip to the RLVSMXN map I would think?
Input to KFMIOP in %MDMAX should be rlmax_w since SY_TURBO = true. RLVMXN and RLVSMXN should have no effect. |
|
|
|
|
Logged
|
|
|
|
NOTORIOUS VR
Administrator
Hero Member
Karma: +58/-7
Offline
Posts: 1056
|
|
« Reply #23 on: April 03, 2012, 09:02:42 AM »
|
|
|
ah true... I suppose both of them would only be used for N/A mapping then. so this is what MDMAX looks like for us then:  |
|
|
|
|
Logged
|
|
|
|
|
jibberjive
|
|
« Reply #24 on: April 03, 2012, 09:56:52 AM »
|
|
|
So, I starting modifying the torque management last night (I was not going to worry about it for a while), and HOLY COW does it make a difference in both around town power and part throttle smoothness. I went a little gung-ho to begin with to see if I get tq management issues, and I'm not sure if I got any (didn't have any points that felt "corrected", but haven't checked the logs other than the normal variables). Let me know any data you guys want to see, I've got 4 FATS pulls as well as a bunch of driving around town. One thing I did notice was that my cold start revved at like 2k RPM for the first 10 seconds, much higher than normal. My changes I made to this file:
KFMIRL - I took the RS4 values interpolated for the S4 scale from nehalem's document, and multiplied everything by a scalar so that the max load is 248 (it's even more aggressive than your file, Berttos, and yours is the most aggressive I've seen). I purposely went aggressive, like I said, to see if I have issues.
KFMIOP - Load axis changed to Berttos', RPM axis modified last two rows up to have some resolution at higher RPM's, used KFMIOP interpolator sheet with the above KFMIRL.
KFZWOP/2 - Load axis shared with KFMIOP, RPM axis also extended on top end, values are modified from stock , but are the same from my previous files
KFPEDR_0_A - back to stock (was previously linearized)
KFPED_0_A - from nehalem RS4 to S4 interpolate sheet
KZMFIOUL, KZMIZUNS, KMFIZUOF - from nehalem RS4 to S4 interpolate sheet
I skipped all of the _UM stuff and throttle body stuff (except for my KFWDKSMX which was already modified).
Very real and noticable gains in smoothness and part throttle. It felt like I added mini K03's for around town power, and there was zero part-throttle bucking around parking lots etc. Really smooth. 2nd gear WOT felt as good as ever, but 3rd gear came on a little less quickly when doing my log ramp up's from approx 2500 RPM. Accel g's were a little down from previous file, as well as boost coming in a tiny bit later. FATS time slightly down as well. Timing was about the same according to zwout. The file felt better/smoother in every aspect, except sheer rampup during 3rd gear log pull.
The whole bricking the ECU thing scares me a little bit, but I used the MTX checksum plug-in after modifying the above, and I've turned the car off more than 3 times since. I also just ran the file through ME7check just to be safe. MTX worked fine for the above maps.
|
|
|
|
|
Logged
|
|
|
|
|
s5fourdoor
|
|
« Reply #25 on: April 03, 2012, 11:48:58 AM »
|
|
|
KFMIRL - I took the RS4 values interpolated for the S4 scale from nehalem's document, and multiplied everything by a scalar so that the max load is 248 (it's even more aggressive than your file, Berttos, and yours is the most aggressive I've seen). I purposely went aggressive, like I said, to see if I have issues.
Yes this is exactly what I did, and I'm sure you believe the hype I wrote about yesterday now that you've tasted the results. Its insane though right? Just goes to show how much time Cosworth / Audi spent tuning the RS4 engine. The part-throttle is a stupid-good improvement compared to the stock scaled KFMIRL table. However, I have also tried this another way and it worked even better. Take the S4_axis RS_4valued (tab 3) KFMIRL. Scale columns 60/70/80/90/99 to original * (248 / 200). Then take column 50 and average the values between columns 40 and 60 to create a new column 50. Now your part-throttle off-boost is the RS4, your part-throttle on-boost is a super-RS4. This leaves only one issue. The RPM surging caused by strange values in KFMRIL columns 0 and 10. Here's my thoughts on that aspect. Either leave the stock S4 column 0, or average the RS4 columns 2 and 5. Remember this is a proxy for power/boost, so there's no reason the idle should have changed because you have a a beast of a turbo setup. The beauty of our setups is we all run the same cams, same heads, same CR, and the same (relatively) timing curves. Nothing's changed about the car's idle. Anyways, I also never figured this point out, but I'm just providing color on my engineering process. |
|
|
|
|
Logged
|
|
|
|
|
s5fourdoor
|
|
« Reply #26 on: April 03, 2012, 12:44:41 PM »
|
|
|
hey, here's what i was saying before: could you give this KFMIOP and KFMIRL a try and see what happens with your logs?
given jibber's previous post, it seems like he already has the other maps modified sufficiently to avoid problems, so maybe its best for him to post logs- would be excited to see NVR or berTT try it too...
|
|
|
|
« Last Edit: April 03, 2012, 12:46:47 PM by nehalem »
|
Logged
|
|
|
|
|
jibberjive
|
|
« Reply #27 on: April 03, 2012, 05:21:19 PM »
|
|
|
Possibly later tonight. What logs are you looking for?
|
|
|
|
|
Logged
|
|
|
|
|
s5fourdoor
|
|
« Reply #28 on: April 03, 2012, 05:29:49 PM »
|
|
|
nothing in particular, just would like to hear some results from trying it out, like general observations. if you take full MarkP style logs that'd be cool too...
|
|
|
|
|
Logged
|
|
|
|
|
nyet
|
|
« Reply #29 on: April 03, 2012, 05:31:30 PM »
|
|
|
This thread has shown a definite dearth of logs that actually SHOW the torque monitoring variables of interest. Hint: spec load isn't gonna tell us anything  that template is utterly useless. |
|
|
|
|
Logged
|
|
|
|
|
