Page 2 of 2 FirstFirst 12
Results 21 to 27 of 27

Thread: LNF KR Histograms - need feedback

  1. #21
    Tuner
    Join Date
    Jul 2013
    Location
    Golden, Colo
    Posts
    109
    You need to address the DAL and the WGDC, that is most likely where the KR is coming from. Go back to the LNF guide thread and start there this thread is a decent start http://www.hptuners.com/forum/showth...F-Tuning-Guide although his DAL is a little high iirc that was when hptuners did not have as many tables available as they do now. 255 seems to be the highest most get before no changes are noticed, so starting around 235-240 in the high load areas is not a bad choice.

    I'd also stick with the OE ACdelco's plugs there isn't really a reason to go step colder on a stock turbo or even a slightly larger turbo.

  2. #22
    Tuner in Training
    Join Date
    Sep 2015
    Posts
    20
    Quote Originally Posted by cobaltssoverbooster View Post
    ill once over the file when i get home but from the few posts i have read you lack the general understanding of the different system operations. as one of my previous posts said...i dont have any time to teach so after i once over i cannot help any further.
    Well, no worries then. This tune isn't that important to me, which is why I was seeking input about the KR histograms. I hadn't seen anyone talk about how the tables behaved, I figured I would ask. I also have no problems with starting from scratch and being more critical about my process. I'm certain with enough road testing and reading I'll eventually get the feel I want. I likely just need to verify boost is stable, then adjust my PE tables.

    Thanks for all the tips, I've learned more about my car's engine in the last 48 hours than the past year.

  3. #23
    Tuner in Training
    Join Date
    Sep 2015
    Posts
    20
    Quote Originally Posted by ECaulk View Post
    You need to address the DAL and the WGDC, that is most likely where the KR is coming from. Go back to the LNF guide thread and start there this thread is a decent start http://www.hptuners.com/forum/showth...F-Tuning-Guide although his DAL is a little high iirc that was when hptuners did not have as many tables available as they do now. 255 seems to be the highest most get before no changes are noticed, so starting around 235-240 in the high load areas is not a bad choice.

    I'd also stick with the OE ACdelco's plugs there isn't really a reason to go step colder on a stock turbo or even a slightly larger turbo.
    I actually had linearized my DAL, and I liked the feel even though it is abnormal. I reset it when I noticed the KR issue.

    I'll be rebuilding it either late tonight or early tomorrow morning since I've eliminated most of the KR at this point.

  4. #24
    Senior Tuner cobaltssoverbooster's Avatar
    Join Date
    Dec 2008
    Location
    Nevada
    Posts
    4,452
    do not boost hard before 70% load. As for your linear dals, it is fine that you want a specific feel but making them linear through the 30-60 % load will cause problems. you can adjust them and fine tune them but when you make them linear you end up with a huge air load request that causes to much cylinder pressure and heat causing early detonation.
    2000 Ford Mustang - Top Sportsman

  5. #25
    Tuner in Training
    Join Date
    Sep 2015
    Posts
    20

    Lightbulb

    Quote Originally Posted by cobaltssoverbooster View Post
    I really could have gone to town on that tune file but since you only interested in one system, i left the pile of other issues i have out so you can just focus on having clearer results later down the road. My schedule is slammed and although your posting interests me i unfortunately cannot help you out any more than this. If your still going on this when my schedule slacks up i will come back and follow it.
    Good luck mate.
    Thank you for this. I just saw it, not sure how I missed it yesterday, but it's great info.. One of the biggest lessons I've learned since my first post: I cannot and should not modify more than one system at a time, and I necessarily must zero-out (or normalize) any corrective tables that may skew my results. The hard part (and due diligence from me) is learning and understanding how certain systems/tables interact..

    A physical process I needed to change was to review logs often, in the car, after every flash. This has yielded me the best results. When I first posted this thread I would make many edits over the course of an hour (like 2-3 tables at one time), try to work out math where necessary and make rash assumptions about expected results. I would do a "full write", because I didn't know better and usually err on the side of caution, and then run a long log, and spend another hour trying to re-tune.. BIG MISTAKE Now I sit in my car, I make one change, flash, scan, run down the street, block, whatever, and then immediately pull over, let the car idle/cool and review the log. Edit, flash, scan, pull. Rinse and repeat. When testing PE changes for example I must have reflashed 2-3 times over the course of 10-15 minutes.. but it was worth it. Now I understand the true value of a dyno when doing tuning. It must save SO much time.

    Thanks again.. I've almost completely eliminated the knock and, so far, have not touched the Main Spark tables. I'm making a few edits based on this, run a few more logs and then make some very minimal corrections to spark and see where it gets me. The feedback on this thread and other referenced threads (and threads references by those) will have made all the difference to my approach when I do test:

    * tune spark only after every other system has been tuned, or expect to re-tune it if other systems are touched again.
    * when tuning spark timing, set the precision to 2 decimal places
    * disable COT for duration of testing
    1. Torque->General->EGT Torque Limit: all 100%
    2. Fuel->COT->Lambda: all 1.0
    3. Flatten "Spark Correction" IAT and ECT to remove any variance temperature would cause (not sure about AFR, since this isn't really environmental "entropy" but something we control?)
    * no radical changes to individual cells, test in small increments, similar to PE tuning, or risk of damage
    * no radical changes to individual tables, ensure final cell values are properly normalized between tables, or risk of damage
    * no radical changes to range of cells, focus on only the cells/power-range that is showing peak knock (using highest KR average as focus of attention)
    * avoid smoothing/interpolation tools, smooth/interpolate by hand

    When I am comfortable enough to test out the actual behavior I will post results, who knows, maybe I drop a valve and everyone here will go "hey, you know, we told you so." Or, more likely, I discover that individually tuning tables is a huge waste of time, and everyone here will go "hey, you know, we told you so." ... So, first in order, I need to finish stabilizing my tune before I go playing around with timing.
    Last edited by wilson0x4d; 10-10-2015 at 09:24 PM.

  6. #26
    Tuner in Training
    Join Date
    Sep 2015
    Posts
    20
    Quote Originally Posted by cobaltssoverbooster View Post
    do not boost hard before 70% load. As for your linear dals, it is fine that you want a specific feel but making them linear through the 30-60 % load will cause problems. you can adjust them and fine tune them but when you make them linear you end up with a huge air load request that causes to much cylinder pressure and heat causing early detonation.
    I think I noticed this today, I had done a pull right after re-interpolating 30-60 from stock and it introduced some silly knock, took 0-50 back to stock, leaving only 60-100 interpolated, also removed tip-in again to rule that out, and adjusted prop gain to curb the ramp-up between 1500-2200 range. If it's eliminated I plan to bring each of those, except for DAL, back to prior to understand the impact, if any (I figured for the purpose of confirming KR reduction editing all three was fine.)

  7. #27
    Tuner in Training
    Join Date
    Sep 2015
    Posts
    20

    Post

    Just an update.. I have continued to avoid editing Main Spark tables. Whether or not this is a proper approach I've no idea.. I'm still learning.. but I have eliminated a significant amount of KR just from MAF Calibration, and then MAF Correction, and then discreet MAF Corrections. For example, I did an initial 10 "MAF Calibrations" (maybe more) until I was within 1-2% variance between logs under "general" driving conditions. Then performed 10 or more "MAF Correction" updates in the same manner I did the "MAF Calibration" updates, but focused more on desired load/rpm/speed ranges.

    I have almost zero knock doing a 0-80mph pull. I get +1-ish on initial launch, Tip In i still disabled (both of them). MAF Corrections can't seem to remove the blip of KR at launch (as far as I can tell.) I have zero knock doing a 80-120mph pull.

    My thinking has been:

    1. settle on a target afr, boost, etc (i am happy with my current settings, including boost PID behavior)
    2. coerce knock (drive uphill at a low RPM, down a hill, through a tight turn.. basically put the engine under unique load that can be repeated easily AND which exhibits true KR event.

    3. replay log, pause (not a 'stop') when true KR event occurs, remain paused until step 7
    4. open ST+LT histo, right click and Clear all data

    5. open Chart Display, hover mouse over Chart putting the white line/marker where just before KR event occurs, I notice histogram does NOT provide a "scrubbing" control (or simple time input box) to do this work, this is the only way i've found -- scrollbar is twitchy at best
    6. carefully move mouse cursor up to the top and off the graph, being careful to keep the white line/marker at where I need it: at the KR event start point

    7. I collect a sample: while watching the histogram, unpause, and quickly pause just after the event completes (2.24 lacks a visual indication of where playback location is, except for when right-side of graph begins loading more data.. i.e. "graph will appear to not be updating, tune is in fact playing back." I look at the "Time" label in the status bar to get a more accurate end when necessary.

    8. I switch to ST+LT histo to "Count" mode and verify sample counts are very low, say less than 20 for the cells showing a large trim. I identify "large trim" by switching to "Max Values" mode to see the peak trim values. I then compare them to averages. High averages and High Peaks with little/no deviation is what I look for.

    9. I switch to "Average Values" mode and ONLY SELECT THE CELLS OF INTEREST (not the whole table), and then press Ctrl-C to copy the selected cells.
    10. In VCM Editor I click on the correlating top,left cell in the MAF Correction table. That is to say: If I copied the histo cell range from 3,3 to 9,11, then in VCM Editor, I would select cell 3,3 in MAF Correction table.
    11. Right click target cell, multiply by half. Verify the correct cells were updated.

    I try and be careful to only edit unique cells (easiest) for a single log (i.e. do not multiply a single cell more than once.) Since most of my KR events appear at specific RPM ranges this is easy for me to do. If I apply corrections to the same cells more than once per session I have to manually average the two results, which is a tedious pain to be doing manually over and over.

    Here are what my KR histos were before MAF Corrections (but, after MAF Calibration), and then after about a dozen MAF Correction updates, each new update more discreet than the last. For my current log there is some KR between 1750-2000 rpm (it's actually mostly around 1900 rpm). I will have corrected this with my next flash (out of time for this weekend.)

    Before MAF corrections:
    kr-before-maf.png

    After MAF Corrections:
    kr-after-maf.png

    I will probably continue doing this until I've built values up for most of the MAF correction table (some of them do look a little generic still.)

    after-maf-tune.hpl
    Last edited by wilson0x4d; 10-12-2015 at 03:57 PM.