Page 6 of 6 FirstFirst ... 23456
Results 101 to 116 of 116

Thread: Whats the torque/ inverse calculation?

  1. #101
    Did the excel work on s197?

  2. #102
    Potential Tuner
    Join Date
    Jul 2020
    Posts
    2
    Quote Originally Posted by murfie View Post
    I've been getting asked this question by a lot of people. So I figured Id just make a post.

    Its popular and yes you can use angle error and modify the torque values then calculate the inverse of the load tables. I do not suggest it as calculating the inverse for the top inverse row the torque is 0 is impossible.
    Heres why:
    Its an inverse relationship which is y=1/x.
    x being air load and y being indicated torque.
    It can also be x=1/y. This is where dividing by that 0 is impossible.

    If you look at the inverse relationship Y will go to infinity before x is 0 and x will go to infinity before y goes to 0.
    If you look at the inverse relationship indicated torque will go to infinity before air load is 0 and air load will go to infinity before indicated torque goes to 0.
    ETC torque/desired torque/ driver demand torque can all go below 0. basically you need to use this as your 1 or the "constant" to establish your base air load and indicated torque.

    Attachment 72087

    The way I suggest to do it is start by modifying the load values. I first find my desired/ETC/driver demand torque in the torque tables. The PIDs read nearly the same when things are not way off.
    To find our values to replace that 1 and get our desired torque values we can multiple the load and torque of the Y axis of the torque and inverse tables. X*Y=1 As long as your maximum driver demand falls somewhere near the bottom of the tables everything should work out. And because ETC goes below 0 we end up with a load floor for that 0 torque row. All you have to do is setup a Graph with these values in the rows and using PID ETC TQ. Then set the columns to the RPM range in your tables. I suggest making them slightly higher than your red line. Then have the table fill with your air load. Drive around trying to hit every RPM and every load to fill in the table. Filter the graph by MPs so you can apply the load to the correct MP. After applying the new load values you will need to click smooth the entire table a few times to get it smoothed out. You can then take these new values and apply them in to the attached sheet to get the new torque values. The torque values in the top table do nothing(except the top row) in this sheet so it cant be used for the angle error method.

    shouldn't take to long for all IPC error to clear up and the car driving much better. .

    Attachment 72088
    So where in HP Tuners software would I go to make these changes what tabs would it be under

  3. #103
    Tuner in Training
    Join Date
    Oct 2015
    Posts
    30
    I've read this whole thread and vacillated between oh I see, to completely lost many times...I know I should get it my attempt to use known torque values today was a disaster...
    Reverted to Stock Edelbrock...with tweaks on timing ect....
    The last few post on tq flow thru the tune helps some. Couple questions..

    Let's say all im trying to do is match the power of a tune that was giving me trans problems holding a gear...I know the load it is in the DA im running with the PD supercharger ( surging has to wait to later)...
    After failing miserablely to scale the torque and inverse/load tables to what I know using peak tq of my combo...
    .I reverted to simply making the spark and lambda settings match, regardless of the refinement in these tables, how they act upon each other....
    If I manage at the same VCT OP angles to achieve the same timing, and lambda values with the tb full open on shift, is it not physically making the same power no matter what the tq tables indicate or run like up to that point? Its all inferred from the MAF seems like so if I get similar maf data, timing, lambda target.and VCT...at shift I can at least test if a gear is holding?.because physically it's actually putting out power as long as throttle body is wide open, spark is same and same fueling?
    If so my trans seems to be fine with just slightly more tq reduction than the custom tune runs aka none.....
    Also couldn't I reverse engineer at least peak power/TQ if I know all the variable aka load, spark, maf flow and vct angles...somehow reverse engineering it? I'm not trying to steal anything the tuner won't take any heat off the tune and im trying to decifer to I need my trans rebuilt the 2nd time, a valvebody, or just a little less power applied on the 2-3 ( blips the tires even with abs module unplugged during the insano hard shift; i was told the trans is going out to rebuild it so I did...now its same thing and im told its normal to leave em alone), so im stuck on my own usual 3 engines and 2.5 trans in 10k miles, old man don't even run at the track just smash in country now and then....

    When reverting back to the 50 states legal tune with several tweaks to match said angles and timing/fueling with a little tq reduction left in...so far I have completed several ( about 5 with no closing of throttle body but is pulling some timing; i presume due to knock sensitivities, numbers are higher in each individual cylinder tab vs stock) 2-3 shifts with a lower blip of tire spin and pulled out 3rd gear fine ( previously erroneously shifting to 4th) even when shifted way to low and pulling hella load....
    Now this tunes load is approaching 2...I followed a recipe and know it should make about 720whp and 633ft lbs of torque literally 600 from maybe 2700 to 5600 lol...their tune is 1.77-1.78 and feels butt dyno much stronger ( though it may be accelerating faster and peaks are same )....
    My attempts to reverse the torque tables to match that 1.77 ish were a disaster just trying to get apples to apples comparison to see if im on my 3rd engine and 2 or 2.5 trans ( valvebody or new tune i don't have money for both at Christmas hehe)...

    Thanks....I have become a very dumb old man....

    Sorry Gen 3 Coyote, 10r80 car, 10.5psi boost PD blower...

  4. #104
    Here's how it works, may differ what you've seen before, but i will simplify this as much as possible and it will help other users, lot of threads and guesses how it works. I had to make it long time ago, but had no time.
    But yesterday was good occasion to create logs to show how it works, i started tune my GT350 with Cobra Jet, cams, 163mm TB without having any data and it took me few hours to make car drivable like stock.

    I realised right after it was released, that TQ inverse calculator is screwed (sometimes sets 0, sometimes crazy high values) and finally not really needed and i don't use it. Also Murfie's calculator is not needed doing it my way.

    Here's how it works:

    Driver Demand (it's ETC Torque Request)->thats' converted by ETC Predicted/Effective/SD/MAF to Air Load -> now Airload it's your Y axis in Torque table, having RPMS and current air load it easy find torque. Torque in torque tables is in real Engine indicated Torque reference (Sheduled Torque)-> which finally is Y Axis of Torque Inverse table-> and Torque Inverse values are Desired load.

    Now having that info it's easy.

    Some important information.
    1. Torque in Torque tables limits ETC Torque Request, that's why many users dealing with raising Drived Demand (so ETC torque request) gets no results because DD is getting higher values than it Torque table.
    Torque table values can be raised, then it takes values from lower rows in Inverse tables (so usually higher desired load). So this way we can adjust also Driver demand feeling.
    2. Torque Inverse table, so Desired Load, if set to low, creates torque limits, bacause Air Load is to high, if i'ts sets to high, it will override Airload and will open throttle more.
    3.Keep in mind, changes in SD model changes Airload for same amount of MAF
    4. X axis in Driver Demand, so pedal position is pid Acceleration position D (SAE), if you need to log where you are.
    5. All values are somehow filtered by many variables you already know it), but that's not so really important

    So to fine tune torque tables assuming SD and ETC data are correct, in most cases, it's enough to log all MP you use with histo set to Y axis Engine indicated Torque reference, X RPMS and Air load as values and interpolate them in inverse tables without touching torque tables. If ETC data is correct and SD is bit screwed, it will work still, you will just adapt inverse tables to what you have. What i'm doing usually, histogram already plots Air Load multiplied by 1.03 (3 percent more) as start point to get desired higher and avoid IPC errors, later on i use airload error percent/2. Remember, before you start, raise tq errors to max to avoid limp mode. After tuning, you can return them to stock

    You can also easy adjust Torque tables first if you like, by putting there logged ETC torque request multiplied by 1.08-1.10 as start point (keep in mind you can do that in few steps, as TQ table can limit logged value from ETC request)

    Things may get bit complicated if ETC model is screwed, but that's simple to fix if you read my other very old thread about throttle tuning. If after play with inverse 3-4 times you are getting higher values of Desired Load and also more higher values than before of Air Load, easy fix is to correct Predicted throttle to bit lower values, no need to recalculate ETC Effective area .

    I'm not writing here how Air Load is calculated, there are other threads about MAF/SD/ETC, but that doesn't really matter in this case if it's not much off. If you get in NA Airload 0.9-1.1 and in boosted eg 1.5 for 0.5 bar of boost or close, it's fine.

    I'll post updates if i have some, but that works for me perfect for years, cams tuning or boosted car is easy with this method.
    So same applies to boosted cars with stock OS and boosted OS like Whipple or Roush made by FP. But remember, if you use stock OS and boost, you have to adjust ETC PR Ratio, so it calculates more air over TB than it does in NA, usually left axis raised to 3 on bottom and then adjust values.

    I prepared 3 tunes with different traps in Driver Demand table TQ and Inverse tables (see MP-15-20), once you will compare traps i set, and logs, you will see it's all easy to understand how it works and how you can check itself.

    50 2nd gear Driver demand traps and inverse
    51 DD same, but some changes in inverse
    52. modified DD and added TQ traps.

    Added also screens of histos i use (mapped - don't look at Y axis as values, are from car, they should be this same as Y values in tune), made it fast today morning

    If there are any errors here, let me know (also language - greetings from Poland)

    Pawel
    Attached Images Attached Images
    Attached Files Attached Files
    Last edited by caniggia; 11-01-2021 at 05:02 PM.

  5. #105
    Senior Tuner
    Join Date
    Jan 2013
    Location
    Hawaii
    Posts
    2,101
    I still think it is very important to keep the torque model relationship inverse from one table to the other. Torque To Load(TTL) and Load To Torque(LTT) and I think HPT's calculator does that very well. If none of the below makes sense, just know the torque model is scheduled torque> desired load and indicated torque> actual load. The relationship you skew in these two tables is desired load/ MAF and actual load/ MAF. Scheduled torque and indicated torque will always be the same.

    As Caniggia mentioned its important to know the flow and it makes it easy.

    You control pedal position. Pedal position gives a torque demand. Driver demand torque in CL control or ETC torque in OL control. CL meaning demand and engine brake torque are kept equal, OL meaning they are allowed to diverge. This torque gets converted to a scheduled torque(torque demand on top of idle torque). Scheduled torque removes the variables caused by temperatures, spark, air fuel ratio. This is then used in the torque model to get a desired load. From a desired load and current RPM a desired MAF. The throttle body model then opens/ closes to get actual MAF to this desired MAF.
    Driver demand is not always the control, in some conditions the ECU takes over e.g.: Idle or torque limits. The ECU manipulates indicated torque reference. Indicated torque reference is equal to scheduled torque. Again variables caused by temperatures, spark, air fuel ratio are removed because of their nature, making these torque values great for consistent control, they make the pedal and the ECU agree on demanded torque control of desired load /MAF at the current RPM.

    You can tell when you are in control and when the ECU has taken over by if indicated/ scheduled torque is following just above ETC/torque demand or not.

    The model is desired load to scheduled torque in one table. air load and indicated torque in the other. This is why I think keeping them in an inverse relationship is important. It helps keep desired load and air load equal. So your pedal and the ECU are on the same page about what is happening with the engine brake torque.

    The demanded torque/ ETC torque is not 8-10% below scheduled/ indicated torque. it will be just a static 40-60 ft lb below, each car model variation is a little different. This is the torque at idle conditions and return to idle. Mostly a set throttle angle and torque is controlled by spark to meet a desired RPM. You want the demand negative in decel, and 0 at idle. This throws people off when dealing with the torque model as most discussions about the tables are not about idle but part throttle accel and decel in the higher RPMS / higher loads. But most people first look at and try to make sense of these tables at idle.



    MAF sensor and engine speed is all you need to get air load.

    MAF divided by engine speed = weight of air per revolution

    "weight of air per revolution" divided by "1/2 number of cylinders" = weight of air per cylinder intake event

    weight per cylinder divided by weight of air in full cylinder displacement = air load
    None of this math matters as you can log both air load and MAF, but its good to know when you are dealing with PID limits.



    SD only gives you air load if you are using a MAP sensor.
    The throttle body model is only to speed up determining the blade position needed to correct actual mass airflow to meet desired mass airflow. Or if you have a MAP sensor actual manifold pressure to meet desired manifold pressure. Its also used to make sure the throttle is operating correctly and not stuck open or closed, but thats not really about the actual control of it.
    Last edited by murfie; 10-31-2021 at 12:18 AM.

  6. #106
    Senior Tuner veeefour's Avatar
    Join Date
    Oct 2016
    Location
    Poland
    Posts
    1,711
    Ignoring the inverse relation in TM is just a horrible advice. PCM checks the LOAD in Inverse tables first granted, but also checks the torque relation at the same time LOL


    Multiply this by 1.03, multiply that by 1.1 - Jesus why we ever care to STUDY the patents if you can multiply everything...

  7. #107
    Thanks for answer Murfie.
    I'm not saying it's wrong to get relationship between both and understand all maths there and all info you posted, helped lot.

    But i spent some time to make small compare.

    I used stock TQ and Inverse values as base and my tested TB data before so it's not messing, nearly finally tuned car except WOT, just returned TQ and inverse tables to stock.

    I made 10 passes using your calculator adjusting tq and recalculating inverses. Still lot of things were off and drivability was far from ideal but getting slowly better and better (see i have in this test car TB, cams, CJ in car, so lot of things are off from start).

    Then i took same start tune and adjusted only inverse in 3 passes and nearly all is inline and drivability went to stock. First pass airload logged multiplied by 1.08 to make it higher than real to avoid IPCs, then 2 passes with airload error percent divided by half.

    Maybe tommorow if i'll have time, will test to put your method on top of mine and will see if drivability reamain same and maybe tq will line up with inverse, but i never needed to do that. If it will, it will speedup process when lot of things are off.

    Cheers
    Eastern and Central Europe American Muscle and Harley-Davidson tuning
    www.hd-customs.pl
    http://www.facebook.com/hdcustomspl

  8. #108
    Senior Tuner
    Join Date
    Jan 2013
    Location
    Hawaii
    Posts
    2,101
    I'd prefer you compare your method to using HPT torque inverse calculator.

    Maybe share how some one can log some kind of error and apply it to that, and have it correct drivability.

    The sheet in this thread was extremely limited. Due to the criticality and rigidness of the axis values. e.g if you have 1.0 load at 440ftlbs, 44 ftlbs has to be lined up with 0.1 load.



    The error is between:

    demanded torque> demanded load> demanded MAF or MAP> demanded throttle angle.

    and

    actual torque(engine brake torque) > Air load> MAF or MAP sensor> actual throttle angle.

    Out of all of these to calculate an error percentage to correct scheduled/ indicated torque values, too apply to the HPT calculator and calculate the inverse table from. It should be pretty obvious which to use in a custom math, although they all could be used as they all mean the same thing.
    Last edited by murfie; 11-02-2021 at 01:24 AM.

  9. #109
    Please see this.

    01. is base whipple tune i just loaded into car, see diiferences in airload 1.5x and desired load 1.2x, you feel it's not smooth in car.
    02. is 2 pass with histos you see on screenshot, airload = desired load, all other things also inline and smoothed WOT to stock feeling, smooth like butter in summer.

    Math used for airload error is 100 * ([2323] - [19076]) / [19076] for fast adjust, you can use (100 * ([2323] - [19076]) / [19076])/2 if you want to get there with smaller steps.

    No tq adjust, no inverse calculation, just copy from scanner and paste as multiply percentage.

    Attached also histo for TB, where calculation is negative airload error -(100 * ([2323] - [19076]) / [19076]). No need to use if you have known TB data, but if you don't, also adjusts it quick. I suggest also to divide it by half and go smaller steps.
    Usually 2-3 passed and all done, difficult setups where lot of things are off, 5-10 passes.
    Attached Images Attached Images
    Last edited by caniggia; 11-03-2021 at 03:27 PM.
    Eastern and Central Europe American Muscle and Harley-Davidson tuning
    www.hd-customs.pl
    http://www.facebook.com/hdcustomspl

  10. #110
    Tuner in Training
    Join Date
    Nov 2021
    Posts
    25
    Hi murfie,

    I've been trying to figure out for a few days but with no success, so I need to ask for explanation.

    Why do you paste the load readings from the histogram with modified torque values (Y axis) to the original inverse table (in the tune) with original Y axis values that don't match the values from the histogram? I can't understand why this is correct. It seems to be incorrect, since you paste load values related to a different torque for each row.

    When you log the load values in histogram with different Y axis and simply copy-paste this to a table with different Y axis - the values will be not correct in the inverse tables. Could you please explain what is the logic here ? I can't sleep after reading this thread

    Could you please also explain what is the purpose of putting in the histogram the Y-axis values that are the multiplied LOAD Y-axis values (from the torque tables) * Torque Y-axis values (from the inverse tables)?

    Why just not make a histogram that has the same X and Y values as the Inverse Tables, log the LOAD and paste it to the inverse tables, then calculate the torque tables?
    Last edited by lambda1; 12-25-2021 at 03:50 PM.

  11. #111
    Tuner
    Join Date
    Jan 2021
    Location
    Northern Virginia
    Posts
    143
    I made this spreadsheet for scaling my torque tables for boost. It uses quadratic curves to extrapolate to higher loads (the stock tables fit these curves with surprising accuracy). Just paste in the stock tables (use the "copy with axis" function in the editor) and enter the load values you intend to scale to. Then use the calculator to get the inverse (make sure the inverse Y axis is scaled up too).

    Ecoboost tables are much more linear, so I'd be interested to hear some thoughts.

    I've been recopying the first three rows of my inverse tables because the calculator likes to zero them out, and I don't think they should be (in fact, I don't think zeros should be anywhere on the inverse table at all).

    Not that my opinion matters, but I agree that the inverse relationship shouldn't be hacked up.

    Torque Calculation.xlsx


    Screenshot 2022-01-27 234002.png
    Last edited by RobCat030; 01-27-2022 at 11:36 PM.

  12. #112
    Senior Tuner CCS86's Avatar
    Join Date
    Nov 2017
    Location
    Austin
    Posts
    1,089
    Quote Originally Posted by RobCat030 View Post
    I made this spreadsheet for scaling my torque tables for boost. It uses quadratic curves to extrapolate to higher loads (the stock tables fit these curves with surprising accuracy). Just paste in the stock tables (use the "copy with axis" function in the editor) and enter the load values you intend to scale to. Then use the calculator to get the inverse (make sure the inverse Y axis is scaled up too).

    Ecoboost tables are much more linear, so I'd be interested to hear some thoughts.

    I've been recopying the first three rows of my inverse tables because the calculator likes to zero them out, and I don't think they should be (in fact, I don't think zeros should be anywhere on the inverse table at all).

    Not that my opinion matters, but I agree that the inverse relationship shouldn't be hacked up.

    Torque Calculation.xlsx


    Screenshot 2022-01-27 234002.png



    I'm curious what data you are looking at. The stock torque and inverse tables for my 2012 GT are almost perfectly linear:

    tq.jpg

  13. #113
    Tuner
    Join Date
    Jan 2021
    Location
    Northern Virginia
    Posts
    143
    These are stock tables from my 2013 Focus. Just to check I ran a 2012 Mustang as well as a 2015 Focus ST. On all of them the quadratic fit is just slightly more accurate than the linear (no more than 1 or 2%, but I'll always go with the most accurate fit).

    I'm not sure why mine is so particularly curved, it may just be because the overall range of HP on the stock tune is relatively lower. If I extrapolate them linearly (R^2=99.8%) the torque values would soar way above what I believe I'm actually making by about 50 ft/lbs
    Last edited by RobCat030; 01-28-2022 at 04:42 PM.

  14. #114
    Murfie & Caniggia sent you guys a PM

  15. #115
    Tuner in Training
    Join Date
    Dec 2023
    Location
    Florida
    Posts
    24
    Thanks so much to R8Bill and murfie for posting this, it has been very helpful in figuring out how to adjust my torque modeling. I wish there was a way to give thumbs up or at least @ them in a post to let them know.
    - Brad
    2014 F150 Raptor
    6.2- 11:1 Compression, Long Tube Headers w/Cats, Volant Cold Air Intake, 180* thermostat.

  16. #116
    Senior Tuner
    Join Date
    Mar 2022
    Location
    Indiana
    Posts
    2,775
    Quote Originally Posted by trick76cj5 View Post
    Thanks so much to R8Bill and murfie for posting this, it has been very helpful in figuring out how to adjust my torque modeling. I wish there was a way to give thumbs up or at least @ them in a post to let them know.
    You can. Hit the star in the bottom left under the username in the post.