Page 5 of 6 FirstFirst 123456 LastLast
Results 81 to 100 of 116

Thread: Whats the torque/ inverse calculation?

  1. #81
    Tuner in Training
    Join Date
    Feb 2019
    Posts
    26
    Quote Originally Posted by CCS86 View Post
    Not all TQ errors are cause by the TQ/inv tables though. Issues with your MAF transfer function and throttle body tables can be to blame. Possibly even speed density tables, if you are using cylair anticipation and filtering.
    Thanks for rsponding. My car is a 2017 3.7L, It's 100% Speed density no MAF at all. I've installed twin f150 ecoboost turbos. I'm trying to start out conservative. I'm maxing out around 7psi of boost on just spring pressure.

    I've been locking in mapped points and tuning the SD tables individually. As I started dialing them in, I began getting all this IPC error.

    I've been using a method I found somewhere on the forum to tune the TQ maps and using the built in calculator to populate the inverse. This helped a lot, but not completely.

    This morning I extended out the torque axis on my inverse table then used the built in Inverse calculator to populate the table. My IPC errors almost vanished, but it threw my SD table way out of wack. I'm going to try and bring it back in line tomorrow.

    One thing I notice is that when I finally seem to get the SD tables dialed in, the engine starts to run away at low loads. Any suggestions on the cause and how to tune this out?

    Thanks for any help, I'm learning as I go.
    Last edited by Boody78; 10-16-2019 at 01:20 AM.

  2. #82
    Senior Tuner CCS86's Avatar
    Join Date
    Nov 2017
    Location
    Austin
    Posts
    1,089
    Quote Originally Posted by Boody78 View Post
    Thanks for rsponding. My car is a 2017 3.7L, It's 100% Speed density no MAF at all. I've installed twin f150 ecoboost turbos. I'm trying to start out conservative. I'm maxing out around 7psi of boost on just spring pressure.

    I've been locking in mapped points and tuning the SD tables individually. As I started dialing them in, I began getting all this IPC error.

    I've been using a method I found somewhere on the forum to tune the TQ maps and using the built in calculator to populate the inverse. This helped a lot, but not completely.

    This morning I extended out the torque axis on my inverse table then used the built in Inverse calculator to populate the table. My IPC errors almost vanished, but it threw my SD table way out of wack. I'm going to try and bring it back in line tomorrow.

    One thing I notice is that when I finally seem to get the SD tables dialed in, the engine starts to run away at low loads. Any suggestions on the cause and how to tune this out?

    Thanks for any help, I'm learning as I go.


    Modifying your TQ tables won't screw up your speed density tables. They aren't dependant on each other.

    How exactly are you going about tuning your speed density tables?

    Have you modified the torque tables outside of extending them out at the top?

  3. #83
    Tuner in Training
    Join Date
    Feb 2019
    Posts
    26
    Quote Originally Posted by CCS86 View Post
    Modifying your TQ tables won't screw up your speed density tables. They aren't dependant on each other.

    How exactly are you going about tuning your speed density tables?

    Have you modified the torque tables outside of extending them out at the top?
    I'm tuning the SD tables by locking in each specific map point and applying the method outline here https://forum.hptuners.com/showthrea...ity-Calculator

    The only changes I made were to the inverse table. I scaled up the last 3 rows on the TQ axis, then used the built in TQ inverse calculator to repopulate the table. On my next log I experienced practically no IPC error, but my fuel trims were way way off from what they had been.

    I did some more iterations of the above method on my SD table and now it's back in line.

    My main issue now is that the engine hangs when I clutch in and the revs don't drop at all.

  4. #84
    Senior Tuner CCS86's Avatar
    Join Date
    Nov 2017
    Location
    Austin
    Posts
    1,089
    Are you using an external MAP sensor to correct SD?

  5. #85
    Senior Tuner
    Join Date
    Jan 2013
    Location
    Hawaii
    Posts
    2,101
    His car has an oem map sensor, he just corrects the load values with fuel trims.

  6. #86
    Senior Tuner
    Join Date
    Jan 2013
    Location
    Hawaii
    Posts
    2,101
    The equivalent would be putting a external maf sensor on it and double checking the load value the fuel trims are getting you too are accurate.

    Do the RPMs drop when you put it in neutral while moving?

    Does that system even have a BOV, if so where is your BOV located in relation to your throttle body? Is it functioning correctly. With out a TIP sensor like the f150s have the ecu is thinking barometric pressure is at the TIP. It could be much higher and the ECU is just matching throttle to calculated airflow from the MAP sensor. Some times limiting the inferred BARO can help if this problem is intermittent. Its most likely a mechanical thing that the TIP sensor usually can figure out, but with out it you may need to modify to optimize for your stock calibration.

  7. #87
    Senior Tuner CCS86's Avatar
    Join Date
    Nov 2017
    Location
    Austin
    Posts
    1,089
    Ah, I forgot he was on an SD car!

  8. #88
    Tuner in Training
    Join Date
    Feb 2019
    Posts
    26
    Quote Originally Posted by murfie View Post
    The equivalent would be putting a external maf sensor on it and double checking the load value the fuel trims are getting you too are accurate.

    Do the RPMs drop when you put it in neutral while moving?

    Does that system even have a BOV, if so where is your BOV located in relation to your throttle body? Is it functioning correctly. With out a TIP sensor like the f150s have the ecu is thinking barometric pressure is at the TIP. It could be much higher and the ECU is just matching throttle to calculated airflow from the MAP sensor. Some times limiting the inferred BARO can help if this problem is intermittent. Its most likely a mechanical thing that the TIP sensor usually can figure out, but with out it you may need to modify to optimize for your stock calibration.
    Thanks for the help guys,

    The system uses vacuum operated re-circulation valves. They are connected directly to the turbo housings, I installed a couple nipples into a TB Spacer I already had for the vacuum reference. I plumbed the re-circ tubing up to the intake pipes about 20" or so before the turbo just like they are in the truck. The problem only starts to show up after several iterations of SD corrections(just when the trims start getting within 5% +/-). When I let off the throttle in gear the car will maintain speed as if on cruise control, and if I clutch in the revs will stay constant or even start to increase. If I use the stock SD map this doesn't happen, but my trims are way lean. I was thinking about possibly trying to blend the top 3 rows of the map with the stock values as this seems to be a low load issue, that's what I was going to try next. The weather is starting to change and I'm not sure how much longer I'll have the car on the road. The first salt truck I see and It's going into storage. I'll have to try and figure out a way to test the re-circ valves as I don't have anyway to put the engine under load without driving it. The problem doesn't happen with stock values and they are brand new turbosmart valves so I didn't consider they might be the issue.

  9. #89
    Tuner in Training
    Join Date
    Feb 2019
    Posts
    26
    It seems to me like something I'm doing in the SD table using this method is causing the throttle to not close under low load.

  10. #90
    Senior Tuner
    Join Date
    Jan 2013
    Location
    Hawaii
    Posts
    2,101
    Injector data maybe? Are you increasing the calculated MAF enough to not be on the low slope in this no throttle condition? Take your breakpoint and convert it to a MAF value, make sure you stay below that with the SD modifications. If needed modify the low slope, lower, to scale the load to a lower reasonable value.

    Make sure you are not above your DFCO disable load also. It really should go into DFCO when you let off the pedal and not have that "cruise control" effect.

  11. #91
    murfie, my stock file was altered by a local tuner that USED to do my tuning. was tuned with sct but after returning car to stock, my supposed stock tune showed low slope of 65 lbs which is not obviously stock injector values. when i enter my data in the inverse sheet, it pushes my new tq values on the top cell to 1475. should i go back to a stock table for my car and use those values and change them all to match in my tq tables ? that seems high but my car has always been boosted.

  12. #92
    Senior Tuner
    Join Date
    Jan 2013
    Location
    Hawaii
    Posts
    2,101
    65 lb/hr low slope is 47# injector data... Stock for 07-12 GT500.

    Yeah you would need to change the axis values on both to what you have.
    Remember when you multiply the axis values, thats not a new value to be used, just a reference point.

    1.6 load times 738 is 1180. That doesnt mean your axis value now needs to be 1180. If you have 2.0 load lined up with 738, you could easily see a value like 1475.

    It just means for the inverse model, if load is 1.55, torque will be 761. If load is 1.65, torque will be 715. Kinda the opposite of what you would think it would do in reality. The model it's self should be fairly flat through the RPM ranges, and this is why I say getting your axis values correct and covering your application is the more important part of this. You don't want 1.0 load meaning 800 torque and you don't want 200 torque meaning 2.2 load.

    You need to realize in the model, the axis values don't need to reflect straight across to the other table. using the 07-12 GT500 as an example, .10 load is an axis value, but in the OP tables that lines up with torque values between 36-42, which those values would be in between 18 and 74 axis cells of the inverse table. Some axis values line up but are offset by a cell between table axis. Either renormalize the model, so the axis values reflect straight across, or choose to multiply by a cell value like the 1000RPM value instead of the other tables axis. Keeping the change across the RPM range fairly flat, the higher the load goes up the more it can change.
    Pay close attention to the difference from one cell to the next when trying to come up with a trend line to extend these axis values higher.

    Its very easy to just extend these models, just using a trend in excel. You can do RPM, load and TQ.
    Example of a GT500 table extended to 4.0 load(50+PSI of boost), 1400lb-ft of torque and 9000RPM.
    Axis values are based on .1 load =55.31716lb-ft of torque in a linear relationship

    Inverse torque model.PNG

    Example of MP14 from a 15-17 mustang GT extended to 3.5 load, 1250lb-ft, and 9250RPM
    Axis values are based on .1 load = 36.87811 lb-ft of torque in a linear relationship. You see it in the whipple tune as well. The inverse calculator just converts the torque values to load directly. You can then fine tune from that or choose not to, it will be with in a percent in most of the model, at most 3-4% in spots.
    15-17GT Inverse torque model.PNG

    inverse calculator.PNG

    Any model you look at will fall into this line and be based around it, all that matters is what the starting point is, and how high does it need to go. Double the load, double the torque direct relationship. Start at 1000NM or 738lbft of torque and just keep cutting it in half or any fraction, that will fall on this line.

    torque values.PNG

    Its how torque is converted to load, and load is converted to torque.
    Just like how MAF is converted to load based on RPM, and vise versa.
    All very simple relationships, used by ford and given by them through their R&D/ Patents. Not false assumptions.

    Most people try to fine tune this model, when thats completely unnecessary, as this is much more abstract than people think it is. It just needs to cover your application, and not be the limiting factor, then the important things can be fined tuned in the actual idle,fuel,air, and spark. Torque/ load limits you don't have access too would be the only other reason this might become difficult as it turns into a scaling thing. 99% of peoples cars are not at those limits or even near them. Some people probably have their calibrations really far off making them think they are, but they are not. Hopefully reading this, makes scaling it seem easy. Warning to auto trans guys, this will require scaling all of your pressures, as they come from this model. Not directly, but indirectly.
    Last edited by murfie; 11-01-2019 at 06:31 AM.

  13. #93
    thanks for putting the why to the how. as for the auto trans, I converted to a 6r80 with the standalone quick6 controller. my ecm thinks my car is still manual

  14. #94
    Senior Tuner
    Join Date
    Jan 2013
    Location
    Hawaii
    Posts
    2,101
    The torque inverse calculator just needs two points to calculate the entire relationship. (MAX/MIN)

    Two points.PNG

    Two points.PNG

    Two points.PNG
    Last edited by murfie; 11-02-2019 at 03:41 PM.

  15. #95
    Tuner in Training
    Join Date
    Feb 2019
    Posts
    26
    So I've been working on eliminating ipc error for a couple months. The only thing that seems to work is actually lowering the values on the torque tables and calculating the inverse. I can not understand how/why this works. The cookbook says to try and get these values as accurate as possible using a dyno to measure real torque. Now I know I'm making considerably more torque then stock, but I've had to lower the values drastically to eliminate the ipc errors, as much as 20-30% in some cases. Doing this seems to have absolutely no adverse effect on the actual torque the engine produces. What gives? I'm happy I'm finally getting my throttle to stay open, but I want to know WHY!

  16. #96
    Senior Tuner veeefour's Avatar
    Join Date
    Oct 2016
    Location
    Poland
    Posts
    1,711
    Torque model follows your torque input and checks for plausibility - your torque input comes from several different places Driver Demand being the most important.

    Basically speaking: if you command some torque with your pedal it first looks in the driver demand and applies the current value - it tries to open your throttle to meet the demand.
    Now torque is Load based and Load is coming from your Speed Density model and if there's too much Load it will think it's making more torque but it does not. It means you are making
    too much load and your SD needs some attention.

    There are 3 ways to fix your torque problems:
    1. Adjust your DD
    2. Adjust your SD model
    3. Adjust your Torque Model.

    Hint: I barely ever touch the Torque Model after extrapolating it for more load. The general trend is to adjust your torque model but I tend to disagree.
    Last edited by veeefour; 12-03-2019 at 11:50 PM.

  17. #97
    Senior Tuner CCS86's Avatar
    Join Date
    Nov 2017
    Location
    Austin
    Posts
    1,089
    Weird, I agree with veeefour. You just want relatively flat TQ/INV tables, scaled out for your load range.

    I'll add that the ETC model can cause IPC error too.

  18. #98
    Tuner in Training
    Join Date
    Feb 2019
    Posts
    26
    Quote Originally Posted by CCS86 View Post
    Weird, I agree with veeefour. You just want relatively flat TQ/INV tables, scaled out for your load range.

    I'll add that the ETC model can cause IPC error too.
    I agree completely that it makes no sense which is why it's so frustrating. It's just the only thing I've tried that actually improved the IPC error. I've tried about a half a million other things and they either did nothing or made things worse.

  19. #99
    as i am re-reading my way thru this trying to understand it and figure out the 1st part of figuring out the load values to extend, why wouldnt you just use the math to get the exact load values for your application ? I mean, we use maf, and load values are a calculation of g/s converted to g/m then divided by rpm. what am I missing ?

  20. #100
    Senior Tuner
    Join Date
    Jan 2013
    Location
    Hawaii
    Posts
    2,101
    the airload PID is reporting the end of that calculation from the MAF value. The MAF PID is limited to reporting 86.68 lb/min. Thats 655.36 g/s. A 16-bit integer can store 2^16 (or 65,536) distinct values. Its also sometimes polling slowly. Airload does not have that limit, and is usually polling much faster. Anything over ~12-14psi on a coyote and you will be above that 86.68lb/min limit and the PID becomes useless. Coincidentally, this is also where most will say switch to E85 and don't use pump gas above it. My guess is they don't trust their transfer, and hence the load value, to put the ECU at the right spot in the calibration. Just log and use airload for this. If you have good injector data, and a good MAF transfer it will be accurate.

    I'm was hoping to see the commonly recommended boost level for pump gas go up with the use of MAP sensors and the SD model, but I fear it may go down as people are having a hard time comprehending how to dial the SD model in as well as a MAF transfer.