Virtual Torque Table Adjustment

[CaudleDynamicsLLC]

This is now relevant to something I keep seeing, but am questioning...
Isn't the "Delivered Engine Torque" PID a calculation being done based off of the virtual torque table (or at least the coefficients in the background that make up the VTT)? If so, then I believe that modifying the VTT to match engine tq PID would be a circular reference? And "delivered engine torque" PID would HAVE to match the VTT since that is what it is calculated from? If there is a difference between a VTT value and the engine tq PID then there must be some underlying discrepancy in the calculation - I have seen mention of injector data, scaled tunes, etc causing skewing and there is an EQ ratio coefficient modifier under the torque model tab that maybe confirms ECM not knowing actual fueling could cause a difference between VTT and engine tq PID.
I thought I saw big differences at first then I noticed the VTT tables are in nm and the datalog I was looking at was in ftlb

One additional thing I haven't been able to confirm yet, when the value in the VTT under MAP conditions is different than the value in the VTT under the Airmass conditions, does the ECM use the higher of the two, or lesser, or average to report as "delivered engine torque"? In my limited checking so far, it seems to use the lower of the two TQ values (?)

[CaudleDynamicsLLC]

...The stock Virtual Torque Table values are 25% higher than the torque values obtained from the Dyno...

know that was a while ago, but if the VTT table is in Nm it would read about 25-35% higher than if it were in ft-lb, im hoping the dyno units were same as VTT units?

[hjtrbo]

There is a pumping loss table that we haven't got mapped to us and perhaps another table to deal with the supercharger on LSA / LS9 engines. Have run some data in matlab using the VTT co-efficients against dyno numbers and have confirmed there is parasitic loses that we do not see in the tune. Long story short, the VTT tables will always read slightly higher than actual.

[CaudleDynamicsLLC]

There is a pumping loss table that we haven't got mapped to us and perhaps another table to deal with the supercharger on LSA / LS9 engines. Have run some data in matlab using the VTT co-efficients against dyno numbers and have confirmed there is parasitic loses that we do not see in the tune. Long story short, the VTT tables will always read slightly higher than actual.

Cool, good to know!

[GHuggins]

To add some more to this. The map side of the torque model always seems to control shifting - how quick it downshifts or up shifts and if you think about it - the VE table is used for transient fueling (quick throttle changes), so makes sense they would use the map side for quick throttle movement shifting whereas the airmass side of the torque model seems to be used primarily for hold pressures or shift pressures in general. This is at least what I've noticed. Had one slipping slightly in reverse of all gears due to not having the idle to 1400 rpm range high enough on the airmass tables to adequately hold the clutch pack and it only took about 30lbft increase to fix it. So doesn't take much to make a big difference.

[mtvainiot]

To add some more to this. The map side of the torque model always seems to control shifting - how quick it downshifts or up shifts and if you think about it - the VE table is used for transient fueling (quick throttle changes), so makes sense they would use the map side for quick throttle movement shifting whereas the airmass side of the torque model seems to be used primarily for hold pressures or shift pressures in general. This is at least what I've noticed. Had one slipping slightly in reverse of all gears due to not having the idle to 1400 rpm range high enough on the airmass tables to adequately hold the clutch pack and it only took about 30lbft increase to fix it. So doesn't take much to make a big difference.

How do I know which side of the torque model I should modify? Or should I usually make same adjustments to map and airmass side? I have seen someone modify the EQ ratio coefficient too. Is it necessary?

[smokeshow]

To add some more to this. The map side of the torque model always seems to control shifting - how quick it downshifts or up shifts and if you think about it - the VE table is used for transient fueling (quick throttle changes), so makes sense they would use the map side for quick throttle movement shifting whereas the airmass side of the torque model seems to be used primarily for hold pressures or shift pressures in general. This is at least what I've noticed. Had one slipping slightly in reverse of all gears due to not having the idle to 1400 rpm range high enough on the airmass tables to adequately hold the clutch pack and it only took about 30lbft increase to fix it. So doesn't take much to make a big difference.

That's more or less accurate. The throttle affects a change in torque by adjusting the pressure drop across it, as it cannot directly influence airflow. So the MAP torque model is the feedforward for throttle position via back calculating through the compressible flow equation.

I've been driving around on a brute-forced attempt at a MAP torque model correction with a Monte Carlo parameter variation (realistically just me being lazy and wanting to throw stones to see which ripples can help can me lol). I haven 't really been able to find any downsides to it yet...

[GHuggins]

If you want it to be super sensitive to pedal movements - downshift quickly for example - then lower it (just keep airmass up so it doesn't lose pressure). On the reverse if you want it to hold gear longer and lug more - increase it. What I've also found as far as the torque model goes and even though it shouldn't be - it still uses airmass with MAF failed or what ever... I've even had some gen 5's have idle instability issues due to the b,c and d maps being different - still can't make heads or tails of that, but what I found on a few.

[verlon]

There is a pumping loss table that we haven't got mapped to us and perhaps another table to deal with the supercharger on LSA / LS9 engines. Have run some data in matlab using the VTT co-efficients against dyno numbers and have confirmed there is parasitic loses that we do not see in the tune. Long story short, the VTT tables will always read slightly higher than actual.

Hansel and Gretel, bread crumbs
Look at the GMW8762
EngTq.png

I've discovered several Torque Loss tables in E38:
* Accessory Tq (vs RPM) - Coolant Pump, Power Steering etc
* Viscous Friction (vs RPM vs OilTemp)
* Pumping Losses (2 maps - Normal and DoD) (vs RPM vs Vacuum)
* Supercharger Tq (vs RPM vs Boost)
* Generator Tq (vs RPM vs Gen-F-Terminal %DC)

E39 and E92 have similar set of tables.

[hjtrbo]

Ooh, very nice. Would be good if there were some E38 A2Ls out in the wild for public consumption.