Lets Talk Torque Modelling / Virtual Torque

[Allen Vos]

Alright guys, I am trying to wrap my head around the torque modelling/ virtual torque on a Gen 4 and want to make sure I understand it as best I can so I am not jumping blindly into it.

I understand that the values in the virtual torque table respond directly to the set amount of torque my engine will produce at whatever the given spark value is. Torque basically translates to a set amount of air going into the engine. That's the simple part.

I have read all kinds of mixed opinions on this and have even come across guys that say to not even bother with it.

From most stuff I have read if you do not change the values to match your torque increase with a cam or whatever you have changed the engine wont produce its full torque capacity.

I know putting the car on the dyno would be the best option however that is not an option for me, so I am wondering what the best way to go about this is?



Here are my thoughts.

My factory LS3 makes 426 foot pounds of torque at approx. 4600RPM

The spark 30 table on my MAP virtual torque table shows 433 ft/lbs @ 4800 rpm which is right in line with what the car is making.

The camshaft I am using along with the long tube headers, X-pipe and CAI made 530 WHP and 500 Ft/lbs of torque @ 4600 RPM with the identical setups on other cars

Corrected to FWHP would be roughly 580 HP and 553 ft/lbs.

So knowing these values I would have to multiply my MAP/Spark tables by 1.3 to come up with 563 ft/lbs at 4800 RPM which falls in line with my particular setup.

My after market cam seems to make peak torque @ 4600 just as the stock camshaft does so that should make things a bit easier to figure out.


Attachment 142651


As far as the idle goes, the stock engine ran around 30 KPA at idle(650 RPM) with 13 degrees of base ignition timing

My current cam runs 57 KPA at idle (800 RPM) with 18 degrees of base ignition timing

I am noticing small amounts of negative torque values in the idle areas for the stock camshaft.

I know that you want negative torque at idle or very close to the actual idle torque to reduce the integral air table from adding unwanted air causing stupid stuff for example my little throttle idle blip I have been cursed with lol

So is the proper way to just transfer those negative torque values from 30 KPA on the table to the 57/60 KPA area on the table?

Attachment 142652


Engine torque can be logged with VCM scanner. Are you able to log torque and plug those values back into the appropriate spots on the torque tables?

I might be overthinking this or be right out to lunch but how do all you guys do this?



As far as the Airmass tables in Virtual Torque goes, how does one figure out the correct torque values for these?

The values are a little bit different than the Map Based torque tables.

On the spark 30 table it shows peak engine torque values between 1200-2000 RPM.

Attachment 142651

That does not seem right to me.

[hjtrbo]

From most stuff I have read if you do not change the values to match your torque increase with a cam or whatever you have changed the engine wont produce its full torque capacity.

That is not true for Gen 4.

Don't worry about erroneous values, its a by product of the coefficients that are used to model the torque output in the engines normal operating range. The engine will never operate in that region you mentioned.

You just need to add a little bit in the airmass and map table to bring your idle torque up.
Select the 3 cells shown, make sure 'link selections' is checked. Multiply by 0.8. Highlight all 3 of those columns and vertical interpolate. Click extrapolate, calculate.
0.8 is a pretty big change, see how it goes. You might need to return it and make a smaller adjustment.

[GHuggins]

Torque model editing is actually extremely simple. Not so simple if you don't have a tool to help, but nonetheless, I fought it for a long time. I will say if the models aren't "balanced" to one another you will have all sorts of erroneous throttle issues. Throttle controls are primarily on the map side. No, dialing in the air models doesn't magically make them correct as some would have you believe. If you're experiencing hesitation issues, dead spots or whatever such as to the point of having bad throttle dips then the solution is most likely there combined with a couple of other tables. Fuel models have to be completely dialed in. Not just the idle area, so don't worry about it right now.

Nathan, tools working great. Just need a way to edit the airmass if you could get the plans to work there I know. I know, it's on the list. I'll wait. Right now I'm doing it by hand making the airmass match the map, but having the map close is worth it's weight....

I would also like to say that if either Smokeshow or DStek come in here and make any comments at all you better sit back and read everything several times in detail. Then read between the lines, above the lines, below the lines, then everywhere else while thinking long and hard about torque and ask questions carefully

[07GTS]

i just try to get the new engine torque (ie camshaft change) torque to read about the same as the stock torque was so then the rest should line up better, airmass usually stays pretty close but the map needs some tweaking, but im manual so dont have to worry about the auto side

[Allen Vos]

That is not true for Gen 4.

Don't worry about erroneous values, its a by product of the coefficients that are used to model the torque output in the engines normal operating range. The engine will never operate in that region you mentioned.

So if I am reading this correctly, I should just leave the MAP based torque tables alone then?

If that's the case what about the idle on the spark-10 map? The idle went to 57 KPA from 30 KPA.

Is that not going to add unwanted air to the throttle body coming back down to idle if I don't change things?

Or is this something I need to address in the Torque follower table under Idle air flow?




Ill definitely try adding a bit in the airmass tables like you suggested, but what exactly am I looking for with regards to how the engine behaves, to determine how close I am with the numbers?

I'm trying to understand how the air mass tables work as the numbers don't make a lot of sense to me.

RPM across the top of the table is self explanatory but the air flow is measured in mg.

I know 1000 mg = 1 Gram

So 100mg as indicated in the air mass chart is only .1 of a gram
and 800 mg is .8 of a gram.

So the numbers in the air mass tables seem way to low or small to have any effect on the engine whatso ever?

I'm not trying to change things unnecessarily but would like to have an understanding how this dang virtual torque works.

Inside and out preferably lol

[Allen Vos]

i just try to get the new engine torque (ie camshaft change) torque to read about the same as the stock torque was so then the rest should line up better, airmass usually stays pretty close but the map needs some tweaking, but im manual so dont have to worry about the auto side

Is the virtual torque not already reading the same as stock torque? Sorry I am not following you on this one

I might be completely missing what you are trying to say but is there any options in virtual torque for automatic or manual transmission?

I always thought those values for auto/standard were defined in tables like integral/proportional idle and spark coast tables?

[Allen Vos]

Torque model editing is actually extremely simple. Not so simple if you don't have a tool to help, but nonetheless, I fought it for a long time. I will say if the models aren't "balanced" to one another you will have all sorts of erroneous throttle issues. Throttle controls are primarily on the map side. No, dialing in the air models doesn't magically make them correct as some would have you believe. If you're experiencing hesitation issues, dead spots or whatever such as to the point of having bad throttle dips then the solution is most likely there combined with a couple of other tables. Fuel models have to be completely dialed in. Not just the idle area, so don't worry about it right now.

Nathan, tools working great. Just need a way to edit the airmass if you could get the plans to work there I know. I know, it's on the list. I'll wait. Right now I'm doing it by hand making the airmass match the map, but having the map close is worth it's weight....

I would also like to say that if either Smokeshow or DStek come in here and make any comments at all you better sit back and read everything several times in detail. Then read between the lines, above the lines, below the lines, then everywhere else while thinking long and hard about torque and ask questions carefully

I think right now before I make any adjustments, I want to try and understand why we make the changes we do on the Virtual torque and how it all works.

That way I can make a good educated decision on what to adjust based on knowledge how this thing works.

[TheMechanic]

It was my impression that torque as it applies to Gen 4 is a calculation instead of a measurement like in reality. There is nothing that can measure torque. No strain gauges. So it uses the fuel volume to determine torque. Top torque measured is equal to top torque calculated by the tables. Correct me if I'm wrong.

[hjtrbo]

I'm getting lost, you're doing that 'post to many times' thing again

The row axis on the airmass tables are in mg of 'cylinder airmass'. I checked your log, you're logging it.

Is the virtual torque not already reading the same as stock torque? Sorry I am not following you on this one

It was, until you added that cam shaft. That has shifted your idle vacuum which means the torque model is no longer valid in that region. In your case we need to get the oem idle torque numbers shifted into your new idle region.

I might be completely missing what you are trying to say but is there any options in virtual torque for automatic or manual transmission?

I always thought those values for auto/standard were defined in tables like integral/proportional idle and spark coast tables?

Auto & manual use the same torque tables.
For auto tunes we can increase the torque values as needed which causes the transmission to apply more pressure. There is no science behind the changes, just a blatant number fudge although we want the numbers to be realistic. On mild to big builds or when adding forced induction it's very important to check the reported torque and adjust the torque tables as required.

All of us have different ways to correct the numbers, but we are all aiming to achieve the same outcome.

[hjtrbo]

It was my impression that torque as it applies to Gen 4 is a calculation instead of a measurement like in reality. There is nothing that can measure torque. No strain gauges. So it uses the fuel volume to determine torque. Top torque measured is equal to top torque calculated by the tables. Correct me if I'm wrong.

Not fuel. It uses airmass, map, rpm, spark and cam angle.

Top torque reported should be more than measured. The reported torque includes parasitic loss. For up to a 1000hp on a WOT pull, dyno wheel torque x 1.3 gets us somewhere in the ball park.

[hjtrbo]

So if I am reading this correctly, I should just leave the MAP based torque tables alone then?

You can leave the map alone for now to keep it simple.

Ill definitely try adding a bit in the airmass tables like you suggested, but what exactly am I looking for with regards to how the engine behaves, to determine how close I am with the numbers?

It needs to look just like smokeshows screen shot in his bed time story post. Add a new chart and set it up just like he did. That screenshot he gave us is worth a lot of money, we are lucky to have it. For those of us with more experience it tells us a lot.

[GHuggins]

Dave and Jake actually got into what the torque model is referencing in the gen V section. The APC or air per cylinder is what's changed and is why it's important to dial it in.

You can't just use a dyno. You don't know the losses to back calculate it. Trust me. I've studied the patents and talked to engineers about it. Heck, Jake is one of the one's that brought it to the GM platform GM is including things such as oil pump drag via it's changing pressures in it's own internal equations for delivered torque. Without knowing the "secret sauce" as Dave puts it, which he does, there's no way to back calculate from a chassi dyno.

Using a virtual dyno's algo, you can back calculate it's peaks via the vehicles speed climb rate, but even then you need all of the slips and losses in the drive line. Plus it's 100% worthless for where you really need to know the information, which is in the idle and drivability areas.

Other things to know or understand

Airmass - if you have an auto - this primarily controls your line pressure
Map - if you have an auto - this controls both rev matching by being more balanced to the airmass tables and upshifting/downshifting or in other words what percentage or rate of throttle application to force the downshift or upshift
Map - is what's mostly changed with a cam install - don't worry about the airmass side too much against what's been stated. Either way THEY BOTH need to be better adjusted.
Map model is also your throttle controller. Peaks will make it open quicker.
Airmass also controls throttle, but not to the extent that the map side does. You want to close the throttle slower, then you usually lower the airmass lower quadrant returning to idle. However on a 4th gen this is primarily controlled by throttle follower, adaptives and min air, but is why dialing in the VE greatly helps with this cause it helps to correct the map side of the model. Not to say the airmass side isn't still doing something...

While tuning in the air models does change delivered torque and it does "correct" the models a little via the APC factor, it does not correct the models on it's own. It does bring them closer to reality!!!

Next little tib bit - hpt has it defined wrong per Dave and after coming to realization with that, it makes much more sense All of those spark tables they put into it really shouldn't be there. This means "certain" parts of the tables are nothing more than the math equations being stretched out. AND like many things GM does, parts of the tables were never dialed in, in the first place, so again, the math interpolation is what's filling in these gaps.

Delivered torque, even though it's not right, works decently on 4th gens. Otherwise study the VE table and make adjustments accordingly to that. You need to dial in the map and airmass tables both. They must be balanced to one another. Otherwise those odd throttle issues will never be straightened out. Use delivered to then keep things in check and shifted correctly.

Other things will affect the torque models or have you believing the torque model is to blame such as the throttle model. The throttle model needs to be right for the throttle body. It uses things such as etc scaler and air pressure in front of and in behind the throttle plate to define its opening. This is what the baro tables in the gen 4's along with manifold volume and several others are defining. I may be mixing 4th and 5th gens at this point some too, but the baro tables for different tb's really seems to affect things.

Another thing to note and is probably why some say no need to mess with the map side as I understand it, both the throttle model and the map models have the ability to "learn" a small amount and "shift". I was once told something around 60Nm's allowed shift, so the closer the better to allow for learning either way....

[hjtrbo]

Good morning Greg,

Dave and Jake actually got into what the torque model is referencing in the gen V section. The APC or air per cylinder is what's changed and is why it's important to dial it in.

I've only been into the gen 5 section once or twice. Time to dive in.

You can't just use a dyno. You don't know the losses to back calculate it. Trust me. I've studied the patents and talked to engineers about it. Heck, Jake is one of the one's that brought it to the GM platform GM is including things such as oil pump drag via it's changing pressures in it's own internal equations for delivered torque. Without knowing the "secret sauce" as Dave puts it, which he does, there's no way to back calculate from a chassi dyno.

These are all known if you can use Ghidra or IDA. One day I'll post them.

Next little tib bit - hpt has it defined wrong per Dave and after coming to realization with that, it makes much more sense All of those spark tables they put into it really shouldn't be there. This means "certain" parts of the tables are nothing more than the math equations being stretched out. AND like many things GM does, parts of the tables were never dialed in, in the first place, so again, the math interpolation is what's filling in these gaps.

Why does Dave think they're defined wrong? They are merely presenting the data in a visual way. Granted there may be a better way to visualise it? I've had a delve in matlab with the coefficients and logged data to generate a nice pretty 3D map, but that was just to get the idea of how it all worked, the average Joe ain't gonna do that. The tables are a good idea if you know how to read them. So many people don't and it causes a lot of confusion and mis information. I'd rather edit a region of a table than fuck around with the coefficients. As an example you can set spark from 15 to 25 degrees in 1 deg increments with a granualar airmass set, do a ramp run and easily use the tables to see what's out. The art is making the adjustment so it doesn't fuck up the slope which would screw with your overrun reported torque.
Heck, i'm 99% confident my earliest posts on this subject are completely wrong. But 'ol mates chatGPT model scooped them up and probably gives it as an answer .

[GHuggins]

Good morning Greg,


I've only been into the gen 5 section once or twice. Time to dive in.


These are all known if you can use Ghidra or IDA. One day I'll post them.


Why does Dave think they're defined wrong? They are merely presenting the data in a visual way. Granted there may be a better way to visualise it? I've had a delve in matlab with the coefficients and logged data to generate a nice pretty 3D map, but that was just to get the idea of how it all worked, the average Joe ain't gonna do that. The tables are a good idea if you know how to read them. So many people don't and it causes a lot of confusion and mis information. I'd rather edit a region of a table than fuck around with the coefficients. As an example you can set spark from 15 to 25 degrees in 1 deg increments with a granualar airmass set, do a ramp run and easily use the tables to see what's out. The art is making the adjustment so it doesn't fuck up the slope which would screw with your overrun reported torque.
Heck, i'm 99% confident my earliest posts on this subject are completely wrong. But 'ol mates chatGPT model scooped them up and probably gives it as an answer .

LOL... Yeah I saw that thread. Just waiting for an AI system to come out to be able to scour the whole web and correct things for me I've been wrong a lot on it too. Just recently really started understanding and coming to terms with it. Always kind of known what they do or how they correlate, just not to the current extent.

Didn't know that's where Dave got them Guess every vehicle platform would have them encoded to correctly calculate their own. Makes more sense now.

Dave said it had to do with the spark tables and GM didn't necessarily dial them in like that. The better way to visualize or plot it and just the way the math is interpolated and sloped as you put it. GM to me appears to have gotten lazy with the newer ones depending on the crew working on them. To me the gen 4's are more finely dialed in (when it was implemented, so they covered their bases better before realizing what they could get away with). Also full throttle and torque control is being utilized in the genV's or newer, so seems to me like they shaped some of the tables to do just that. I can completely change how the throttle reacts for dips as to whether or not it opens or closes it via the torque model alone in one of them.

[GHuggins]

Now, what would be really nice is if there was a way to populate the positive timing tables and then hpt make an all together new 0 and - timing tables without having those interfere with the positive timing tables like they do now.

[07GTS]

is it safe to say gen4 if u want a smoother idle or throttle correction to make the torque value in VTT between the spark values -10, 0, 10, 20, 30, 40 greater ? so if the spark 10 and 20 deg tables have say 15nm torque between them for the idle cell area it will only add within the 10 deg spark range ? but if it wanted the same 15nm correction and that was in the 30 deg table (10 20 30 deg tables span 15nm between them for idle) it would add too much and overshoot ? hope that makes sense or im just thinking about it all wrong now

[hjtrbo]

Dave said it had to do with the spark tables and GM didn't necessarily dial them in like that

Agree with that. I think it's highly likely the coefficients would have come from a best fit algo after they characterised the engine. Jake would know.

Didn't know that's where Dave got them Guess every vehicle platform would have them encoded to correctly calculate their own. Makes more sense now.

Maybe, or he pays for it? Fark knows. Getting the a2l is the ultimate but those are not public and extremely hard to get.

Here is what I could find easily for a L76, would need to dive deeper to see what else there is:

Driveline efficiency


Pumping loss


Oil viscosity friction


Supercharger parasitic drag loss (not fitted)


Air con torque load


Air con inertia torque adder


Accessory drive loss

[GHuggins]

See, that's the problem. Delivered torque is primarily coming from the airmass side or the "steady state" side, however the map side plays into more than anyone realizes. The map side usually requires lowering whereas the airmass side may just require tweaking. I've been using a tool that I wanted built based off of what I was seeing with the VE to delivered torque correlation to correct the map side as good as possible and then using what it's showing me to correct the airmass side in the same general manner and then I raise the airmass side back up to get the torque I'm looking for. At the moment this seems to be working. What I was doing before this was just plugging the logged torque back into the tables. This was smoothing up idle, fixing blips, fixing off idle hesitations and so on. I still primarily use that method on gen4's then add a small percentage back to it.

[GHuggins]

Agree with that. I think it's highly likely the coefficients would have come from a best fit algo after they characterised the engine. Jake would know.


Maybe, or he pays for it? Fark knows. Getting the a2l is the ultimate but those are not public and extremely hard to get.

Here is what I could find easily for a L76, would need to dive deeper to see what else there is:

Driveline efficiency


Pumping loss


Oil viscosity friction


Supercharger parasitic drag loss (not fitted)


Air con torque load


Air con inertia torque adder


Accessory drive loss

So if I'm understanding this correctly in this one specifically, there's a 17% loss just in the driveline? I assume this isn't including tires to rollers, so you would have to use a hub style to be accurate....

Don't think you would need to worry about ac stuff. That's defined on it's own. Would also think accessory would have alternator load somewhere unless that's no load free spin?

[hjtrbo]

is it safe to say gen4 if u want a smoother idle or throttle correction to make the torque value in VTT between the spark values -10, 0, 10, 20, 30, 40 greater ? so if the spark 10 and 20 deg tables have say 15nm torque between them for the idle cell area it will only add within the 10 deg spark range ? but if it wanted the same 15nm correction and that was in the 30 deg table (10 20 30 deg tables span 15nm between them for idle) it would add too much and overshoot ? hope that makes sense or im just thinking about it all wrong now

If you pick a cell and plot it for each spark point you will see it has a curve to it. My thoughts whether right or wrong are keep the curve a similar shape whilst tweaking it.


Here are 2 curves, the blue 1 is OEM and the orange one is the lowest airmass cell x 0.8 then vertical interpolate with link selections turned on. You can see the curve stays similar but about 10Nm has been added. Added benefit is when clicking the calculate button the changes stick, it plays nice with the coefficient generator. Ignore the high values, I just picked a random cell for this example. This so far has been working well for idle. I have a different method for WOT.