Silverado VT issues

[performancemastersllc]

So I have searched and searched for a proper way to dial in virtual torque only to find bits and pieces that only sort of make sense. I am having multiple issues with this cammed 2014 silverado 5.3 that i cannot figure out:

1. Idle with ac on, it will sometimes idle smooth and out of nowhere will jump up a few hundred rpms, back down about 100 below target and settle on target for a few seconds and repeat. most of the time it idles perfect after startup and starts to screw up after its been driven. fine in gear, only does this in park/neutral.

2. foot off the brake in gear(no throttle) tends to slowly climb rpms and basically drives itself until about 2k.

3. will do this at different speeds and different gears: steady cruising at around 10% pedal, tip in to about 20-30% and the throttle opens, timing retards, and map shoots to about 98kpa.

4. when cruising, taking foot off the pedal and letting it decel, it will jump between open and closed loop and hold around 1500rpms(like cruise control is on) and will slowly move up and down about 100rpms each way.

I would think this would all have to do with virtual torque and idle torque, maf and vve are pretty well calibrated, but i have taken torque out at the lower rpms(due to the cam) where all these problems occur and it will get a little better, take more out and it gets worse. I have yet to see any of the torque source pids say anything but driver, axle, or idle.

Main thing im looking for is a clear explanation of what torque scanner parameters match which tables and how everyone is getting these right on gen 5s. Also how to decipher whether to adjust airflow VT or map VT.

Posting my current tune and a log.
Thanks

[performancemastersllc]

in case anyone has similar issues, i have figured out problem number 3. Disabled torque converter lockup in all gears and that issue goes away completely, drives like a completely different truck. I guess because of the cam, it cant handle any acceleration whatsoever in lockup at lower rpms. Still trying to work in some lockup for steady cruising and get the release speeds right.

[smokeshow]

So I have searched and searched for a proper way to dial in virtual torque only to find bits and pieces that only sort of make sense. I am having multiple issues with this cammed 2014 silverado 5.3 that i cannot figure out:

1. Idle with ac on, it will sometimes idle smooth and out of nowhere will jump up a few hundred rpms, back down about 100 below target and settle on target for a few seconds and repeat. most of the time it idles perfect after startup and starts to screw up after its been driven. fine in gear, only does this in park/neutral.

2. foot off the brake in gear(no throttle) tends to slowly climb rpms and basically drives itself until about 2k.

3. will do this at different speeds and different gears: steady cruising at around 10% pedal, tip in to about 20-30% and the throttle opens, timing retards, and map shoots to about 98kpa.

4. when cruising, taking foot off the pedal and letting it decel, it will jump between open and closed loop and hold around 1500rpms(like cruise control is on) and will slowly move up and down about 100rpms each way.

I would think this would all have to do with virtual torque and idle torque, maf and vve are pretty well calibrated, but i have taken torque out at the lower rpms(due to the cam) where all these problems occur and it will get a little better, take more out and it gets worse. I have yet to see any of the torque source pids say anything but driver, axle, or idle.

Main thing im looking for is a clear explanation of what torque scanner parameters match which tables and how everyone is getting these right on gen 5s. Also how to decipher whether to adjust airflow VT or map VT.

Posting my current tune and a log.
Thanks

What you've asked for is essentially a mechanical engineer's dissertation...it's pretty gnarly stuff for the layman to tackle. I'll give you an overview though, or at least some info based on what the issues appear to be in the data.

For the first topic, you might read this for some background: https://forum.hptuners.com/showthrea...l=1#post616505. I posted that a couple weeks ago to give a basic who's who of idle calibration. It's based on older engines/technology but the physics remains the same. That said, based on what I see in the data I think it's good to remember that your vehicle has a variable displacement AC compressor which means different torque loading based on how hard it is being driven. It also multiplies the number of torque model operating points that have to be spot-on to avoid those RPM flares. Looking at the beginning of the first RPM flare at 14:13.200, you can see the controls preparing to do exactly what I described in the previous post. If you look at the predicted torque, you can see it jump up 12ft-lbs really quickly. The throttle and MAP are following this request. So, predicted torque is essentially a feedforward estimate at the present operating conditions, or an 'unmanaged' torque potential based on the output of the calibrations you've entered into your torque model. Now the predicted torque is only an expectation from the controller, basically a target to be able to reach. Then the immediate torque comes in and provides final arbitration over actual desired torque delivery based on any other requestors that may want less torque. It may help to think of predicted torque as a slow path, ie airflow and the immediate torque as a fast path control like spark or fuel. You can make torque much faster with a change in spark than you can with a change in airflow because air moves slowly, especially at idle. Now the goal here is to have a nice, smooth AC clutch engagement. Customers don't want to feel it and the torque reserve makes this possible. So you bring in more airflow, pull out some spark via the immediate 'managed' torque request and you have a reserve ready to ramp back in as the AC clutch comes on. Typically this achieves a negligible impact to idle RPM but in your case, as you mentioned...the problem is the torque model. A snapshot from your file...
torque model inflection.PNG
The point the cursor is hovering over shows the problem - an inflection in the spark response. It shouldn't spike up, especially not when your idle speed is too high. And it isn't a transition to main spark because the state variables you've got in your log don't change to 'torque'. Since you've likely noticed there is no idle spark table, you know that idle spark value is coming from the torque model. There are quite a few nuances to get the idle spark value from the torque model as it is technically the inverse of that model, which is technically a blended array of models where the APC or airmass model is more or less primary. So the immediate torque request is targeting -26ft-lbs at 732 RPM and 0.186 grams in the cylinder. Now attaching your airmass torque model vs a stock 2014 Silverado airmass torque model:
VT_mod.PNGVT_stock.PNG
I can see how it may make sense to decrease the modeled torque output at lower RPM. Your cam moved the torque curve up, so it'll be a little soggier on the bottom end. Have to keep in mind though, this isn't a VE table...you've got mass on the Y axis, not pressure. While torque will look a lot more like VE with MAP on the Y, it's not the same for air mass. Those tables depict the torque output with the assertion that all of that air mass is actually getting smashed into the cylinder, manifold pressure notwithstanding. And, since you'd be doing it with less pumping loss, the torque output is actually higher in your case with a reduced DCR/lower effective displacement at low RPM. What ends up happening with your model is when the inverse APC model is parsed for a spark value, it doesn't move as far down in terms of spark because you've told it that it makes less torque with the same air mass compared to stock - now in order to meet that same -26ft-lb torque request, it is able to select a higher spark advance (or higher air mass, depending on low level arbitration). Looking at that spark advance creeping up and with an already low APC, the controller is adding spark and driving your idle up.

While there is no suitable way to calibrate this on the street (yet), you should work on that airmass model more. I'd put that back to stock and then add a little more torque in the area where you've reduced actual low-end torque. If you could actually get that air into the cylinders, that cam would be doing it with less pumping effort. The MAP model may benefit from revision too, but one at a time.


This ended up being far longer than I anticipated, so I'm gonna stop with just question 1 for now lol. I have a feeling making those changes will end up helping numbers 2 and 4 though due to the likelihood of hanging the throttle open rather than adding spark to meet a higher torque value. I also have a solution to 3 as well, but for another time. Hope this helps.

[TriPinTaZ]

What you've asked for is essentially a mechanical engineer's dissertation...it's pretty gnarly stuff for the layman to tackle. I'll give you an overview though, or at least some info based on what the issues appear to be in the data.

For the first topic, you might read this for some background: https://forum.hptuners.com/showthrea...l=1#post616505. I posted that a couple weeks ago to give a basic who's who of idle calibration. It's based on older engines/technology but the physics remains the same. That said, based on what I see in the data I think it's good to remember that your vehicle has a variable displacement AC compressor which means different torque loading based on how hard it is being driven. It also multiplies the number of torque model operating points that have to be spot-on to avoid those RPM flares. Looking at the beginning of the first RPM flare at 14:13.200, you can see the controls preparing to do exactly what I described in the previous post. If you look at the predicted torque, you can see it jump up 12ft-lbs really quickly. The throttle and MAP are following this request. So, predicted torque is essentially a feedforward estimate at the present operating conditions, or an 'unmanaged' torque potential based on the output of the calibrations you've entered into your torque model. Now the predicted torque is only an expectation from the controller, basically a target to be able to reach. Then the immediate torque comes in and provides final arbitration over actual desired torque delivery based on any other requestors that may want less torque. It may help to think of predicted torque as a slow path, ie airflow and the immediate torque as a fast path control like spark or fuel. You can make torque much faster with a change in spark than you can with a change in airflow because air moves slowly, especially at idle. Now the goal here is to have a nice, smooth AC clutch engagement. Customers don't want to feel it and the torque reserve makes this possible. So you bring in more airflow, pull out some spark via the immediate 'managed' torque request and you have a reserve ready to ramp back in as the AC clutch comes on. Typically this achieves a negligible impact to idle RPM but in your case, as you mentioned...the problem is the torque model. A snapshot from your file...
torque model inflection.PNG
The point the cursor is hovering over shows the problem - an inflection in the spark response. It shouldn't spike up, especially not when your idle speed is too high. And it isn't a transition to main spark because the state variables you've got in your log don't change to 'torque'. Since you've likely noticed there is no idle spark table, you know that idle spark value is coming from the torque model. There are quite a few nuances to get the idle spark value from the torque model as it is technically the inverse of that model, which is technically a blended array of models where the APC or airmass model is more or less primary. So the immediate torque request is targeting -26ft-lbs at 732 RPM and 0.186 grams in the cylinder. Now attaching your airmass torque model vs a stock 2014 Silverado airmass torque model:
VT_mod.PNGVT_stock.PNG
I can see how it may make sense to decrease the modeled torque output at lower RPM. Your cam moved the torque curve up, so it'll be a little soggier on the bottom end. Have to keep in mind though, this isn't a VE table...you've got mass on the Y axis, not pressure. While torque will look a lot more like VE with MAP on the Y, it's not the same for air mass. Those tables depict the torque output with the assertion that all of that air mass is actually getting smashed into the cylinder, manifold pressure notwithstanding. And, since you'd be doing it with less pumping loss, the torque output is actually higher in your case with a reduced DCR/lower effective displacement at low RPM. What ends up happening with your model is when the inverse APC model is parsed for a spark value, it doesn't move as far down in terms of spark because you've told it that it makes less torque with the same air mass compared to stock - now in order to meet that same -26ft-lb torque request, it is able to select a higher spark advance (or higher air mass, depending on low level arbitration). Looking at that spark advance creeping up and with an already low APC, the controller is adding spark and driving your idle up.

While there is no suitable way to calibrate this on the street (yet), you should work on that airmass model more. I'd put that back to stock and then add a little more torque in the area where you've reduced actual low-end torque. If you could actually get that air into the cylinders, that cam would be doing it with less pumping effort. The MAP model may benefit from revision too, but one at a time.


This ended up being far longer than I anticipated, so I'm gonna stop with just question 1 for now lol. I have a feeling making those changes will end up helping numbers 2 and 4 though due to the likelihood of hanging the throttle open rather than adding spark to meet a higher torque value. I also have a solution to 3 as well, but for another time. Hope this helps.

Would you agree that the VVE model is still effectively telling the ECU how much of the incoming air is efficiently used for combustion? If so, wouldn't it make more sense to reduce the the VVE in the idle areas if you have a cam with more overlap than OEM? Then this would equate to less torque and you would tell VT that you're making less torque for the given Airmass/MAP at idle and it the ECU would command more spark to meet the Idle Torque demanded?

[smokeshow]

Would you agree that the VVE model is still effectively telling the ECU how much of the incoming air is efficiently used for combustion? If so, wouldn't it make more sense to reduce the the VVE in the idle areas if you have a cam with more overlap than OEM? Then this would equate to less torque and you would tell VT that you're making less torque for the given Airmass/MAP at idle and it the ECU would command more spark to meet the Idle Torque demanded?

Yeah that's correct. The first part about recalibrating the VVE, anyway. There is an order of operations that should ideally be followed to yield the best result. You can't model torque without airflow; that's what the APC term is.

I feel I should clarify: installing a bigger camshaft and shifting the torque curve higher in the RPM range reduces maximum available torque at a given lower engine speed, namely at idle. It does not mean it requires that much less torque to idle. Neglecting the pumping loss benefit, there is still a lot of parasitic loss simply from a spinning engine and accessories that requires torque to overcome. Those losses go nowhere with a different camshaft. If you needed 30 ft-lbs indicated torque to idle with the stock camshaft, you still need the same torque (read: APC or airmass) to idle at the same RPM with the bigger camshaft. But with the intake valve closing later, you have less effective displacement to make that torque so you must utilize a larger percentage of the potential torque to do the same job. This is why big cams always need higher MAP to idle. They've got less to give down low.

Funny story, I worked on the GSE-T4 Firefly engine that Chrysler brought out a few years ago. Me and my team were on a development trip in Dillon, CO battling a problem with the new engine. It was a turbo 1.3L rated at like 180hp, so it was a pretty powerful little thing. But when we went to Dillon at 9000ft elevation and turned all of the accessories on in the car, the little 1.3 was literally wide open throttle just to make enough torque to idle in gear. And it wasn't enough exhaust to spool the turbo obviously, so it was basically a flintstone act to get the thing moving from a dead stop. That one required both hardware and software fixes lol. It's the same concept though. Less available displacement but at altitude.

[TriPinTaZ]

Yeah that's correct. The first part about recalibrating the VVE, anyway. There is an order of operations that should ideally be followed to yield the best result. You can't model torque without airflow; that's what the APC term is.

I feel I should clarify: installing a bigger camshaft and shifting the torque curve higher in the RPM range reduces maximum available torque at a given lower engine speed, namely at idle. It does not mean it requires that much less torque to idle. Neglecting the pumping loss benefit, there is still a lot of parasitic loss simply from a spinning engine and accessories that requires torque to overcome. Those losses go nowhere with a different camshaft. If you needed 30 ft-lbs indicated torque to idle with the stock camshaft, you still need the same torque (read: APC or airmass) to idle at the same RPM with the bigger camshaft. But with the intake valve closing later, you have less effective displacement to make that torque so you must utilize a larger percentage of the potential torque to do the same job. This is why big cams always need higher MAP to idle. They've got less to give down low.

Exactly regarding your point about a higher MAP at idle. This is actually the engine being less efficient and pulling less vacuum at idle due to the overlap. So now instead of 35KPA you're idling at 60KPA and the 60KPA values in the VVE table are too high. Maybe we're both trying to say the same thing differently?


Now looking at the Virtual Torque, your idling at a similar or slightly higher airmass and a higher MAP area in the VT Tables. So the ECU is thinking its making too much torque at idle and starts pulling timing, this is how you get the negative idle timing issues with cams. You have to tell the VT Table that the new idle areas in VT are not making as much torque as the OEM calibration says. Now the VT MAP values are way off and the VT Airmass still need to be lowered. So by lowering these values the ECU allows the appropriate amount of timing at idle. Then they need to be blended like the OEM logic for increasing throttle/RPM. You don't necessarily have to greatly change the tables, in fact as little as possible is best because you start affecting other things if you change them too much.

This is how I have found the Gen V stuff to work in my experiences. I'm certainly open to more knowledge. Just sharing my experiences.

[smokeshow]

Exactly regarding your point about a higher MAP at idle. This is actually the engine being less efficient and pulling less vacuum at idle due to the overlap. So now instead of 35KPA you're idling at 60KPA and the 60KPA values in the VVE table are too high. Maybe we're both trying to say the same thing differently?


Now looking at the Virtual Torque, your idling at a similar or slightly higher airmass and a higher MAP area in the VT Tables. So the ECU is thinking its making too much torque at idle and starts pulling timing, this is how you get the negative idle timing issues with cams. You have to tell the VT Table that the new idle areas in VT are not making as much torque as the OEM calibration says. Now the VT MAP values are way off and the VT Airmass still need to be lowered. So by lowering these values the ECU allows the appropriate amount of timing at idle. Then they need to be blended like the OEM logic for increasing throttle/RPM. You don't necessarily have to greatly change the tables, in fact as little as possible is best because you start affecting other things if you change them too much.

This is how I have found the Gen V stuff to work in my experiences. I'm certainly open to more knowledge. Just sharing my experiences.

I see what you're saying. Efficiency in terms of cylinder fill capability. Yes, the cylinder will be less able to ingest air at low RPM with the large camshaft. More efficient though in terms of pumping loss, which is the idea behind the Atkinson cycle. A large camshaft, Otto-cycle engine approximates the Atkinson's efficiency because its expansion ratio is greater than its compression ratio. But that's another topic. The point is you still need that same trapped air mass to generate the torque to idle. That air mass is an output of the VVE and MAF. So the air modeling and measurement absolutely needs to be accurate in order to generate an accurate torque model.

A piece of information that is probably not well known: only the airmass torque model provides torque estimation for consumption by other pieces of software in the controller during normal run conditions. The MAP model is expected to be more prone to error due to air turbulence in the manifold. The only time the MAP model is used for torque estimation is during startup, where the manifold is full and the air begins stationary; after that, the airmass model takes over and the MAP model is actually filtered towards it. I can see how it would appear in HP Tuners that they are equivalent, but the airmass torque estimation is the primary. Further, each model has an inverse. The inverse of the airmass torque model using the torque request, airmass, cam positions, etc as inputs generates the feed forward terms for closed-loop control of idle: desired torque, desired throttle, desired airflow, desired airmass, desired spark, etc. The MAP model has limited functionality - the inverse handles desired MAP and by extension desired throttle (as both pressure ratio and flow are used in the compressible flow equation for throttle position determination) and the non-inverse handles limited torque estimation during cranking that I mentioned before.

I'm sort of skipping some finer details, but the key takeaway is this: an accurately calibrated airmass torque model is critical for idle stability. It receives engine operating conditions to estimate torque and the inverse takes the torque request and produces the new desired operating conditions. The correlation between the inputs to the model and the torque estimate itself is a linear relationship and specific to the engine. Barring changes to the cylinder head, compression, etc, the fixed airmass will produce a fixed torque (ignoring for now pumping losses, as I mentioned earlier). With a larger cam and the later IVC, there is less effective volume, so more MAP is necessary. If you decrease the airmass model torque, a product of airmass and spark (and of course RPM, but assume it is constant), you are changing the calibrated relationship between the torque value and the model inputs (or the inverse's outputs) that generate it. It does not make sense to decrease the torque product of airmass and spark without another variable also changing, like compression or chamber design. In fact the indicated torque increases ever so slightly because of decreased pumping loss.

This is harder to explain than I anticipated

[cmitchell17]

It sounds like to me ya'll are saying that the reason you get the low/negative timing at idle with bigger cams is only because the later intake valve closing lowers your effective displacement, but what I can't understand is the loss of airmass from big cam overlap/late intake valve closure metered by the MAP and or MAF sensor? So does this loss in "effective displacement" or loss in airmass/flow show up in dynamic APC? Now if it does show up, then technically we wouldn't have to alter virtual torque right?

So smokeshow, in your opinion when does virtual torque need to be changed? Seems to me for any lessen airflow restriction type mode (ported intakes, ported heads, etc.) seems like stock virtual torque would never (or should never) be changed. Now gets kind of questionable for resonance tuning kind of mods like headers. For compression increasing it seems obvious that it would need to be adjusted for that. It seems like to me a tuned set of headers that is scavenging good at a certain RPM would show up in APC right? and therefore not need a virtual torque adjustment?

[smokeshow]

It sounds like to me ya'll are saying that the reason you get the low/negative timing at idle with bigger cams is only because the later intake valve closing lowers your effective displacement, but what I can't understand is the loss of airmass from big cam overlap/late intake valve closure metered by the MAP and or MAF sensor? So does this loss in "effective displacement" or loss in airmass/flow show up in dynamic APC? Now if it does show up, then technically we wouldn't have to alter virtual torque right?

It's not making sense because it doesn't, that's what I'm saying. The change to airmass and thus torque required to maintain idle with no change other than a cam is nearly negligible. A change in the modeled torque must be accompanied by a change in spark, airmass, or both. It's like adding a number to one side of an equation; the math doesn't work.

So smokeshow, in your opinion when does virtual torque need to be changed? Seems to me for any lessen airflow restriction type mode (ported intakes, ported heads, etc.) seems like stock virtual torque would never (or should never) be changed. Now gets kind of questionable for resonance tuning kind of mods like headers. For compression increasing it seems obvious that it would need to be adjusted for that. It seems like to me a tuned set of headers that is scavenging good at a certain RPM would show up in APC right? and therefore not need a virtual torque adjustment?

You have to change the torque model if the modifications you are making are not represented by the inputs to the model. For a simple comparison, consider the old conventional volumetric efficiency tables. That model is just VE incorporated into the ideal gas law, pv=nrt. Boost the engine, VE may not change much. Double the IAT, same result. But change the camshaft and the entire table needs to be edited. The camshaft isn't one of the input variables, so the model output must be adjusted. In the case of virtual torque, if you were increasing airmass, torque would go up. If you increase spark, torque would go up. Changing nothing to the inputs and you don't expect the output to change either. Now like the conventional VE/camshaft example, a reduction in pumping loss due to less manifold vacuum is not represented by the torque model inputs, so it will likely shift torque up a bit. Same for headers. If scavenging is happening, that's a reduction in pumping loss. Torque goes up slightly because of lower parasitic losses, potentially requiring some edits. Depending on the headers and cam, it may increase APC as well...but the MAF will pick that up, feed it into the model and generate the increased torque estimate for you.

The challenge when you do need to edit the torque model is representing the physics properly so you can make the correct changes without undershooting, overshooting, or going the wrong way completely.

[cmitchell17]

So first I want to say how much I appreciate the answers, some of these questions I have I felt would never be answered and I have a lot of confidence in your answers, hopefully I can get around to actually understanding it though haha

I think I get what you are saying to put it simply that mods you do to increase air will be correctly picked up by the MAF and appropriately reference a new, increased airmass cell and produce a higher torque lookup from that table. However what we can't simply do is model the "artificial" torque increases that are caused by reduced pumping losses. With that being said the slight "artificial" torque boost of we get at idle with a big cam from the pseudo Atkinson cycle/late intake valve closure (even though it seems like this effect would be negated by a associated change in exhaust timing as well) is what causes us having to "make up for" with a decrease in virtual torque. I however, am kind of confused still on why we decrease VT with a cam in the idle ranges? If we are "artificially" making more torque with a cam why are we not increasing it?

[SultanHassanMasTuning]

whats your cam specs?

[smokeshow]

So first I want to say how much I appreciate the answers, some of these questions I have I felt would never be answered and I have a lot of confidence in your answers, hopefully I can get around to actually understanding it though haha

I think I get what you are saying to put it simply that mods you do to increase air will be correctly picked up by the MAF and appropriately reference a new, increased airmass cell and produce a higher torque lookup from that table. However what we can't simply do is model the "artificial" torque increases that are caused by reduced pumping losses. With that being said the slight "artificial" torque boost of we get at idle with a big cam from the pseudo Atkinson cycle/late intake valve closure (even though it seems like this effect would be negated by a associated change in exhaust timing as well) is what causes us having to "make up for" with a decrease in virtual torque. I however, am kind of confused still on why we decrease VT with a cam in the idle ranges? If we are "artificially" making more torque with a cam why are we not increasing it?

Which model are you referring to? VT is sort of a blanket term - the MAP and airmass models are very different and must both reflect the physics of the changes to the engine to coordinate achieving the correct feed forward desired torque as well as preventing steady state torque error.

Maybe we should try to focus this on being a data-driven kind of discussion so it's not so high level and abstract.

[Buck]

Let me dumb things down, I believe Smokeshow and Taz are both saying the same thing completely different... In the example of a camshaft replacement the engine now idles at a different pressure and RPM, possibly. Referencing the VT Airmass model we find that torque value is now in a potentially higher value numerically. So adjust the model accordingly. We are not necessarily raising/lowering torque but making the model reflect our new idle location. So if we knew approx. where our engine idled pre-modification, imagine if you will... your VT Airmass model torque numbers on a transparent sheet overlaying your VT Airmass model on a sheet without torque numbers only X/Y axis. Now slide your transparency to the appropriate new idle location and adjust accordingly in your model... If it was only that simple in our software.

Maybe I'm totally off but thats what I gathered from this thread