VVT calibration
01-05-2018, 05:44 AM
#1
VVT calibration
This is intended to be a high-level DIY guide to calibrating late model GM controllers for variable valve timing using HP Tuners. If you've ever wondered why GM started using a strategy as cumbersome as virtual volumetric efficiency in the first place, you can at least somewhat point the finger at VVT. It is well known that retarding your intake cam timing can increase high RPM torque. This is in fact what the VVT control surfaces do in stock GM calibrations. Notice here that as the RPM increases the cam position values increase (cam retards). Also note that the intake cam angle is referring to the camshaft angle as a whole since all lobes rotate together...so exhaust valve phasing effects on volumetric efficiency are baked into the intake coefficients.
The small area that is excessively retarded in the low load cruise RPM range is there for emissions and fuel economy reasons, but I won't detail that here.
In order to accurately calculate speed density cylinder air mass for VVT, using the previous strategy of the explicit/conventional VE surface the controller would require a new VE surface for every tiny infinitesimal movement of the camshaft. This isn't realistic from a controls standpoint mostly because of the sheer amount of controller memory space this would require. So GM uses a calculation to factor in not only RPM and air pressure into a given VE surface, but any number of other 'modifying' parameters such as cam timing. The result is a zone-segmented and (hopefully) continuous virtual surface (more specifically in this case, a 2nd order response surface) providing a calculated cylinder air mass output that changes as a function of the independent variable inputs and, to a lesser extent, the covariance of the individual independent variables.
If this sounds confusing, its because it is...but luckily calibrating for VVT is much easier than understanding it. You might be thinking though, why would I need to go through all this trouble if my car/truck seems to run just fine with my new TSP VVT camshaft? In short, its because GM logic ignores speed density input to the air mass calculation during steady state operation. A properly calibrated MAF sensor is the trusted airflow measurement source under steady state conditions and the MAF doesn't care what the cam position is. So that VVT angle dependency seems to go right out the window. However, during throttle/MAP/RPM transitions or MAF-less vehicles, the MAF isn't used as much (or at all) so this all becomes necessary. As an aside, a more astute calibrator might also ask, why put in all the effort if I always command the same cam angle at all points in an RPM column regardless of the cylinder airmass? (pictured below) The answer is that those values are a target angle, not the actual cam angle. A quick rev of the engine will result in an actual cam angle that significantly lags its target, resulting in an error in the air mass calculation. The cam phaser can only react so quickly, and indeed the ramp rate itself is also calibratable, so the logic uses current cam position in the VVE determination rather than the target position.
Now that we've listed some conditions concerning the why's and how's of VVT and its effects on the VVE surface, we should probably look at quantifying them into data filters that can be used in the HPT scanner. I use the scanner filters pretty heavily because I'm lazy, but also because it can make quick work of a job of this size. I won't be explaining how the VVE calibration itself works here as I am assuming that you are familiar with VVE if you're attempting to calibrate for an aftermarket VVT camshaft. Some of the operating conditions we've listed are: 1) RPM & MAP (and corresponding zone number); 2) steady state operation (again, if you've calibrated any fueling ever, this is familiar); 3) properly calibrated MAF, if present (used as a coarse VVE cal shortcut; I'll explain this later if there is interest); 4) actual cam position. I expect you're aware of plotting STFT or AFR error versus RPM & MAP in the scanner. So, you'll want to generate some scanner filters from these conditions.
#1 isn't necessarily a scanner filter, but more of a consideration you should be aware of for VVE calibration. As in the screen shot below, the VVE zones aren't divided up into a nice orthogonal grid. In the low MAP/RPM ranges, the zone boundary values are stepped in an odd manner to shift resolution in the VVE surface. This is done to optimize the VVE surface for areas that need greater resolution. This screen shot is from my 5.3 Colorado, but you'll notice that factory supercharged cars like the ZL1 Camaro have this same type of zone segmentation in the boost/WOT areas. I am mentioning this because the VVE accuracy is dependent on each individual zone. The logic includes a hysteresis value for both pressure and RPM when crossing zone boundaries, so it is possible to plot zone 1, zone 6, or zone 7 STFT into zone 0. There won't be a huge impact on fueling but it can produce large step discontinuities especially if there is fueling error contribution from multiple sources ie. cam position AND pressure. If you've wondered why GM's factory VVE surfaces sometimes have unrealistic staircase-looking things going on in the middle of a normal operating range, this is why. The only way to fix this apart from calibrating 30 VVE zones multiplied by 20 cam positions (600 VE surfaces) is to align your zone boundaries and avoid plotting STFT data near the boundaries. To my knowledge, zone ID is not available on the CAN hence why we cannot set up a filter in HPT for it. Though not available in all HPT definition files, common zone hysteresis values are 25 RPM and approximately 2.5kpa. This means that if you have zone 7 for example that stretches between 1000-1750rpm and 25-45kpa, it is best to avoid plotting STFT data within 25rpm and 2.5kpa of the edges of the zone. How you do this is up to you.
#2 Steady state operation, again up to you. I use a couple filters that ignore STFT or AFR error during and following transients above a certain threshold.
#3 If a MAF is installed and calibrated*, it is possible to plot VVE airflow against MAF airflow to rough in the majority of the VVE surface. Again, I can detail this later if there is interest.
#4 Since this is all about VVT, actual cam position is the main filter here. To capture fueling error at an Intake Cam Angle of 1 +/- 0.5 (from 0.5 to 1.5) degree, the HPT scanner filter would be as shown below. The variable ID for Intake Cam Angle is [2172.161] and I included [50121] (commanded AFR) for reference. This filter will plot all fueling error values in the histogram/graph versus MAP and RPM for cam angles between 0.5 and 1.5 degrees retarded and commanded AFR above 14.6:1.
So, it's fairly simple to set up from a filter standpoint. Choose which cam angle you want to collect data for and plot the fueling error vs MAP and RPM in the scanner. Once you have populated enough cells you then correct the error on the VVE surface after selecting the camshaft angle and hit the 'calculate coefficients' button.
Setting the cam angle can be done in the bidirectional controls (not sure if this is available on ALL VVT-capable controllers) or in the calibration itself. In the calibration, you'd ideally set the entire surface to a single cam angle and collect the fueling error data. In this case, the value is set to 5 degrees retarded for the entire operating range. No matter what, the target angle will be 5 degrees. It is still a good idea to keep the filter active in the scanner for something like [2172.161]>4.5 AND [2172.161]<5.5 because there are certain off-nominal scenarios where the controller will ignore this surface as the target and actual position will not be the same. Collect as much of the error as possible and correct the error in the VVE surface at 5 degrees.
The small area that is excessively retarded in the low load cruise RPM range is there for emissions and fuel economy reasons, but I won't detail that here.
In order to accurately calculate speed density cylinder air mass for VVT, using the previous strategy of the explicit/conventional VE surface the controller would require a new VE surface for every tiny infinitesimal movement of the camshaft. This isn't realistic from a controls standpoint mostly because of the sheer amount of controller memory space this would require. So GM uses a calculation to factor in not only RPM and air pressure into a given VE surface, but any number of other 'modifying' parameters such as cam timing. The result is a zone-segmented and (hopefully) continuous virtual surface (more specifically in this case, a 2nd order response surface) providing a calculated cylinder air mass output that changes as a function of the independent variable inputs and, to a lesser extent, the covariance of the individual independent variables.
If this sounds confusing, its because it is...but luckily calibrating for VVT is much easier than understanding it. You might be thinking though, why would I need to go through all this trouble if my car/truck seems to run just fine with my new TSP VVT camshaft? In short, its because GM logic ignores speed density input to the air mass calculation during steady state operation. A properly calibrated MAF sensor is the trusted airflow measurement source under steady state conditions and the MAF doesn't care what the cam position is. So that VVT angle dependency seems to go right out the window. However, during throttle/MAP/RPM transitions or MAF-less vehicles, the MAF isn't used as much (or at all) so this all becomes necessary. As an aside, a more astute calibrator might also ask, why put in all the effort if I always command the same cam angle at all points in an RPM column regardless of the cylinder airmass? (pictured below) The answer is that those values are a target angle, not the actual cam angle. A quick rev of the engine will result in an actual cam angle that significantly lags its target, resulting in an error in the air mass calculation. The cam phaser can only react so quickly, and indeed the ramp rate itself is also calibratable, so the logic uses current cam position in the VVE determination rather than the target position.
Now that we've listed some conditions concerning the why's and how's of VVT and its effects on the VVE surface, we should probably look at quantifying them into data filters that can be used in the HPT scanner. I use the scanner filters pretty heavily because I'm lazy, but also because it can make quick work of a job of this size. I won't be explaining how the VVE calibration itself works here as I am assuming that you are familiar with VVE if you're attempting to calibrate for an aftermarket VVT camshaft. Some of the operating conditions we've listed are: 1) RPM & MAP (and corresponding zone number); 2) steady state operation (again, if you've calibrated any fueling ever, this is familiar); 3) properly calibrated MAF, if present (used as a coarse VVE cal shortcut; I'll explain this later if there is interest); 4) actual cam position. I expect you're aware of plotting STFT or AFR error versus RPM & MAP in the scanner. So, you'll want to generate some scanner filters from these conditions.
#1 isn't necessarily a scanner filter, but more of a consideration you should be aware of for VVE calibration. As in the screen shot below, the VVE zones aren't divided up into a nice orthogonal grid. In the low MAP/RPM ranges, the zone boundary values are stepped in an odd manner to shift resolution in the VVE surface. This is done to optimize the VVE surface for areas that need greater resolution. This screen shot is from my 5.3 Colorado, but you'll notice that factory supercharged cars like the ZL1 Camaro have this same type of zone segmentation in the boost/WOT areas. I am mentioning this because the VVE accuracy is dependent on each individual zone. The logic includes a hysteresis value for both pressure and RPM when crossing zone boundaries, so it is possible to plot zone 1, zone 6, or zone 7 STFT into zone 0. There won't be a huge impact on fueling but it can produce large step discontinuities especially if there is fueling error contribution from multiple sources ie. cam position AND pressure. If you've wondered why GM's factory VVE surfaces sometimes have unrealistic staircase-looking things going on in the middle of a normal operating range, this is why. The only way to fix this apart from calibrating 30 VVE zones multiplied by 20 cam positions (600 VE surfaces) is to align your zone boundaries and avoid plotting STFT data near the boundaries. To my knowledge, zone ID is not available on the CAN hence why we cannot set up a filter in HPT for it. Though not available in all HPT definition files, common zone hysteresis values are 25 RPM and approximately 2.5kpa. This means that if you have zone 7 for example that stretches between 1000-1750rpm and 25-45kpa, it is best to avoid plotting STFT data within 25rpm and 2.5kpa of the edges of the zone. How you do this is up to you.
#2 Steady state operation, again up to you. I use a couple filters that ignore STFT or AFR error during and following transients above a certain threshold.
#3 If a MAF is installed and calibrated*, it is possible to plot VVE airflow against MAF airflow to rough in the majority of the VVE surface. Again, I can detail this later if there is interest.
#4 Since this is all about VVT, actual cam position is the main filter here. To capture fueling error at an Intake Cam Angle of 1 +/- 0.5 (from 0.5 to 1.5) degree, the HPT scanner filter would be as shown below. The variable ID for Intake Cam Angle is [2172.161] and I included [50121] (commanded AFR) for reference. This filter will plot all fueling error values in the histogram/graph versus MAP and RPM for cam angles between 0.5 and 1.5 degrees retarded and commanded AFR above 14.6:1.
So, it's fairly simple to set up from a filter standpoint. Choose which cam angle you want to collect data for and plot the fueling error vs MAP and RPM in the scanner. Once you have populated enough cells you then correct the error on the VVE surface after selecting the camshaft angle and hit the 'calculate coefficients' button.
Setting the cam angle can be done in the bidirectional controls (not sure if this is available on ALL VVT-capable controllers) or in the calibration itself. In the calibration, you'd ideally set the entire surface to a single cam angle and collect the fueling error data. In this case, the value is set to 5 degrees retarded for the entire operating range. No matter what, the target angle will be 5 degrees. It is still a good idea to keep the filter active in the scanner for something like [2172.161]>4.5 AND [2172.161]<5.5 because there are certain off-nominal scenarios where the controller will ignore this surface as the target and actual position will not be the same. Collect as much of the error as possible and correct the error in the VVE surface at 5 degrees.
01-05-2018, 05:50 AM
#2
The amount of resolution you calibrate for here is up to you. If you have an aftermarket VVT cam and a phaser limiter installed that prevents cam retard beyond 20 degrees and wanted to rough in the VVT calibration, you can perform the fueling error correction for data collected with the cam angle at 0 for the entire surface and again at 20 so long as you use an interpolation in excel to fill in the gap for angles 1 through 19. There is no way to interpolate this in HPT save for already having the correct coefficients, so it must be done manually. You CANNOT ignore the values in between because the coefficient generation still takes those values into consideration when performing its polynomial line fitting. If you DO interpolate in excel, the downside is that there is a potential for higher-order errors to creep into the VVE surface. These will be noticeable as undesired peaks or valleys in the surface as you click through the cam angle range in the VVE editor. However, if you go through this procedure for each cam angle in 1 degree increments, this issue will not present itself.
Performing this procedure for each individual cam angle isn't realistic or necessary. There will be certain cam angles that you probably won't ever set as a target value, for example 0 degrees at 6000rpm. So, there is no need to gather data out that far at that cam angle. The question then becomes, how far is far enough? The answer is in the response rate of the cam phaser versus the rev rate of the engine and the rate of increase of cylinder airmass. For example in the case of a high rev rate, you might be idling at 800rpm at 0 degrees cam angle, stab the throttle until the engine reaches 4000rpm where the target angle is 6 degrees but find that the actual cam angle has only reached 4 degrees at that point. So, you'd find the intake cam angle error as 6-4=2 and this becomes the lower range that you should calibrate for during a rev-up scenario. For a 6 degree target at 4000rpm, you need fueling error data as low as 4 degrees at 4000rpm. If your cam target angle isn't linear with RPM, this also needs to be tested for decreasing revs as well. Note: this test is best done at cold oil temps as phaser response will be slowest then. If you do not change the target angle versus cylinder airmass, this does not need to be done for rate change of airmass.
In addition to the phaser response rate, I would suggest hitting all cells in a given zone if your engine/vehicle combination will spend any significant amount of time operating there. For example, in zone 11 picture below, if you were to launch off idle, your cylinder airmass may hit 0.75g at 1400rpm before revving the engine out of that zone. Even if that was the fastest the engine would respond during a transient and the engine would normally never touch the 1000 and 1200rpm cells, those cells in the 1000 and 1200rpm columns will still weigh on the coefficient generation and change the VVE airmass. Not addressing this can result overly lean or rich fueling at 0.75g and 1400rpm even if there is 0% error in that cell when you generate the VVE coefficients. It's pretty counter-intuitive, but that's how VVE and the zone resolution works. The solution would be loading the vehicle down and forcing the RPM low enough at that air mass to get a comprehensive error collection of the entire zone.
This can all be done either on a dyno or on the street so long as you pay attention to maintaining steady state operation when collecting error data. Not only is fueling going to mess with the STFT on a transition, but with VVT the cam angle error will as well. Hopefully this shines some light on why tooners who blow out a VVT calibration on the dyno in an hour are full of ****
Any questions related to VVT or VVE calibration or if I've missed something, let me know...
Performing this procedure for each individual cam angle isn't realistic or necessary. There will be certain cam angles that you probably won't ever set as a target value, for example 0 degrees at 6000rpm. So, there is no need to gather data out that far at that cam angle. The question then becomes, how far is far enough? The answer is in the response rate of the cam phaser versus the rev rate of the engine and the rate of increase of cylinder airmass. For example in the case of a high rev rate, you might be idling at 800rpm at 0 degrees cam angle, stab the throttle until the engine reaches 4000rpm where the target angle is 6 degrees but find that the actual cam angle has only reached 4 degrees at that point. So, you'd find the intake cam angle error as 6-4=2 and this becomes the lower range that you should calibrate for during a rev-up scenario. For a 6 degree target at 4000rpm, you need fueling error data as low as 4 degrees at 4000rpm. If your cam target angle isn't linear with RPM, this also needs to be tested for decreasing revs as well. Note: this test is best done at cold oil temps as phaser response will be slowest then. If you do not change the target angle versus cylinder airmass, this does not need to be done for rate change of airmass.
In addition to the phaser response rate, I would suggest hitting all cells in a given zone if your engine/vehicle combination will spend any significant amount of time operating there. For example, in zone 11 picture below, if you were to launch off idle, your cylinder airmass may hit 0.75g at 1400rpm before revving the engine out of that zone. Even if that was the fastest the engine would respond during a transient and the engine would normally never touch the 1000 and 1200rpm cells, those cells in the 1000 and 1200rpm columns will still weigh on the coefficient generation and change the VVE airmass. Not addressing this can result overly lean or rich fueling at 0.75g and 1400rpm even if there is 0% error in that cell when you generate the VVE coefficients. It's pretty counter-intuitive, but that's how VVE and the zone resolution works. The solution would be loading the vehicle down and forcing the RPM low enough at that air mass to get a comprehensive error collection of the entire zone.
This can all be done either on a dyno or on the street so long as you pay attention to maintaining steady state operation when collecting error data. Not only is fueling going to mess with the STFT on a transition, but with VVT the cam angle error will as well. Hopefully this shines some light on why tooners who blow out a VVT calibration on the dyno in an hour are full of ****
Any questions related to VVT or VVE calibration or if I've missed something, let me know...
The following users liked this post:
Tuning_addict77 (11-22-2019)
01-05-2018, 06:02 AM
#3
Thanks for this smokeshow.
1. For the values related to intake and exhaust position, are you saying the values shown are retard values?
2. Or are they related to the cam? Intake = advance and exhaust = retard?
3. If #1 is true, how do you advance the cams?
1. For the values related to intake and exhaust position, are you saying the values shown are retard values?
2. Or are they related to the cam? Intake = advance and exhaust = retard?
3. If #1 is true, how do you advance the cams?
All lobes rotate as one, so retard on intake is retard on exhaust.
Trending Topics
01-11-2018, 08:26 AM
#8
Read
This will take some time to read and digest.
it will be very interesting since i have VVT
I will need my tuner to print out this stuff
without telling him he got the tune wrong.
or screwed up the adjustments. I don't want
to **** him off, he spent a lot of time working
on the VVT settings. I feel/think the truck
runs awesome.
it will be very interesting since i have VVT
I will need my tuner to print out this stuff
without telling him he got the tune wrong.
or screwed up the adjustments. I don't want
to **** him off, he spent a lot of time working
on the VVT settings. I feel/think the truck
runs awesome.
01-22-2018, 10:22 AM
#10
I can add and remove but nothing less than 0.
Thoughts?
How does ramp rate affect this? The way I have digested it is "how fast the cams get to the configured values". I might be wrong though.