Can ETC Angle Errors Cause Fueling Issues?

[CCS86]

Hey guys,

In steady state my fuel trims look really nice, but I am chasing down transient trim swings (rich and lean). I'm trying to see where to focus my energy and think that my injector calibration may very well still be off (MU52s on a 4 bar Coyote). But today, I did notice an interesting correlation between ETC angle error and some of these fuel swings.

My gut says that the fueling is tied to MAF, injector, and maybe SD calibrations. With ETC angle error being more of a throttle response issue. I know that just because it correlates, doesn't mean it's the cause. But I wanted to float the idea across anyway.

Any help is appreciated!

ETC Error vs STFT.png

[veeefour]

Regarding fuel swings - You may have torque related fuel enleanments. Turn all FUEL corrections OFF under TORQUE MANAGEMENT/TORQUE RATIO.

[CCS86]

Don't those settings only allow for a commanded enleanment for torque reduction purposed? Anything commanded should not cause a STFT error.

[CCS86]

I have been searching and reading on methods to correct ETC throttle angle error and just haven't found a consensus at all.

Some people say you need to plot IPC Wheel TQ Error, not ETC Angle Error. For whatever reason, my values for Wheel TQ Error are always 0.0.

If we are plotting ETC Angle Error against Effective Area and ETC Vacuum, I can't find a concise explanation of what to do with the data. Some talk about applying half your average error back to the relevant cells in the Predicted Angle table. The is mention of updating the Effective Area table, but no real steps about how.

In general my throttle response actually feels great (far better than with the Roush OS I used first). I want to correct these mainly to improve fueling spikes. I see some giant negative error when shutting the throttle fast. As well as other positive errors during roll-on:


ETCError.png

[Devildog1325]

If you have correct injector data and you have correct throttle body data I would leave them alone and work on your torque tables. The information is out there on how to correct this and people go about it different ways so you will have to do allot of reading and trial and error to see what is easier for you, sometimes it takes a little bit to soak up all that information to be able to apply it correctly, I doubt it will happen over night. Regardless the torque and torque inverse tables and driver demand tables will definitely cause tip in and tip out problems on a boosted car more so than what people think. I went through the same thing chasing my tail with transient fueling thinking that was the problem. Usually the issue is that everything looks fine as your cruising as far as fuel trims and etc error but accelerating you may be commanding more torque faster than the engine can produce it causing a lean tip in or you may have a spike in your driver demand table and not realize it after you have adjusted the Torque tables. The torque tables may not be off bad enough to cause a huge wheel torque error but enough to cause a problem that look like a transient fuel issue.

[CCS86]

Can somebody post a log that is basically free from ETC angle errors, during abrupt pedal movement?

The more I look at the data, the less I agree with what I have read about "correcting" these errors.

I see essentially no steady state ETC angle error. Couldn't the ETC Angle Error just be the delta between desired and current throttle angle? If that's the case, there will always be brief periods of error during fast throttle movement. The ECM uses a PID control, which is by nature reactive, to control the throttle. If there is 0 error, there is 0 movement. Error is what drives the movement. No you obviously want to close the gap as quickly as possible without setting up oscillations.

When I look at the Angle Error data from a long drive, it does not appear to be focused near the average cell values at all. I really wish VCM Scanner gave us some basic statistical analysis tools. Say I have an average error of 3* in one cell, the max might be 14 and the min -17. How are you going to make any progress if each cell shows a huge range of reported values?

Looking at a fast lift off the pedal, where I see an ETC error of -15*, the throttle moves from 81* to 10* in 0.2 seconds. That's 355* a second, while having to accelerate from a stop and decelerate to a stop, plus any filtering/smoothing. Not bad.

[Devildog1325]

From my experience yes you will always have some throttle angle error on abrupt throttle movements. If you feel like it is transient or injector related then go down that path but looks like according to the screenshots you have posted you are having some oscillation accelerating.
https://forum.hptuners.com/showthrea...with-VMP-gen-1
Here is a link that murfie helped me out with transient fueling but ended up not being the problem....it helped but didn't fix it, best of luck!

[CCS86]

Thanks for the link Devildog.

It is interesting that in generating those transient correction maps, murfie recommends filtering based on RPM slope. I can't really agree with that. When you shut the throttle quickly in gear, the RPMs do not change quickly at all. I think filtering based on either pedal position slope, or throttle angle slope.

Looking in my logs, the other thing I see, is a slight time delay between the throttle action and the resulting transient spike. This "lag" will make it hard to get the correction data mapped to correct cell. I think it might actually be more accurate to zoom in on a specific instance of tip in/out to find the RPM and MAP values occurring, then look at the fueling dip that follows to find the magnitude of correction needed.

I'm still unclear about the units of these gain tables.

What makes you say that I have oscillation during acceleration? Can you point that out?

[murfie]

Thanks for the link Devildog.

It is interesting that in generating those transient correction maps, murfie recommends filtering based on RPM slope. I can't really agree with that. When you shut the throttle quickly in gear, the RPMs do not change quickly at all. I think filtering based on either pedal position slope, or throttle angle slope.

Looking in my logs, the other thing I see, is a slight time delay between the throttle action and the resulting transient spike. This "lag" will make it hard to get the correction data mapped to correct cell. I think it might actually be more accurate to zoom in on a specific instance of tip in/out to find the RPM and MAP values occurring, then look at the fueling dip that follows to find the magnitude of correction needed.

I'm still unclear about the units of these gain tables.

What makes you say that I have oscillation during acceleration? Can you point that out?

The point of the filing by RPM is to capture the accel/decel event. If you filter by throttle angle you will not capture much data and where fuel is incorrect (after throttle open or close) will be filtered out.

You are capturing air with MAP, you are correcting the fuel errors the O2s are to slow to correct. The RPMs are changing to quickly for the feed back to be accurate.

[CCS86]

The point of the filing by RPM is to capture the accel/decel event. If you filter by throttle angle you will not capture much data and where fuel is incorrect (after throttle open or close) will be filtered out.

You are capturing air with MAP, you are correcting the fuel errors the O2s are to slow to correct. The RPMs are changing to quickly for the feed back to be accurate.

Maybe you are trying to capture different data than I am. Here is a snapshot of the transient decel dips I am dealing with. I go from stead throttle cruise, jab the pedal, then return to the cruise position. There is a small rise in RPMs during the tip-in, but almost no fall in RPM during the "decel" period (really tip-out) where my fueling swings rich.

6b
6b-fuel dips.png

After changing my decel gain values in the 3000 rpm row, between 6 and 12 in hg, by +0.031, here was my next log:

6c
6c-fuel dips.png

Pretty big improvement! But other stuff seemed to get worse. I tried a number of other revisions where I cut the time constants all in half to see the affect. I eventually zeroed the whole gain table for accel and decel, just to see the behavior change. The tip-out dips returned, but interestingly, the car seems to have globally leaned out.

I would really love to demystify these tables, as they are clearly very important.


Stock transient tables:
Stock-transient.png


Transient gain tables zeroed:
Transient gain zeroed.png

[murfie]

Not sure you got why I applied the filter. Accel and decel describe which way the RPMs are going. So if you filter a positive slope( greater than 0) for RPM you are seeing fuel error in the Accel table. By a negative slope(less than 0) you are in the decel table.
That filter just makes it so you apply the correct error correction to the correct table.

Making a filter based on time constant table after throttle tip in or tip out would be beneficial to narrow in on the error even more, If you can find an easy way to apply an entire table in the filters please share. For now you just need to assume your air model( MAF, SD, TQ/load) is accurate and the only major error left is transient. Once everything is +/- 5% with nice rich/lean/rich oscillation of the wide band signal you are probably fine. Time constants may vary if you have a different size manifold that moves the injector locations drastically, or if you have something like meth injection with an injector further away from the cylinder, Maybe even if the manifold is made of a different material that heat soaks more or less, you may need to increase the time constant. Other wise you probably don't need to change it.

With a cold engine the fuel is less likely to evaporate and pool more so it needs its extra time constant and gain for a period of time after startup. I believe these get added to the normal gain and time constant based on the ECT and engine run time. While logging make sure you are not getting values added in based by watching your engine run time and ECTs.

Basically for now, as you already figured out, zooming in on the error and viewing data range only is about as good as you are going to get with filtering by the time constant.

[CCS86]

Not sure you got why I applied the filter. Accel and decel describe which way the RPMs are going. So if you filter a positive slope( greater than 0) for RPM you are seeing fuel error in the Accel table. By a negative slope(less than 0) you are in the decel table.
That filter just makes it so you apply the correct error correction to the correct table.

Making a filter based on time constant table after throttle tip in or tip out would be beneficial to narrow in on the error even more, If you can find an easy way to apply an entire table in the filters please share. For now you just need to assume your air model( MAF, SD, TQ/load) is accurate and the only major error left is transient. Once everything is +/- 5% with nice rich/lean/rich oscillation of the wide band signal you are probably fine. Time constants may vary if you have a different size manifold that moves the injector locations drastically, or if you have something like meth injection with an injector further away from the cylinder, Maybe even if the manifold is made of a different material that heat soaks more or less, you may need to increase the time constant. Other wise you probably don't need to change it.

With a cold engine the fuel is less likely to evaporate and pool more so it needs its extra time constant and gain for a period of time after startup. I believe these get added to the normal gain and time constant based on the ECT and engine run time. While logging make sure you are not getting values added in based by watching your engine run time and ECTs.

Basically for now, as you already figured out, zooming in on the error and viewing data range only is about as good as you are going to get with filtering by the time constant.

I do see why you wanted to use the RPM filter, but just don't think the approach will work for the conditions I am looking at. On the little throttle blips I posted above (probably in 3rd gear). From the time I start backing off the pedal, the RPMs actually climb another 200 RPM before peaking. Then during the 1.5s of transient richness, RPMs only fall 100 rpm. With a decel slop that gradual, I just don't think it is possible to isolate that data from within a long log, with that filter. I think you will get other data in there, which will muddy your corrections.

Trying to filter for pedal or ETC angle slope instead will get you a lot closer. But still, the transient fueling reaction is somewhat delayed from the initiation of decel/tip-out. That time decoupling makes the whole concept of using a filter less useful (without a tool specifically for delaying your data sample). This isn't a huge deal, as it's easy for me to find these areas in the log. I can look at a few, and note the relevant stuff (inferred MAP, rpm, fueling error), making sure the car is fully warm, and not too soon after start.

In my case, I have a Department of Boost manifold. I don't think the injectors are moved significantly, but the manifold material is aluminum and the dynamic of a PD blower will both change some things. Plus, the MU52 injectors on a 4 bar system are kind of a wildcard.

What steps do you take to verify your air model? The remote tune I started with already had a modified MAF curve. I decided to return first to the stock 2012 GT500 MAF curve, since I am much more confident in that data, than I am the injector data. By tuning the injector settings (not always best practice), I have gotten stead state fueling quite good without touching the MAF curve yet. I did actually find that there was a pretty big IAT dynamic going on. IATs will climb to 140* during a length of idling. I built a table to isolate this and found that at low flow, I needed 10%-13% corrections at 140*.

I was hoping that the Coyote Cookbook would have good info on tuning transient fuel, but there is essentially nothing.

[CCS86]

I think it is possible that by tuning my injector BP and low flow setting, some of this tip-out dip can be recovered. My fuel load falls below the BP during these tip-outs, but even idle tends to be above it.

To gather better data on that, I really need to stay in closed loop during decel and prevent DFCO. I have been unsuccessful so far, and think DFCO is skewing my data.

[CCS86]

I really wish there were PIDs to report current transient fuel comps. By knowing exactly the magnitude, shape, and time of those compensations, you could pretty easily tell how significant they are to any issue you are having.

I asked support if they exist and could be added, but haven't heard back.