ETC Angle Error Doesn't Make Sense

[murfie]

Throttle angle equals an effective area. pressure difference across an effective area is equal to airflow. With a know entry pressure, a throttle angle will be a mass airflow. Throttle body model as simple as it can be explained.

Calculated MAP + ETC vaccum = Entry pressure. Entry pressure should be assumed as barometric pressure.

Yours doesn't add up. MAF values are driving the calculated MAP too low. Manipulate your low mass airflow values of your transfer, along with your low injector slope, as to not disturb your lambda ratio, but bring your calculated MAP value up.
Keep them both in inHg or PSI, its not rational to look at them in different units. looking for 14.7 PSI or 29.92inHg, unless you are at higher elevations.

Two pressure sensors and an accurate MAF transfer (not just one that fixed fuel trims), are needed to derive an accurate throttle body model. Outside of that you will play with it on every car you use it on, with no confidence in its accuracy.

[CCS86]

Throttle angle equals an effective area. pressure difference across an effective area is equal to airflow. With a know entry pressure, a throttle angle will be a mass airflow. Throttle body model as simple as it can be explained.

Calculated MAP + ETC vaccum = Entry pressure. Entry pressure should be assumed as barometric pressure.

Yours doesn't add up. MAF values are driving the calculated MAP too low. Manipulate your low mass airflow values of your transfer, along with your low injector slope, as to not disturb your lambda ratio, but bring your calculated MAP value up.
Keep them both in inHg or PSI, its not rational to look at them in different units. looking for 14.7 PSI or 29.92inHg, unless you are at higher elevations.

Two pressure sensors and an accurate MAF transfer (not just one that fixed fuel trims), are needed to derive an accurate throttle body model. Outside of that you will play with it on every car you use it on, with no confidence in its accuracy.

I agree that throttle angle is the biggest driver on effective area. But, if they were equal, why would the factory have tables that show effective area vary (for a given throttle angle), based on ETC Vac?

I agree that MAF accuracy is important, but this is not a hacked up MAF curve. I have used an actual rail pressure sensor to remap the inferred rail pressure tables. I have used an actual MAP sensor to rewrite all the speed density tables, so that calculated MAP matches actual. I am using FIC injectors with good calibration data. With all these fundamental set, an OEM GT500 MAF curve produces very nice fuel trims, no hacking needed.

Here is the rest of the channel data for these two snapshots. There is a slight difference in the MAF flow (less than 2 cells away on the MAF table). So, I can't see a way for MAF curve inaccuracy to be responsible for the giant change in ETC Vac. The IAT is one of the biggest differences, so I wonder if there is a hidden table describing some temp based changes.

etc-vac2.jpg

[murfie]

Its actually the opposite, a range of 2-3* throttle angles is narrowed to 0.1 in^2 effective area range. So that a given airflow can fall between throttle angles. Then you should have a pattern that the smaller angles increase the pressure difference, and larger ones decrease it. making it track diagonally across the table.

The calculated map + ETC vaccum = barometric pressure wont work when you are at the bottom limit of ETC vaccum axis. The model is meant to show at what throttle angles it transitions from low MAP to higher map.

Throttle body model.PNG

[CCS86]

Its actually the opposite, a range of 2-3* throttle angles is narrowed to 0.1 in^2 effective area range. So that a given airflow can fall between throttle angles. Then you should have a pattern that the smaller angles increase the pressure difference, and larger ones decrease it. making it track diagonally across the table.

The calculated map + ETC vaccum = barometric pressure wont work when you are at the bottom limit of ETC vaccum axis. The model is meant to show at what throttle angles it transitions from low MAP to higher map.

Throttle body model.PNG

I don't follow.

The ETC Vac in the on the left of 6.6psi is pegged to one side of the ETC vac range, but in spite of that, ETC Vac + Calc MAP = much closer to baro than it is in the right side, where ETC Vac is 3.3 psi (middle of its range). How do you explain that?

More importantly, how can you explain such a large variation in reported ETC Vac, with nearly matching engine operation parameters, same tune, same drive?

[CCS86]

I did another test today. I took a baseline log with the current tune on my personal car. My ETC error is near zero in most operating conditions, but I found that at around 75% pedal, I get a few degrees of reported error.

I modified the tune by multiplying all values in the Effective Area table by 0.5. This was the only change. Then I drove around in a similar way and compared the data.

Here is a side-by-side plot:

ETC EA x 0.5.jpg

I could definitely feel the change in throttle response and return to idle (it wasn't for the better). But, it was still totally driveable. I also noticed IPC TQ errors, where previously there were none.

But, it's clear to see that it had little to no impact on the ETC angle error. It also appears to have had no effect on the reported Effective Area.

ETC Vac is a bit different though, at least percentage-wise. I'm not sure if that result is significant or not. I'll look at a few more data points.

By far the biggest effect is on what I call 'MAF Ratio' and 'Load Ratio'. These are two maths I made that take [Actual MAF]/[Desired MAF] and [Actual Load]/[Desired Load]. Cutting all Effective Area values in half, universally raised the MAF Ratio and Load Ratio (roughly 30%).

[murfie]

It's feed forward model used to predict.

It's not feed back used to correct error.

All it does is correlate a throttle angle to a change in air mass, so that of more torque torque is demanded the correct load gets achieved.

[CCS86]

But, I'm not asking if it is a feedback used to correct error, I'm asking: how can we explain such a large variation in reported ETC Vac, with nearly matching engine operating parameters, same tune, same drive?

[murfie]

You keep talking about ratios of different things having to do with the controls demand->actual feed back loop. The throttle body model is not part of that, so I don't think you understand that it is a forward table, possibly you don't understand what that means. You are not going to correct it with a ratio of anything. It's not logging and tracking error that comes out of it.

When there is a large change from demand to actual, say 200ftlb, the ECU says I'm not waiting for the feedback loop to correct this. TB model how much do I move the throttle to achieve this 200ftlb, the TB model says 45*. The ECU moves it. What ever error comes of that is corrected in the feedback loop. That error does not make it back to the TB model to try again or even to correct the TB model in the future, at least it shouldn't if the model is even some what accurate. If it's way off it might trigger the ECU to query it's prediction again, but having the feed forward fighting the feedback is fairly obvious in how it drives.

Using Barometric pressure, ETC vac., And calculated map, with in the confines of the axis of the model, in an open loop approach is the only way to "tune" it if you are not going to just build an accurate model and fix the areas you actually have error.

Greater than 13-14 inHg vaccum, which is getting into very low MAP, you are in sonic flow. Air velocity is very high and changes drastically with small throttle movements, making predictive throttle movements inaccurate or unnecessary. So the model is not used at low MAP.

[CCS86]

The ratio of [actual MAF] / [desired airmass] is very useful.

Like you said, the TB model determines how much to open the blades, for a given torque request. If your TB model tells it to open the blades too much, this ratio goes above 1 and you are much more likely to get an IPC TQ error. If the TB model doesn't command the blades open far enough, this ratio drops below 1. It is a great guide for tuning the TB model.

What is this "feedback loop" you keep referring to? I see no evidence that there is any closed loop correction here (besides a PID controller for the actual angle of the blades). If your calibration is wrong, the engine will run in steady state with large ETC angle error, large deviation in [actual MAF] / [desired airmass], etc. IPC TQ errors will cause the blades to close, but that is a safety mechanism, not a proper feedback loop correcting for airflow inaccuracy.

Do you know how ETC Vac is calculated? You said that is is Baro - Calc MAP, but I have data showing that not to be the case. I also have data with matching RPM and throttle angle, with very different ETC Vac and this doesn't make sense to me.

[murfie]

https://patents.google.com/patent/US5526787A/en

ETC VAC comes out of comparator 26 B - P.

B = barometric pressure

P = Manifold pressure

This is line 28 that goes into table 50 stores values which indicate the desired throttle angle Θ given a particular effective throttle area. The TB model.


Patent shows how effective area is calculated as well.

(D*Sqrt(T))/K

D= desired MAF

T= air temperature

K= Barometirc pressure * 22.8789*2.6458*(sqrt(((MAP/baro)^1.4285)-((MAP/Baro)^1.7143))

ETC PR.PNG

ETC PR FIlter.PNG

Throttle Flow model equations.PNG

(2/y+1)^(y/y-1) = .5283

.5283= 15.807inHg/29.92inHg = 14.1inHg ETC vac. showing the model is for non choked flow. I mistakenly called it sonic flow.

The feedback loop is error from any of the measured values... sensor MAF> air load> engine brake torque(with applied AFR, spark, ECT, ATC), and any of the desired values... torque demand> load demand> MAF demand. with the goal of keeping demanded torque and engine brake torque the same. This gets ignored between WOT start and WOT end values, and throttle control is purely TB model no correction between actual MAF to desired MAF.

[CCS86]

I have used that ETC P Ratio table to very good effect on these boosted cars.

Let me be more specific: I am struggling with tuning a car right now (2014 GT, PD Blower). It's a bit bi-polar. Sometimes the throttle behaves perfectly, and other times it goes wild. The biggest driver for this behavior seems to be the reported ETC Vac value. After the Twin 69mm started misbehaving, we threw on a Ford twin 60mm and the issue remains. That seems to take the TB electronics/mechanics/and TPS out of the equation.

Can we agree that ETC vac should correlate well with air load? I believe that when air load is low, ETC Vac is high, and vice versa.

Here is some chart data showing traces for only ETC Vac and Air Load. My car is on the bottom; problem car on the top.

ETC Vac vs Load Comparison.jpg

My car tends to hit minimum ETC Vac (0.2 inHg) @ about 1.0 air load. This problem car seems to hit 0.2 inHg @ ~0.7 load.

But, look at what happens about halfway through this chart. The problem car refuses to get ETC Vac back up to max values where it should be. It can get down to 0.05 airload and still only read 10 inHg ETC Vac. Something is clearly wrong here, but I don't think it is calibration related.

Here's a section of log showing very steady pedal, and wild throttle behavior:

Surge.hpl

[murfie]

I hope you understand how to follow units or convert units from science class, or this may not make sense.

lb/min divide by revolutions per minute gets you lb / revolution.
lb/ revolution divided by 4 intakes/revolution gets you lb/ intake. 4 because half the cylinders perform an intake on a 4 stroke engine. if this was a V6 it would be 3. a single cylinder it would be 0.5.
This value is the "airmass" that both calculated MAP and air load are derived from. It requires 4-5 decimal places to see properly.


For air load:
"airmass"/ intake divided by 'standard airmass'/ intake. ECM 44205 - engine displacement is 'standard airmass'.

For calculated MAP:
"Quadratic" * (airmass ^2) + "slope" *(airmass) + "offset" = Calculated MAP.

In order to properly fix your speed density coefficients, you need to plot a bunch of points (X,Y) and use polynomial regression. "airmass" being X, and an accurate manifold absolute pressure sensor reading being Y. This will get you the correct coefficients to get the ECU the correct calculated MAP, from the MAF "airmass" it is seeing. In theory someone could use a calculated MAP from the ETC vaccum to fix SD coefficients.
MAF has a limit of 86.68lb/min, so using air load to get to airmass is needed. Its simply multiplying air load by "standard air mass."

If your calculated MAP + ETC vaccum is not coming out close to Barometric pressure, your speed density coefficients are wrong. I'm refereeing to MAP values between 15.8 inHg and 29.9 inHg. This is the part of the SD that needs to be fixed and lined up correctly, no pun intended. Sometimes it takes increasing or decreasing the entire MAF transfer and fuel injector slopes together, to get things closer to where they should be. Its all a matter of actual intake port VE that is happening. It shouldn't be well over 100%, but it shouldn't be low 80's either. aligning 1.0 load at peak torque 4500-5000RPM with 29.92 MAP is a best guess case scenario. WOT air load should have a dyno torque curve shape to it.

Are the cams phasing into an unmodified MP?

This surge log you posted, is a MAF sensor driven problem. The raw signal should not go from smooth to noisy, back to smooth like that. Fan wash? bad sensor?

[CCS86]

I have tuned every enabled mapped point's SD tables using a physical MAP sensor.

I don't agree that this surge is caused by a noisy MAF signal. Look closer, it is the erratic ETC that is causing that MAF variance:

Surge.jpg

Here are two points from the same log. The MP blend is very close to the same. The data on the right shows an airload of about 1/3 of that on the left, yet the reported ETC Vac is significantly less. This makes no sense to me:

ETC-Vac-mismatch.jpg

[CCS86]

I am having the owner scrutinize the bypass valve operation. If that thing is hanging up and occasionally not opening fully, I think that could be at the root of all these ETC issues.

[murfie]

You are experiencing an issue, where you are making it use the TB model in choked flow, an area that is know for its inaccuracy. What ever the cause is, the calibration being used is not setup correctly to help avoid such issues. You are probably on your own with this calibration strategy, and good luck to anyone else who may be trying to following it. Reminds me of the tuning schools Ecoboost tuning guide.

Logged vs calculated.PNG

[CCS86]

You are experiencing an issue, where you are making it use the TB model in choked flow, an area that is know for its inaccuracy. What ever the cause is, the calibration being used is not setup correctly to help avoid such issues. You are probably on your own with this calibration strategy, and good luck to anyone else who may be trying to following it. Reminds me of the tuning schools Ecoboost tuning guide.

Logged vs calculated.PNG

Can you explain what you mean by this? How am I "making" it use the TB model in "choked flow"? When you are cruising at 0.200 load or less, actual engine vacuum will always be high. The issue here is that the calculated ETC Vac doesn't match the actual engine condition. Have you done the calculations that show this is a choked flow condition?

I have tuned probably 20 different PD blower Coyotes using the same calibration strategy, and all have amazing throttle response.

Did you look at the second image in post #93?

[Grim5.0]

ok ive been holding off but need to ask or throw this out there..
running a stock TB, stock/OE maf curve, corrected fuel pump pressures with external sensor, corrected SD for a my18 mani and getting near perfect baro calculations when allowing the air model to learn = great running setup. swap in the gt350 TB and calculated baro goes south, really south in a hurry and IPC error limp mode abounds. If i lock the baro learn to actual (29.2-29.4 typically) everything stays as it was with the stock TB and it drives great, ipc under 2-3 ftlbs if ever and etc errors matching stock tb

Im not sure what to correct the SD model too. external map and calculated are almost always in line. even during transients. is this a possible multiplier issue? or something else?

[murfie]

Can you explain what you mean by this? How am I "making" it use the TB model in "choked flow"? When you are cruising at 0.200 load or less, actual engine vacuum will always be high. The issue here is that the calculated ETC Vac doesn't match the actual engine condition. Have you done the calculations that show this is a choked flow condition?

I have tuned probably 20 different PD blower Coyotes using the same calibration strategy, and all have amazing throttle response.

Did you look at the second image in post #93?

Yes I did see your image, same info as post 80 that prompted me to reply. I don't know how its being caused, but the problem is being driven by it. Baro-MAP= ETC vaccum. See "surge" vs a stock file attached, or any car that is going to drive normally. Why that car is not following that logic, I don't know. Flash back to stock, and reflash modified file, or use the stock strategy file and copy your changes over to it. pedal commander box. auto blip box. FMEM mode? Any number of reasons, but its probably a software thing.

ETC vacuum should sit at 13-14inHg, the bottom of the model, when in low MAP(choked flow), it should not be actively moving at 7-8inHg.

BAROMETRIC minus MAP.PNG

barominMAP.PNG

[CCS86]

ETC vacuum should sit at 13-14inHg, the bottom of the model, when in low MAP(choked flow), it should not be actively moving at 7-8inHg.

That's what I have been saying from the beginning and you kept vaguely disagreeing; claiming that I have done something wrong to cause this, without specifics.

I'm happy to entertain the idea that I am taking the wrong approach, but empirically, that doesn't seem to be the case. I drive my PD blown Coyote every day, have my WOT start/end at 595/599, and have very high standards for throttle control. You can't accidentally find that.

But, since you are convinced that your approach is superior to mine, I'd love to look at that data myself. Can you post up a driving log of a positive displacement blower Coyote that you have tuned?

[CCS86]

BTW, the customer found that the lower manifold was clipping the bypass butterfly of his VMP Gen3R. I am not a betting man, but would put a wager on this being the source of these issues.