Yes those are the parameters some factory tunes have these filled in with numbers and some don't. I wonder if it would be better to tune the controller with the key on engine off or while driving that way there is resistance from airflow?Originally Posted by CCS86
Like I described, the only reason there is "spread" in my log, is because I have forced a 0.7* minimum opening. Otherwise, desired and actual track very closely.
If you want to understand the changes I made to Effective Area and Predicted Angle, just read my posts in this thread. I described them pretty thoroughly and even posted an Excel tool that I built.
I'm not using "throttle %" in any of my work. I use Pedal % and convert to AD counts. Otherwise the throttle PIDs are desired/actual angle.
There is no correlation linking a "spread" between desired and actual angle, to ETC Angle Error. You can see it in the screenshot where my actual caps at the 0.7* minimum. There is a big divergence there, between desired and actual, yet the ETC Angle Error barely changes and nearly goes to 0.
I am definitely not wiping out the top side of the throttle body tables for my resolution in low openings. I don't see any real benefit to be had. I want my DD system to work nicely without forcing it via WOT start/end. I don't think the issue here with ETC Angle Errors, is a result of interpolation error. I think it is because we don't fully understand how ETC Angle Error is calculated.
Here's another interesting log snippet.
I have generally gotten a great return to idle characteristic using TQ based decel, and stock GT dashpot settings (which appear to try to disable dashpot in spite of it still showing up as the control mode). I have tuned the idle airflow table and idle angle table.
If I free rev the engine to low or high rpm, it will return smoothly to idle. The nagging issue, is when I am in gear, decelerating for a long time, rpms have dropped below 2000 rpm, then I kick in the clutch, it dips well below the idle set point. If I didn't have my minimum throttle opening set to 0.7*, it would stall. Dropping from above 2000 rpm, it works perfectly.
Idle-Dip.jpg
Both ETC TQ Request and Scheduled TQ show big growth when the clutch is pressed in, peaking at the minimum rpm of 600. But where does it think this extra torque is coming from? The ETC opening is very steady. In fact, it wants to close the throttle even more, despite being 200 rpm below the idle setpoint. My idle airflow table shows 0.872 lb/min for 800 rpm, and the angle table shows 1.06* opening for that MAF rate. So why is the desired ETC opening 0.59*?
Last edited by CCS86; 07-30-2018 at 07:31 AM.
One more, where I give the pedal a tiny blip to perturb it, Dashpot takes over causing oscillation and very different TQ behavior. Then finally, RPM Control kicks back in and smooths things out:
Idle-RPM-ctrl.jpg
I don't think we have all tables to adjust on the Copperhead pcm. If we do, I haven't seen anyone that knows how to adjust them correctly. I do know that on my 2014 GT500, the crank relearn does not work, & I can't pull codes with HPTuners, SCT pulls codes & resets crank no problem, so something is different for sure.
Ok let me try to show you again. You already know I can't explain, I'll try to show you how I do it, it will get confusing and you probably won't understand. I had a Roush today so I decided to use the throttle body model you posted on the first page to see how bad it was an if I could fix these errors on a roush vs fixing them on my N/A twin 67.
To start, Your 2d throttle body model has to appear the same when you compare the two. Excel is good for keeping the 2d model shape because you can interpolate the data 2 or 3 times in between the columns to find what you need.
It's usefull to have less interpolation where the most cell hits are at. If you keep the interpolation, then it needs to be on point.
When your Throttle Angle Erorr is positive, it means your ETC Area needs to increase with out increasing your Throttle Angle Predictive.
When your Throttle Angle Error is negative, your Predictive Throttle body is too needs to decrease without decreasing your ETC Area..
You will know when you have decreased/increase far enough and your columns need to change. It could be typing ".001" and hitting the add button a few times, it could be rearranging your columns. Just a few thousandths in your columns and cells makes a difference. Even if your error says 5.
Throttle body error is your ETC Area predicting the difference of your Throttle Predictive and ETC Area tables and how far they're off based on that. Throttle body desired is telling you what would have been a better value in the cell that specific vacuum and area.
If I were to check the tables a few times, 5.76 + 1.06 = 6.82. Finding 5.76 in my predictive table. Finding where .2377^2 in my area table is and comparing the axis. Then getting the true ETC area of 5.76 throttle angle, the ETC area for my throttle body desired 5.78, finding the true predictive angle of my ETC Area .2377. One of those is going to equal 6.82. Yes I do check those. I will look up my desired throttle angle, I will find where the ETC area that angle SHOULD be. Then using the ETC area I just found, I get the throttle angle of that.
If you have a Throttle Angle Error, it's the answer to part of your formula. You can use it simply as a sign to increase or decrease. If the error is large, increase by .1 to .5, if the error is small. Increase by hundredths or thousands.
throttleerror2-1.jpg
throttleerror3-1.jpg
I chose this angle to show you that throttle angle error will fix your throttle body. ETC Area x _______ = 1 degree of throttle error.
We know 5.78 - 5.76 is not 1.06. If your throttle angle and throttle desired are close like 5.76 and 5.78, it should be quite obvious you need to change your throttle area table
So in your experience throttle angle error has more to do with the difference in the throttle body model and not the torque tables?
They all correlate. A torque error happens when your throttle body is not at the correct ETC Area to allow the proper amount of air required to achieve engine brake torque.
Its possible to get rid of torque errors by increasing driver demand in the correct cells. If you add to driver demand and your requested torque does not increase. Your throttle body model needs adjusted.
That means that if you're getting torque errors at 1500 rpm 12% throttle. Check your Engine Brake Torque and your ETC Requested. If your EBT is more than your ETC Request, add to the correct cells in driver demand.
If you added to driver demand and see no increase in your ETC Request, then your throttle body model needs adjusted.
I'll post some pics and examples to show you what throttle angle error is. You can have throttle angle error and have 0 torque errors. But you will not have torque errors with out having throttle angle errors.
OP was having idle problems as most do with oval throttle bodies. I still suggest create a column with an axis value less than .32*. Your throttle body minimum is. 0625 and your throttle body needs a calibration for .0625* to .32*.
The same rule goes for idle fluctuations. You need to find that happy sweet spot in interpolation because if you request too much, driver demand will attempt to request the value you have by opening the throttle more. When it can not reach necessary torque. The throttle body closes. Throttle angle decreases as well as driver demand. Over and over and over again. That's 1 cause of idle fluctuating a few hundred hundred RPM.
Everything I have tested and logged disagrees with your observations ThatWhite5.0. I wonder why we are seeing such different results.
- Changing the driver demand table has no impact on IPC TQ Errors
I have heavily tuned my Driver Demand table. I started with 0 IPC TQ errors, and still have none. If I modify my TQ Inverse table however, I create IPC TQ Errors.
- Changing the effective area in relation to the predicted angle table does not seem to change my ETC Angle Errors.
Just this morning I tested this again. From the baseline Twin 65mm TB tables, I first multiplied the Effective Area table by 1.2 (a 20% increase), and logged. I then reverted that table and multiplied by 0.8, and logged. These two versions are 150% different from each other. The average ETC Angle Error at idle was 0.62* and 0.070* respectively.
It sure doesn't seem like ETC Angle Errors are a result of these tables being "out of sync".
I am still leaning towards there being a relationship between the Engine Torque tables, for a given RPM and load, and the airflow measured by the MAF, in being the source of ETC Angle Errors. But I have not tested this yet.
I reverted to the original TB tables and tried scaling the engine torque table for MP0, 1500 and less rpm, 0.3 load and below, to 75%. Result is the same. Idle ETC Angle Error is around 0.6 - 0.7*
https://forum.hptuners.com/showthrea...-angle-(SURGE)
I’m not sure this will help but maybe the info in this thread could be of use if you haven’t read it already, it sounds somewhat similar to the issue you’re trying to solve.
Thanks for the link!
He seems to have the same idea as my theory: ETC errors are causes by a mismatch of Engine Torque Tables and actual airflow.
However, in my test today, I made a 25% change to the torque table for idle conditions and saw essentially no change in ETC error.
What were the absolute values of your 25% change? Maybe the absolute difference was too minute to make a noticeable difference?
On a side thought, there’s an ETC Idle table, can’t remember the name off the top of my head, but maybe test your luck there if nothing else nets any positive results.
I have tuned both the ETC idle angle and idle airflow tables. Logged idle shows that the values match beautifully. Same with the TB tables. They point right to the values I log.
Here is the TQ table change. First I went -25%, then on this one I went +30%:
TQ-change.png
Here is a section of idle, first with the AC on, then with AC off. There is an interesting correlation between a math parameter I created (ETC TQ - Engine BK TQ) and ETC Angle Error. Both are net positive together, and spike up and down with each other.
TQ-Idle-Diff.png
Would it be possible to get a screen shot of your current Predicted Throttle Angle and Effective Area tables to look at with that last log you posted?
I see more of a correlation to your scheduled torque(airflow,stoich, MBT based). Especially when you look at how it settles after the spikes. Your math seems to settle to nearly the same difference. Sch TQ and angle error both settle relatively high or low to their average. The two constants in this are fuel and air flow, as at idle it is not using the MBT or borderline value spark is being controlled by torque request, or controlling torque to achieve an RPM. Disturbances to RPM are feed back into the spark PID to correct spark and maintain 800 RPM. load changes will increase and decrease airflow, spark will compensate. You want to watch spark as you make changes to your torque tables.
The idle airflow tables are the base airflow you want to see at idle. There is no actual ISC valve so the throttle is just opened or closed for base airflow, then RPM is controlled by spark.
Attachment 82105
TB-tables.png
I can see the correlation to Scheduled torque too. As far as the scheduled torque value goes, unless it is using "Calculated Engine Load" (~29%) and not "absolute load" (~16%), like I thought it was, the value of ~70 ft-lb doesn't make sense for my torque table.
Interestingly, looking at idle spark between tunes with the MP0 torque 30% increase, there is no change. So maybe the resulting spark has more to do with the idle spark and idle air tables? With AC off, my spark is averaging 10*, but the table for idle spark vs ECT says it should be 17*. Maybe there are some other spark modifiers getting stacked on, or maybe my idle air table is high and causing the lower spark? My short term idle adapt is hover around only 0.08 though, which suggests my idle air table is pretty good.
Idle-Spark.png
Your attachment didn't come through.
Last edited by CCS86; 07-31-2018 at 08:02 AM.
Reply With Quote