-
4 Attachment(s)
Solved the high rpm hang issue with some help from layouts posted before. What the empirical data showed was that if the Torque value was too low, it would hang. If it was too high, it would hunt. The only thing that isn't solved for that is possible idle air adapt. If I make any move now it just flops from surging to hunting and then goes back to surging when I try to creep back the other way. Gonna stop on that as ok for now as its just a quick hunt and then idle but needs addressed. Curious what you think? also I've added the layout you crafted for me so the data will look the same. This is what I think the IPC error is indicating tho need a refresher on how to apply a fix. When the Scheduled torque exceeds the other numbers, thats when I see the IPC error start and then grow by exponents. Scheduled torque is DD, yes? Basically what I see the trend doing is flattening out that spike we put in to make up for the maf sitting in turbulent/negative void of the after cooler. I compared the log to the one right after you gave me this layout and it's dramatic. All that to say, the car is 100X more drivable. you'll see the car is doing 90 a couple places and it just didn't feel like it was working that hard. very easy to gain speed and then loose engine speed on throttle out.
-
The log gives me a network error when I try to download it.
For the idle you want to lower the air flow value in the RPM ranges it has oscillation. Raise if it is hanging above desired idle RPM. This effects the calculated Load and Calculated TQ value at idle so your DD(scheduled TQ) table values are closer to equal to the calculated values at low RPMs.
IPC error usually happens when DD(ETC or desired TQ) is too far above what calculated(engine brake TQ) is. Calculated is Tq calculated from MAF(at idle or no throttle you are seeing negative calculated values, idle air flow values should fix this). DD Torque should either pull this up or down based on what the pedal is doing. If DD values climb much faster or higher than calculated indicates you run into limits. you have to shape DD table around the values of calculated you see. How far off you want DD from calculated is basically driver preference for pedal feel.
IPC error is adjusted in the torque/ inverse tables as they are used for the plausibility check. These tables are a theoretical torque and load relationship that is not possible in the real world to achieve. This makes tuning them challenging. Each MP relates to a intake and exhaust cam position. So you are tuning for the different cam position combinations if the engine was running stoich fuel ratio with MBT spark, and under standard air and coolant temperatures. The IPC wheel torque error is not applied directly to any table for correction, but gives you a general idea of where values are wrong and if the values are off by a lot or a little bit.
-
2 Attachment(s)
Here's the log from the tune on the car right now. I've driven it several days in a row since this log and it's starting to settle into some of the old habits again. Hunting at idle and occasionally hitting a hard limit at 2500-3000rpm. The idle hunt I think is exactly what you described - lower the idle values until it evens out. The cars at the dealer for the air bag recall. Once it comes back the plan is to log this tune again and compare the two to see what's going on at 2500rpms. Thanks for walking through all the definitions again. I feel like the car made some big break through with the trq4 file - but it's still showing some room for improvement. Once I get that a bit more figured out I want to sort out some confusion on the VCT. I'm going to post a tune that I found on here (not in repository - it was in a thread). Does what he did with the intake open, exhaust close do what you were suggesting?
-
4 Attachment(s)
I downloaded the log from you previous post tonight. You may be having trouble tuning the idle as the DD table has cell 600 then 1000 and your target idle is 800 right in the middle of those two cells.
Looking at your return to idle while in neutral I see The two cells in the DD table are about double what they should be. The Throttle body model is used because of this and modifying DD table wouldn't help with out drastic changes. You may need to increase idle air flow. .4 Lb/min and 10%-14% load is low for a normal idle.
Attachment 67860
Besides seeing load fluctuate a lot (indicates a weak idle) between 10% and 14% while idling at .4 Lb/min I think in gear it settles down by increasing the idle air to .9 lb/min and 19% load. Load is a lot more stable at this idle air flow value.
Attachment 67861
Right now your desired RPM at idle is 800RPM. your idle air is .46 airflow. This makes the ECU target 1.5* blade angle. This makes the ECU work to close the throttle blade more than necessary at idle. It evens out once it determines it cant obtain that.
Attachment 67862
Heres what I suggest for modifications to the idle airflow tables.
Attachment 67863
For your 2500-3000 RPM issue I noticed it does not stay on the OP mapped point around those RPMs. I think you are not quite meeting the enable % load requirements under 3200RPM and are falling back to best drivability mode. It could also be the traction control system as the pedal and throttle are closing, but not consistently.
The file you found shows VCT changes for WOT and that is what I was talking about. I also think those changes in that file would help with the added FI.
-
1 Attachment(s)
I like where your thinking is going with both of these areas to tweak. So the order of updates is this: Move idle airflow to your suggestion and tweak as necessary. I've attached a screen shot of the table that drives load enable for OP - I reduced it by 5% so that 3000 rpms is .93 load which is closer to what the car is showing at WOT - do that second, I think. That may just move the issue to another rpm band but that would be progress. Finally - I think DD can go back to stock and the idle spark can go back to stock which would smooth everything out. Start tweaking the VCT similar to the comparison file.
-
1 Attachment(s)
Well, it did stop the hunting. it worked kinda like you thought it would. by telling it that .9lbs min stablized load at 18% but rpms hung at 1100. if you watch the throttle blip it drops to 800 as commanded but gives it up because load is 10% and kicks the car back up to 1100rpms to satisfy the .9lbs/min which is stable. So this says to me that the torque is too low in that region of torque/inverse causing it to miscalculate load vs MAf. 1st tho, going to drop the air flow to the middle of the values and see if idle RPMs drop in response.
-
2 Attachment(s)
This is another solid improvement in driving feel. I'd say that the idle is still a little weird but it doesn't hunt around and if the cooling fan kicks on it doesn't flip out. I didn't get to try the a/c or work on the VCT. I'll do those after this next round of storms passes. What do you think? I feel like I got a good grasp on what the logs point to for corrections now. This is the last log of 4 tonight. DD to stock. idle was tweaked down a bit. torque was raised ever so slightly. OP load requirement dropped by 5%. No limit triggers found trying all the accel modes that usually trigger it. I'll drive on this tune for a couple days and see what the adapt algorithms do with it.
-
The idle is generally smooth & follows the inputs dialed in. Its not 100% for all occassions tho. It can hunt a bit on some cold starts & will surge if driven on before it settles sometimes. Once warm its rock solid but the spark is rolling ever so slightly on the ngauge. I think that goes back to calculated load being in the 11% range and the pcm trying to raise it by adaptive reasoning. I'll get a log of this tomorrow but I think you'll agree that the solution is to get load closer to 16% with air flow at 800rpms as .46lbs/min.
The OP load limit actually caused a different outcome. It moved the limit point to 4th gear instead of 3rd. All I feel in third is the shift to OP.. doesnt bounce in & out. This is an improvement as its a lot more load consistent in 4th. It took trying to accelerate moderately hard at 30 in 4th. Not something im in habit of doing but a minor tweak will banish that.
-
I can see the improvement in log at the return to idle.
Id say make your MAF period at 450u .57-.58 and 475u to .51-.52 area instead of the .48 to .61 jump in the curve and your idle should improve even more at the cold start in OL. its about 7% to 15% lean right now because of the hole in the MAF curve. This will even help the incorrect load reading with out having to mess with the TQ/load tables.
The problem with the stock DD table is you engine brake torque( calculated of MAF values) is higher than your indicated torque. Indicated torque is spouse to be MBT timing, Stoich AFR, and standard iat and ECT all with the same amount of air. As it is not possible to reach MBT before knock in most WOT cases the indicated torque should represent the theoretical maximum torque the engine could be making. So seeing engine brake TQ higher than indicated means power is being limited by the ECU usually through spark. DD table should represent the actual torque the engine makes so this table should be quite higher than stock in the WOT range.
If you compare engine brake torque and load you can see they are directly related to each other as both are calculated from the MAF value. you can see how they are related to MAF by looking at RPM and MAF. If RPM is climbing and MAF is not load falls. If RPM and MAF are raising at the same rate Load is steady. If RPM is raising slower than MAF load is raising. You can filter the log and tune the DD table based on load. tuning the DD table for a smoother load will help you get the most from boost and better drivability.
-
2 Attachment(s)
This is a bit longer log as I was running errands. Able to do a lot of sweeps on a torque/inverse that worked up from older logs and then refined. The drivability is incredible right up til the load limit switch into OP is barely crossed. then it bucks surges bucks surges. I moved the load limit down to see if it got worse. which it did. So do I raise the load limit for OP instead of lower it. Why does this torque/inverse - which is pulled from indicated torque - buck worse than when I just smothered the 2500 region in too much torque as per previous tunes?
-
Look at your desired MAF and actual MAF. The DD table needs to describe the torque spike where boost comes in or you will have an area where the engine is getting boost but the DD is not keeping up with it and you get bucking and surging. this is usually right around the .9 load region where you have the transition to OP.
-
1 Attachment(s)
You jogged my memory and I went back to the part of the thread where this was first brought up. I made two histograms based on throttle percent. It wouldn't give me a good output for voltage so I did some rough math to break the histogram down somewhat like the dd table looks like. Clearly the car isn't going to like trying to suck in 3-5lb/min more than what is possible. So the third chart is the percentage off and that would be applied to DD table, yes?
-
You would need to filter the data by what the throttle was doing using its slope. If you were increasing throttle or above a certain point and desired was not greater than actual you need to increase DD in that area. If you were releasing or below a certain point and desired was not below actual you need to decrease DD at that point.
Your DD table does not have to be exact but it needs to be close for drivability. The little bit it's off can be used for the pedal feel, but go too much and you get hangs and surges.
-
I kinda follow what you're saying. If I applied the values above the DD is too radically different than stock. it should be mostly stock until the boost kicks in. So the values in the lower right quadrant would change but it's not the right change using the unfiltered results. So I'm trying to understand the slope filter. I want to put the accell as the variable and set it to slope. That part I get. I don't get what the operator/value should be after that. Is there a histogram I can look at? I'm a visual person so seeing it and then doing it work best.
While pondering this - I realized there's an easier way. the deviation between the three DD's I've set up are very minor across the majority. The latest version has very good upper left and if I go back to where the DD wasn't causing surging it's mostly only in the lower right - there's a bunch of shared values with stock. So I'll merge the two across the values they share that go lower left to upper right.
-
I would put it together and show you, but I'm not at my computer.
You need one chart for where values are too high and one were they are to low.
The slope is the rise over run. The time you choose is the length of the run. You can visually see the slope in the graph of the pid. If it's going up it's positive slope, down it's negative and flat is 0 slope. You can use slope greater than 0 and slope less than 0 to get only the data the pedal is increasing or decreasing. You would then add OR then the pedal position is greater than 70 or less than 20 as the slope will be zero at WOT or no pedal input.
-
I appreciate you describing it as best you can. I think this is what it would look like as written above. I don't know what value the slope should be so I put 100. With this setup the values come back about the same in the area I'm working on and it applying the percentage decreases the values below stock. That can't be right.
[73.156.slope(100)]>0 or [73.156]<20 or [73.156]>70
-
oh---this might be part the problem. torque for mp0 was off by one row and didn't match it's inverse. Let me take a drive with that fixed!
-
2 Attachment(s)
I'm frustrated for the night. I went back a couple tunes and updated the maf to the values that idle well. It still idles well but that didn't undo what ever is causing the limit. its back to be a hard limit in 2nd gear at 2500 rpms and progressively less so in each sequential gear. Wish I understood this better - it's not a surge, it's a hard limit at the OP transition. once you pedal past it, it's fine but any dithering at that point triggers the limit and it's brutal how quick it kills output then rekicks it back to OP. worse than fuel cut.
-
4 Attachment(s)
Decel filter
Attachment 68004
Accel filter
Attachment 68005
Pay attention to the units. In these pics I had it as number not percentage for the PID> 48 so it was not including that. You can also filter it more by adding AND [19077.73]<[50040.73] which is desired is less than actual or vise versa for decel. This helps separate the data to target trouble spots.
MAF % error
Attachment 68006
Throttle voltage to ADC conversion
Attachment 68007
The greater than 70 is just an example. The slope at 50ms or whatever seems to match a factor of the polling speed you are using.
Video of accel filter at Positive slope or above 48%. Viewing only zoomed data I saw the throttle flattens at 49% thats why I choose 48%.
https://youtu.be/yE9B1fHYznw
Video of same spot but through Decel filter. Any negative slope or below 20%.
https://youtu.be/quqfhDMesOk
2 seconds in to each video the throttle starts closing thats why Accel shows less data than decel. But from this you can see that applying those values would be unreasonable.
They show very similar data. Take note of the difference of actual and desired MAF. the engine brake torque starts to go above ETC torque. Indicated torque crashes because of this. This correction would be applied to the DD table by paste special - % multiply or multiply by half if you are being cautious. Also as always after applying corrections make sure the table is smooth with no unreasonable transition between cells.
-
3 Attachment(s)
I pored over this last log because something wasn't making sense. The DD by all rights looks impressively better compared to the one for op5. So I started running the log at very slow increments watching the sources. WHY WHY WHY is it going into OL power enrich and then the thing leans out. This looks like what if feels like in the car. Where is the trigger for this?
One clue I found. Watch ETC eff area. it goes off the chart when OL power enrich is triggered. or vice versa. if etc doesn't go off chart the car stays in OL and the values stay smooth. So when I say the car is hitting a limit - this is what it looks like. the throttle blade control essentially disappears and the car dumps fuel til the etc returns below some threshold. Looking at prior logs it was there, just not as bad as when I started dialing in the Indicated torque.