05 Centri 3v Log Review & WOT Surging

[ZOOMEEP]

@ B E N
I'll bump up the load with failed MAF table. I have Autolite HT0's gapped to .032" I believe. I also have Accel coilpacks. I've been told the Accels can be problematic and I should get OEM coilpacks.

How do I get my lambda values below 1? I adjusted my base fuel table and blended it as you suggested earlier but my lambda hasn't changed. Does it only change during open loop? Should I adjust the "OL Fuel TPS Thresh"? It currently above 40% for any RPM under 5000.

I am writing the tune iteratively but I'm getting very frustrated. R26 means "revision 26" and I keep a detailed notepad log of all the changes of every revision. I also save every revision. I probably have 20 revisions with LTFT and DFCO turned off and I'm just trying to dial in the MAF transfer. I've done STFT, wideband and now LTFT+STFT adjustments and it doesn't seem to be getting any closer. I'm thinking either I have bad data, my workflow is off, or something is wrong in mt maths somewhere. I've likely done around 100 pulls and I'm STILL screwing with MAF transfer. I'm balancing learning this, with work and family so I expected it to take a bit, but I've been blown for 2 years and I've driven the car maybe 1000 miles total trying to datalog and tune it. I've also done a remote tune with Lasota, and visited 2 local tuners. No one got it running right, and no one had suggestions for how to fix it. It's running better than ever with the HPX5000 MAF and larger BOV, which were design changes I made myself At this point I feel it's my own inexperience holding it back from running great.

What are the dozen parameters you generally focus on? I'll narrow my focus to them.

@Robcat
I reviewed your R26 tune and adjusted my torque table. The IPC torque error was better in some spots. No worries on the long novels, I'll read and digest them. Plus I can be long worded too. I appreciate sharing the knowledge and info.

[B E N]

You will go through more revisions in learning, at some point the changes you need to make click and it goes quickly.

I keep meaning to do a write up on this but "life gets in the way". It would be good for me to at least post some notes though, so I don't forget.

Hopefully some of the 3v gurus will chime in, but this is what has worked for me.

Start with a stock tune for your car. This list is a complete summary of what I would change with the knowledge I currently have.

Don't mess with the throttle body table unless your changing the throttle body in some way.
You need good injector data for whatever injectors you are running, this can really screw you up.
You need to make sure your fuel system is doing what it is supposed to (log fuel pressure through a run, make sure it is rising as appropriate)
If you are deleting IMRC you need to make all those changes off the bat (write up elsewhere on this site) and ensure you are upkeeping those changes.

Maf: airflow vs voltage set the initial numbers using math, or a known chart for your particular maf housing, use stft+ltft to dial it in. This sort of becomes a moving target, so keep this histogram in your logs always, when you see numbers off by more than 5 paste special multiply the whole thing. This is your main fuel commander, so if you have cruise lambda that is high or low it's the first place to look

load with failed maf This is your tip in fuel command, its what tells the pcm how much fuel you need when you move the throttle, before the maf reading catches up. If you have lambda readings that arent doing what you want when you hit the throttle modify this (more air means more fuel)

manifold volume this governs how quickly the PCM will force a change based on a throttle input. Smaller numbers mean a faster change, too small isn't good, with a centri blower you might need a bigger number than stock, I am not sure, try both directions. I use 3, I might go up to 5 on an iteration and see how it cruises.

max throttle angle vs rpm you can just change this to 82 in all cells to get your tune going. Fiddle with it later if you feel the need.

fuel stoic make sure this is setup for whatever fuel you are running. Ford set it up for ethanol free, which you probably are not running.

fuel base already covered in this thread

rpm limits pick your poison on this, the 3v with a blower wants to make power past the factory redline, but the rods are shit... you pick!


There are a lot of tables that govern what spark does, make sure you are pulling spark when in boosted load. Being conservative here isn't bad. A stock 3v won't take detonation.

vct spark these tables need to be extended to wherever your max load is (I ran mine out to 2.0)

borderline knock same as above

mbt spark same deal as VCT spark, make sure you are pulling spark (mbt is maximum brake torque, so perfect world spark before the PCM starts pulling timing for other adders)

Dial these in using spark source and plotting a histogram. You can look at the min, max and average values to see which one the computer is referencing.

Torque modeling

engine torque/inverse plot ETC throttle angle error in degrees on engine speed vs air load. Use those numbers to adjust the torque, use the built in calculator to find the inverse.

ipc rom/enable disable both, set max wheel torque to 300000, you will eventually get your wheel torque errors down, but having the car going into limp mode while your trying to log is a pain in the ass.

scheduled torque max set this to 10% or so above your max torque from the torque/inverse.

driver demand my suggestion is you leave everything below ≈148 alone and only mess with the higher input stuff. This will keep the car easy to drive at low speeds.


That's it. My car with ported Whipple, headers, high flow cats, 62lbs injectors, 5" maf tube, on e85 drives around like stock, is within 3% of commanded fuel values at all times, gets good fuel mileage, passes Colorado (same as california) emissions flawlessly and melts tires at will.

If you have cams that is a different ball game. I cannot advise you on that part.

Keep it simple, until you find a deficiency you can't correct.

I went through approximately 250 iterations to get it this way. Maybe this guide will lessen your pain. I know there are 14 parameters there, but several are set and forget

[ZOOMEEP]

I'm 100% in the weeds on these torque tables.

I've tried RobCat's suggested edits. I've also tried RobCat's excel calculator based on an ST, which gave me torque values around 1500 lbft.

I've read though several threads and tried several other user's excel spreadsheets. Nothing works or makes sense.

I found the ETC and Inverse torque calculators in HPT, the "calculate inverse" button is greyed out so I can't even click it. Am I using it right?

[RobCat030]

That spreadsheet isn't exactly the most intuitive. I make them with my own understanding in mind but if I was an excel whizz I could probably make them work better. I derived the torque tables I provided you earlier using the same spreadsheet, so TQ values of 1500 indicate improper usage (again, my fault).

In order for the ETC calculator to show that button, you need to make the changes inside the calculator, not in the table in the software. A bit frustrating, I would say.

I'll say again (not with the intention of being argumentative or redundant, but because I think it's important to understand), that you're trying to throw a physics based error into one of these models to make the car happy. You can't correct the physics error itself without changing the code, that is just a fact. Both inverse calculations (torque and ETC) when using the HP calculators use curve fitting (at least I'm mostly sure about this). The space between the two curves is essentially the space for your physics error. For example, looking at the torque tables with arbitrary numbers (simplified):

1) Driver Demand requests 400TQ at a certain accelerator position and RPM

2) Inverse table translates to a desired load of 1.3 to generate the requested TQ (greater than barometric pressure)

3) Load is translated to a MAF flow rate, throttle opens to the predicted angle to generate the effective area that will produce said load

4) Physics error occurs due to centri blower generating a throttle inlet pressure higher than barometric pressure, ETC Pressure Ratio function fails, MAF is greater than requested

5) Load is calculated from measured MAF (higher) and sent to Torque Table for Independent Plausibility Check (IPC), torque table output is higher than ETC Requested torque, lets say 500 lbs.

6) IPC error is added and throttle closes to correct.

If your torque table output was 400 lbs like you requested in the Driver Demand (at a different load then you requested in the inverse table, because the blower forced a higher load than you asked for), then the IPC should be happy, and the throttle shouldn't close. Enough bending of the relationship between torque and torque inverse should accomplish this. You could also modify the Driver Demand table so that when the blower begins generating boost, the DD follows the torque/load output of the blower as to not generate an error. I haven't had luck with that method since it felt like I was essentially letting the blower drive the car instead of me.

This is how I see it, it's very possible that I'm dead wrong. Another method you might try is this: start with the torque table I sent you, don't touch it. Stock ETC tables. Follow your Driver Demand area where IPC error piles up and only lower the inverse in those areas, don't recalculate anything.

[ZOOMEEP]

No worries Rob, I figured the 1500ftlbs was my error. I definetly don't think you're being argumentative or redundant, I appreciate the help. I don't mean to sound argumentative either. I'm just frustrated at how long this is taking me to figure out.

That explanation helps me understand the relationship between the tables, but I'm lost on how to calculate, data log our even guess at corrections. Do you have equations or excel tables you commonly use?

@ Ben, I tried your method using your scanner histogram...

"Engine torque/inverse plot ETC throttle angle error in degrees on engine speed vs air load. Use those numbers to adjust the torque, use the built in calculator to find the inverse."

Is there some filtering or math built into that histogram? My data log only populated a few cells and didn't have many cell counts. I couldn't generate a log with enough data to work with to try this method.

[B E N]

Lets simplify further: pretend for a moment the spanish oak system is a vacuum secondary holley

LWFM is accelerator pump shot/squirter tuning
TQ/inverse is secondary jetting (sort of)
Fuel base is primary jetting.

My guess is you are crossing what LWFM and TQ inverse are supposed to be doing. Torque inverse can come directly from VCM scanner.

All driver demand does is tells the PCM where the throttle position wants to be. Its like your pedal to butterfly arm ratio in a carb. In the case of spanish oak You can leave it completely stock and the car will drive fine, once you hit full throttle driver demand is bypassed. You can tweak to preference after the rest of the tune is working well.

Filtering the inverse load calc:
[50091.156]>10 AND [50091.156.avg(300)]<20 AND [50091.156.slope(300)]>5 AND [50010.242]>140
I think you could go further for refinement if you wanted to, but this worked for me. Tune modification will be iterative, keep this histogram running, modify torque table with the data any time you see a significant change. I think the built in calc automatically updates inverse any time you mess with torque.


What I did for LWFM that worked fastest for me:
Plot EQ ratio from your wideband in lambda against engine speed and air load. Set the column and row values identical to ECM 44768 table.
Set up an excel spreadsheet with what you want to see: in this case drop in your desired fuel table, drop in an area for what your getting from the data logs, use excel math functions to quickly output the difference. Try this: LWFM calculator.xlsx Disclaimer: that was never meant to be a public document, it isn't set up to be pretty. You have to understand there is no direct correlation between base fuel table and load, base fuel is set up for throttle position. I don't care, all I want from LWFM is for my tip in fueling to give me the desired result, don't let the math flaws get in your head. Simply take the outputs from excel, paste special: multiply to your LWFM table. This works less well at low load inputs.

Don't try to adjust low load LWFM off of long term data logs, it just becomes a hornets nest. Do logging sessions dedicated to LWFM tuning. As soon as you get your MAF dialed in do LWFM next. Find a remote road with a generous speed limit where you can get on and off easy and there isn't going to be a lot of traffic. You don't need to do 120mph to get this stuff to work, second gear pulls are fine.

You can also unplug your MAF and tune LWFM this way, but it needs to be close to start.

Trying to math out the numbers for torque/inverse is fraught with stress and peril... It is fine to set up your initial table this way, after that It's much easier to just let VCM scanner tell you what you need to know. If you don't have access to a dyno to generate real numbers you are much better off setting it up by histogram, hptuners logging is pretty powerful, use it.

Simplify what you are doing. Get it running decently, save that tune somewhere it won't get lost, then start worrying about the minutia. If you screw it up you can just reload your basic functioning tune.

As your LWFM, and torque inverse get closer to true they will both change a bit, you are adding or subtracting fueling with both of these, they do not work independently of each other.

Don't make 18 changes to things at once. Change one table, drive, see how the changes actually effected your logs. That's the easiest way to understand what a parameter does.

[murfie]

I disagree with the TQ/ inverse part of the carb analogy.
The torque models primary purpose is to make sure everything is operating normal. Especially the throttle, and making sure its not stuck open or opening way moe than the driver is expecting. It does other things like compensate for friction with temperatue changes and dynamic accessory loads, but those are small. After the ECU has exhausted its other perffered methods of torque control it can modify torque using fuel control, but it would not be used for long before a full fuel cut happened. Not putting the engine at unsafe lean conditions for extended periods of time when setup corectly.
These are all things an old carb engine didn't control. You might be able to compare an idle screw to a very small part of the model, but thats about it.

Your ETC toque and driver demand torque will be the same
Your scheduled torque and indicated torque will be the same
You also have engine brake torque, this is an estimate of engine brake torque based on current conditions.

The important part of this is to get ETC torque/ Driver demand torque to agree with engine brake torque under steady state conditions. What this means is your accelerator pedal has an idea of what the engine conditions mean, and your throttle body has an idea of what the engine conditions mean. They need to agree, when they don't you get dead pedal or surge. When the ECU sees too much error between these it assumes the throttle is stuck open or opening more than it is requested to be, and you will get put in a limp home mode. The pedal position WOT start value is the point at which the error is used in a feedback loop to hold the demanded torque at the estimated engine torque and when they are allowed to diverge a little in openloop control. You have this set low, so you are seeing how far off your torque values are at different points. Its just a matter of correcting it in the model.

Throttle angle error is telling you how far off the throttle angle is. It is not telling you how far off the torque values are. Adjust the torque in the inverse calculator with the torque error between demand torque and engine brake torque. Reduce the error between demand and actual and it will drive much better.

Traction contol coming in and out can also be confused for WOT surging. You can set it to 0 if your traction control button(TCS) in the car doesn't seem to be turning it off.

[B E N]

Throttle angle error is telling you how far off the throttle angle is. It is not telling you how far off the torque values are. Adjust the torque in the inverse calculator with the torque error between demand torque and engine brake torque. Reduce the error between demand and actual and it will drive much better.

I think I was trying to say the same... I should have been more clear, the filters I pasted on here are in reference to the layout I posted earlier in the thread. Inverse load calc is ETC throttle angle error plotted against engine speed and air load. For whatever reason I thought that was self explanatory, and it was very vague.

You are right about the carb analogy, that was a stretch. Your explanation is much better.

[murfie]

Torque modeling

engine torque/inverse plot ETC throttle angle error in degrees on engine speed vs air load. Use those numbers to adjust the torque, use the built in calculator to find the inverse.

[

This is clear. I don't think using ETC throttle angle error in degrees to adjust torque values makes any sense. Does it work? maybe, seems to have worked for you. 1* throttle angle = 1 % error...

What I'm saying is adjust it using the percentage difference between two torque values, so that your model makes them(driver demand and engine brake torque) match each other at the same steady load or RPM. Use error between two torques to adjust a torque model. You will be taking your engine brake torque and making it match your driver demand, if that helps setup the user math.

The other way is make you driver demand table match what your engine brake torque PID is reporting.

[B E N]

This is clear. I don't think using ETC throttle angle error in degrees to adjust torque values makes any sense. Does it work? maybe, seems to have worked for you. 1* throttle angle = 1 % error...

I see, I struggled with that as well. It is really an approximation, 90 deg "max" sweep, and a non linear relationship don't really coincide with 0-100%. I think your way is more elegant/efficient would be easy to set up with a math function.

[ZOOMEEP]

I'm reading what you guys are saying, but I'm not fully understanding.

BEN,
In the equation...
[50091.156]>10 AND [50091.156.avg(300)]<20 AND [50091.156.slope(300)]>5 AND [50010.242]>140
What are "50091.156" and "50010.242"? I don't understand what this equation is calculating or doing. And I don't understand why when I log, I drive around for 20-30 minutes and this histogram might have 6 or 7 values in it total.

Torque inverse can come directly from VCM scanner? As in I should log air load against Torque (rows) and rpm (columns)? Isn't torque inverse just the air load required to make that torque?

For LWFM I have a histogram in VCM Scanner set up to log air load against throttle position and RPM. I should be able to copy paste that into editor correct? The table rows and columns in scanner match the table in editor.

Hi Murphie!
I read through your thread of torque tables and tried your method, I don't think it works for a boosted application correct?

Your ETC toque and driver demand torque will be the same (where is ETC torque? Is this a PID in scanner? Can I just log it then copy paste it into driver demand in editor?)
Your scheduled torque and indicated torque will be the same (indicated is engine torque in editor correct? where is scheduled torque?)
You also have engine brake torque, this is an estimate of engine brake torque based on current conditions. (my car has no AC and I converted it to EPAS. Do I need to update this to reflect having less accessory load?)

Also would any of you guys be willing to do a google meet and screen share? I feel like I'd learn a lot with a conversation vs many messages. I'll follow up here with lessons learned for other users to reference and I can kick you a Venmo for you time.

[B E N]

Filters are summarized here: https://www.hptuners.com/help/VCM-Scanner/filters.html

[50091.156]>10 AND [50091.156.avg(300)]<20 AND [50091.156.slope(300)]>5 AND [50010.242]>140
50091.156 is the address for the APPS
>10 you have to exceed 10 on the apps position (basically be on the throttle, so you aren't logging full decel and idle) this value should probably be higher, but it looks like its filtering a lot for you.
AND AVG 300 means it has to have been over that value for over 300MS
AND SLOPE 300 helps quiet the pedal movements you get in street driving.
AND[50010.242]>140 Means coolant temp has to be over 140.

For your torque inverse try murphy's method of graphing engine brake torque against engine speed and air load. Then those are just copy and paste values.

For LWFM what spanish oak is looking for is the air load that is happening before the maf can catch up to changes, it uses this table to understand what is going on without sensors. Higher air load means add more fuel. Use the LWFM calculator I posted above, re read that little paragraph and see if it helps.

I am happy to help at some point, we can get on google meet. Just have to find a time that works.

[B E N]

Did you read and attempt the steps above???

[matt5liter]

messaged you B E N