Xtreme ECU and 6.2L LS Modeled Fuel Issues

[loca5790]

Hello All, hope everyone is well.

 

I'm scratching my head here on this one.  I've gone through all the settings I can check and the car wants to idle at around 100% VE and it won't stay running above 3000 rpm without maxing the VE out.

Injectors are bosch 2200s.  I have flow data dynamic and static matching with them.  I used ID2000 data and compared the link (done at 4bar) vs their standard at 3 bar vs my data for my injectors and they are close, my dead times are little higher than ID which sure that doesn't help buy not enough to correct what I'm seeing.

Injectors have been taken out and reverified although they are brand new.

MAP has been verified it is piggybacked with my factory E67 ecu, it did take some slight scaling to get it to sync or map calibrate and match-up with the factory ecu.

Intake temps match factory ecu and ECT matches factory (these are there own stand alone thermistors for each ECU).

Fuel pressure was checked with a mechanical gauge and verified.

Please see attached tune.  Would love to hear feedback!!!  Trying not to ditch modeled VE.  Next thing is to put additional widebands in the exhaust to verify.  Albeit the car does smell rich at idle with a global stoich it has a decent cam with adequate overlap.  Traditionally on other systems I would shut off correction below 12-1500 rpm and tune it by sound.  Add fuel until rpm starts dropping and then remove fuel until rpm starts dropping to find my bounds and bias it towards the rich side.  Checking the wideband doing this same thing.... the values are logical.

 

G4+ Xtreme CTSv - Modeled VE Issues.pclr

[Adamw]

Ive had a quick look over the tune and nothing leaps out at me as obviously out of place.  What I would do initially though is change fuel mode to modelled (rather than multi-fuel) treat it just like it is running on a single fuel, as that then takes a whole lot of settings out of the equation which will make it easier to diagnose.

Can you give us a short log of it running?

How come the fuel tables in the attached map dont have 100%VE in the idle area like you say but it appears it was idling when you saved that map?

[loca5790]

Hey Adam,

 

Correct, the fuel map shown is my base map.  I did not save the map that it took to idle it right because it made 0 sense and I was trying to figure it out.

 

I'll grab an idle log and attach the tune associated with that log tonight.  About to pull it out of the garage.... or attempt to.

[loca5790]

G4+ Xtreme CTSv - Multi-FUel Off - Cold start 1 - Lambda 2 AND 1 off - Modeled Not GOod - Good Rev and Drive.pclrWideband half time.llg

 

So apologize if the log isn't good, firs time logging and still trying to figure it out.  Need to do some more reading.

 

Anyway.... also apologize my Lambda sensor crapped the bed.... so I installed my daytona wego IV sensor kit in the tail pipe.  It fell out on a rev so you will see that.... then I went for a short drive and it set a Fault Code and I kept trying to clear it and check wiring at end of the log and it apparently took a crap too.

The tune of the drive it also attached.  You can see it is a very high modeled percent and I turned multi-fuel off too as you suggested.

[loca5790]

Anyone take a look?

[Adamw]

A bit hard to tell as there is only one small part of the log where just  one lambda is working which is at idle where you have a inj PW of only 0.66ms, I suspect the injector is not very linear down there so is not a great indication of what will happen at higher PW's.  Even though you have a VE of 75% in that area , CLL is pulling 20% fuel out to meet the target lambda so that would suggest your VE isnt far off normal in that area.

[koracing]

The pulse width is pretty similar (inj actual pw around 1-1.5ms at idle) from a car I tuned with old ID2000s on pump gas on a 2.2L 4cyl to what you're getting.  The odd part is that you're having to use so much VE to get there modelled.  I would think either your injector flow is off, or possibly the engine displacement is off.  Most of the other model settings appear correct based on information given.  I've only really had this same issue for those 2 reasons in the past. 

One odd thing I did notice with your injectors was this: 

Based on the note to the right, however, I don't know if that would affect the injectors in Saturated mode.

Is it possible your injectors are the CNG 2200 bosch injectors that have issues flowing less over time when used with liquid fuels?

[loca5790]

Hey Ko and Adam thanks for the feedback!!!

That injection setup was an artifact I was messing with the table to see how it worked and saved it.  It was 8.0 and .01 when ran but they are high imp injectors.

The car is idling on traditional right now at 1.2ish ms.  Actually my traditional map after 3 stabs at it is within about 10% for all the operating ranges I've been through.

Displacement is 100% a 6.2.  This is the crate motor that was purchased.  https://www.chevrolet.com/performance-parts/crate-engines/lsx/lsx-376-b15

The injectors were from a good friend who buys batches of injectors and has an ASNU bench and matches/flows all of my injectors.

Direct data from the injectors that were brand new installed.  After I had this issue, I removed them and took them to another friend who has a flow bench who also verified this information.  I use these injectors on alot of cars.

To further elaborate on this issue now that I have the vehicle running on traditional and running well. I reverted back for try 2 today.  Attached is attempt from today.. NO log just watched live.

VE in idle region was 50ish percent compared to where it idled traditional map.  Airflow the same, tps the same and spark the same.  The car was so lean it would not run.  Tried throwing more in it and got nothing so I gave up and reverted back.  *worth a mention I checked after I warmed the car up on traditional.*

Interesting thing to note... I will investigate this more tomorrow, but ... I had this car running previously on the factory ecu.  I have since changed the cam to a slightly larger cam with 8 degrees less LSA and about 10 degrees more duration, new heads and moved the MAF to a 4 inch diameter pipe from a 5 inch.  The MAF is located in a long straight section like it was previously. I cannot get 6D length distance from a bend but it is a solid 10+ inches form a bend in the same location in the pipe.  I scaled my old MAF by 70% so a 30% reduction because there was a 36% reduction in cross-sectional area.  I use base MAF curves I build for each sensor type and scale like this and am always within 10% so I errored high here as I typically do especially with knowing the changes.  Let's say that the cam/heads changed idle by 10% which we can do and say I'm off by about 15%......... high but doable.   Where I am going with this is that my factory ecu was calculating .28 g/cyl.  If I was off 15% that puts the high error at .322 g/cyl.  The link was 15% rich and was calculating .336 g/cyl... if you remove the 15% here is the magic.... my factory ecu was spot on because it was .28 g/cyl.    Where I am going with this is that it took a whopping 82% VE to do this.  If we pull 15% off that 82% that puts us at 70% range VE at idle.  The factory ecu would be idling in the 50% range in the same condition which results in a roughly 40% error in the VE tables between the two.

Some background, my base VE table works in 95% of cars within about 10% on factory ecus.  I have used these same injectors with the same data, on the same fuel pressure and it idled in the 50% range VE. 

 

For reference I attached the traditional map as well.  Please note the bottom of the traditional map I fall in what I've seen as the non-linear range of the injector which typically ends up being about 1 ms or lower with it getting nasty and reallying showing it's head around .4 ms.  I compensated for it by fudging the numbers high. 

Like I said I will take a look at it again, but it is taking a significant more VE than expected and what I have seen required on other setups.   Is there some calculation that is being applied to try and scale or "normalize" it?

CTSv -Cold Start Fuel 1 - Ign Correct - Map Offset - Trad 3 - Transient - Ign1.pclr CTSV - Ign Correct - Map Offset - Modeled - Transient - Ign1.pclr

 

To be clear I will get a log tomorrow on the modelled again and get it idling and reving.  Didn't mess with a lot because I was trying to rule everything else out but since it is running well on traditional I think we are pretty stable.

[koracing]

Maybe just lie to the ecu lol.  I'm not sure what else to tell you. Something seems off in the VE calculation on the ecu and it will have to come from link to resolve if that's the case.  You can tell it the injectors are smaller or the enigne is larger and it should improve those values, but I don't see anything that shouts out as the primary issue.  This would definitely be an at your own risk where your results may vary.  It is possible to go through and do all of the flex fuel dual fuel/ignition and 3D startup tables in traditional mode, but ideally the modelled system should work best.

[loca5790]

Thanks KO,

 

I already did all th e flex fuel/ignition and 3D dual startup tables for flex in traditional.  I would like to move to modeled but I tried again today and you can see attached.  The issue is if I have to fake data to make it work I don't trust what is going on in the background that I cannot see so unfortunately I will not use it. 

 

The Modeled is the tune used for the log modeled.

 

Trad 5 was the drive today and it was within 10% up to 7 pounds of boost and 6k rpm.  Got a huge portion of the map varying RPM and Load.

Modeled LS3 Base.llg mike andy - Axis Correct - Ign Correct - Map Offset - Modeled Base Start LS3 - Transient - Ign1.pclr mike andy - Axis Correct - Cold STart Fuel 2 - Ign Correct - Map Offset - Trad 5 - Transient - Ign1.pclr

[Adamw]

I suspect your MAP sensor calibration is wrong.  It is obviously pretty close at atmosphere when the engine isnt running, but it doesnt look realistic at all when it is running under vacuum.  You only have 16kpa at 4000RPM and 25% throttle, and it goes down to 4kpa (near absolute zero) in moderate over-run which is near impossible.    

Most of the GM sensors cant even go below 20Kpa.  

Whats the part number for the sensor?

[loca5790]

Adam!! Astute observation, the BAP and the MAP would not match.  In order to calibrate them I had to offset my map curve by 6 kPa or about 1 PSI which matches exactly how much off it is when driving.  I had to scale my spark and fuel tables to account for this.

Here on the cold start on Traditional tune you can see my MAP is reading 98 kPa on factory ECU.  It is broke out to both ECU's btw.

 

On the link you can see it is about the same and I had to scale the MAP to get it to the same curve.

 

Here is started and idling at 126F ECT.  Both sensors are within 5-10F between Link and factory one is at back of engine and one is at front. (4 min and 6 seconds from start)

 

Here is the link at about 1 psi difference.... I'm assuming this is due to a slight grounding difference between the factory ecu and the link.... not sure the cause but I had to offset the tables.  Ill go into how I addressed that difference in my maps here in a second.

 

 

 

[loca5790]

It won't let me post another picture but at the lower bound of the sensor the link and the factory ecu drift further apart.  At the decel the link was at 4 Kpa but the factory ecu was at 16 kPa.  

 

During driving at max boost factory logged 146 kPa and link 140 kPa.  This lines up with the idle numbers and all other driving numbers.  Idle link is 52ish kPa and factory was around 60 kPa.  The 6-8 kPa delta holds true under all conditions other than the one you pointed out.  

 

In order to scale my MAP axis to account for this error when running my axis for load via MAP was shifted by 7 kPa across the board from my original tune to my new tunes once I identified this offset.

 

https://www.summitracing.com/parts/nal-12592525?seid=srese1&gclid=Cj0KCQiA8dH-BRD_ARIsAC24umYEJaykPE9MeNTgaAA0MQCHaB6SZCODECRjy2RWrfQqlXy97rvNvL8aAr9UEALw_wcB  is a link to the map sensor.

 

Factory calibration is -11.25 offset and 312.5 linear.  The way to calculate this is as follows: 312.5 is the 5V number and -11.25 is the axis cross point at 0.

Volts	Map
0	-11.25
0.5	21.125
1	53.5
1.5	85.875
2	118.25
2.5	150.625
3	183
3.5	215.375
4	247.75
4.5	280.125
5	312.5

 

This is the curve generated off that method.

If you reference my table you can see it is shifted on the volt axis.  I also have a couple kPa error in there because the factory curve actually started at -11.25, I will redo that table for austerity measures.

 

Old table

+

New Table - I still may have to scale this table.... will check...

 

This table was created by calculating y=mx+b then solving for the limitation of no decimal places allowed in axis divisions in the link.  The table was broken into even steps rounded to the nearest whole number and then solved for x because decimals places could be used.  9 was used because it was the first positive number in the table.

 

Anyway, the shifting of the VE map to correlate with the discrepancy in the map reading between factory and the link should have been acceptable for trial purposes, but this again depends on how the model is calculating VE and brings us back to how does it work.

 

I am investigating bringing my MAP in across CAN and using that as a scalar on my Fuel/Ign tables but would have to be ANV table.  Would love to see CAN variable be able to be assigned to hard coded axis such as TPS, IAT, CLT... I think you understand the desire.

 

Ill try again with the updated axis and rescaling all my fuel/ign tables.

 

 

 

This curve is to demonstrate the error between the tables.  The X represents my first table and the triangle is the new table created.  The triangle is the factory curve.

I have to be honest here, bit by Holley's genius once again.  I used their curve labeled as that map sensor part number in their software figuring how could one mess that up............ wellllllllllllllllllllllllllllllllllll it is wrong.  My mistake, thanks Adam... hopeful this fixes it.

 

[Adamw]

Note for a linear sensor you can just use the 2 point cal 4/5/6 instead of having to calculate every point.   For example something like this would cover your new values: