MAF Scaling - Open/Closed Loop Fueling info
 

MAF Scaling - Open/Closed Loop Fueling info
MAF Scaling - Open/Closed Loop Fueling info
MAF (mass Air Flow) Sensor Scaling - Open/closed loop Fuelling

Disclaimer : I am not a tuner. I am simply trying to provide information and links to help you. Use this information at your own risk.


Hopefully this will be helpful to others who wish to have a go at scaling their MAF sensor.


If you find errors let me know and I will try and correct them.


MAF Sensor

The MAF sensor in mounted on the intake after the air filter. Its job is to measure the rate of flow into the engine.

If you change your intake ,you will need to rescale your MAF sensor or at least use the scaling provided by the Intake manufacturer. Changing just the filter will not require MAF rescale.

Indications you need to scale MAF are high Long Term Fuel Trims (LTFT) when running on Petrol (gasoline) say over 5%.

Large differences between Actual AFR and Commanded AFR in Open loop (Wide open throttle)

Warning: OFT tunes (and some other E85 tunes) on E85 do not use flex fuel sensor they use fuel trims to account for variations in E%. Do not try to scale your MAF while running an OFT E85 tune. If you want to scale MAF go back to Petrol (Gasoline) unless you have a very consistent mix of E85, most E85 varies in ethanol content from about 60-90%. If you do scale maf on an E85 tune be careful to check your fueling after the maf scale as it may require adjustment.

When your rescaling your MAF, keep in mind your trying to eliminate the LTFT compensations. If your LTFT is around -5% then your MAF will need adjusting by about the same amount. So if your calculations or the utilities want to make -20% change check your calculations or input data something is wrong.

Remember your changing the fuelling to your engine, make small changes and check the results.

Your probably aiming for LTFT around +/- 2% or so. Their will always be a bit of variance due fuel formulation (ethanol content E10) temperature , pressure ect.

Comments by Jamesm from HRI Tuning to contact HRI tuning contact @jamesm

Important if you have changed injectors or fueling components.
----------------------------------------------------------------------
Originally Posted by jamesm
Thanks for posting this. I'll just make two (very important) notes: first, what you're modifying here is an input to a load calculation, not a fueling calculation. Though it may not seem to matter for simple closed/open loop MAF scaling where the goal is mostly to minimize trims and get AFRs hitting targets, it's extremely important to understand conceptually. The MAF doesn't only dictate fueling, it is an input to a load calculation which dictates nearly everything the engine does in operation (like timing, port/di ratios etc). The results we see from this process with regard to fueling are only one side effect of this.

Secondly, the process as described here will work reasonably well for a vehicle with stock fueling components. To properly calibrate a vehicle with aftermarket fueling components requires process separation between port and direct injection modes. This is something that catches a lot of people off guard, particularly those who are used to tuning WRXs and think an 86 will accept the same process. There are a number of ways to accomplish proper calibration of a vehicle with aftermarket injectors, one of which is to calibrate the MAF against a fueling error log generated while running only direct injection (which in the common case has not changed), then calibrate the injectors against that known-good MAF (using the scaler and latency). Ultimately, the goal is to have extremely low fueling error on both sides of the coin. If not all sorts of awfulness ensues, like idle dipping, stalling when coming to a stop etc. I actually apply this process even when tuning cars with stock fuel systems, because I've found the stock calibration to be imbalanced by as much as 10-15%, and at that point there is a drivability win associated with correcting it.

--------------------------------------------------------------------------------------



Closed loop = When fuelling is calculated from inputs from MAF and O2 sensor (generally idle and cruise)
Open loop = Fuelling calculated from MAF sensor and OL Fuelling table targets (Generally Wide open throttle or quick throttle movements)


OFT or Tactrix logs – Fuel system status indications


1= Warmup not up to temp
2=Closed Loop,
4=Open Loop or de-celeration fuel cut
8=Open Loop due to system failure



Simplified explanation Open loop/Closed loop fueling

Usually open loop is used on initial start and during warmup, until the 02 sensor reaches its operating temperature and produces a reliable signal , at which point the ECU switches to closed loop for idle. Warmup mode is Fuel System status =1

Under heavy acceleration or abrupt throttle movement the system will generally go open loop. Fuel system status =4

During cruise and idle it will generally be in Closed Loop mode Status = 2

When the engine is running in closed loop, it is using the signal from the O2 sensor to regulate the AFR in conjunction with the MAF sensor. There is a feedback "loop" between the O2 sensor and ECU so the ECU continuously makes adjustments based on the O2 sensor data.

The correction applied by the ECU due to the O2 sensor input will be in the form of STFT, if the STFT stays negative or positive consistently for a period of time it will be written to LTFT. There are several bands of LTFT based on MAF sensor flow rates (roughly related to engine RPM) where the ECU stores the LTFT. This is a continuous learning process and other offsets and compensations are applied due to IAT, Pressure , coolant temp ect.

In open loop the O2 sensor isn't used and the ECU sets fueling based on stored maps(OL fueling AFR target maps) the MAF sensor input and LTFT trims learned when engine was running in Closed loop mode.

So if you MAF scaling is out of wack you will build up large LTFT when on Closed loop mode these will be stored and then applied as corrections when you go Wide open throttle.(Open loop mode).

This would be fine if the MAF is “off” by a consistent amount throughout the air flow range.

However if the MAF scaling is out by a considerable amount you will build up large say over 5% LTFT while your cruising around in Closed loop.

Then when you go Wide Open Throttle (or just more aggressive throttle movements) this stored LTFT is applied as a correction to the fuelling targets in Open loop mode and throws off your AFR.

Proceedure

Collect data logs logging required parameters required by utilities below generally includes

Collect logs at similar temperatures or it will throw off the calculations. A few degrees does not matter but if you collect one log at 10 C and another at 30C its going to cause problems with calculation. Best temperature is around 20-25C as but that's not always possible as this is whre no temp compensation is applied to MAF readings.

When MAF scaling, ensure you remove any data when the coolant temp is below 70C (160F) as fueling compensations are active and will skew data. If you don't log coolant temp, then don't log until at least 5 minutes of actual driving. Hint: Log coolant temp
(thanks @Kodename47)

Fuel system status ol/cl
AFR (actual measured by O2 sensor
Commanded AFR ( from OL Fueling table data)
LTFT
STFT
LOAD
MAF Volts
Throttle position
IAT
RPM

Once you have logs use @vgi utility and it does all the work for you, just remove and unneeded header lines from log.
http://www.romraider.com/forum/viewt...p?f=32&t=10481

Scale using CLOSED LOOP data first.

Once CLOSED LOOP portion of MAF scaled move onto collecting and scaling for OPEN LOOP

Pre-Scaled Intakes

Some intakes such as Perrin come with base MAF scales to suit
see attached files at end of post below for

Perrin 2.75 intake
Perrin 3" big maf intake

Other scaling supplied by forum members


AFE Takeda Intake (Gready)

Very Informative Post here from @Wayno showing graphical representation of different MAF scales and differences between MAF sensor in brz/86/frz likely due to change in sensors since original cars

http://www.ft86club.com/forums/showp...&postcount=353




Overcoming stock MAF limit with direct replacement Forrester MAF sensor

http://www.ft86club.com/forums/showthread.php?t=106496


base scaling for replacement forrester/wrx maf sensor
We can use impreza (forester ) stock maf sensor to our car
without any modification.

part number is 22680AA380

same dimension, same harness
but much higher maf reading limit ( about 130%)
can use more boost without Speed density tuning


base scaling for replacement forrester/wrx maf sensor in standard maf housing
http://www.ft86club.com/forums/showp...64&postcount=5






MAP Scalings for Alternative MAP sensors for Boosted cars (Manifold Pressure Sensors)


Omni 2.5 Bar MAP

The OFT300 (Vortex supercharger) map uses the 2.5 BAR Omni and has CEL-P0068 disabled in the map
ok for 2.5 Bar OMNI MAP sensor ECUTEK scaling in BAR

Manifold Pressure Sensor multiplier in bar per volt. = 0.500
Manifold pressure sensor offset. Value is measured in bar. = 0.100
MAP Sensor voltage max = 5.0v
MAP sensor Voltage min = 0.0v

Omni 3 Bar MAP Sensor (Metric in Bar)
0.617 - MULTIPLIER
0.03 - OFFSET
scaling limits 0 - 4.75v

Omni 4 Bar MAP Sensor
Manifold Pressure Sensor Scaling - (imperial PSI)
Multiplier - 12.087
Offset - 0.168
Metric/Bar

bar/volt .833
offset (bar) .012

Manifold Pressure Sensor Limits - (CEL still comes on )
High Input CEL Above - 5.00v
Low Input CEL Below - 0.00v
Disable CEL P0103 and P0068






Links to MAF scaling utilities and information


Link to vgi’s MAF scaling program (very good easy to use)

http://www.romraider.com/forum/viewt...p?f=32&t=10481

The above utility by VGI is probably the most comprehensive and easiest to use. It will also do Closed loop and Open loop scaling in one go.

-------------------------------------------------------------------------------------------------------------
Couple of tips from @Kodename47 when using VGI's utility

Can I point out that to ensure you add the max/min correction % into the OL section. You will always get a big error shortly after the throttle opens and you do not want this to be used for corrections. The tool is set to 200% by default, what I'd do is load in the log like that, look at the data at the top of the columns and have a quick check over what seems a good limit to set. Clear the run data and load in the same log with the limit set to a sensible value, I usually shoot for 7.5-10%. That should help with the lump you get around 3v.
I also set OL min and CL max to 3v.
----------------------------------------------------------------------------------------------------------------
@Kodename47 spreadsheet to smooth and compare MAF curves to OEM maf curve
https://dl.dropboxusercontent.com/u/...omparison.xlsx

Kodename47's Tuning Tool Thread
http://www.ft86club.com/forums/showthread.php?t=75710

Info Closed loop MAF scaling:
http://forums.nasioc.com/forums/show....php?t=1427448

Info Open Loop MAF Scaling:
http://www.ft86club.com/forums/showp...&postcount=995


Jamesm’s screencast on Injector and MAF scaling

[ame]http://www.youtube.com/watch?v=NkqY9JVRA4M&feature=youtube[/ame]




"http://www.youtube.com/watch?v=NkqY9JVRA4M&feature=youtube"


YIKES Spread sheet for maf scaling
http://www.romraider.com/forum/viewt...hp?f=32&t=6987
In the Yikes CL MAF scaling spreadsheet
A/F correction #1 = STFT
A/F Learn #1 = LTFT
Futher instruction in the spreadsheet.




Fuel system status in logs
1= Warmup/ O2 not up to temp
2= Closed loop
4= Open loop
8= Fuel system fail/error

MAF= Mass air flow sensor
LTFT= long term fuel trim
STFT = short term fuel trim
O2 = oxygen sensor
IAT= Intake Air temperature

Might I add that vgi's scaling tool is designed to be able to scale open loop even if LTFTs are being applied so if you collect enough data you can do a complete OL and CL scale in one. Meaning you should be able to get the car running on point in one go without having to go and re-flash/reset the ECU.

what unit is LTFT measured in? %?

LTFT and STFT are in percent

This is great - I've been wanting to scale my MAF and this will cut down on the reading time for me. Thanks!

I've been thinking about writing a web app that would do what the spreadsheets do, just haven't had any time at all to do it.

One of the features would be the ability to load multiple logs and have it work with them all.

Once youve adjusted the MAF, previous logs become irrelevant, I don't see much need for multiple logs. Have you tried vgi's one? Its java, so quick and simple and does all you need it to. Just a 20-30 min log for CL and 2/3 WOT pulls in 1 log and that's good enough for a full MAF curve. The technique to the tool is:
Set CL max V to 3
Set OL min V to 3
Do the CL with no smoothing and update the MAF scale, then upload the new scale to the OL portion. With that adjusted, apply smoothing as necessary.

I meant multiple logs from before the scaling, and not needing just WOT or just part throttle. The app would look at throttle position and figure out what is WOT and what isn't and do all the work for you.

That way you could use a few logs that are fully mixed instead of needing one single non WOT log and multiple just WOT single pulls. I personally find it hard to get a good solid 30 minute log of mixed driving unless I go way out of my way to drive around just to get a log. Using multiple files I could grab 10 minutes here, 10 minutes there and have it all work out.

12.1 dangerous? really?

Read it in contex. That quote is referring to a FI vehicle (most likely a WRX or STi) with a top mount intercooler or front mount intercooler, not a BRZ with a NA motor.

You seem to be mistaken, fuel system status will let you know if you are in CL or OL.

Thanks for posting this. I'll just make two (very important) notes: first, what you're modifying here is an input to a load calculation, not a fueling calculation. Though it may not seem to matter for simple closed/open loop MAF scaling where the goal is mostly to minimize trims and get AFRs hitting targets, it's extremely important to understand conceptually. The MAF doesn't only dictate fueling, it is an input to a load calculation which dictates nearly everything the engine does in operation (like timing, port/di ratios etc). The results we see from this process with regard to fueling are only one side effect of this.

Secondly, the process as described here will work reasonably well for a vehicle with stock fueling components. To properly calibrate a vehicle with aftermarket fueling components requires process separation between port and direct injection modes. This is something that catches a lot of people off guard, particularly those who are used to tuning WRXs and think an 86 will accept the same process. There are a number of ways to accomplish proper calibration of a vehicle with aftermarket injectors, one of which is to calibrate the MAF against a fueling error log generated while running only direct injection (which in the common case has not changed), then calibrate the injectors against that known-good MAF (using the scaler and latency). Ultimately, the goal is to have extremely low fueling error on both sides of the coin. If not all sorts of awfulness ensues, like idle dipping, stalling when coming to a stop etc. I actually apply this process even when tuning cars with stock fuel systems, because I've found the stock calibration to be imbalanced by as much as 10-15%, and at that point there is a drivability win associated with correcting it.

Thanks James I will add your comments to original post.:thumbup:

I just used @vgi program for MAF scaling and it sure makes things easy after wrestling with "yikes" and doing manual calcs for OL scale. Would have saved me a ship load of time.

The good news was that my scale was almost identical to @vgi program.

But I learnt a lot in the process so its all good

Thanks @vgi and @jamesm :thumbup: :thanks:

*sigh*

To use the existing tools you need to divide logs into OL and CL and do different things with each. I haven't used the Java one, but from what I've read you still need to separate the logs.

My idea would take any logs, with a mix of OL and CL and do all the work for you. You could in theory have one log that is an hour long with half a dozen WOT pulls and a bunch of mixed driving, or 10 logs with 5 minutes of mixed driving each, plus WOT logs, etc.

To use raw logs like that now you need to manipulate them in excel (including the time stamp column).

I'm not saying the app I was thinking of making would fundamentally change the underlying logic, it would basically just do both OL and CL at the same time from one or more log. It could also be written to include validation on how "good" the logs are, etc.

Realistically the entire process of tuning a car (in terms of base calibration) is entirely deterministic and much better suited to a computer than a human being. I've worked on ruby scripts in the past that do exactly what you speak of, but have never had time to clean them up and test them extensively due to actually having to tune cars all the time :). One of these days i'll get around to it.

I see, maybe vgi can make it happen if he feels like it.

Sent from my Galaxy Nexus using Tapatalk

So is vgi application enough for maf scaling? Does it mean i could install a Perrin CAI 3" or any other intake and use the apps to scale the maf sensor?

For someone using OFT tunes is it a good idea to rescale the maf just to have a more accurate maf table specific to my own car?

(I'm asking these question because my car run slightly on the rich side at wot, low 11)

Yeah you could but perrin supply maf scales for their intakes and that would be a much better place to start from

http://www.ft86club.com/forums/showt...ghlight=perrin

Ummm OK.... I think you just had a go at me over nothing.
Your tools looks to be the best yet as I had scaled my MAF using other less polished tools, haven't actually tried because I have not been home.

Hold the train right there... I wasn't knocking your tool in any way, if you read an earlier post you'll even see that my issue is finding a time/place to get the needed logs in the right format.

The desire to build the app like I want it is to suit my personal laziness around manually manipulating logs, not because the existing tools won't work. If I can import half a dozen mixed logs and have it sort it all out without me needing to edit anything in excel it not only takes out some potential for human error, it also makes it simpler to test and use.

The *sigh* was because the post it quoted was taken out of context for what I wanted to do with my app, not a reflection on the current tools.

Before i knew of vgi's app I made my own to do closed-loop scaling from multiple OFT logs at once. It's console-only but if you're interested and know C# I could share the code. I was going to throw it up as a web app eventually -- maybe we can work together.

C# is what I do, porting to a web app would be trivial!

The java tool does not need to have different logs for OL and CL. I use the same one for both.

I'm not sure if it can add multiple logs, I'm not sure it would be hard to get it to do that though.

Adding my MAF rescale tool here too:
https://dl.dropboxusercontent.com/u/...20Rescale.xlsx

Who else here has found the temperature compensation for the fueling is off by a good amount. While logging with about 90-95 degrees IAT's vs. 70-75* IAT's I found that fuel trims can vary about 4-5% for any given MAF voltage. Anyone else notice this? Fuel trim go up in hotter temps but get trimmed back in lower temps, it seems.

Yes noticed similar mostly at idle where iat climhs in traffic fuel trim in lower band moves 4 to5 percent however need to look at if the temp comp is for sensor error at differnt temps or ita actually to compensate for the density change in the air.

the sensor is trying to measure mass of air colder air more dense so it most likely needs to add fuel so moves fuel trim positive.

not sure will have to think about it

I did look at it just haven't used it due to not been at home. I knew same log could be used which was what I was referring to with the fuel system status being the differentiation rather than throttle input but the other guy was so adamant so I said you can add it if you feel like it. Anyways, I will remind myself not to make any comments, thanks.

Apparently it is the other way around, maybe it is adding fuel for cooling in hot weather?

It looks like the computer makes too much compensation for temps in the base fueling causing the AFR to go rich in lower temps but lean in hotter temps. Once the AFR registers fueling adjustment is made via fuel trims.

Bleh, brain fade moment of not realising increasing fuel trim because it is running lean. Have you tried changing the temp compensations?

Not yet. I only discovered this recently, so I'm just going to keep this in mind while I continue to dial in the maf scale. I'd want the fuel trims to be a bit in the + side on hot days and a little in the - end in cooler night temps across the board aside from idle. Since idling IAT's tend to be high even when IAT's while the car is moving are on the cooler side. Once I have the MAF scale to where I like it I'll start to mess with the temp compensation. Just curious if anyone has done anything about it. Also if you didn't know that there was this variation in fuel trims due to temperature change you'd never be able to get your MAF dialed in and chasing your tail until you realize this.

The simple solution is to zero out the IAT compensation table while scaling the MAF. You should really do that to all compensation tables while calibrating for that very reason.

On ECUtek you fill the table with values of 1, RR is 0.

The IAT compensation isn't a simple hot/cold +/- as it reverses as air volume increases too. I would return the stock table when you've done calibration.

This is very important... And why it's crucial that you filter out high IATs when doing the calibration.

I havent touched the temp compensation table. Unless Vishnu Tuning changed it from stock values, I dont think it has been changed.

Well I've done a couple set of CL and OL logs, and input them into vgi's tool (thanks btw for that). and one thing I noticed is my CL corrected MAF scale has hardly any corrections, maybe two points that are barely off. But when I do OL and there are a lot of corrections, and the corrections are mainly in the middle of MAF scale curve where as the CL corrections are only at the beginning and the end. My question is should I be combining the cl/ol corrections together? Also does this sound normal?

Yes sounds normal.

might be worth just applying the CL corrections, then check the OL again after that. Yes sometimes you have to average them.

The CL corrections should be somewhere around what your LTFT are the OL corrections will be the difference between the AFR measured and AFR commanded (in OL fuel tables).

Make sure you do the logs at similar temperatures, wide temp variations will cause problems. If you could do them around 20-25 C this is ideal as that is where their is no temp compensations applied not always possible though.

Thanks, I have been using Airboy's spreadsheet to find my AFR error %, what are you using for that? And unfortunately I have been doing all my logging after work when its about 32C, so maybe I should redo everything in the morning. After I get the MAF scale down is it worth looking at injector scaling with stock fuel system?

Initially I used the "Yikes" spreadsheet for CL scaling and just calculated the OL manually.

When VGI utility came out re-did some more logs and just imported them into his utility (tactrix logs) with the OL fuel tables and MAF values , in less than 5 minutes it did what took hours and came up with same results.
now I'd just use vgi utility for both ol/cl its magic.

As long as you do logs at same temps it should be fine, just not some at 10c and others at 30c for the one calculation.

Can I point out that to ensure you add the max/min correction % into the OL section. You will always get a big error shortly after the throttle opens and you do not want this to be used for corrections. The tool is set to 200% by default, what I'd do is load in the log like that, look at the data at the top of the columns and have a quick check over what seems a good limit to set. Clear the run data and load in the same log with the limit set to a sensible value, I usually shoot for 7.5-10%. That should help with the lump you get around 3v.

I also set OL min and CL max to 3v.

Seeing as this is most relevant thread I thought I'd add this here. For those that have been wondering how to do the PI/DI scaling. This is a spreadsheet tool I've put together:
https://dl.dropboxusercontent.com/u/...libration.xlsx


If you have any form of MAF scaled maps, use that to create 2 identical maps. 1 with 100% PI and the other with 0% PI, although I leave the standard ratio above 5200. Go and log data on each as if you were doing CL scaling.


On my tool, input the MAF scale and use this for @vgi's tool as well. Then copy the corrected MAF scale (only the g/s with no smoothing) into my spreadsheet for both DI and PI results into the New PI/DI columns and voila....


As the MAF tool adjusts all values below the lowest recorded MAFv by the same correction factor, I find it's best to remove any duplicate error values in the low RPM range. Then use the % difference chart to work out where to set the max and min range as to avoid erroneous data having an impact on the injector scalar. You then can select which injector set to adjust if necessary. I would choose the range with the most positive trims/error, or whichever is the higher line on the graph. I would always look for the ideal correction to be at the higher voltage range as at lower voltage the PI system can have latency error as well. This will become obvious if the lines drift apart or get closer the nearer you get to 0v.


I've also included a couple of columns for you to paste the DI fuel rail pressure for any adjustments you may want to make later.


The last 2 tabs are for you to keep an eye on how much MAF variation your map has from the OEM map. This will allow you to see if there are other issues, like if you're bumping up the whole curve etc. Paste your MAFv range into the OEM adjusted column and it will calculate the correct g/s. Then paste your current scale into the current ROM columns.


Any questions then please ask.

I just went and took a nice hour long log to start scaling my MAF and then went to use Vgi's java app only to find I didn't log RPM. It might be worth noting that in the first post under procedure @steve99
It was stupid of me but I literally just logged what you had listed.

Also can someone confirm what does OFT call the commanded AFR? Is it Equivalence RC?