Stock MAF scaling issues and corrections.

Jpierro79

Senior Member
First Name
John
Joined
Mar 6, 2019
Threads
33
Messages
398
Reaction score
264
Location
Myrtle beach South Carolina
Car(s)
2018 type R 2015 c300
Country flag
By now many of you have noticed that the fk8 tuned with bolt ons can have large afr fluctuations at higher rpm and boost levels. I am not talking about the catalytic regen issue where the afr command goes 14.7 at wide open throttle. That regen issue is separate and only hondata can correct.

Data logging a lot of you will notice very large afr swings. I’ve seen swings larger than a full point leaner and richer. Example target at 12.5 and it might swing from 12.5 to 13.5 to 11.5 and back and forth. Sometimes even more. On a dyno it can cause wavy power graph.

This seems to occur at two voltage points in the MAF scaling. One at 4.56 volts which isn’t that bad and another at 4.77 and this is much worse. A lot of off the shelf maps hang in this area of airflow at higher rpms. It is not from the intake design and it is not from the Prl turbo inlet as many have suspected. I have tested many different afr ranges and with and without turbo inlet from prl and with or without mishimoto intake and it still occurs. So don’t go buying another expensive intake to fix it cause it won’t. Only hks intake without the sensor corrector and eventuri will work on stock scaling because they use a larger diameter housing and have a different MAF scaling to run. They do not reach the higher voltage running the bigger diameter MAF housing.

If it is possible to fix this by only changing the values in the MAF scaling I will release it publicly. I also will have another member test it on their car. It shouldn’t require large changes enough to effect trims or throw commanded afr off. I am currently running a scale that does seem to be working. The only thing I noticed is if you max out the MAF it runs a little lean above 6600rpm. Hondata stated that would happen a long time ago. So in order to fix this after the MAF scaling proves correct I will add the needed Volumetric efficiency corrections as well.

My goal is to create a MAF scale that does not throw off other values in the ecu. Incorrectly scaling the MAF can actually result in up to 20 bar pressure drops in fuel even though the pump duty cycle is below 88 percent. This happens cause the measured amount of fuel in grams by the MAF sensor can be incorrect and the ecu calculated amount of fuel needed is now wrong.
Although the pressure drop wwas my really hurt the car it could result in some power loss. We aren’t talking significant but to me it will not be correct.

Fixing this requires more then just changing one or two values or even just raising it because you can raise it 1 percent above 4.56 volts but the end result is negative trims that fluctuate a lot. It will still lean out a bit above 6500. People running port injection do not have this issue cause the ecu is in serious amount if negative trims. Running upgraded turbo and fuel system with stock diameter MAF housing don’t see it as much either because it goes right passed the range where there’s an issue. If you add or subtract just one value you then throw off the ecu value differential between the value before or after.

Before you swear your car is special this has been confirmed by over a dozen cars different years and different set ups. Guys running ethanol on stock fuel system don’t see the higher MAF voltage as much due to reduced aircharge levels. Guys running methanol are less prone to it. I’m working on both with and without methanol.

Here is a pic of 93 octane tune with afr fluctuations at 4.77 up to 4.94. Also look at blue short term trim line.
71681AC1-250F-4AE2-A47D-E72645B71FE3.jpeg


Here is another that has at both points but isn’t as drastic. This is closer to base map. Again look at short term trim line
image.jpg


Here is a map. Methanol actually reduces the swings by a little bit. Still the short term trim goes up and down
image.jpg


Here is a map someone tried to correct just trying to correct the 4.77 problem area only. . You can see it can go the other direction a full point. You can see right when it goes rich the trims drop.
image.jpg


Here is corrected scaling but with up to a half point leaner at above 6600. You can clearly see a way better mixture control. You can see the blue short term trim line is far smoother also. There are negative trims because this is a methanol long but you can see the afr control is much closer compared to other maps. Yes there are negative trims from methanol but you can see that the short term fuel trims are more inline.
image.jpg



My end goal is to have it that flat to redline but I will need at least one more car with a different set up to test. I’d prefer 2 cars. You can use your current tune but will need to import MAF scaling values and VE correction values. A 91 or 93 octane car stock fuel system stock turbo with at least an intercooler intake and downpipe. I already have some one that can test ethanol.

Those interested please send me a direct message only. I will consider customizing the tune as a thank you for testing.
Testing the values can in no way harm any car. The changes in MAF scaling are literally less than .3percent. That’s .003 differential. I do not have current ve values I have just started testing that now. Stock cars or cars with only downpipe will not be able to maintain the volume of airflow in order to test. Eventuri intakes or hks intakes can not be used for testing. A drop in turbo fuel system upgrade car with eventuri intake could have the same issue. I can correct this as well cause once I correct the stock MAF scaling it’s just a matter of bring all the values up same percentage .
Thank you to those interested.
Thank to those who already sent me logs to help me verify this is not an intake design issue.
When I make it public I’ll post a base map that people can either copy and paste off of or import into their tune.

 
Last edited:

FK8_K20c1

HONDA The Power of Earth Dreams Technology!
First Name
Jonathan
Joined
Jan 11, 2018
Threads
27
Messages
1,192
Reaction score
719
Location
Indiana
Car(s)
FK8 Type R
Build Thread
Link
Country flag
Good read. I had this issue happened to me and the tuner recommended to change my fuel pump ?. Got ahold of a different tuner and he got it corrected by scaling the maf.
 
OP
OP
Jpierro79

Jpierro79

Senior Member
First Name
John
Joined
Mar 6, 2019
Threads
33
Messages
398
Reaction score
264
Location
Myrtle beach South Carolina
Car(s)
2018 type R 2015 c300
Country flag
Good read. I had this issue happened to me and the tuner recommended to change my fuel pump ?. Got ahold of a different tuner and he got it corrected by scaling the maf.
Thank you.
Many tuners don’t know you should not need to scale the MAF on the car to tune it. It’s a target fuel strategy closed loop. WOT Open loop fueling you rescale the MAF not closed loop unless there’s an issue. All fk8 stock MAF scaling is
Incorrect from factory. So is Volumetric effeciency. You should not need to scale the MAF with an intake unless the MAF housing is a different diameter.
The cars will run without 99 percent of the people noticing it. So if people have had pressure drops and inconsistent power it can be this. When trims go all over the place like the early datalogs power becomes inconsistent and choppy. We’ve all seen up top fi8 dyno and they get choppy up top.
Here’s a good example. Even after being smoothed it’s choppy.
2B61802B-ADCE-4B17-8CDC-239D775BEB64.png

A lot of dyno smoothing hides it. I had originally figured it was just the car having boost fluctuations chasing aircharge but when the power fluctuations didn’t line up and people were complaining about see wild afr swings I looked into a theory someone else had. After figuring out it was not due to intake or turbo inlet I knew it had to be the scaling. Especially when it did it at the exact same spot In 8 different logs. They can get much worse actually here is one. It can throw off boost control sometimes
781AF985-AC22-4F07-9B98-AEEED522EBA2.jpeg


I’m pretty sure people wouldn’t like their cars fueling 14.0 on over 22 psi.
 

longthaitran91

Suiton20
First Name
Long
Joined
May 15, 2018
Threads
24
Messages
459
Reaction score
186
Location
Austin, TX
Car(s)
2017 honda civic si
Country flag
How has this gone unnoticed for so long? I don’t own a type r. I’m surprised the maf is slightly off. Oversight from Honda?
 

elmerzasty

Senior Member
Joined
Sep 12, 2018
Threads
8
Messages
166
Reaction score
103
Location
Poland
Car(s)
FK8
Country flag
How has this gone unnoticed for so long? I don’t own a type r. I’m surprised the maf is slightly off. Oversight from Honda?
Everything is possible, but I think it is unlikely given the importance of MAF readinds in the Bosch world (almost everything this ECU does is based on aircharge which is determined by the MAF) and the K20C1 being built since 2015 in the FK2R.

You have to remember that only a fraction of all maps that the ECU uses are exposed to the user in tuning solutions like FlashPro. Regarding the MAF in particular, I am fairly sure that the ECU has logic to account for flow irregulariaties under certain conditions (turbulence), because actual flow may differ even though MAF voltage is reading the same same, depending on different RPM, IAT, BP etc.

I am not sure if this MAF issue that the OP is investigating presents itself on 100% stock cars? I mean stock hardware AND software. Because it is possible that with everything stock there are no issues at those MAF voltages, because the voltages are not reached or are reached only under conditions accounted for in other maps of the ECU.
 
OP
OP
Jpierro79

Jpierro79

Senior Member
First Name
John
Joined
Mar 6, 2019
Threads
33
Messages
398
Reaction score
264
Location
Myrtle beach South Carolina
Car(s)
2018 type R 2015 c300
Country flag
How has this gone unnoticed for so long? I don’t own a type r. I’m surprised the maf is slightly off. Oversight from Honda?
Because many people either go right passed the range or are below it. It has to do with the math calculations. Honestly a lot of factory tunes are done through simulations. The stock type r never hits this range for a reason. Higher up in the scaling of any sensor on a car has less accuracy. It’s the nature of the beast. I have been a diagnostic tech a long time. This is actually very common with modified cars.
The reason no one has made a larger MAF housing to fix the range the MAF operates in is that when the type r exceeds the MAF reading it stays in closed loop automatically corrects the fuel. Because the MAF scaling is off the swing is from the ecu trying to correct the value and trying to use the MAF value. That’s why it swings back and forth.
Being when you have enough power your always gong to hit this range without changing the MAF housing as a technician I found this is unfair cause stock settings should work in this scenario.
Keep in mind that the fluctuations shown differently each time due to constant closed loop state and you won’t always hit the values exactly the same way.

A tuner can fix it on a single car but that scale won’t work for all cars. This is why I am testing on more than one car to confirm set up and stick and aftermarket intakes.
 
OP
OP
Jpierro79

Jpierro79

Senior Member
First Name
John
Joined
Mar 6, 2019
Threads
33
Messages
398
Reaction score
264
Location
Myrtle beach South Carolina
Car(s)
2018 type R 2015 c300
Country flag
Everything is possible, but I think it is unlikely given the importance of MAF readinds in the Bosch world (almost everything this ECU does is based on aircharge which is determined by the MAF) and the K20C1 being built since 2015 in the FK2R.

You have to remember that only a fraction of all maps that the ECU uses are exposed to the user in tuning solutions like FlashPro. Regarding the MAF in particular, I am fairly sure that the ECU has logic to account for flow irregulariaties under certain conditions (turbulence), because actual flow may differ even though MAF voltage is reading the same same, depending on different RPM, IAT, BP etc.

I am not sure if this MAF issue that the OP is investigating presents itself on 100% stock cars? I mean stock hardware AND software. Because it is possible that with everything stock there are no issues at those MAF voltages, because the voltages are not reached or are reached only under conditions accounted for in other maps of the ECU.
Iat is only a reduction MAF doesn’t need anything but and I take temp. A camaro zl1 runs 100 percent fueling after 3000 rpm off MAF and iat sensor in the MAF. Bosch has nothing to do with it. The MAF sensor is not Bosch. The tuning is not Bosch. Honda just tested it in the lower MAF voltage range. This is very common on many cars when modifying. The Bosch med 17 values are put in by people who are more focused on emissions. I’ve talked to manufacturers. They actually wish they could do such extensive testing but they can’t. It costs too much. Every engine will fuel completely different. This makes the fact it’s made by Bosch pointless. I. Motor sports they’d laugh at the ecu you think is so great.
This would not be an issue if the car was open loop like other cars but it’s full closed loop. Your assuming there’s no issue without testing. The only fueling maps we are missing for fuel us the base volumetric efficiency which can be adjusted by the VE correction and ect temp correction. Since all the sensors remain the same there’s no need to have iat and map scaling and pid changes on our cars. The fuel pump hysteresis is the only thing you need for an upgraded fuel pump which is what I wish hondata would make available so more manufacturers can develop fuel pumps.
I begged hptuners to take on this car cause as a tuner they release all available charts and compensation. Even alterations on manifold volume and manual transmission torque limits. Yes there are manual transmission torque limits on many cars. It’s calculated load by gear.

You have to fully grasp how all sensors effect the car and what sensors are used when. Bosch has nothing to do with this. A car ecu is only as good as the values put in it.
 

elmerzasty

Senior Member
Joined
Sep 12, 2018
Threads
8
Messages
166
Reaction score
103
Location
Poland
Car(s)
FK8
Country flag
The MAF scaling table that we have exposed in FlashPro Manager is typically done on a flow bench using a housing from the car. It only represents air flow vs voltage of a known MAF in a known housing. A theory that Honda did not put enough effort to properly calibrate the MAF at 4.56 and 4.77, but at the same time put enough effort to properly tune all the values before, in between and after them, and all of that because of emissions, does not seem plausible to me.

I am aware of IAT2 reduction of boost pressure and it is not what I referenced. I will admit that I have not seen all the maps in the MED17 so I cannot say for sure if in the MED17 there is a particular MAF compensation table in relation to IAT2, but that was not the gist. My point was that the air flow the engine sees is reliant not only on MAF reading (and hence MAF scaling table), but on everything that happens on the intake path causing variations and - most importantly - turbulence. Thus a compensation table (or tables) exists in the ECU to account for any potential mismatch of what the MAF is reading versus what the actual airflow is at the intake valve.

For example when the same MAF and housing is used on different engines (typical in VAGs) it will require different compensation in a given configuration.
 

Centripetal

Senior Member
Joined
Nov 19, 2017
Threads
0
Messages
388
Reaction score
308
Location
Boston, MA
Car(s)
2006 RSX Type S, 2017 Civic Type R
Country flag
This is a good post and has a lot information that begs for discussion.

Many tuners don’t know you should not need to scale the MAF on the car to tune it. It’s a target fuel strategy closed loop.
Open-Loop is how a lot of hptuner people do it. I get it, keep the control loop out of the equation to calibrate the A/F.

Swings are normal in a closed-loop system because of error (inaccuracies) in system. The swing will be larger if there is error that hasn't been calibrated out.

Therefore, I pose this question: If fuel trim is a measure of said error in the loop, why can't you calibrate the MAF to correct the error?

Incorrect from factory. So is Volumetric effeciency. You should not need to scale the MAF with an intake unless the MAF housing is a different diameter.
While I agree with you for the most part but it is not completely true. Unless you really mean characteristic length, which is sometimes diameter. There is much more to it than I'm going to layout but there are a few things to consider. We have to consider where the MAF measures the velocity of the air: center-line and distance from the orifice.

The FK8 intake operates in ranges with different flow characteristics, which can been seen in the MAF transfer function (map). The Reynolds number indicates what type of flow is occurring. Generally speaking, a Reynolds number (Re) bellow 2100 is laminar and flow above 10,000 is turbulent. The area in between is transitional, which I'm going to hand wave away.
Screenshot from 2020-10-20 08-39-25.png

Under light load, the flow is laminar (a). Laminar flow is great because it pressure difference is directly proportional to velocity. This is the area (0.5V-1.7V) of the MAF transfer function is mostly linear, where the Reynolds Number is below 2100. The velocity profile for laminar flow is nice and uniform.

When the Reynolds number is above 10,000 the flow is turbulent (b). Turbulent flow is it is more resistant of flow as pressure difference increases. It also increases resistance as Reynolds number increases. This is why the slope increases on the MAF transfer function as voltage increases (2.85V-4.5V, 10,000 Re - 50,000 Re).
laminar-and-turbulent-flow-maf.png


If this is accounted for in the MAF transfer function (map) and MAF measure on center-line, why is this important? Well, the answer is it cannot account for all situations. For example, you have to consider that characteristic length is not necessarily pipe diameter. For example, the characteristic length around a bend will be larger, therefore increasing the Reynolds number. The first FK8 Eventuri intake had the MAF right by a bend. I think we all know that had issues with fuel trims.
Screenshot from 2020-10-23 09-49-36.png


Another thing to consider is MAF position and Reynolds number. Consequently, in certain situations, the velocity profile can change with Reynolds number and velocity. As you can see in the diagram below, the center-line velocity can increase or decrease with Re. Note this is in the transitional range of flow type.
Screenshot from 2020-10-23 10-21-46.png

If the MAF is in a different location than stock, its velocity profile will change, therefore changing the MAF transfer function/map. For example, the newer revision of Eventuri intake has much more pipe before the MAF, which means the velocity profile will be more developed. If the MAF measures the velocity slightly off center line with a fully developed profile, there will more error.
660px-Development_of_fluid_flow_in_the_entrance_region_of_a_pipe.jpg


In conclusion, there is a lot more than diameter to consider when changing an intake.
 
Last edited:

ayau

Senior Member
Joined
Aug 14, 2016
Threads
45
Messages
1,713
Reaction score
1,295
Location
USA
Car(s)
Subaru
Country flag
Honda put a lot of time and money designing the factory intake and having the proper MAF scale. Way more than any aftermarket intake company will.

It's already been proven that the fk8 factory intake isn't a bottleneck for a relatively stock car. In fact, this is true for many modern DI turbo cars. Why make things more complicated by using an aftermarket intake? I suppose if you want that woosh noise.
 

turbociv910

Senior Member
First Name
taylor
Joined
Jun 24, 2019
Threads
22
Messages
993
Reaction score
961
Location
wilmington nc
Car(s)
2020 SGP Civic Type R, Audi Q5
Build Thread
Link
Country flag
Honestly a lot of factory tunes are done through simulations.
source?

You dont think honda had a k20c strapped to a engine dyno for weeks making the tune right?
 

Centripetal

Senior Member
Joined
Nov 19, 2017
Threads
0
Messages
388
Reaction score
308
Location
Boston, MA
Car(s)
2006 RSX Type S, 2017 Civic Type R
Country flag
source?

You dont think honda had a k20c strapped to a engine dyno for weeks making the tune right?
Since he’s too busy repeat posting, I‘ll weigh in.

Simulation is based off assumption, which will only get you so far. Most of engineering is realizing your wrong and structuring experiments to figure out what data you missed. I highly doubt they would test on just one engine. Usually, you want statistically significant sample (n = 30). In addition, the quality and safety standards would never let you go off just a design without any verification and validation.
 

CJ1320

New Member
First Name
Chris
Joined
Sep 9, 2020
Threads
0
Messages
2
Reaction score
0
Location
nazareth PA
Car(s)
2020 Civic Si Sedan, 1995 Integra GSR, 1993 Civic Cp. B16 swap
Country flag
This is a good post and has a lot information that begs for discussion.


Open-Loop is how a lot of hptuner people do it. I get it, keep the control loop out of the equation to calibrate the A/F.

Swings are normal in a closed-loop system because of error (inaccuracies) in system. The swing will be larger if there is error that hasn't been calibrated out.

Therefore, I pose this question: If fuel trim is a measure of said error in the loop, why can't you calibrate the MAF to correct the error?


While I agree with you for the most part but it is not completely true. Unless you really mean characteristic length, which is sometimes diameter. There is much more to it than I'm going to layout but there are a few things to consider. We have to consider where the MAF measures the velocity of the air: center-line and distance from the orifice.

The FK8 intake operates in ranges with different flow characteristics, which can been seen in the MAF transfer function (map). The Reynolds number indicates what type of flow is occurring. Generally speaking, a Reynolds number (Re) bellow 2100 is laminar and flow above 10,000 is turbulent. The area in between is transitional, which I'm going to hand wave away.
Screenshot from 2020-10-20 08-39-25.png

Under light load, the flow is laminar (a). Laminar flow is great because it pressure difference is directly proportional to velocity. This is the area (0.5V-1.7V) of the MAF transfer function is mostly linear, where the Reynolds Number is below 2100. The velocity profile for laminar flow is nice and uniform.

When the Reynolds number is above 10,000 the flow is turbulent (b). Turbulent flow is it is more resistant of flow as pressure difference increases. It also increases resistance as Reynolds number increases. This is why the slope increases on the MAF transfer function as voltage increases (2.85V-4.5V, 10,000 Re - 50,000 Re).
laminar-and-turbulent-flow-maf.png


If this is accounted for in the MAF transfer function (map) and MAF measure on center-line, why is this important? Well, the answer is it cannot account for all situations. For example, you have to consider that characteristic length is not necessarily pipe diameter. For example, the characteristic length around a bend will be larger, therefore increasing the Reynolds number. The first FK8 Eventuri intake had the MAF right by a bend. I think we all know that had issues with fuel trims.
Screenshot from 2020-10-23 09-49-36.png


Another thing to consider is MAF position and Reynolds number. Consequently, in certain situations, the velocity profile can change with Reynolds number and velocity. As you can see in the diagram below, the center-line velocity can increase or decrease with Re. Note this is in the transitional range of flow type.
Screenshot from 2020-10-23 10-21-46.png

If the MAF is in a different location than stock, its velocity profile will change, therefore changing the MAF transfer function/map. For example, the newer revision of Eventuri intake has much more pipe before the MAF, which means the velocity profile will be more developed. If the MAF measures the velocity slightly off center line with a fully developed profile, there will more error.
660px-Development_of_fluid_flow_in_the_entrance_region_of_a_pipe.jpg


In conclusion, there is a lot more than diameter to consider when changing an intake.
 

CJ1320

New Member
First Name
Chris
Joined
Sep 9, 2020
Threads
0
Messages
2
Reaction score
0
Location
nazareth PA
Car(s)
2020 Civic Si Sedan, 1995 Integra GSR, 1993 Civic Cp. B16 swap
Country flag
Honeycombs anyone

20210418_222642.jpg


20210418_222521.jpg

 

 
Top