how to scale MAF from scratch
May 28, 2012, 01:44 PM
#1
Newbie
Thread Starter
Join Date: Sep 2011
Location: Bavaria
Posts: 29
Likes: 0
Received 0 Likes
on
0 Posts
how to scale MAF from scratch
Hello everyone.
I'm into ECU disassembling/remapping for around a year, mainly using quite a "bastard engine" in my Galant for testing purposes:
- rebuilt 6-bolt bottom end (Galant VR4)
- rebuilt 4G67 cylinder head (smaller rooftops -> CR ~9.2:1 instead of 8.4:1)
- VR4 camshafts (latest evolution, same profile as evo3)
- evo3 16G turbo
- evo4 MAF with open K&N filter
- Evo 9 injectors
- Evo 7/8 ECU
(and much more that isn't essential for this topic)
From the beginning of tuning this engine I had to figure out a lot for myself by testing or disassembling and trying to understand the code (perfect for long, LONG, winter weekends
).
One such thing that gave me some headaches for a while was the MAF scaling.
I've found 3 different scalings for the evo4 MAF and lots of people were posting in different forums, the evo4 MAF could be used with evo 7/8 scalings...
As neither of the 3 scalings nor the 7/8 scalings worked i decided to scale the MAF from scratch. Maybe someone will face the same issue, so i just made some screenshots from my old maps and wrote this. Maybe someone will find it helpful, maybe not, or maybe someone will find this way completely wrong and will help me improving it. Any comment or constructive critics is welcome!
Here we go...
Things we need:
- An Engine running any kind of karman-vortex MAF and Evo ECU (i'm using 7/8, but the procedure should be the same for any other flashable evo ECU)
- WBO2 sensor, loggable through the ECU or directly via serial connection
- notebook to log the ECU and WB02 (e.g. evoscan)
- a place where we don't bother anyone when we will rev up the engine again and again...
First we will have to reset the fueltrims, so we disconnect the battery.
Then we will disable closed loop either by the peripheral bits or by setting the "min coolant temp for closed loop" up to 130°C or any other value it will never exceed.
Then we need some "cleaned" MAF scaling & smoothing tables we can start with.
I started with the scaling curve from an evo7 MAF (if i remember correctly...) but anyone will do the job. The important part is, to flatten out the smoothing table. I chose 120 which is the "baseline" of many OEM smoothing tables and turned out to be quite usable:
Before we flash these new values, we let the car idle for a while to start with a warm engine. Then we flash the modified maps and turn the car back on.
Now with the logging program we will monitor the MAF values (Hz), AFR Map and AFR from the WBO2. If the AFR is WAY off the line we should turn the car off, increase/decrease the whole MAF scaling table and try again until we get a clean 14.7 AFR at idle.
If the WBO2 values match the AFR Map at idle, we start to increase the engine rpm step by step, holding the Hz-values we also have in our MAF scaling table for ~3-5secs (to get an average/stable AFR value).
First start with 2-3 values above the idle we just matched to 14,7 AFR by moving the whole scaling curve. In my case as idle Hz was ~35-40Hz i went up to 50, 75, 100 and 125 Hz.
have a look at the AFR Map and AFR WBO2 values at each Hz value - if it matches, leave the "MAF scaling" entry as is, if not increase/decrease the value for the given Hz.
flash the new map, rev to the Hz-value(s) you just cchanged and compare AFR map to AFR WBO2.
Repeat this until ~5500-6000rpm, open the bonnet for better cooling and always have an eye on your coolant temps!
This procedure works up to ~700Hz - anything above only works with some load!
If we have matched all AFR WBO2 values to the AFR MAP values at any given Hz value, the curve might look like this:
The values above ~700Hz can be manually interpolated by extending the curve we already have and will be adjusted later
No, it's not a smooth curve, but that's perfectly normal for now!
Now as we have extrapolated the curve above ~700Hz manually, we can redo the whole thing with LIGHT load. A quiet industrial area on a weekend with long straight roads would be perfect. let the car roll and try to match the Hz values again to fine-tune the table.
These changes should be very small - if the values are completely off the line, your injector scaling might be wrong! Mine were scaled slightly too large even if i used the same scaling i read out of the ECU the injectors were running with!
If the lower values are OK we can start to go up to the higher Hz values (and also higher boost levels). A driver or instructed co-driver that tells when the desired Hz rate is reached is essential for this (or you will be logging and flashing for days...).
This way we should have the MAF scaling table completed on half a day, maybe 1 day if the injector scaling was also wrong.
Now the smoothing table comes into play:
As you might have recognized from the logs the load doesn't increase as smooth as normaly and sometimes at lower loads, where the MAF scaling isn't a smooth curve, the engine seems to act a bit rough on slow throttle changings.
This is because of the "bumps" we have in our MAF scaling curve - The values in this table seem to be directly used for load calculation by the ECU (please confirm). So by having "bumps" in the curve, the load "jumps" even when the airflow is smoothly increased.
Therefore we will now use the smoothing table.
Flatten out the MAF scaling curve and add/substract every value you change at the MAF scaling table, equally to the MAF smoothing table at the same Hz value!
So if you have to SUBSTRACT 23 at 100Hz from the MAF scaling table, ADD 23 to the MAF smoothing table.
Do this for the whole table until you get a smooth MAF scaling curve. you can use the main form from an original map as reference.
After this the MAF scaling might look like this:
and the new MAF smooting table:
Now go out on the road again, check if the AFR WBO2 readings are matching the AFR MAP at the given Hz values. If fine tuning is required, apply it to the MAF smoothing table.
Only apply changes to the MAF scaling table, if a whole Hz-range is off the line and needs to be increased or decreased and a smooth curve is obtained.
My current MAF scaling & MAF smoothing tables now look as follows:
Hypthesis (unconfirmed! any comment/advise welcome!):
The MAF scaling table directly changes the ECU-calculated load!
This could also be used to optimize/affect load-based routines of the ECU - To decrease overall load, decrease the MAF scaling table and add the same value to the MAF smoothing table. (if you do this, you have to change ALL load-based tables to match the new loads!!)
I had significantly higher loads overall, when my smoothing table "baseline" was ~100 and overall MAF scaling values were higher - idle load was ~50-60, so all lower cells at e.g. fuel or ignition tables were useless... now im at ~30 at idle which perfectly matches most idle-relevant routines with standard-values.
I'm into ECU disassembling/remapping for around a year, mainly using quite a "bastard engine" in my Galant for testing purposes:
- rebuilt 6-bolt bottom end (Galant VR4)
- rebuilt 4G67 cylinder head (smaller rooftops -> CR ~9.2:1 instead of 8.4:1)
- VR4 camshafts (latest evolution, same profile as evo3)
- evo3 16G turbo
- evo4 MAF with open K&N filter
- Evo 9 injectors
- Evo 7/8 ECU
(and much more that isn't essential for this topic)
From the beginning of tuning this engine I had to figure out a lot for myself by testing or disassembling and trying to understand the code (perfect for long, LONG, winter weekends
).One such thing that gave me some headaches for a while was the MAF scaling.
I've found 3 different scalings for the evo4 MAF and lots of people were posting in different forums, the evo4 MAF could be used with evo 7/8 scalings...
As neither of the 3 scalings nor the 7/8 scalings worked i decided to scale the MAF from scratch. Maybe someone will face the same issue, so i just made some screenshots from my old maps and wrote this. Maybe someone will find it helpful, maybe not, or maybe someone will find this way completely wrong and will help me improving it. Any comment or constructive critics is welcome!
Here we go...
Things we need:
- An Engine running any kind of karman-vortex MAF and Evo ECU (i'm using 7/8, but the procedure should be the same for any other flashable evo ECU)
- WBO2 sensor, loggable through the ECU or directly via serial connection
- notebook to log the ECU and WB02 (e.g. evoscan)
- a place where we don't bother anyone when we will rev up the engine again and again...
First we will have to reset the fueltrims, so we disconnect the battery.
Then we will disable closed loop either by the peripheral bits or by setting the "min coolant temp for closed loop" up to 130°C or any other value it will never exceed.
Then we need some "cleaned" MAF scaling & smoothing tables we can start with.
I started with the scaling curve from an evo7 MAF (if i remember correctly...) but anyone will do the job. The important part is, to flatten out the smoothing table. I chose 120 which is the "baseline" of many OEM smoothing tables and turned out to be quite usable:
Before we flash these new values, we let the car idle for a while to start with a warm engine. Then we flash the modified maps and turn the car back on.
Now with the logging program we will monitor the MAF values (Hz), AFR Map and AFR from the WBO2. If the AFR is WAY off the line we should turn the car off, increase/decrease the whole MAF scaling table and try again until we get a clean 14.7 AFR at idle.
If the WBO2 values match the AFR Map at idle, we start to increase the engine rpm step by step, holding the Hz-values we also have in our MAF scaling table for ~3-5secs (to get an average/stable AFR value).
First start with 2-3 values above the idle we just matched to 14,7 AFR by moving the whole scaling curve. In my case as idle Hz was ~35-40Hz i went up to 50, 75, 100 and 125 Hz.
have a look at the AFR Map and AFR WBO2 values at each Hz value - if it matches, leave the "MAF scaling" entry as is, if not increase/decrease the value for the given Hz.
flash the new map, rev to the Hz-value(s) you just cchanged and compare AFR map to AFR WBO2.
Repeat this until ~5500-6000rpm, open the bonnet for better cooling and always have an eye on your coolant temps!
This procedure works up to ~700Hz - anything above only works with some load!
If we have matched all AFR WBO2 values to the AFR MAP values at any given Hz value, the curve might look like this:
The values above ~700Hz can be manually interpolated by extending the curve we already have and will be adjusted later
No, it's not a smooth curve, but that's perfectly normal for now!
Now as we have extrapolated the curve above ~700Hz manually, we can redo the whole thing with LIGHT load. A quiet industrial area on a weekend with long straight roads would be perfect. let the car roll and try to match the Hz values again to fine-tune the table.
These changes should be very small - if the values are completely off the line, your injector scaling might be wrong! Mine were scaled slightly too large even if i used the same scaling i read out of the ECU the injectors were running with!
If the lower values are OK we can start to go up to the higher Hz values (and also higher boost levels). A driver or instructed co-driver that tells when the desired Hz rate is reached is essential for this (or you will be logging and flashing for days...).
This way we should have the MAF scaling table completed on half a day, maybe 1 day if the injector scaling was also wrong.
Now the smoothing table comes into play:
As you might have recognized from the logs the load doesn't increase as smooth as normaly and sometimes at lower loads, where the MAF scaling isn't a smooth curve, the engine seems to act a bit rough on slow throttle changings.
This is because of the "bumps" we have in our MAF scaling curve - The values in this table seem to be directly used for load calculation by the ECU (please confirm). So by having "bumps" in the curve, the load "jumps" even when the airflow is smoothly increased.
Therefore we will now use the smoothing table.
Flatten out the MAF scaling curve and add/substract every value you change at the MAF scaling table, equally to the MAF smoothing table at the same Hz value!
So if you have to SUBSTRACT 23 at 100Hz from the MAF scaling table, ADD 23 to the MAF smoothing table.
Do this for the whole table until you get a smooth MAF scaling curve. you can use the main form from an original map as reference.
After this the MAF scaling might look like this:
and the new MAF smooting table:
Now go out on the road again, check if the AFR WBO2 readings are matching the AFR MAP at the given Hz values. If fine tuning is required, apply it to the MAF smoothing table.
Only apply changes to the MAF scaling table, if a whole Hz-range is off the line and needs to be increased or decreased and a smooth curve is obtained.
My current MAF scaling & MAF smoothing tables now look as follows:
Hypthesis (unconfirmed! any comment/advise welcome!):
The MAF scaling table directly changes the ECU-calculated load!
This could also be used to optimize/affect load-based routines of the ECU - To decrease overall load, decrease the MAF scaling table and add the same value to the MAF smoothing table. (if you do this, you have to change ALL load-based tables to match the new loads!!)
I had significantly higher loads overall, when my smoothing table "baseline" was ~100 and overall MAF scaling values were higher - idle load was ~50-60, so all lower cells at e.g. fuel or ignition tables were useless... now im at ~30 at idle which perfectly matches most idle-relevant routines with standard-values.
Last edited by r4p.t0x; May 28, 2012 at 01:49 PM.
May 28, 2012, 02:39 PM
#2
Evolved Member
iTrader: (8)
Join Date: Nov 2010
Location: Georgia
Posts: 752
Likes: 0
Received 0 Likes
on
0 Posts
good step by step write-up
I had also succesfully done it to my previous evo3, but I haven't touched smoothing table at all, only the scaling. WB02 readings were dead-on at afr map after it
way easy for tuning
I had also succesfully done it to my previous evo3, but I haven't touched smoothing table at all, only the scaling. WB02 readings were dead-on at afr map after it
way easy for tuning
Last edited by stunt2; May 28, 2012 at 02:45 PM.
Thread
Thread Starter
Forum
Replies
Last Post
Raceghost
ECU Flash
22
Jun 1, 2016 12:30 AM
