Help with scaling Bosch HFM MAF mdot/V table 1

[ChriswithaC]

I've been searching all over car forums and chip-tuning forums for help with this, but I don't know why it didn't occur to me to seek the information here.


Here is the situation: I am tuning my own turbocharged car and have had to move to a larger Mass Airflow Sensor because the stock one was maxing out.

So... I have access to a 512 column table which converts the 0-5V signal of the MAF to a mass flowrate in Kg/hr. I am using the same sensor but in a larger housing in order to flow more air without maxing the signal out.

Currently I have simply scaled the expected mass flow for a given voltage by the change in area, and adjust the MAF Offset so the curve still crosses Y-0 at 1V.

old housing = 70mm
new housing = 82.5mm

%difference area = 5345.62mm^2 / 3848.45mm^2
= 1.389.

Scaled the whole map by 138.9% and adjusted the offset.

......this seems FAR too easy. This is "standard practice" for chip tuning from what I've been told by others, but I have spoken to a tuner with a LOT of experience with Motronic and he indicates this is cheating and that Bosch engineers use a much different way to alter the map.

His words were, "You cannot increase percentage that map! It is based on exponential function!" (he's Greek, hence the broken English.)

After plotting the MAF linearization curve in Excel, it looks like the trendline that fits best is a 5th or 6th order polynomial.

So I'm calling out for some help, advice, and theories as to how the correct method would be to scale the MAF voltage vs. mass flow.


I would really appreciate some help on this guys.

Thank you in advance,

Chris

 

[gruntguru]

I can't fault your original suggestion. I doubt the scaling factor will be exactly 1.38 - it will depend on the velocity profile across the section and the region of that profile occupied by the filament. (ie the ratio of velocity at the filament to average velocity across the section may be different for the two housings.) Another consideration is the small change in Reynolds number which may effect the coefficient and the shape, but probably only slightly.

Can you simply run the engine with the 1.38 factor and compare the mixture at a known operating point? If needed, change the scaling factor then check some other points.

Engineering is the art of creating things you need, from things you can get.

 

[yoshimitsuspeed]

I can't help you with the original question but surely at this level of tuning you must have a wideband AFR gauge right?
This should allow you to safely test any theories.

What exactly do you mean by chip tuning? Do you plan on doing the math then getting a chip or your ECU reprogrammed with those changes? I'm not real familiar with chips or reflashing ECUs.

Is this a very unique setup? I know generally you would be able to find the most knowledge on a forum specific to your car and I would think they would either be able to answer your question or point you in a better direction.

What kind of car/motor/ECU?

 

[ChriswithaC]

Thanks for getting back to quickly guys. I really appreciate it.

@gruntguru: I'm currently running the car with the MAF scaled at 38% larger and it seems to be running well from what I can tell, but I am running the stock fuel injectors at the moment to minimize the variables while tuning the MAF. Unfortunately because of this, they run out of fuel very early with the much larger turbo that I have fitted, so it's difficult to get any data in the higher flow areas of the MAF table.

I do not have any logs from before I installed the new turbo setup so I have no numbers to compare. So I'm trying the best I can with the 205cc injectors. With this car being turbocharged, it makes tuning rather difficult because at any given RPM I could be in vacuum or at 15 PSI.

Is there any way that I could work with the trendline formula to work through the scaling? How would you suggest I go about working through this?



@yoshimitsuspeed: My car does have a built in WB AFR sensor so that makes tuning much easier and much safer, however, if the curve for the MAF is out it will throw off everything else with the tune. With later Bosch Motronic ECU's they are Torque-Based which rely heavily on MAF as the primary source for calculating engine load. Engine load is used for Wastegate Duty tables, Ignition advance tables, Fuel tables, protection tables.....etc.


I REALLY would like to nail this down as close as possible. The car runs good as it is, but I was to keep the O2 correction to a minimum. I'm trying for a very OEM+ tune, despite the fact that I am making over 2x the HP from when the car rolled off the assembly line.


Chip tuning is just a term that's used to describe tuning a cars ECU to do what you want. Tune for more power. Write out certain check engine lights. etc....

While tuning my particular vehicle is nothing new for a number of software tuning companies, the concept of self-tuning is very new. The software I have is the first of it's kind to actually allow the end-user to edit maps based on their own setup. You're given a base file that goes with a specific engine/injectors/MAF and you go from there. It allows me to change injectors to whatever i want, alter ignition timing, alter desired AFR, alter boost, change the throttle pedal response curve, 2-step launch control, changing rev limiters, .......many other things.

The car I'm working with is a 2005 Audi A4 1.8T. Bosch ME7.5 ECU. Garrett GT2871R turbo.

Have a look at my build thread if you like

http://www.audizine.com/forum/showthread.php/366320-The-Build-A-GT2871R-build-documentary

 

[patprimmer]

The concept of tunable substitute chips has been used or years on Honda ODB0 and ODB1 ECUs

The two concepts are to dyno tune or road tune.

To be done correctly both require a readable wide band oxygen sensor with display and det cans.

Det cans can be some sort of microphone or even a knock sensor hooked to an amplifier and speaker or a stethoscope with a solid metal transducer or echo plate bolted to the block.

The Honda system uses a MAP not an MAF and mathematically derived base tunes work OK as a starting point so long as extreme caution is used to test and develop the tune.

On the Hondas we normally substitute the chip with an emulator to work out a good tune, then burn a chip and install it in the socket on the ECU main board.

One of the most basic steps is to install injectors with sufficient fuel flow.

Unless your ECU has fine enough resolution, it may be difficult to reduce fuel flow for efficient idle once you install large injectors and that is often the factor that sets the upper limit of your injector size.

You can increase the range of injectors by also tuning fuel pressure, but that creates an additional complication and response time issues as the fuel pressure takes time to adjust so fuel pressure tends to lag behind demand.

The best approach is to run the biggest injectors that will deliver acceptable fuel for idle at the lowest duty cycle the ECU will support.

I would calculate for a:f of about 12:1 for the base map at no boost and 11:1 a:f on boost, then work from there by tuning at no boost then gradually delving further into boost while tuning with the wide band AND plug readings until it had detonation free WOT with no misfires or loading of plugs.

Mathematical methods are good for starting points, but eventually you need to read the engine and make t happy.





Regards
Pat
See FAQ731-376 for tips on use of eng-tips by professional engineers &

http://eng-tips.com/market.cfm

for site rules

 

[ChriswithaC]

Pat,

I'm pretty familiar with the tuning process and have a very good grasp of the tuning process. I also have much larger injectors that I plan to put back in one I get this MAF scaling sorted out.

I completely agree with the "Mathematical methods are good for starting points, but eventually you need to read the engine and make it happy." philosophy. That's my plan, however, the one item that I would like to keep as my "mathematically fixed constant" is the MAF because I don't plan on changing it again. The 82.5mm MAF should flow well beyond the amount of power I could make on this engine.


I feel that getting the correct curve for the MAF response is pretty important and should be fairly straight forward. There may be some areas of the map that need corrections for turbulence, but I have a map for that based on Engine Load and RPM, so I'm not worried. I just need to get the scaling issue figured out.

 

[yoshimitsuspeed]

I agree with Pat on the detcan.
If you wanna go fancy I have heard great things about the J&S safeguard. It will actually pull timing if it detects knock and also has a gauge so you can see when and where it's pulling timing and how much.
I got a basic knock light from Summit but I don't really trust it. I will need to confirm it works properly with headphones before I trust it for pushing my tune. It currently lights up between 4k and 6k rpm but I really suspect it's an engine frequency triggering it, not knock. I am lightly boosting 11:1 compression though so can't be sure till I confirm with another method.
The other tool I would suggest is a pyrometer. The knock sensor will help you see if your timing it too advanced or if boost is too high/octane too low, the pyrometer will help you see if the timing is too retarded or if you are doing something else that could melt pistons or valves.
WIth a wideband a knock monitor or safeguard and a pyrometer you should be able to guess on a rough tune then safely test it to see if it works. Set your timing on the conservative side then start bumping it up till you see early signs of knock and then bump it back enough to feel safe.


You could also do a couple bench tests to confirm you are close to your target with the MAF.
Maybe hook up a fan and test the output on the old maf then the new one. Hook up a household vacuum and get readings, then maybe a big shop vac or something that moves a lot of air. See how the outputs align with your calculations and if something is way off make adjustments from there.
In the end you will have to dial it in on the car because no one is going to know the exact effects it will have till they try it.

 

[ChriswithaC]

I really appreciate the advice with how to tune.....but I've got that covered. This isn't my first rodeo when it comes to tuning. My Audi is equipped with a wideband O2 sensor as well as an amazing knock sensing system.


MAF scaling, however, is new to me. Luckily for me I have access to data that allows me to compare my MAF element between two cars already in which Audi uses the same element just in a different housing. I also have the 1x512 value transfer function tables for both vehicles.

What I have ended up doing is simple calculating the % difference in values between the two known MAF curves and then scaling them by the ratio of area change from my custom MAF and the largest MAF I have a known curve for in order to minimize error. I have also applied some very simple curve smoothing to try and feather things out again.

I haven't had a chance to try out the new curve yet.

Anyone have any thoughts on how boundary layer and how it might be affecting readings? My MAF sensor is located at the center of the MAF tube. Just wondering if this isn't why I'm seeing some strange LTFT's

 

[Lou Scannon]

MAF sensors are extremely sensitive to upstream and downstream pipe configuration. Changing the pipe configuration or repackaging the sensor in any way necessitates evaluation and mapping of the new flow transfer function. It would be very unusual indeed, in a real world automotive application, to repackage a drop in sensor into a larger diameter tube and get a scalar transformation of the transfer function. I say that because it's almost certain that the original transfer function was influenced by the original piping configuration, and I can't imagine that it's possible to change the diameter of the MAF tube without also changing the upstream and downstream piping configuration.

 

[ChriswithaC]

They are indeed the same MAF element in different housings right from the factory. The transfer functions are not simply just scaled though. The Bosch Motronic software also has a 3d table that corrects for flow areas with strange turbulence and flow irregularities in addition to the 1x512 Mass Flow vs. Volts table.

My MAF sensor is in a fabricated housing at the moment with no flow-straightener grid like the OEM MAFs, so I believe this might be contributing to some of my difficulty with tuning. I have some honeycomb material on order that I plan to make a flow-straightener to put before the MAF element. I wish had some room to move the MAF a little further down on a straight section of tube before my turbo, but unfortunately that just isn't possible.

 

[Lou Scannon]

Just to make sure we're on the same page, the housing does not constitute the inlet and outlet piping configuration in its entirety. Any cross-sectional area or shape changes upstream or downstream of the MAF unit are part of the configuration. If you introduce a MAF unit into a piping system that it was not calibrated for the calibration will be off.

 

[ChriswithaC]

That MAF sensor in a particular housing uses the same 1x512 table for flow vs. voltage regardless of what piping system they are in. I have verified this.....trust me.

The 14x14 correction tables, however, between each are are very different. This is because of the reasons you're describing.

A couple of examples:

The Audi 2.7T RS4 uses the same 512 column table that the 4.2L S4 V8 uses but a VERY different airbox/inlet/outlet configuration. The 14x14 correction tables are different.

The 180HP Audi TT 1.8T uses the exact same MAF housing and sensor as my stock 170hp A4 1.8T used.....again, 512 the same, 14x14 correction table much different. Even though the TT engine is transversely mounted and the A4 is longitudinally mounted. Very different inlet/outlet configurations though.

The 225HP Audi TT 1.8T MAF uses the very same 512 table as the VW R32 VR6 engine. 14x14 not even close to the same



The 1x512 tables in the Bosch Motronic ECUs are a relationship between the air mass measured by the sensor e\ extrapolated for the size of the cross section that the sensor is in. That relationship does not change from car to car, setup to setup. What changes are the turbulence and inconsistencies between the various uses which require corrections using the 14x14.

 

[Lou Scannon]

OK, sounds like we're on the same page then, as long as you're using the correction tables applicable to your installation.

 

[ChriswithaC]

We are indeed on the same page. Thank you for chiming in. I appreciate the help

I have had quite a bit of luck tuning the correction table, but that will all go out the window after I add in a flow-straightener.

..........is there a technical term for this?? I have tried to find an actual technical term for this, but all i come up with is "flow straightener." Sounds like a layman's term. I can't help but think I would find better information if I was using a more technical term.

I come up with thread after thread of guys cutting them out to make more power. I just laugh to myself...

 

[Lou Scannon]

Flow straightener is the most common term I've seen used when discussing a honeycomb mesh applied to a mass flow sensor.
In reality the main function is to reduce turbulence so that the signal is "quieter". If you can log data at a high update rate you can see the difference in signal with and without the "straightener". By smoothing out the actual flow, less electronic/digital filtration is needed for processing the signal, hence more timely data to the ECU for calculating fuel, spark etc.
A wire mesh accomplishes the same thing; perhaps with a little more restriction. I have developed and applied hot wire flow sensors where restriction was not an issue, and used perforated sheet metal for a "flow straightener", as this provided the easiest manufacturing and assembly method.

Re your last statement, let's hope the flow straightener was not an appreciable restriction in the original application. If mods have increased the air flow considerably, it could present some appreciable restriction at the upper end of the flow envelope. I have observed that the transfer function is hardly affected by the straightener, only the signal to noise ratio.