ScoobySteve

Thanks. I used to live within walking distance of UCI but have since moved to Tustin. Despite my very low post count, I'm actually an old-timer of sorts. Had my 944 since '97 and joined here (but never posted) in 2004.

Are you a UCI alumnus?

Stephen

marcusw03

my ex went to medical school there, shame we didn't get hitched she would have provided more income to my 944 projects

MrTom

So why did you do this? You converted your car to mass air system? Can somebody explain this to me?

V2Rocket

the stock AFM measures airflow and assumes a given quantity of air based on a known air mass per volume and temperature. it measures this with a "barndoor" which is attached to an electrical wiper arm with makes contact with a track in the AFM, depending where the arm touches the track the car figures out how much fuel to use. the problem with the barndoor is that it isn't the most accurate measuring system; it works well enough but fluctuations in temperatures and other variables can alter the signal which can alter performance. the barndoor is also restrictive in allowing airflow through which is bad for making HP in part-throttle situations.

also, the shape of the inside of the AFM is detrimental to performance as it is a rectangular space which then goes to a round tube. this sudden transition causes turbulence in the air tract which isn't really what you want. again it works "well enough" but it can be improved.

a MAF measures the mass of air at a given temperature and calculates how much volume is actually coming in. this is a more accurate way of metering fuel.

the MAF always has a constant opening so air can pass through freely without having to squeeze past a barndoor. the MAF is also a round shape connecting to a round tube which smooths airflow.

all in all it lets the car breathe better at all RPMs, and while peak HP gains may be negligible, across the midrange improvements should be seen simply by smoothing out and opening up the air tract.

whalebird

I have alway been under the belief that the control units in our cars cannot understand air-mass. It is a cumbersome conversion beyond the control units ability. Having done this conversion to many motronic based porsche and other cars. The expensive kit and all the hardware was removed and put back to stock because there was no gain at all. In fact cold start and drivability suffered.
I am interested because the architecture is clearly better. If there were good results, I would do it in a heartbeat.
The air density problem is my only reservation (There is an altitude sensor on our cars). It seems a signal could easily be converted. I have just had poor results from such attempts.

V2Rocket

Isnt the altitude sensor inside the cabin next to the DME though?

Anyways, I dont think that the Motronic actually needs to understand air-mass. Obviously a better solution would be to program the DME for use with a MAF, but a simple solution like we have here is to use a MAF that calculates in its own way and sends out an electrical signal of how much air is passing through. We use the stock AFM plug to simplify wiring. The late DME understands a 5V max signal, and the MAF puts out a 5V max signal. This should be enough to calculate pretty accurately the amount of fuel needed...

Like I said though, as soon as I find a cheap late DME (why they are so expensive is beyond me, there are more of them than there are early DMEs) I will do this conversion and let you all know in this thread of my results.

whalebird

Yea. the altitude is in the cabin. Has Denso on it. Just wondering if that could be manipulated to further enhance this conversion. I would really like to do this conversion. It seems like it would be easy, but there are uncalculated variables. I wish I knew more about it, but I was told by a Porsche engineer that the conversion is futile - the maps will not corralate. Screw that guy though- I would be happy to give it a whirl. The MAF sends TOO MUCH info to our ECU. If it were to little I could be a sceptic, but a motronic ECU is one of the most basic - they have been around for centuries. There are a lot out there nowadays and I would love to revisit this. I want to do it
I have two ECUs for my 924s. I really want to tamper with my spare. My experiance in the field is that chips and MAF conversions don't work, or at best, is a compramise. So, I have just not focused on that as a means for tuning an alredy fine running car.
Good luke and by all means keep this thread alive. I think if it works out that it could be the best bang-for-buck deal on there cars.

whalebird

Sorry to all - my spelling suks

blown 944

My understanding..this is from someone currently working on this situation.

The Motronic is using 3 tables to interpret load and voltage related to such. When one is changed the whole voltage curve changes dramatically. Kind of like trying to keep a nice warm shower going when there are 3 other showers going in an old house. Easy to do with one but not so much with 3 at eh same time.

To get a smooth transition between a MAF signal and what the DME wants requires a lot of time adjusting all 3.

MrTom

This whole thing has me hooked. Any reason why you chose the Ford MAF? I have a collection of others I would like to give a whirl at. I am going to wait for the snow to melt and have at this. Will the DME run off a wideband okay? I know volvo LH2.4 gets pissy.

I am going to go back and read this post 9 or 10 times just to soak it all in.

MrTom

Thumbnails are not working man. Once you pulled the IAT out, where did you wire it back to? I am sorry if this is shown, I cannot see the pics well lol.

I find it amazing that the ECU has the ability to read a MAF. In my mind the car just would not run. I commend you greatly for this man seriously.

Edit:
I have a spare JAW wide band I would like to use. Does the LC allow you to adjust the AFR? Did you use the wideband as a means of baby sitting your AFRs or did you tune this at all?

http://www.14point7.com/JAW/JAW.htm

ScoobySteve

I've been a bit of a Motronic aficianado for a while.

In many ways, most of the operating principles of modern engine management are based off things pioneered in the Motronic. So here's a bit of information I've collected from various sources over the years about how it works and issues that there might be related to the adaptation of a MAF to a system originally designed for AFM. I'm by no means an expert but I do like to think I have a knack for understanding these types of things and my information sources are fairly varied. Below I'll note somethings I have a fuzzy understanding of with a [fuzzy].

Despite its ground-breaking nature, the Motronic system is fairly simple, from a logical standpoint. The actual code that runs the engine is less than 4k. That's tiny. Consequently, understanding how it works is not brain surgery.

The simple way to look at the fuel side of things is a chain of calculations. Think of it like a big flow chart where the end result is the injector duty cycle. The first thing that happens is a very basic calculation involving RPM and the AFM reading [fuzzy: I believe this is hardcoded]. From there the ECU looks in the main fuel table (unless it's at WOT or idle). This table has rpm on one axis and load on the other. AFAIK [fuzzy] the load determination is also based on the AFM. It finds the appropriate spot on the table and adjusts the previous calculation. So the fuel table is not a table of absolute numbers, but a set of compensating factors. From there the ECU checks the IAT (which is also a table) and compensates again. Then it checks the O2 sensor (if so equipped) and compensates again based on yet another table of values [fuzzy: though this one may be based on just rpm]. So at the end of this you end up with a duty cycle that's the result of a bunch of calculations.

Here's where the trouble starts. A MAF and an AFM do not measure the same thing. A MAF (more or less) measures the number of molecules passing by it per unit of time. The AFM measures the volume of air passing it by it per unit of time. That's a problem because at different temperatures the same volume of air can equate to a different number of air (and more important, O2) molecules. The guys at Bosch knew this and incorporated a temperature compensation system with the IAT sensor. This is unnecessary as the MAF auto-corrects for temperature. The problem is that you need to deal with this when going to a MAF because if you don't, you'll end up with a car that runs great at one temperature and not so well at the extremes.

There are really two ways to properly deal with this. IMO, the most "elegant" solution is to edit the IAT correction map in the ECU to all 0s which will result in it basically ignoring the IAT sensor. The other solution is to just feed the IAT sensor input a constant reading.

The second set of problems with fitting a MAF is finding one that both has the right shape curve to be similar to the AFM and is scaled properly. Most of the piggy-back devices do exactly this. In electrical engineering circles this is often referred to as a transfer function. A 0-5V (in our case) signal comes in and it "maps" it to an output that closely resembles what the AFM would be sending at the same airflow levels. The original poster had an interesting idea of finding a MAF whose output was close enough to the stock AFM (at a given temperature). That's one solution. The other is to use a device similar to the SplitSecond ARC-2 -- which achieves largely the same result -- it lets you change the output curve until is closely resembles what the AFM would read. Probably the best unit for this is the one sold by Lindsey Racing which just lets you directly edit a "translation" table.

Now, there's also another option. Some of you may be aware that on the late DMEs the AFM has a "straight" output as opposed to the curved output of the earlier AFMs. Did Porsche completely redo the ECU to account for a new AFM sensor? Nope. The ECU has a correction map. It converts the linear output of the later AFM to look like the non-linear output of the early AFM. Apparently the Maxtronic editor [fuzzy] gives you access to this translation table so you can adjust it for whatever MAF you want.

Lastly, there are some other considerations. One is the sample rate and sensitivity to turbulence. I don't know what frequency the DME actually samples the AFM at. BUT, I do believe the MAF actually has a more smoothed output (at least the physical device) so I think it should work better. I would still apply a simple low-pass filter to help reduce noise. Secondly there is the issue of the DME's input impedance and general electrical compatibility, but I think that's fairly minor.

If I were to do this (and the tinkerer in me wants to) I would start with getting a good tuning / measurement platform. That means Maxtronic with the live tuning board and a wideband O2 sensor. Beyond that, I think the biggest consideration is just finding a MAF that is physically compatible and properly sized for the HP being made (it's not a bright idea to use a MAF that at WOT is only outputting 1.5V). With the proper tuning of the DME I think you could get just about any MAF to work correctly (and by work correctly I mean be tunable to deliver your selected A/F) without serious idle or other drivability issues.

Lastly, there have been discussions about how much power MAFs make and drivability issues. As far as power, the AFM box causes a lot of turbulence and pressure required to keep it full open is not trivial. With the system properly tuned it *should* make more power. There is a reason why AFMs are no longer used on new cars. The problem is that most piggyback devices represent a step backwards in fuel metering / tuning accuracy. For drivability (somebody mentioned cold-start idling)...I can't think of any technical reason why this would be the case. The DME uses a coolant sensor to determine cold-start enrichment. This enrichment is independent of the air metering. The only thing I can think of is that some MAF sensors themselves might not like the extreme cold and might have accuracy issues...but if that's the case it can be tuned for in the IAT correction map.

Anyhow...just my thoughts (from the last few years).

V2Rocket

thanks for that info

barks944

Scooby Steve: You have made an almost identical set of conclusions to what I did. I'm from the UK and ive been doing some research on doing the same thing. The route I was thinking of going down was to use a pic chip and some ADC's and DAC's to read both the MAF output and the IAT output and send back a signal to the DME that would essentially remove any temperature variance from the MAF signal. I was also thinking that it would be the extremes of temperature that the MAF setup would struggle with, hence the need to compensate for temperature.

My car is a 1984 model N/A which is why I couldnt do any changing of the DME code, however since then ive got my hands on a late model DME. Maxtune looks interesting, anyone here running it on a lux? It would be a much more elegant solution to modify the DME maps for the MAF and would be great to be able to do on the fly tuning.

Rogue_Ant

Although the code footprint is small, dis-assembling it isn’t a trivial matter. The DME was programmed completely in Assembly language, and even with the small footprint you looking at close to ~800 lines of code. Furthermore you have to try to figure out the code without any of the source-code comments or variable names.

The DME uses a lot of factors to get injector duty. The Base Pulse Width calculated by the DME will produce a ~14.7:1 air fuel ratio. That BPW then gets adjusted by a bunch of other inputs to derive the final injector duty.

Basically the AFM is a volume sensor and a MAF is a Mass sensor. The DME needs the input of air-temp, and barometric pressure to properly determine air mass entering the engine. Bosch got around having a true barometric input by just assuming the DME is at sea-level, unless the altitude compensation module tells the DME that the vehicle is over 1000M in elevation. At this point the DME simply subtracts ~6% from its fuel delivery.

When converting to MAF, you need to either flatten the DME compensation on chip or just simply provide the DME with a static voltage in place of an IAT. However the IAT is also used for other things such as warm-start enrichment, idle control valve duty-cycle and temp based ignition retard, I believe. So this truly needs to be done on chip instead of a static voltage signal. Also the altitude compensation needs to be removed or disabled.

I don’t believe any MAF or AFM has a truly straight output vs flow. They are all curved. Do you have the transfer function for this AFM?
I don’t think the Maxtronic allows user-access to the transfer function. Need someone with Maxtune to confirm.

The DME samples the AFM input at roughly 88hz (turbo DME). Impedance shouldn’t be a concern.
However the MAF does not have a ‘more smoothed output’ compared to the AFM. And this is a problem. Even if the MAF sensor follows a similar curve to the AFM, the voltage output by the MAF will be much more ‘turbulent ‘. The MAF sensor is effected by individual pulses produced by each cylinder opening and closing. This causes the signal output of the MAF to be very 'jumpy'. A software filter in the DME is necessary in order to do a MAF conversion properly.
Furthermore the AFM has a sort of built-in enrichment for changes in throttle opening. When opening the throttle the AFM swings a little far then settles at the appropriate position for current airflow. This provides the DME with a little bit higher fueling requirement during throttle changes. When converting to a MAF, this extra fueling doesn’t happen, causing poor tip-in throttle response. So changes need to be made in the DME to provide acceleration enrichment that isn’t dependent on the AFM over-swing.

With a decent piggy-back you can probably get a lot of different MAF sensor to ‘work’. However to get it to truly work properly requires internal changes to the DME code, not just adjusting the DME maps.

Cold-start the DME uses coolant temp for enrichment. And drivability issues are a result of the issue I mentioned earlier.


-Rogue