Tomas L

The disadvantage of converting the MAF signal to match the AFM signal is that you will not get any greater airflow capability.

To clarify my MAP comment. I think the disadvantages are greater for a MAP when compared with a MAF in a streetdriven car. With the DME/KLR I would not even consider MAP and even with an aftermarket EFI I'd still preferr MAF. My point here is that you can use a MAP sensor in addition to the MAF and use some information from the MAP sensor to correct the MAF signal in certain conditions.

Tomas

Chris White

“To clarify my MAP comment. I think the disadvantages are greater for a MAP when compared with a MAF in a streetdriven car”

I guess it depends on how you process the MAP data! The Tec3 uses a MAP along with a manifold air temp sender to quite accurately calculate the amount of fuel needed.

I am not sure why you feel the MAP based system is disadvantaged compared to a MAF system

Chris White

Transaxle

With a modern EMS like the TEC3 you could easily calculate the injection time from different parameters. With the old DME it is difficult. It uses the AFM signal as a load indicator to get the injection time at a certain engine speed. There is no real-time calculation. It is just a slow vintage 8051 ...

If you like to keep the motronic, the MAF path is easier. With a modern and fast EMS it is better to use MAP as all modern engines do.

The AFM is not a very limiting device in the engine. It is used in the 968 TS and TRS without modifications. If you use a compatible MAF it will not limit the maximum power under "normal" circumstances (3l, 18PSI, 7.000 rpm).

As said before, the DME is not able to do realtime fuel calculations. It is just able do calculate the index to a matrix or curve and to use that value to control the injection.

In the DME are 3 enties for idle, part-throttle and full throttle. Only the part-throttle is a matrix with engine speed and engine load as index. The full-throttle entry is a curve over the engine speed. It is so very easy to tune the engine without using the AFM/MAF signal at full load.

At our 3l 16V engine we can swap the MAF and the AFM without additional modifications. Why then use a MAF? Main reason for us was the fact that the AFM is a mechanical device that ist getting old and unreliable. Other advantages are the faster response, less restrictive airflow and lower weight of the MAF

Tomas L

I have written more detailed on this subject earlier so I will keep it short here, search my earlier posts for a more detailed description of MAF vs MAP.

Yes to make an accurate calculation of air mass flow into the engine you need to know airtemp, manifold pressure, rpm and volumetric efficiency.
The biggest problem is air temp, common sensors for this application has a response time of appr. 10 secondes. This may be ok in a naturally aspirated car but for a turbo where the temp rise can be more than 50°C in a second it's not acceptable. The other problem for a daily driven car is that VE may change over time, OEM:s with their more advaced EFI:s may compensate this but the normal aftermarket EFI:s will probably get out of tune with time.

I have not done extensive research so I may be wrong on this but I have yet to see a modern turbocharged car equipped with MAP in stock form. MAP is mainly used as a cheaper alternative to flow measurement.

This is not correct.
The DME reads flow from the AFM, it divides that flow with rpm to get what Bosch calls "load". In part throttle the DME has a map, rpm vs load, where it looks up what could be described as a sort of lambda value. It takes that value and multiplies it with load to get the amount of fuel to be injected.
The full throttle map is just a special case of the PT map, it should be considered as an extra column in the PT map. The calculation is still the same, load multiplied by map value gives injected amount.
The CPU is just an old 8051 but at 12 MHz, and with programs written in assembler, it's still able to do a lot of calculation during on revolution at 6000 rpm (0,01 s).

Since the AFM flow vs output curve is exponential it's hard to define the flow capability, but if you run air flow for a 3L engine at 18 psi at 7000 rpm through it I sincerely doubt that it has any accuracy in the measurements. If you have a good VE, an engine like that should deliver at least 450 hp at the flywheel. If the AFM really can handle that amount of flow with any accuracy then Porsche must have made a big error when putting it on a 250 hp car.

Regards,
Tomas

rage2

Actually, with boosted operations, every MAP based system has calcuations for that temperature rise when boost climbs. So unless ambient (outside) temperature changes 50C in a second, what you described doesn't happen. If it did, we'd have a lot of rich running motors . Oh, and it'd affect NA motors too.
After tweaking my AFR, I didn't touch it for a year and a half. I dyno'd again and got the same results. I do know that VE changes over time because of carbon deposits, etc, is theoretically possible, but I know all Hondas use a MAP system, and they do not suffer from VE changes this at all even with 300,000km on the clock. I've fully disassembled the Honda ECU programs, and they're from from being advanced EFI's hehe, no VE corrections over time from what I can see.
My SLK 32 AMG (3.2L supercharged) uses a MAP system. The 5.4L V8 Kompressor motors (CL55/S55/SL55 AMG) also uses a MAP system. I'm fairly certain the MB V12 twin turbos (CL600/S600/SL600 and CL65/S65/SL65 AMG) also uses a MAP system, never seen one to confirm .

Tomas L

Yeah, it's possible to reduce the effect of temperature changes by guessing the temp change from the boost change but that will not take the IC:s actual cooling in consideration. The temp rise will be very different depending on the IC:s cooling capability.

Don't hey have a closed loop learning capability? Most OEM EFI:s have had that for some time.

It will be easier to adapt MAP to a supercharged engine than to a turbocharged. The heating is more predictable and there is normally no IC (I don't know if the MB supercharged engines have IC).

Tomas

Transaxle

That is interesting. Do you have any detailed information about the algorithm?

My experience is based on an simulator which allowed me to see which addresses in the eprom are currently read by the CPU. That gave me the impression that there are 2 separate columns and one map.

The first column is used when the idle-switch at the throttle is closed. The cell in the column is selected by the engine speed. When the switch opens, the map is read. The cell in the map changes with the AFM signal and the engine speed (rpm). When the full throttle signal from the KLR goes to 0V, the other column is used. The cell again is selected by the engine speed.

When I change the values of the data, it seems to give me direct control of the injection time. I could not measure any significant changes by the AFM signal. The signal also does not change to much at full throttle (+/- ~0.3 V).

I was wondering why two special load situations (idle & full-throttle) are stored in separate columns. I would have stored all into a single map. Do you have any details about the ML3.1 program?

(Becomes a little OT now - but it's an interesting hobby - isn't it ? )

Tomas L

No, not really. It's just bits and pieces I've picked up here and there. I'm currently trying to diassemble this part of the DME code but it's a lot of work remaining.

By looking at the values for different loads in some part of the PT fuel map you can see that the fuel change is very small for a quiet large change in load. This gives the conclusion that it is not the map value alone that determines the fuel amount.
You can also try accelerating with full throttle from idle in first gear and then in fifth. In first gear you will not have any boost until very high rpm, and in fifth boost will come much earlier. If the DME was looking only at the full throttle map without taking actual air flow in consideration, then the engine would run very rich at 4000 rpm in first gear.

I think the division into three different maps has somethink to do with emission control. The DME will not run in closed loop at full throttle. And for the idle map they maybe wanted more resolution in the low rpm range.
But still they could have used the first and last row in the big maps for this.

Tomas

Tomas L

A little more OT. I was searching for info on Mercedes Kompressor models and found info on a 2000 230 SLK Kompressor.
It was using an Hot Film Mass meter but more what really turned my head was that MB still is using roots superchargers!!! At the risk of being flamed for an eternity... Didn't they die with the cavemen??? ;-)

Tomas

Transaxle

The "Full-Throttle" signal is generated by the KLR unit. As far as I could understand this signal is send when the throttle is opened > 65° and the correct boost level is reached. Until then the PT map is used with the AFM load.

It would be very helpfull to get a look at the disassembled code. The guys at APE must have a source version since they could relocated and resize the data tables. This should be set to public-domain for vintage cars...

Please don't mix the actual MB with the old MB. They don't build good cars any longer. They're specialized in building cars as cheap as possible and selling them as expensive as possible. They call it "shareholder value". They don't care about "customer value".

rage2

The cooling will be fairly constant over different engine speeds/loads/boost, which when you're setting up the 3D fuel map really takes care of already. When there's heat soak, that's when there MAY be an issue. That's when the Coolant temperatures fuel correction maps come into play. As for guessing the temp change from boost, there's no guessing, the compressor map is there for everyone to see fairly close what the outlet temps would be. All that doesn't matter anyways, as it's fairly constant like I said earlier, once you get a set point tuned, it's gonna be mighty close.
Yes, but that's just for part throttle. VE changes at anything over a preset MAP pressure (~700mbar) runs straight off the fuel maps, with no correction.
There's an air to water IC, which gets heat soaked even more than an air to air IC.

Point is, I've done WBO2 traces of both my MAP based cars, turbo vs supercharger, on the street, on the track, and I've never seen a huge deviation in A/F ratio from different temps, humidity, pressure, etc. This is from the most basic MAP based system (SDS) where there's not even a air temperature correction map for the end user to adjust (it's a fixed map) all the way to the ultra complex SLK's Bosch Motronic system.

rage2

hehe, the 2.3L Kompressor motor is actually from "old MB", the more reliable one. Motor's basically a supercharged version of the old 190E 2.3-16 .

As for MB being cheap, man, it's so true unfortuantely. I'd hate to own an S class out of warranty.

azmi951

i was driving a brand new S600 (not mine) like 200 miles, showroom new and the POS had its ecu go out. This caused it to think it was overheating and turn on every means of cooling......(read as the heater was on full blast in 80 degree so cal weather). The dealership said they replaced the ecu and it worked fine. I think it was a sensor or something little, what are the chances the dealer has an ecu for a new S600 in the parts room?

MildMax951

Keep going, your whole project as well as the DME internals makes great reading, I look forward to each new addition.

Chris White

The Manifold Air Temp sensor that comes with the Tec3 is a lot quicker responding than 10 seconds. I data log mine quite often. Also - the temp sender location is after the intercooler – anywhere else is useless.
Using a predetermined cooling data for an intercooler? That’s pretty inaccurate. Even our reasonably good sized intercoolers suffer from heat soak on the track. The heat transfer numbers can change dramatically based on ambient temps and humidity.
As for ongoing changes in VE – they are pretty small and I would even arguer that I doubt that they can be measured on a 951. A cheaply made NA engine may see some changes due to carbon deposits but a well tuned 951 motor does not get carboned up.