rage2

Yea, but it's not instantaneous... which is what Tomas is saying. On the hondas and on my SDS, the temperature readings seem to lag by 1-2 seconds. Even the TEC3 I tuned has this behavior.
Not really pre-determined, more like already factored in during tuning. When you dial in the fuel tables, you're already factoring in the cooling effects. As for heat soak, you need pretty serious heat soak to vary the temperature enough to skew your AFR... heat soak happens over a period of time (ie on the track), so the temp sensor sees it too. It's when there's real bad heat soak (ie stopped in traffic after a hard run) when you need to make use of water temperatures fuel corrections.

To deal with the lag of the temp sensor, the ecu "guesses" temperature rise as boost rise enough so this does not happen to compensate for the slow responding temp sensors. Like Tomas said, a hard hitting turbo can change air temps by 50C in under a second.

Tomas L

The full load signal only depends on throttle opening (65° as you said), there is no boost relationship. This information comes from the factory workshop manual so it should be correct.

As Rage2 already said, the exact response time is not that important, it's the fact that are quite slower than the changes the are supposed to measure. Because of this it's not possible to depend only on the temp sensor, the EFI has to figure out the actual situation, not just rely on the sensor readings.
Temp sensors have a time response that is logaritmic, if you instantaniously change the temperature in the media that the sensor measures, then the first 10% change in the sensor output will come relatively fast, the next 10% will take longer and the last 10% will take forever.
According to Bosch datasheets their fastest temp sensor has a response time of less than 5 s. This is defined as the time it takes for the sensor to show 63% of the temp in the media after a change and with a gas speed of 6 m/s.
A MAP sensor has a response time of 1-10 ms. These sensors doesn't have the same logaritmic behaviour so the time measured is the time between 10% and 90% measurements.
A HFM5 MAF has a response time (to 63%) of less than 15 ms.

To compare this, one engine revolution at 6000 rpm takes 10 ms. Both the MAP and the MAF have a response time in that magnitude. A MAP sensor has a slight advantage in response time and due to the fact that it measures closer to the cylinder which will give it even more speed advantage.
The great advantage of MAF is that it directly measures air mass flow into the engine. Since we know that we want to mix 12.5 kg (or lb) air with 1 kg (or lb) fuel we can calculate the amount of fuel needed. This means that if you know the flow caracteristics of the injectors, you can program a MAF system that, in theory at least, will give the correct A/F ratio without any adjustments, when mounted on an engine. In reality there may have to be adjustements due to engine type and because of tolerances in the components. The bottom line is that you can sell a MAF system that is pretty much bolt on without any adjustments (fuel part only, ignition is another story) regardless of what modifications your engine has, a MAP system has to be tuned for each modification.

Does this make sense? It sometimes happens that I read my own post the next day and don't understand it...

Transaxle

I know that part of the manual - but my measurements were different. I just put a voltmeter to the FT signal as described in the manual and watched it when I drove the car. The impression was that it needs both - at least 65° but also a certain boost.

Transaxle

Btw....you may be interested in some pictures of a small adapter I built to make measurement a little easier...

phlip

Wow, you made a breakout box for 951. That is way cool, I am a tech in Dallas Tx and I have been wanting something like that for a long time.
How much to make me one?

eastendr

Nice breakout box !

I think this is one of the most interesting threads that I have read for a long time.

My own experience consisted tuning the Marelli and Pectel management system on the Ford Cosworth 16v 2.0L engine. These are native MAP installations.

I'm still waiting for my MAP II Kit (nearly a year now) to get on with tweaking my 951.

Although I've got the hardware diagrams for the DME/KLE somewhere, is there any published information of the map locations in the eprom ?

What software are you guys using to manipulate the map and other parameters ? (The Pectel software is excellent, what's the equivalent for the DME ?)

How big is the boost retard map in the KLE ? Is the knock sensor correction data available ?

Has anyone located the ATS in the inlet manifold instead of near the (AFM) inlet ? Is the ATS correction analog or mappable ? Is there any TPS speed related enrichment ?

I agree that the disassembled program should be available for what is now an 'antique' EMS. I'd certainly be interested ...

Tomas L

That seems strange, if that's the case, the Porsche test procedure would be useless. Are you sure it's not just a time delay?

Check out the Motronic Editor at:
http://www.mskar.org/
You may have to write your own spec files to define where the maps are located.
As you may know there are three different version of each map, which is used is depending on code plugs outside the DME.

Version 1
Fuel idle 1538h
Ignition idle 13ACh
Fuel part throttle 148Ch
Ignition part throttle 12CAh
Fuel WOT 1544h
Ignition WOT 1376h

Version 2
Fuel idle 18A2h
Ignition idle 17C8h
Fuel part throttle 17F6h
Ignition part throttle 16FAh
Fuel WOT 18AEh
Ignition WOT 17A6h

Version 3
Fuel idle 1ABAh
Ignition idle 19E0h
Fuel part throttle 1A0Eh
Ignition part throttle 1912h
Fuel WOT 1AC6h
Ignition WOT 19BEh

Observe that these addresses is not the start of the data area, first comes information about the size of the map, then how the axes are scaled, then comes the data.
The revlimiter is at 1171h.
The KLR is a darker area were it is much harder to find information.
This is partly due to the fact that the first 4kb of the KLR:s memory is located inside the microcontroller and therefore very difficult to read. All engine specific mappings should be in the outside EPROM but since most of the program code is unaccesible it's not possible to diassemble the code.
Good question, there are a lot of maps in the DME code (try and search them with the Motronic Editor), but I've not yet seen any information what their purpose is.
No, the TPS is connected to the KLR and is used to control boost as a function of TPS and rpm. The only thing that the DME sees is the full load signal that the KLR sends to the DME when the throttle opening is more than 65°.
I've been curious about this. My APE chips has the maps in different location and even has a different number of maps compared to the stock chips I've seen. I was thinking that there may be some version of the stock chips that used this configuration and that APE had used that as a base.

Tomas

Sam Lin

Roots-type blowers didn't die with anyone, they're still very widely used in industry, they've got their downsides, but within their limits they're a very cost-effective device to get pressure.

Sam

Tom M'Guinn

Transaxle,

Where did you get the connector for the breakout box? I've made a makeshift box, somewhat similar, but gave up trying to find the factory connectors. Thanks

Pauerman

Thomas,

Do you have any experience with the "Motronic Monitor" offered by Atlantis?

The images of the analyzer look amazing! I see that the ROM Info page of the demo shows an option for 24 or 28 pin chips - is this program only compatible with 911 Motronics?

I wonder if this is similar to the Motronic Monitor offered by Vitesse.

Vic

Tomas L

I've tried the Motronic Editor a little a while ago but not that much so I'm no expert.
It's compatible with the 951 but you may have to write your own definition files describing where the data maps are located. In the last version I've tried you needed to do that. I think Bengt Sweden has written definition files, he may be willing to share them.

No, this program only displays the maps in the motronic code and allows you to edit them in your computer. Then you have to program an EPROM with your edited code and put in your car.
The Vitesse program AFAIK displays engine data from the motronic as the car is running but you cannot change anything. Correct me if I'm wrong John.

Tomas

MildMax951

Would any one be interested in the editor I wrote for the 28 pin DME
for the 951 only ie it isn't configurable.

here's a screen grab



A Tomas said this is purely an editor ie you have to read the contents of your DME eprom , load it into the editor, change it and save the file which you then have to write back to the eprom so it is a fairly slow process. I wrote it more for fun and to see if I could decode the data structure at the beginning of the individual maps, the throttle posn's and rpms you see are not hard coded they actually come from the datastruc before each map.

That said when I had a custom chip made for my car on the dyno we didn't use it at all! The Tuner had a real time ROM emulator and I just gave him the actual Offsets of the map we wanted to work on. He twiddled a few values to check we got a change in the torque reading on the dyno confirming we were in the right location.

Marc Skarshinski's editor that Tomas gave a link to is far more comprehensive and configurable.
I'm offering mine with the source code (written in VB6) just for geeks like me to play around with (i would include the std 28 pin DME dump file)
and see what values the factory used where. IF YOU USE IT TO TUNE YOUR CAR AND DAMAGE OCCURRS I TAKE NO RESPONSIBILITY

if you look at my code you will seet that my Offsets agree 100% with the ones that Tomas posted.

Transaxle

Wouldn't it be a good idea to develop such a motronic editor in public domain? e.g. under GNU licence?

I doubt that there is a comercial market for these cars - so why don't put all experiences together?

For the breakout box : I destroyed a 944 DME and a 944 enginecable to get the connectors. I don't know how to get these from Bosch seperatly.

I can't answer all questions in detail now since i'm on the way to the F1 race at the nürburgring - 'll be back on monday...

MildMax951

Transaxle, enjoy the race you lucky man!
Come Sunday I'll be glued to the TV

Take this up again on Monday

fast951

Tomas, you are correct. The Motronic Monitor (MoMonitor) displays the values as seen or processed by the DME and KLR. However, in order for the MoMonitor to work, you must use the chips supplied with it, as the have code to work with the MoMonitor box.
We will be releasing the Motronic Editor, which allows you to view and modify the data in "live" mode. Same as a standalone computer.