Marelli 15M ECU - Understanding the correlation between the different parameters

[Meinolf]

Hi,

recently I have been trying to understand the different maps, their interaction and the resulting fuel injection values on my V11/15M. Beard and Paul have provided valuable input, but I still felt that there's a lot of ground to be covered. I have been using a Lambda datalogger, but found that just changing the values of the base fuel maps falls short of the mark.

So, I decided to built a simulator for the ECU and its environment consisting of sensors and actuators. The purpose is to isolate the influence of the different inputs to the ECU and their impact on the fuel injection duration.

Using a spare ECU and the cabling I built this simulator.


The intention being to live up to highest standards the entire construction is CAT V compatible.


Accordingly the main power switch is rated at 35A.


The ignition switch could be down-sized


Some load resistors and potentiometers connected to the wiring loom.


Using RPM-Emulator (another piece of software written by Beard) to generate the timing signal. I'm picking up the signal generated by the software on the audio-out jack of the notebook. The signal looks good on the scope, but massively interferes when connected to the loom. Need to switch to an shielded cable here.


GuzziDiag connects to the setup following the SOP. Connect powersource, turn on main switch, turn on ignition. Values are picked up, errors are recorded in the ECU log.

Eventually I hope to be able to clarify
- what is the purpose and influence of the different table entries in the BIN for the same parameter. Engine temperature is one example.
- when and how long is the acceleration map active
- what is the duration and influence of the cold start map
- ...

Cheers
Meinolf

[fotoguzzi]

Cool!

[guzzisteve]

Now that is wild.

[Chuck in Indiana]

The main switch looks like it would power Frankenstein's monster.  ;D :BEER:
Nice.. ;-T

[Meinolf]

Hi,

several revisions later the simulator is up and running stable. As you can see I am still using the vanity free-approach, form following function.

I spent quite a few hours measuring the interactions of the different parameters and am slowly contributing to the knowledge base.



Cheers
Meinolf

[lazlokovacs]

that is truly brilliant!!!

[Unkept]

Not sure how I missed this, fantastic work! You are my hero of the day.  ;-T ;D

Can you simulate a map for me? My bike is going from Race Ti fuel map, stock crossover and airbox with Ti mufflers to ??? fuel map with pod filters, FBF crossover, and Ti mufflers...

I have already prepared with earplugs, next will be the fueling adjustments.  ;-T

I understand that it's mostly to find and identify which functions change/affect what in GuzziDiag though. Love the project.

[scra99tch]

Wow nice work. I will be watching intently becuase I have been messing around with my acceleration map on my 15m. 

I have that ecu on my jackal and quota.  The jackal would get 50mpg and the quta only 28-32.  I ended up getting close to 45 on the quota by changing the acceleration map.  But developed a very sluggish acceleration which is intermitant.

I posted these maps to show the difference between the two bikes.

http://wildguzzi.com/forum/index.php?topic=59717.msg982071;topicseen#msg982071

[scra99tch]

Oh and the plugs look great.  Not too hot and no major fuel smell running down the road. 

[leafman60]

Wow, talk about a committed enthusiast !

This is interesting to follow.

[Meinolf]

Hi,

Can you simulate a map for me? My bike is going from Race Ti fuel map, stock crossover and airbox with Ti mufflers to ??? fuel map with pod filters, FBF crossover, and Ti mufflers...

I understand that it's mostly to find and identify which functions change/affect what in GuzziDiag though. Love the project.

as to the first question, not really. I am measuring injection signal duration with a scope at whatever settings the MAP and sensors have, but have found discrepancies between the times shown by GuzziDiag and the measured ones. I need to validate this on the motorcycle itself. The actual injectors may well have different times than the load resistors I am using on the testbed.

And yes, the idea is to use the testbed to fill in the blank areas in the understanding of how everything interacts. And be sure of one thing, everything is interacting :-)

Cheers
Meinolf

[Meinolf]

Hi,

I have that ecu on my jackal and quota.  The jackal would get 50mpg and the quta only 28-32.  I ended up getting close to 45 on the quota by changing the acceleration map.  But developed a very sluggish acceleration which is intermitant.
http://wildguzzi.com/forum/index.php?topic=59717.msg982071;topicseen#msg982071

I looked into your posting. Did you change the acceleration map or, following Paul's comment, the start enrichment?

One of the recent findings is that the start enrichment map has new axis. One is engine temperatur (that's unchanged) and the other one (which is new) is # of revolutions since start. The default value is 3984. Divide this by rpm and you have the time in seconds a value in the table is used (download the latest XDF http://www.von-der-salierburg.de/download/GuzziDiag/xdf/15M_GuzziDiag_V2.28.zip, it has the new values).

This was an outcome of measuring the influence on injection duration at varying engine speed. The graph below shows 6 curves. X-axis is time, y-axis is injection duration. The top curve is at 1krpm, the bottom one at 6krpm. The ratio is 1 : 1/2 : 1/3 : 1/4 : 1/5 : 1/6.



The values also show that the impact of the start enrichment map is very limited. Even when idling at 1krpm the mixture enrichment effect disappears after ~30s. Taking into account that the values in the table for temperatures above 10°C are very low you will see a measurable influence only quite briefly.

The acceleration map only kicks in when opening the throttle rapidly. The open question yet is, what exactly is rapidly? I haven't figured out a simplistic and repeatable means of opening the poti with a defined speed. But, the measurements have already shown that what I would consider to be a slight throttle opening in a average driving situation has no impact on the mixture.

Cheers
Meinolf

[scra99tch]

I tried the start enrichment and leaned it out a bit only result was a little hesitant to catch when colder 40F or below.  

Before messing with the start enrichment my bike would spew white smoke and smell of gas until it was fully warmed up.  Even then my friends would say it was running way to rich down the road.  I think I'll adjust the lower end of the accel enrichment a little and see if that fixes the hacking and sputtering when I take off from a dead stop.

[Meinolf]

Hi,

the resistance of the NTCs used to measure air and engine temperature is between 0.1-100kOhm. Initially 10kOhm potis were used in the simulator, but this cut off the selectable temperature range at ~4°C. 100kOhm potis and switches were added, now the entire temperatur range, beginning at -20°C, can be dialed in with sufficient accuracy.



And as I selected a bare-metal approach (influenced by vanity free, a buzz word in the data center world) an old voltmeter was added to the setup.

Cheers
Meinolf

[Moto Fugazzi]

Very cool.
I'd be interested to see the results of how air and engine temps affect fuel and timing. My V11S seems to run leaner (according to my AFR meter) on very hot days.
Ken

[Meinolf]

Hi Ken,

I've looked into this some weeks ago.

The findings are that engine temperature indeed has a significant influence on the mixture. My measurements show that injection time decreases by 64% when the engine temperature moves from +5°C to +125°C. My apologies for the text being in German, the discussion mainly happens in the german www.guzzi-forum.de forum. Which anybody who cares to register can probably follow using Google Translate. Here's the link to the discussion on GuzziDiag http://www.guzzi-forum.de/Forum/index.php?board=76.0



The influence of the air temperature on injection time is much lower. Which is, looking at the correction factors in the respective trim tables, comprehensible. The values are flat starting at 22°C.



This was discussed in more detail here http://wildguzzi.com/forum/index.php?topic=59717.msg1006600#msg1006600

Cheers
Meinolf

[Chuck in Indiana]

That would certainly explain the value of the brass temp sensor holder..

[Unkept]

That would certainly explain the value of the brass temp sensor holder..

 :+1

[molly]

64% is a big swing it is only 25% on the 5AM ecu.

[Mayor_of_BBQ]

I don't understand a single thing you guys have said, but I'm glad someone with the know-how is cracking into this and thanks for raising the bar!  ;-T ;-T

[pyoungbl]

If I understand this discussion correctly, it seems that air pressure and air temp are fairly minor variables or maybe better stated as variables where the range might not make much difference.  How does engine oil temp fit in?  My experience with FI on a Ducati was that engine oil temp made a huge difference in fuel mileage.  It was common to block the radiator in cold weather or in rain so the engine would get and stay up to at least 160F.  If the engine temp fell below 140F the fuel mileage could drop by as much as 10 mpg.  This happened on both my ST2 (water cooled) and Multistrada 1000DS (air/oil cooled).  I guess I'm looking for confirmation that my Stelvio (air/oil cooled) is subject to over cooling too and would benefit from a thermostat in the oil line.

Peter Y.

[molly]

For some reason ecu designers like to add excess fueling in cold temperatures and to cold engines when previously with carbs this was taken care of with a simple choke lever or even with early fuel injection a high idle lever.  Now in the name of stricter emmission controls more fuel  is burnt then is necesssary.
I have trimmed out a some of the excess on my 1200 Sport and the bike runs just as well in cold conditions and I'm saving at the fuel pump too.
The more sensors are fitted the more tables are added to the engine map and their interaction gets more complicated hence people scratching their heads trying to understand how it all works in harmony, I'm not sure it does at times.

[Meinolf]

Hi Peter,

If I understand this discussion correctly, it seems that air pressure and air temp are fairly minor variables or maybe better stated as variables where the range might not make much difference.

Yes, that is my understanding as well. However, both can be set right, as the values in the 15M are way off. Both load points and values in the trim tables.

I guess I'm looking for confirmation that my Stelvio (air/oil cooled) is subject to over cooling too and would benefit from a thermostat in the oil line.

Sorry, I can't answer that one. I neither have a Stelvio nor have I worked on the 5AM ECU.

Cheers
Meinolf

[rocketcaptain]

Sorry, I can't answer that one. I neither have a Stelvio nor have I worked on the 5AM ECU.

Cheers
Meinolf

But I'll bet you have some good ideas on this one...  http://wildguzzi.com/forum/index.php?topic=68460.0
Any help or insights would be appreciated.

[Moto Fugazzi]

Hi Peter,

Yes, that is my understanding as well. However, both can be set right, as the values in the 15M are way off. Both load points and values in the trim tables.

Cheers
Meinolf

Will you be developing a map for the V11 based off of all of this data? I'd be interested in purchasing a that map. Paul and others have been helping me with maps for my V11S, but I really don't know how much to adjust in all of the different tables on Tuner Pro.
Ken

[Meinolf]

Hi Ken,

I'll gladly share my findings and my map (for free). As mentioned earlier, however, the mainstream and quite detailed discussion takes place in the german forum.

The reasons why I started looking into the details of the way the ECU works are threefold. I know Beard (the guy who cracked the protocols and disassembled the code of the different ECUs and wrote the GuzziDiag-suite) and Paul (who is the field engineer in this enterprise) and wanted to support them with the limited means and knowledge I have at my disposal. Secondly I wanted to learn something along the route, I am a seriously challenged person regarding electrics & electronics. Thirdly I started wondering why a 230kg, 80PS, 1liter V11-engine consumes more fuel than a 1400kg, 80PS, 1liter car during a comparable driving situation. The bike having the advantage of better aerodynamics also.

My conclusion on 3. was that Guzzi didn't really invest much time into using the capabilities of the ECU and the sensors. Many facts seem to support this. Such as useless trim tables, no fuel cut-off during deceleration or the acceleration map being used either half-heartedly or not at all. Which, I guess, is due to the time and money Guzzi could afford (not) to spent 15 years ago on fine-tuning the V11 engines. The result is like average temperature in a clinic - it's not optimal but not half-bad either.

Taking into account the possible variations of the engine, environmental factors, different fuel quantities, state of maintenance, etc an average setting of the maps seems ok. However, my target is to find an optimal setting for my engine. And whether that will be optimal for any other engine is the question. I doubt it.

So, you are welcome to use the findings (with a grain of salt) and apply it to your engine.

Cheers
Meinolf

[molly]

Here is how a Buell uses the various inputs to provide the final ejector time.
Taken from the Buell tuning guide http://www.ecmspy.com/tgv2/guide2.shtml which is one of the best articles I have read on FI tuning.

 Maybe a similar diagram could be applied to a Guzzi ecu if anyone ever figures out how it all comes together.

[Meinolf]

Hi Molly,

that's good feedback and a great link. Frankfurt, the home of the ECMSPY project, is only some kilometers away from my home.

As it happens I did start drafting a description of what is known about the 15M. Not even close to the quality of ECMSPY but what the heck, I'm just starting.

Here you go https://db.tt/ENvMSozh

Any feedback is welcome
Meinolf

[Moto Fugazzi]

Hi Ken,

I'll gladly share my findings and my map (for free). As mentioned earlier, however, the mainstream and quite detailed discussion takes place in the german forum.

The reasons why I started looking into the details of the way the ECU works are threefold. I know Beard (the guy who cracked the protocols and disassembled the code of the different ECUs and wrote the GuzziDiag-suite) and Paul (who is the field engineer in this enterprise) and wanted to support them with the limited means and knowledge I have at my disposal. Secondly I wanted to learn something along the route, I am a seriously challenged person regarding electrics & electronics. Thirdly I started wondering why a 230kg, 80PS, 1liter V11-engine consumes more fuel than a 1400kg, 80PS, 1liter car during a comparable driving situation. The bike having the advantage of better aerodynamics also.

My conclusion on 3. was that Guzzi didn't really invest much time into using the capabilities of the ECU and the sensors. Many facts seem to support this. Such as useless trim tables, no fuel cut-off during deceleration or the acceleration map being used either half-heartedly or not at all. Which, I guess, is due to the time and money Guzzi could afford (not) to spent 15 years ago on fine-tuning the V11 engines. The result is like average temperature in a clinic - it's not optimal but not half-bad either.

Taking into account the possible variations of the engine, environmental factors, different fuel quantities, state of maintenance, etc an average setting of the maps seems ok. However, my target is to find an optimal setting for my engine. And whether that will be optimal for any other engine is the question. I doubt it.

So, you are welcome to use the findings (with a grain of salt) and apply it to your engine.

Cheers
Meinolf

Thanks Meinolf! How do I obtain a map from you for my V11 Sport? I should have my bike ready to go next weekend and would like to install a new map asap. I also have an AFR gauge that I use to monitor how its running.
Ken

[scra99tch]

I am currently trying to modify my engine Temp Trim but it will not let me save the values I entered.  Does this have something to do with them being a Linked Scale?


And why does it say "nicht korreckt" on my Scale Engine Temp (8bit) table?