1985 BMW 745

[mattias]

Your box.com links go to a "login required" page, either fix that or better yet use the built-in reporting tool in VemsTune (see Help menu) which is made for this particular purpose, even with possibility for others to add files and comments.  Too bad other ways can't be forbidden..

[DevInAz]

Thanks Guys I uploaded a log and config through the sharing center, Here is the link.

http://vems.hu/vemstune/sharingcenter/reports.php?cmd=view&key=k1t4Ag

Gunni - my target map is the same that comes with VEMS.

http://www.box.com/s/9d2f58805546f979901b  - log
http://www.box.com/s/238460568ecdc0095fd2  - config

[GintsK]

You have some details what could lead to bad mileage.

1) lumpy lambda signal and uneven cylinder power. It can be because uneven air distribution, uneven fuel distribution.... Check your injectors, air leaks...
2) You have crazy warmup curve.
3)For me VE map do not look like from tuned engine. Lambda do not match target in your log file. You can use more axis where engine spend most of time.
4) For this old skool engine ignition advance at cruising and off-boost areas are retarded.
5) also you do not want to trigger any acceleration enrichment at slow throttle movements. It just eats money.


Crank VE=20 ?!

[DevInAz]

hey GintsK,

1) The Fuel injectors where rebuilt and flowed a few months ago and I have a brand new fuel pump that is delivering stable fuel pressure. I haven't done a compression and leakdown on the engine yet.
2) I'm not sure what else it should look like... I added fuel to the warmup in each section over the winter until it started quickly each morning. The warmer out it got I started tuning the higher temp points. Now thats its about 60F every morning it doesn't use much warmup anymore.
3)  For the fuel map I drove at stable speeds and set everything to the target lambda then once everything seemed close I used the tune by statistics until it didn't change anything.
4)  For the Ignition table I used a map Mattias told me to use. He told me I could probably get more timing but since i'm not really a tuner I don't know how to tell how far is too far for cruise timing so I left everything alone. Mattias said his map was for a m20 which is about 1 liter smaller than this engine so I know it's retarted a bit but i'd rather not kill my daily driver because of my lack of tuning ability.
5)  I'll tune the accel out a lot more.

Thank you for sharing your knowledge!

Where do I find the Crank VE? what would you expect this to be closer too?

[DevInAz]

Here are my updates with a new config:

http://vems.hu/vemstune/sharingcenter/reports.php?cmd=view&key=k1t4Ag#comments

[GintsK]

You have new config there. Try it out. No dangerous changes.
I change many things including spark and lambda tables. Spark at boosted regions is retarded anyway. But this is not for distance changing.
 IIRC my advance values was like 22deg at 1.5bar@3800 with RON98. And of course more advanced at lower boost.

Gints

[DevInAz]

The new config from GintsK has been loaded. I put about 20 miles or so on this tune at freeway speeds while running a log and then ran tune by statistics. It pulled fuel out on the top end and added in a decent amount at cruising speeds below 60kpa.

Here in Nevada our highest fuel at the pump is 91AKI which I think is somewhere are 94-95RON. Since I don't have any knock detection or anything how can I tell when too much ignition is being used? Should I do a drive and pull some plugs? I can't find any reading material about tuning techniques for this. People only seems to talk about tuning when making power.

I had to change the Crank VE down to 45 or it takes a ton of cranking before the engine starts.

Tomorrow i'm going to be installing my 6 COP from a e65 745. No cam sensor will be run for a while so waste spark will be the method of choice.

[GintsK]

I would search for some additional misconfigurations or failures. I touched just lambda map a little bit at 1200rpm cruising  region. So your fueling should stayed as before.
also crank VE=45 tells somewhat is wrong. E.g. injector dead time settings way off. Or leaky injectors or... This value usually is close to table value at low rpm 100kpa VE.

Regarding knock: google for detcans. You even can find useful info in this forum.

Gints

[DevInAz]

After looking at the target VE table changes thats why the regular VE table was changed.

My crank VE was changed again down to 20 because my car wasn't starting without giving the engine more air with the gas pedal. it now starts well again.  I can't explain why it does this but for some reason at Crank VE 20 it starts correctly.

I wiring in my COP today and the car is running well. It is using waste spark.

[lezsi]

A few years ago I was playing with these ancient injectors and they seemed to be very hard to tune with VEMS (although it was the low voltage pwm flyback strategy those days). 
Then we were going for modern injectors double this size, and everything got much better suddenly: idle quality, throttle response, economy.

If my memory don't fail, these engines were running very lean in stock form, I mean 1.03-1.05 lambda from idle to cruise, and even on low rpm/boost, and hardly getting into the 0.8x range at high rpms (and 0.6-0.7 bar of boost), but relying on knock sensors.

Looking at your lambda values in the log file it seems very sluggish to me, the "resolution" in time seems very poor, I used to see much smoother graphs almost like your MAP or RPM signal looks like, without the lean glitches you have very often.
It should be very hard to tune VE based on such a feedback, and I'd forget automated tune based on such logs, but correct this problem first.

[fphil]

at your lambda values in the log file it seems very sluggish to me, the "resolution" in time seems very poor, I used to see much smoother graphs almost like your MAP or RPM signal looks like, without the lean glitches you have very often.
It should be very hard to tune VE based on such a feedback, and I'd forget automated tune based on such logs, but correct this problem first.

Looking at your lambda values in the log file it seems very sluggish to me, the "resolution" in time seems very poor, I used to see much smoother graphs almost like your MAP or RPM signal looks like, without the lean glitches you have very often.
It should be very hard to tune VE based on such a feedback, and I'd forget automated tune based on such logs, but correct this problem first.

Nice hint. Indeed no performing control loop or tuning from noisy data.
I was looking for logs which would show the expected standard for noise level when running Vems (e.g. lambda lean glitches )
Do you get some Lezsi?

[lezsi]

at your lambda values in the log file it seems very sluggish to me, the "resolution" in time seems very poor, I used to see much smoother graphs almost like your MAP or RPM signal looks like, without the lean glitches you have very often.
It should be very hard to tune VE based on such a feedback, and I'd forget automated tune based on such logs, but correct this problem first.

Looking at your lambda values in the log file it seems very sluggish to me, the "resolution" in time seems very poor, I used to see much smoother graphs almost like your MAP or RPM signal looks like, without the lean glitches you have very often.
It should be very hard to tune VE based on such a feedback, and I'd forget automated tune based on such logs, but correct this problem first.

Nice hint. Indeed no performing control loop or tuning from noisy data.
I was looking for logs which would show the expected standard for noise level when running Vems (e.g. lambda lean glitches )
Do you get some Lezsi?

OK, check this log:
http://quasar.dynaweb.hu/~lezsi/vems/v3.3_n002211-2010.07.31-15.59.42.vemslog

This is not a fine-tune or something, just a piece of log from the past which I could came up with in a minute 

The point here is lambda reaction, which is very smooth in my opinion. It also tends to be a bit lean/rich when hitting the throttle but it's just playing around with accel. enrichment. 
Still, the only point in log where the lambda goes really high (> 1.2) when it shows a deceleration phase, right after injection cut-off. These parts shows the lambda latency of the system at those revs/load.