M3.3.1 413

Discuss Bosch (Porsche, BMW, Volvo, etc) tuning topics here. Request definitions, discuss parameters, etc.

Moderators: robertisaar, dex

Triangulé
Posts: 36
Joined: Wed Jun 06, 2018 10:34 am

M3.3.1 413

Post by Triangulé »

Hi guys,

We are building an E30 with an m50b25 turbo for track use.

The car is really too rich to run properly.

the origjnal ecu used an 609 chip and have work before with an 623 improved cheap for NA bought on ebay. Now we add the turbo we don't know if the ecu can work on the 609 and 623 chip or if it need the 609 absolutely.

are these 2 revision using the same maps adresse location ?

i will go step by step if you accept to help me guys
:D

is the 609 revision need a specific xdf file ?
Hairyscreech
Posts: 196
Joined: Tue Jun 20, 2017 3:19 am

Post by Hairyscreech »

They should both work just fine on the 413 as there are practically no changes to the hardware even between the 6 and 8 cylinder M3.3.1 Ecus. (we have seen a 6 cylinder ECU try to pulse the 2 missing Inj/Ign drivers if loaded with an 8 cylinder BIN).

The best way to be sure is to check out the BINs yourself using the info in the big development thread on here or post both of them up and see if someone has the time to take a look and verify they are the same.

Do you have the ability to read both the chips?
Triangulé
Posts: 36
Joined: Wed Jun 06, 2018 10:34 am

Post by Triangulé »

Well, i have open the 623 and the 609 (with the 413 xdf provided on the other big thread) both have been read from the Eprom directly, but nothing is similar. that is really strange, because the car was running good with the 2 chips on the same ecu.

I don't know what to think about that.
If there is nothing similar between the 2 files, the ecu should not work right ?
Triangulé
Posts: 36
Joined: Wed Jun 06, 2018 10:34 am

Post by Triangulé »

Ok i have spent some time on the car yesterday and the ecu is reacting well to the 623 chip.

The block engine is stock, original piston, head gasket. We just change the rods bearings.

We have started by a pressure of 0.35 bars and the car is running good.

Just some mistakes :
_ the idle target seems not working in the XDF, i have to adjust the idle with the IAC map.
_ AFR at the idle is good at the begening, but after a little time it start to become poor.
_ after 6000 rpm, the engine become very lean
_ on the 10 first percentage of the throttle, the AFR is lean (16). Is it coming from the "accl enrich" or the PT map ?

i have to adjust now the ignition.
I have to say i'm not really comfortable with that, it's my first mapping on a turbo conversion.
So i have started by remove 4 degree after 4000 rpm, and 3 degree between 2000 and 4000. I don't have any idea what is this engine able to handle. Probably safe enough for starting but far from ideal.

next week we will increase the turbo pressure to 0.5bars.
What is the max safe pressure for this set-up ?
Hairyscreech
Posts: 196
Joined: Tue Jun 20, 2017 3:19 am

Post by Hairyscreech »

The idle target will work if the IAC table is within the correct range for the engine, the M50 maps will be too much air flow for the M20 at idle, the IAC table only alters the IAC valve but the idle speed tables adjust ignition timing to alter idle as well.
The ignition is varied at idle around the set point in the map as a fast way to correct speed fluctuation and the IAC is varied to control the idle if the small tweeks of the ignition cannot do it, it is a slower response and was part of the reason I had issues with the idle bouncing on mine, the idle ignition was vastly different to the PT map so caused a bounce between the two.

What do you mean by your idle fuel becoming poor?

After 6000RPM? What is your rev limit set to and how lean are we talking? Could be a number of things.

For the last point it all depends what you are doing in that first 10% throttle.
If the throttle is steady then you are in the PT map, keep the throttle steady for about 10 seconds to be sure you are in the PT map.
If you are just cracking the throttle open and seeing a lean pulse then you are looking at the acceleration tables a s the culprit.

I would not go turning the boost up until you are comfortable with all the settings and have no lean areas, a friend may have just killed £2500+ of M20 bits by going too hard too soon. These motors can make really strong power but if the tune is off can also be molten/cracked scrap very fast.
If your boosting an M20 beware of head temperatures, get a sensor to monitor the metal temperatures around cylinders 4 and 5.

Going back to the chip numbers, you have one chip from a pre 93 US M3 and one chip from a post 93 M3, one has EWS and one does not so the code will be offset as shown in the different XDFs and the Excel sheet showing the bins.
Triangulé
Posts: 36
Joined: Wed Jun 06, 2018 10:34 am

Post by Triangulé »

Hi Hairyscreech,

First of all, thanks for the help and time you take to reply me.

I see i have to be more precise.

The car is an E30 but engine come frome an E34 from 1992.
We are in Europe, so all maps are european version.
The engine swapped from the E34 is the M50b25 (195 Flywheel HP), with for sure the ecu (413) coming from this car. There is no M20 anymore in this car.

When the M50 was an NA engine, the idle was also bouncing, but i think it probably come from the IAC rubber pipe, where the connector was broken with the time. Now it's solved, we glued it.
The idle is now really stable, when we release the pedal, the AFR is around 14 but fall to 15 and more after 5 or 6 secondes.
Maybe the ecu is switching from the PT map to the idle fuel map, this should mean that i have to raise the idle fuel map a little.

I have now the right idle rpm target, so making the idle a little more rich should be ok. I can heard some "pops" sound in the exhaust when idiling.
that's a kind of cool stuff :)

Under 6000 the AFR is around 12 but after 6000 this is falling to 16 in one shot. Sometimes the engine "cut" the power. this is really strange that is falling so suddently.
The rev limiter is set-up to 7400 rpm. That was like this on the ebay file we used as a base to start this project.
When the car was NA, i never remember the engine cutting after 6000.
I saw a thread here were a guy explain that he get the same problem, and solve it by adding a speed signal to the ecu. so i don't really know what to think. I didn't check any AFR before the turbo conversion.

The maps also show us that the VANOS is switching back at 6000 rpm.
I have no idea what this could do on the AFR.

For the 10% throttle, ok i will do the test that you suggest.
The turbo start to boost around 2500 rpm.
Triangulé
Posts: 36
Joined: Wed Jun 06, 2018 10:34 am

Post by Triangulé »

Ok, 10% throttle solved.
Idle AFR solved.

Turbo pressure is now 0.5bars.
The engine is now "cutting" at 5500 rpm at WOT.
If we use 75% of the Throttle, engine can go 7000 rpm without problem.
I try to modify the igniton WOT map to rescale it to have a larger load available, but there is no difference.

Today i will check the fuel pressure at WOT if stay stable or not.
The pump is a 255L/h walbroo, this look enough, but the size of the pipes from the tank still are the originals ones.

Do you know how the 413 ecu, switch from PT to WOT map ?
Is there a map to manage this ? or it an algorythm defined by load and rpm?
I wonder if this is not managed by the "LAMBDA OFF" maps ?
olafu
Posts: 186
Joined: Tue Jul 26, 2016 12:35 pm
Location: Finland

Post by olafu »

623 bin:

There is throttle position tables for idle/pt/wot transition. I have no idea where is rpm axis data for those tables... But rpm's are about 1800, 2000, 2400, ???? for idle-pt and 2000, 2400, 3200, ???? for pt-wot.

idle-pt 4x1 table, 0xD53C
pt-wot 4x1 table, 0xD540

What MAF you are using? MAF signal fault modes: Max air flow is RPM dependent, and there is multiplier for that. Did you get MAF DTC:s when engine cuts? If MAF fails (or ecu thinks so) load signal comes from limp mode table (tps/rpm "ve" table) in 0xDBB2.

And, did you have vehicle speed signal for ECU? If speed signal is missing, fuel is reduced about 30%, when RPM goes above set threshold (6000rpm stock, located at 0xD27E).

0xD27B is RPM threshold (2520rpm at stock) to activate speed signal check, so, overwrite it with "FF" (10200rpm) and ecu will never check it... Restricted mode and DTC 42 is gone.
Better solution is proper speed signal, because we don't know if there is something "stationary run mode" or like that. At least idle control has different values, when car is moving or not. DTC 42 has also a load threshold, but i can't remember where it is. But it is in same lump with those RPM limits. 0xD27D is time threshold, low value = DTC 42 appears sooner and vice versa. 0xD27A is minimum valid speed.

We did jump wire from cam signal to pin 42, so ecu thinks car is always moving, if engine runs. Works fine, but idle speed is higher.
I never got enough time to find where is "idle speed when car is moving" -table... :?

There is several idle tables for different conditions. You can find them by tracing (if it is "cold/dead" in tracing, it should not be changed). They are in a row in a bin, so, if you know one, other one is close.

609 and 623 has very similar tune, but at least cold start tables are slightly different. Data is also stored in other place, offset is about 4100. Program part is different, but i don't exactly know how.
Triangulé
Posts: 36
Joined: Wed Jun 06, 2018 10:34 am

Post by Triangulé »

Hi Olafu,

Thanks for these very interesting informations.

We have changed the TPS sensor, and get even better result.
Now it's not lean anymore.
The engine is now really too much rich (aroung AFR9 to 10) after 5500 rpm but not cutting.

I saw your previous message about the v signal.
I have still not check if the speed signal is working in the ecu.
I didn't made the wiring myself, so no idea what have been done.

About the WOT fuel tables, seems is not a "injection time" directly but more a compensation number ? Is the ecu taking the value from the PT table to add the value from the WOT table for the full throttle ? or are the PT and WOT tables totally independant from each others.
I have decrease the WOT table value, and finally got even richer AFR.
Last edited by Triangulé on Tue Jun 19, 2018 10:57 pm, edited 1 time in total.
olafu
Posts: 186
Joined: Tue Jul 26, 2016 12:35 pm
Location: Finland

Post by olafu »

WOT tables, as also PT and idle tables, are not actual injection time. It's correction factor. 128 is "basic" value.
Triangulé
Posts: 36
Joined: Wed Jun 06, 2018 10:34 am

Post by Triangulé »

Ok I just tested to change the WOT map and still same (AFR 10).
Seems it's not changing anything.

We think the engine is cutting at 5500 because it's too rich.
When we increase the turbo pressure, it still too rich but the engine is not "cuting" anymore, and can go upper.

On the PT map, the load is from 1.0 to 8.2.
I know 8.2 is not enough for full throttle, but as the ecu is switching to WOT map, do i really have to change the PT map ?
Triangulé
Posts: 36
Joined: Wed Jun 06, 2018 10:34 am

Post by Triangulé »

Ok so we have made another test tuning the WOT at a very low value.
The AFR still reaching 9.

So i have tested value from 75 to 160 (middle point at 128) and all of them give me too much fuel.
As soon as we push the trottle from 50 to 100% open, the AFR increase from 12 to 9 instanously.

So this seems showing that the ecu is realling switching to the WOT map.
The car run very well on the PT map from 0 to 75% throttle.

So i need your help to put some hypothesis on the table :
_ need to switch off the lambda probe for tuning the WOT ?
_ the MAF is at the max value ? what could happen ?
_ the PT map is saturated, over, not enough value for hi load ?
_ another map for full throttle is not tuned properly because of the new injector size ?
_ V signal missing ? Should make the AFR leaner right ?
Triangulé
Posts: 36
Joined: Wed Jun 06, 2018 10:34 am

Post by Triangulé »

Hi Guys,

Still working ! and have an idea !

The original E34 engine was Automatic gear box but now converted to manual.

There is some specific maps for the A/T.
How the ecu can know is have to work on automatic or manual maps ?

I'm surprised there is no Advanced or Retarded VANOS maps for the A/T.

As all the modification i've done on the normal WOT table are not working, the ecu maybe still think he is an A/T, and probably not switch to the right map if the speed signal is also wrong.

The ecu have probably to switch differently from map to other if he know with speed from the AT is engaged.

The guy who probably swap the engine in, decided to connect the switch of the gear box lever to get it always in neutral and then allow the car to start.
I'm not sure if there is a specific management or maps for neutral.
olafu
Posts: 186
Joined: Tue Jul 26, 2016 12:35 pm
Location: Finland

Post by olafu »

We have A/T car with manual gearbox. Works fine, all A/T stuff is removed, but wiring harness still has A/T controller connector.

How did you actually re-scale your injectors? What values you are changed from ROM? All load dependent tables must be rescale.

missing v-signal makes your engine run about 30% leaner when revs are above 6000 and fuel comes back when revs comes under 6000rpm. It NEEDS v-signal or disable v-signal check.
Triangulé
Posts: 36
Joined: Wed Jun 06, 2018 10:34 am

Post by Triangulé »

when the engine was NA, this was also perfectly working.
We went to the dyno and get the full power, without any problem.

I will check the v signal with the scope today.

I'm interested to know how the ecu software is programmed to recognize if it is a manual or A/T. This is maybe different software or wiring or hardware change somewhere to switch an input.
olafu
Posts: 186
Joined: Tue Jul 26, 2016 12:35 pm
Location: Finland

Post by olafu »

But how did you rescale the injector flow in ROM?
Triangulé
Posts: 36
Joined: Wed Jun 06, 2018 10:34 am

Post by Triangulé »

I've change it directly in the tunerpro table, not in the hex file ?
starting by the last number.
Is it working ? I mean is tuner pro change also the hex file like for data values ?

We have checked, there is no speed signal.
I need now to found the signal from the cluster to connect it to the ecu.
olafu
Posts: 186
Joined: Tue Jul 26, 2016 12:35 pm
Location: Finland

Post by olafu »

I was never changed injectors. But i can try to explain:

By changing injector scaling/constant, you are actually changing a load calculation factor. Raw max. load is 255 in decimal, and it's 12.75ms in "load" value, if we use x/20 conversion.

Now, if you change your injectors, say, for example big as twice of original. Injector constant is now set to 0.50. This will decrease your load value to half from original, in same conditions. Now we have more capacity in "load" value.

EVERY table, what is "load" dependent (D5 and D6 decriptors), are out of tune.

As example: Ignition tables. Max load is 8.2ms. If your K-value ("injector constant") is half from original, there is now 4.1ms load value, in same conditions, what produces 8.2ms load in stock "injector constant". ECU will now read ignition angles from 4.1ms column. (actually interpolated value from between 3.9 and 4.6ms colums)... Same conditions, so, there is now intermediate load's timing values in use, when engine is running at "full" load (if everything else is stock)

Also, if there is any correction, what is direct adder for injector pw, it must be changed. And different injectors propably needs different injector opening time table.

i think Evilm3666 or aboutton can help you in this.
Triangulé
Posts: 36
Joined: Wed Jun 06, 2018 10:34 am

Post by Triangulé »

Hi olafu,

First thing thanks for your time and reply that help to go ahead.

The ecu get this afternoon a speed signal from the cluster.
This is working now the ecu is not cutting anymore at 5500, and can go to the rev limiter without problem.
So you was right ! thanks for this big help.

The WOT still very rich and what you said about rescaling, show me that i havent do the right thing on the details.

So if i understand i have to know first what is the max load my engine is reaching (with an Ostrich i suppose or inpa could maybe gave me a value ? or estimate it) and then rebuild all the map with the new max load to use the best definition as possible. this will help to have a more stable AFR and more precise tuning ?

This mean now the ecu is not reading the value in the right colum.
olafu
Posts: 186
Joined: Tue Jul 26, 2016 12:35 pm
Location: Finland

Post by olafu »

Good! We struggled with that v-signal problem in my frend's rally car about 2 years ago. Our solution was cut v-signal wire, and it was temporary stuffed in to cam signal pin chamber with screwdriver. This temporary solution is still in use :lol::lol:

Need more specs of your build:
What is your injector constant now? Injectors?
And what air mass meter? Stock or something else, and is it pressure side or intake side of compressor?

INPA and many other diagnostic software can show engine load. And also you can see it from Tunerpro with tracing. But if you can get S50B30US turbo bin and suitable XDF for it, you can look how everything is done.
Triangulé
Posts: 36
Joined: Wed Jun 06, 2018 10:34 am

Post by Triangulé »

hello Olafu.

Ok i think i understand what you explain for the rescaling.
So the before theorical max load was 8.2.
the new engine will made 300 Hp instead of 200 so i have to add some margin.

I'm checking with my friend the injectors references. But we have calculated a k = 0.463

original are 165 cc (0280150415) or 120.4 g/min n-hept
New ones 360cc (0280150431) from saab or 260 g/min n-hept


8.2 x 0.463 = 3.79 if i add 30% -> 4.92 max to be sure, i have set the table to a max at 5.6 ms.

On the other hand i had to modify the gap between each column, in order to start low enough and at a coherent value.

1.0 -8.2 become
1.2 to 5.6

maybe i have to go to 0.5 instead of 1.2.
I also moved the values to get them aligned with the original load location.

MAF is the original one. We get the advice from some drift tuners, and they told us it's ok up to 1 bar. I have not check yet the output voltage to see if he is saturated or not.
it doesn't look ideal, but the it is located at the input of the inlet manifold (pressure). I don't know if it's a problem for a low pressure turbo application.


I really wonder how the ecu switch from the different maps.
Because if this is load related, and i don't think this will be so advanced for such old ecu, but if this is, then the ecu is maybe not able to switch to the WOT and use the end of the PT map for the high load and this part is really too rich. this is maybe explaining my problem.

Another point is the "O2 sensor adaptation in the XDF" shoud not made any difference on WOT right ? because it's an open loop area.
olafu
Posts: 186
Joined: Tue Jul 26, 2016 12:35 pm
Location: Finland

Post by olafu »

New max load would be good to oversize. I can't say by sure, but i think so.

You must compress original data to left side of table, and then you get right side of table "empty" for boost tunes. You lose some resolution.

If you want to change gap between columns, you must modify table header data in hex.

There is at least 30 tables, what you need to rescale, when changing "injector constant". I don't know how, you need to ask someone's help, who was did turbo tunes.

Idle, PT and WOT switching is only throttle position/rpm related. PT to WOT is same TPS value through whole RPM scale, but Idle to PT has different TPS values in different RPM.


I think stock maf is at limits with 300hp.

Stock bin has 950kg/hr limit to air flow. If flow is bigger, ecu throws "MAF: signal high" fault code, and ecu goes tps/rpm based limp mode.

Stock MAF table ends to 1321kg/hr, so it would be good idea to raise that MAF fault limit. And also MAF/RPM ratio.

New MAF high values (1322kg/hr = never reached) in HEX:
0xD23E = 4D
0xD23F = 10

EDIT:
New MAF/RPM ratio (0.3) values in HEX:
0xD244 = 8E
0XD245 = 3B
This ratio allows 1321kg/hr flow with about min. 4400rpm.
Stock 0.23 ratio allows 1321kg/hr flow with about min. 5740rpm.
Triangulé
Posts: 36
Joined: Wed Jun 06, 2018 10:34 am

Post by Triangulé »

Only thing i can tell you for the moment is that the engine is too rich (AFR 9) after 5500 on full throttle.
If we release a little the pedal, AFR go up (13) the power come back .
It's look like it's just too rich.
but not everywhere at wot only after 5500.
Also my friend push the turbo pressure from 0.35 at the begening to 0.6 now and it was worth with less pressure.
So this look a map tuning problem.


Anyway, i have made the change for the maf limite.
modifyed the hex value as you offer.
Let me check that on the road.

I also made a PT map flat ant a WOT map flat, to see how this is reacting.


we are waiting the ADS interface from the US to see the defaut code.
Sometimes we can go to the local garage to read the ecu, but they can't accept a everyday visit :(.

Step by step we will do it.
olafu
Posts: 186
Joined: Tue Jul 26, 2016 12:35 pm
Location: Finland

Post by olafu »

Where goes "tracing plot" when problem starts? Is the "load" value always same, or will it jump to somewhere?
Triangulé
Posts: 36
Joined: Wed Jun 06, 2018 10:34 am

Post by Triangulé »

well i think we have to buy the ostrich box, it will be hard to solve this problem with trial and error testing. but my friend didn't realize that will be so much time to make this tuning working. i don't want to let him


I have close also the evaporation valve, that is not connected anymore to the tank.

I found some interesting information on the website of Alientech (ECM tintanium software) they publish a new driver for the 413-623. the value of the injections maps PT and WOT are in lambda on the E36. I suppose it's the silver label ecu ?
there is also 2 ECT correction maps for the injection.

I also wrote a message to the 2 guys you recommand to me for the rescaling of the 30 maps necessary.


EDIT:
Get some news from my friend this morning that made a test home, the car is now working great at WOT with AFR between 12 and 11.
So this seems working !!
So i don't know if this is coming from the rescaling or the MAF limit raised.
that's pretty nice improvement.
Now i have to adjust the maps to get a fine result, all is really row.
When this will be better, we could go to the dyno to finish the work on the igniton table that i decreased probably too much.

I still wonder how you discover the adress for the maf upper limit (limp mode threshold) ? ahahah really nice and clever job !

Ok first thing i have to modify is that now the AFR is becoming poor (15) when we push the pedal suddently.
I probably have to increase the (acceleration enrichissement at 0xD776), but what mean "thottle change" is it a % under an amount of defined time or something like this ?
Because for short number change the increase is really big what is not really coherent.
olafu
Posts: 186
Joined: Tue Jul 26, 2016 12:35 pm
Location: Finland

Post by olafu »

0xD776 is actually a load filter table.

"Acceleration enrichment:

Triggered from rapid or less rapid LOAD increase...

0xDC15 D0xD5 4x4: Acceleration enrichment quantity by RPM and LOAD
0xDC31 D7 4x1: Coolant temperature correction for quantity
These are multiplied by each other and determines how much is maximum enrichment. 255*255 = about 22.5ms

0xDC3B D0xD7 4x4: Acceleration enrichment fade out rate. 0 = Enrichment stucks on. Bigger values = enrichment lasts faster.

0xDC57 D0xD7 4x4: Acceleration enrichment delta LOAD attenuation/sensitivity? Bigger values = less enrichment, lower = more enrichment. 0 = unstable pulse width, acc. enrichment seems to kick in from noise.

0xD4F9 = Delta LOAD transfer function? (known earlier as a TPS transfer function, but in normal operation seems like it has nothing to do with TPS). It has significant effect to acceleration enrichment, but i think it can be global transfer function to "LOAD" change rate and shoult not be edit...

Rapid change of the TPS does not affect to the pulse width, except in MAF failure mode. (Alpha N)"

All addresses are from header data start address, so, actual table is not starting at this point.
Triangulé
Posts: 36
Joined: Wed Jun 06, 2018 10:34 am

Post by Triangulé »

Ok first i will try to get coherent tables and try to understand what is the goal of each "acceleration map" because there is a lot of them :)

Where all theses adresses are coming from guys ?
Did you bought a Damos file or did you disamembly the microprocessor code software ?
I can recognize big maps on the hex, but small one like this ??

OxDC63 is also an acceleration map ?

For the MAF limit i found :

0xD23E : 4D10 = 907 max airflow for the original MAF
17CD = 1324 never reach



i've try the car today, with pretty nice results.
It pull very well, even if still have to adjust few things.
This look that the map have a lack of resolutions, probably because of the big injectors. Each step make a lot of difference. but i still have to work, and this will be ok.

AFR are goods, just the light load is a bit too lean for road use, but i'm not sure if this is the PT table or the acceleration enrichissement.
the car will go on his first track cession this sunday, this will give another infos.
it's really difficult to tune on the road, as you need to go too fast for high loads, and it's really hard to look all the gauges and drive it in the same time :)
too much power. Even 2 guys in the car it's not easy.
olafu
Posts: 186
Joined: Tue Jul 26, 2016 12:35 pm
Location: Finland

Post by olafu »

Triangulé wrote:Where all theses adresses are coming from guys ?
Did you bought a Damos file or did you disamembly the microprocessor code software ?
I can recognize big maps on the hex, but small one like this ??

OxDC63 is also an acceleration map ?
By testing with scopemeter and engine signal generator.

Almost all map start addresses are referred to "address table" what starts from 0xE680. In total 148 addresses in 623-bin, so there are 148 "maps" what are referred to "table of tables" and some other ones which i don't know they have been directly referred to anything.

0xDC63 affects sensitivity for enrichment.
I don't exactly know how, but i think it is some kind attenuation table or "minimum increase rate of load" -threshold for acceleration enrichment activation. It's D0xD7 so it is referred to RPM and CLT. All values are very low, so it can be inaccurate for tuning..

That "MAF high" -limit shares same x/3.1562283 conversion factor as used in the MAF transfer table, and it's LSB first. So, 4D and 10 is actually 104D in hex and it is 4173 in decimal. 4173 / 3.1562283 = 1322.15 and never reached, because last two bytes in MAF transfer table are 4B and 10, so it is 104B in hex, =4171 in decimal. 4171 / 3.1562283 = 1321.51 in kg/h.
Hairyscreech
Posts: 196
Joined: Tue Jun 20, 2017 3:19 am

Post by Hairyscreech »

I'm getting very close to having the arduino logger working at least for RPM, Load, MAF, CLT, IAT, TPS and Lambda which should be enough to tune with.

I have been struggling with similar issues getting the tunes right with this hacked onto an M20 due to the difficulty of getting the acceleration right with such limited data.

I tested it on the car last week and got it outputting some data to tunerpro via an ADX file so I will keep going tonight and post an update later, think it could help you a lot as it would mean data logs of laps could be taken in tunerpro.
Triangulé
Posts: 36
Joined: Wed Jun 06, 2018 10:34 am

Post by Triangulé »

thanks guys for your real reverse engineering work.
It so impressive, that you spent so much time to go deep in the (necessary) details. A lot of XDF available only have the basic and easy findable maps , what is a little bit limited to tune advanced engines.
The M.3.3.1 is more than 25 years old, but finally not so well know.
that strange.
E30 with M50 are becoming more famous and also because it's an easy swap, so for sure, that's right, became interesting to know more.

About the MAF Factor, you have x/3.156 but my XDF revision give x/4.6.
it's maybe a mistake. it doesn't really matter, i will correct it.

Yes it still hard to get something properly working at WOT.

We also need to have some measurement diagnostic tool like at least the ADS interface to have number to know where we are on the maps.

Before going to the dyno, i would like to know better what i need to touch to finalize precisely the tuning in order to minimise the time on the dyno.
As i said, with this kind of power, it's impossible to tune all on the road.

Do you think it's possible to datalog the status of the knock sensor or knock maps to have an idea of what is happening on the ignition side.
I don't know if it's realiable.

Our car look a monster for me, but what an interesting project.
olafu
Posts: 186
Joined: Tue Jul 26, 2016 12:35 pm
Location: Finland

Post by olafu »

Triangulé wrote:thanks guys for your real reverse engineering work.
It so impressive, that you spent so much time to go deep in the (necessary) details. A lot of XDF available only have the basic and easy findable maps , what is a little bit limited to tune advanced engines.
The M.3.3.1 is more than 25 years old, but finally not so well know.
that strange.
E30 with M50 are becoming more famous and also because it's an easy swap, so for sure, that's right, became interesting to know more.

About the MAF Factor, you have x/3.156 but my XDF revision give x/4.6.
it's maybe a mistake. it doesn't really matter, i will correct it.

Yes it still hard to get something properly working at WOT.

We also need to have some measurement diagnostic tool like at least the ADS interface to have number to know where we are on the maps.

Before going to the dyno, i would like to know better what i need to touch to finalize precisely the tuning in order to minimise the time on the dyno.
As i said, with this kind of power, it's impossible to tune all on the road.

Do you think it's possible to datalog the status of the knock sensor or knock maps to have an idea of what is happening on the ignition side.
I don't know if it's realiable.

Our car look a monster for me, but what an interesting project.
IMO, Ostrich 2 emulator is ultimate tool, because it supports tracing in Tunerpro. You can directly see what memory addresses ECU is using. So, it doesn't matter if normalization factors are slightly wrong.

3.15 MAF factor are according by the earlier XDF file from Tunerpro site, so if it was wrong, author did propably corrected it.

Those knock maps are somewhere in SRAM, one map for each cylinder. I don't know how big they are. I'm planned to find them by running ECU on the test bench and by tuning knock limits... :roll: INPA can continuously poll RAM or ROM max. 30 bytes at time. It shows data in HEX.
Triangulé
Posts: 36
Joined: Wed Jun 06, 2018 10:34 am

Post by Triangulé »

The car was on the track this sunday with 33 degree C temperature outside.
It was a really hot day.
The car run all the day with 0.4bar max and WOT AFR set at 12.
All was really great, the old E30 was able to compet with more modern BMW V8 and M2 than are more than 400 HP ...
With a short final gear and a really good driver ... the low boost M50 had quick the ass of these expensives cars.

What we have to improve :
_ the waste gate is really too slow to open, this push us to limit the pressure to 0.4, otherwise the boost rise 1bar when the turbo is charging around 2500 rpm. we need a turbo boost management.

_ the oil temperature is reaching 130° (MOTUL 300V) probably need a radiator
_ oil catch tank is filling quickly, we think this come from worn valves gaskets.
_ dynamic air filter will probably help to get fresh air in the turbo inlet.
we have remove 1 front light, but this will probably work better with an optimized input shape.
_ the maf is over his maximum limit, anyway the AFR is stable.
I just worry for this winter, ecu will not see the difference.
_ we want to push to a stable pressure of 0.6 bar, but we don't know if this will be realiable for a track use. we already tested it, but on the road the way we push the engine is limited to few secondes.

i think it's not really necessary to improve much the scaling of the maps.
AFR are really stable now over the rpm, as we also have decrease the fuel pump pressure to 50 PSI. More pressure request a more precise map between each points on the fuel tables, this is the same problem if you increase the injectors size, you need to have better precision, what is really not easy with the standard tables size.
So for bigger injectors than what we have here, i think re sizing the whole maps will be a must.

idle work perfectly well, as the inlet manyfold have no leak ... and this is the hard part of the story.

On the track the engine is runing between 5500 and 7000 rpm.
Consumption was 40 liters/100 km.

We still have to go to the dyno and see what we can improve on the igniton maps and acceleration maps too.
olafu wrote: Those knock maps are somewhere in SRAM, one map for each cylinder. I don't know how big they are. I'm planned to find them by running ECU on the test bench and by tuning knock limits... :roll: INPA can continuously poll RAM or ROM max. 30 bytes at time. It shows data in HEX.
Did you try some external " knock detector" like the KS3 ?
I wonder how this could work properly, because the original ECU use 6 differents maps as filter of the frequency and shape, probably because the way the sound arrive to the sensors is different from one cylinder to another.
I don't know if there is a way to use tunerpro and the Ostrich 2 to get a tool with some kind of bar graph, but the great point should be to have the real filter from the ecu, and to be able to really see what the ecu is seeing in live mode.

Thanks for your helps guys, next post will be after the dyno session, i think.
It's time to work on the details to understand the last points.
we are really happy to make this car running !
olafu
Posts: 186
Joined: Tue Jul 26, 2016 12:35 pm
Location: Finland

Post by olafu »

If you can use both knock sensors for external knock detector, it can work well. But ECU needs signal from knock sensors, if it is missing, ecu goes limp mode and reduces timing.

0xD288 affects DTC:s 69 and 70. By overwriting it with "FF" ecu doesn't notice missing signal from knock sensors. I dont't actually know what is this, but limp mode and fault codes will disappear... Worked well in my tests, but you use it in your own risk.

By rising knock control load threshold map to "FE" you can easily disable knock control and knock prevent functions, but this combined to untuned ignition timing can easily destroy your engine.

It's very dangerous to switch off engines own safety systems.
ozzyybr
Posts: 4
Joined: Tue Feb 13, 2018 9:11 am

Post by ozzyybr »

Congratulations on the job.
Could you share your XDF file with this new information you discovered?
olafu
Posts: 186
Joined: Tue Jul 26, 2016 12:35 pm
Location: Finland

Post by olafu »

https://www.dropbox.com/sh/1qhnxk084dr4 ... U7y3a?dl=0 <- dtc triggering and knock part. And some other information.

More information is in that other, big m3.3.1 topic.
ozzyybr
Posts: 4
Joined: Tue Feb 13, 2018 9:11 am

Post by ozzyybr »

olafu wrote:https://www.dropbox.com/sh/1qhnxk084dr4 ... U7y3a?dl=0 <- dtc triggering and knock part. And some other information.

More information is in that other, big m3.3.1 topic.
Thank you very much
I believe that unfortunately this will not help me. My car uses the 715 chip
olafu
Posts: 186
Joined: Tue Jul 26, 2016 12:35 pm
Location: Finland

Post by olafu »

Use some offset. I think 715 has same maps but different place/offset. When you found one, you found other ones.
Hairyscreech
Posts: 196
Joined: Tue Jun 20, 2017 3:19 am

Post by Hairyscreech »

I think there is only about 4 different offsets/map positions that fit the different versions of the M3.3.1 (early non EWS nonM, Early non EWS M, Late EWS nonM and Late EWS M)

Which means the info in the big motronic thread will guide you on finding the maps in the 715 chip.

{I am not counting the Alpina ECU as a different version, all they did was move the whole map block further up the data section and left it the same offsets/layout for the maps and the late EWS chips.}
Triangulé
Posts: 36
Joined: Wed Jun 06, 2018 10:34 am

Post by Triangulé »

hi guys,

After our second track day session we are facing few small problems.

_ idle come now too high (1200 rpm) and engine is slow to drop revs.
_ engine is not runing good anymore around 4000 rpm.

Idle is 1000 when cold and 1200 when hot. but very stable.
I have tested the TPS sensor and ECT sensor, and try to understand what in the maps could set the idle at a so high target.

I wonder if the problem is not coming from and air leak, but when we disconnect a vaccum hose (like break assistance) the engine stop.

If the MAF is dead and send no signal, how is the idle managed ?
MAF is at the his original location, just before the inlet manifold, this mean he is getting the hot air from the turbo.
olafu
Posts: 186
Joined: Tue Jul 26, 2016 12:35 pm
Location: Finland

Post by olafu »

Triangulé wrote:hi guys,

After our second track day session we are facing few small problems.

_ idle come now too high (1200 rpm) and engine is slow to drop revs.
_ engine is not runing good anymore around 4000 rpm.

Idle is 1000 when cold and 1200 when hot. but very stable.
I have tested the TPS sensor and ECT sensor, and try to understand what in the maps could set the idle at a so high target.

I wonder if the problem is not coming from and air leak, but when we disconnect a vaccum hose (like break assistance) the engine stop.

If the MAF is dead and send no signal, how is the idle managed ?
MAF is at the his original location, just before the inlet manifold, this mean he is getting the hot air from the turbo.
Any DTC's?

If MAF fails, ecu will still control idle normally. ECU goes limp mode, "load" comes from tps/rpm based limp mode table, and "load" is corrected by estimated position of idle control valve.

Try to block that idle valve air hose. If everything is ok, there is no enough air flow for running engine in throttle body, and engine will stop. Fault is rectified to idle air valve or wiring, or something indirect idle air control problem.
Triangulé
Posts: 36
Joined: Wed Jun 06, 2018 10:34 am

Post by Triangulé »

We had a DTC on the MAF, but my friend didn't notice witch DTC reference it was.
It's maybe coming from a test when we have disconnected it.

Yesterday we try different thing :
_ TPS is working good and give the right position on the diag.
_ ECT sensor seems ok and give a coherent temperature on the diag.
_ remove the IAC motor (it is not block inside, and seems working properly)
The Valve inside the IAC stay naturally open around 1/3.
_ disconnect the IAC motor plug (idle is then much more lower)
_ Connect again the IAC, the idle was unstable (moving from 1000 to 2500 )
_ disconnect the ecu and wait 15min, then we get a perfect idle back ... at 750 rpm.

So the problem seems coming from the ecu management of the IAC.
Is unplugging the ECU same as resetting the adaptative settings of the ECU with the diag software ?
Very strange, look like a bug on the ecu.

This morning when we have started the engine, the idle didn't increase as it should do for cold condition.
so this should mean the ECU think the engine is hot and the ECT sensor value don't move properly.

We bought the ADS cable, but i don't know witch diag software to use with this old ecu, and the serial port USB converter is a nightmare to get it working.


the other problem at 4000 rpm is still there.
So i wonder if the MAF is not dying in the same time.
But we have no references values,
I don't want to modify the maps to correct this problem.

When the car run well the AFR is around 12.5 at 4000 rpm, but when we have the problem, after is around 14.5.

edit:
I get the chance to make a road test this morning.
The problem around 4000 come from too much air.
The turbo pressure rise too much (over 0.7 bar) on a short time and the AFR become poor at this moment.
We didn't get any problem 1 month ago, but weather have change a little.
So this probably mean my enrichissement map is not good enough.

For the idle, it's now stable around 750, i don't know what happen.
Triangulé
Posts: 36
Joined: Wed Jun 06, 2018 10:34 am

Post by Triangulé »

ok some news.

The MAF is dying.
We bought a new unit (probably from china) an get totally another result.
The engine now is very lean and request a new mapping.

We probably made all the set up on a old dying maf ...

The good news :
_ idle go back under 1000 rpm.
_ no more step at 5500

hope this cheap ebay stuff is good also.
Hairyscreech
Posts: 196
Joined: Tue Jun 20, 2017 3:19 am

Post by Hairyscreech »

Be careful that you are not tuning to a new but still incorrect MAF.

If you can I would validate the cheap MAF against a known good working Bosch MAF.
Triangulé
Posts: 36
Joined: Wed Jun 06, 2018 10:34 am

Re: M3.3.1 413

Post by Triangulé »

hello guys,

I come back after a while.
What happen ?

We were facing strange cooling liquid behaviour, it was increasing ... and get also high pressure on the engine bottom.
Original head gasket burnt on the 5th cylinder. In fact, this engine already get some head gasket trouble by the past, we found that the water pump propeller break and a big piece was still travelling in the block reducing cooling capability on the head gasket. the water pump was changed, the block engine haven't been cleaned ... Hole from the engine block to the head is only around 10 mm diameter. So the water pump piece travelling was enough to reduce the cooling and burn the gasket localy.
Frankly speaking, i think a low boost turbo can work with an original head gasket, but we have decided to enjoy this problem to switch for a new cometic 3mm gasket. This allow to reduce the compression ratio from 10.3 to 8.5.
Others problems we discovered in the same time were:
_ the LCA between the exhaust cameshaft and the inlet one, was wrong about 2 tooth (Around 15°),
_ in the same time, the VANOS was wrongly installed, and cannot moved (avanted to retarded).
_ The hoses were all more or less old around the idle valve. leaking
_ Wastegate and dumpvalve are cheap copy from china, they were all leaking too. Not really adaptated to a track use.
So we had a lot of work reworking everything.

From what i can say now, the engine:
_ loose torq under 4500 rpm , it's probably mainly due to the decreased compression ratio.
_ power after 4500 to 7000 tr is much better
_ the engine time response loose some reactivity (compression ratio decreased, but also need some acceleration enrichissement tuning )

things are encouraging and we want to upgrade now the turbo pressure up to 0.8bars,(instead of 0.5) that should provide around 300 horsepower.
i have to check the injectors duty cycle first and also the MAF output. This will limit our power wish, and then request another parts investment.
so step by step.

The car is already really powerfull, and dangerous to drive under the rain, even on a straight line, tires can slip.
The 30 years old chassis is enough up to 220/230 HP but request more improvement now, maybe a limited slip differential.
We also think to enjoy the new torq with a longer final gear. the top speed was limited on some track like "Le Mans".

Well now, the job i have to do :

_ acceleration enrichissement, there is 4 or 5 maps related. how to proceed ? We have try to modify the AE Quantity, but engine run really bad now.
_ VANOS tuning, did you already tryed something interesting ?
_ idle, i have some trouble with the AFR, even increasing the fuel map, the AFR still too high (16).
is the idle always working with the lambda probe target i mean in a loop mode ?
Last edited by Triangulé on Sun Mar 24, 2019 12:15 am, edited 3 times in total.
Triangulé
Posts: 36
Joined: Wed Jun 06, 2018 10:34 am

Re: M3.3.1 413

Post by Triangulé »

hello guys,

I've try to modify the 0xDC15 adress, but it's not good, the engine is really difficult to start, really too rich.
is this map runing at idle and cranking ?

The chinese TIAL waste gate copy is seizing and all the screws from the cover go out by themselves.
The production quality look not so bad, but the design is not really well thinked and we loose so much time to solve that.

The gasket between the turbo and the exhaust manifold (from china also) have burnt (it's a new one).
I think we must have a problem on the gasket surface, this must be not flat enough.
It's probably at the manifold also, not thick enough to handle high temperature.
This leak is probably slowing the turbo response.

The idle still too lean, and i have too understand how the ecu managed that.
The lambda loop is unactivated, i'm scared this will destroy my work on the PT map.
JAS
Posts: 30
Joined: Tue Apr 23, 2019 9:13 pm

Re: M3.3.1 413

Post by JAS »

Read this tread,good info!
Having a problem with my MAF,when tracing i MAF-table it stops around 2,4-2,5V and don´t want to go any higher.
Did a testrun and measured voltage directly on MAF,it´s about 4,7-4,8V when boosting 0,7bar and 5000k rpm.
Is there something in the binfile blocking MAF-signal from reaching higher?
Downloaded stock 413Redlabel 623chip file.

Or could this be how ecu acts when reaching maximum allowed airflow about 960Kg/h?
Read in tread how to raise maximum allowed airflow but there are severals figures?which ones are correct?

Stock bin has 950kg/hr limit to air flow. If flow is bigger, ecu throws "MAF: signal high" fault code, and ecu goes tps/rpm based limp mode.

Stock MAF table ends to 1321kg/hr, so it would be good idea to raise that MAF fault limit. And also MAF/RPM ratio.

New MAF high values (1322kg/hr = never reached) in HEX:
0xD23E = 4D
0xD23F = 10

EDIT:
New MAF/RPM ratio (0.3) values in HEX:
0xD244 = 8E
0XD245 = 3B
This ratio allows 1321kg/hr flow with about min. 4400rpm.
Stock 0.23 ratio allows 1321kg/hr flow with about min. 5740rpm.

Or
For the MAF limit i found :

0xD23E : 4D10 = 907 max airflow for the original MAF
17CD = 1324 never reach
Triangulé
Posts: 36
Joined: Wed Jun 06, 2018 10:34 am

Re: M3.3.1 413

Post by Triangulé »

hi JAS,

Happy to see you joining my thread, for the m50 turbo conversion fans :)

From what i know, and if i'm not wrong, the maf voltage can go up to 5.0volts, but the ECU input can only accept 4.5 volt.
That mean you are already saturating the input and the ECU is unable to see how much air is going inside after 4.5.

I don't know why you can't go upper than 2.5v.
My mapping of the MAF is not working, everytime i try to update it, tunerpro is crashing.

You are maybe reaching the limit value, before the ecu can go upper.
try to change it in O23D.

I ll check it on my side also, we just bought the Ostrich box to improve our tuning capability.
At the moment the car is stop. We are now runing at 0.8bars and we have kill one injector (they are old second ones) the 5 other are still good.
olafu
Posts: 186
Joined: Tue Jul 26, 2016 12:35 pm
Location: Finland

Re: M3.3.1 413

Post by olafu »

JAS wrote: Thu May 09, 2019 10:35 am Read this tread,good info!
Having a problem with my MAF,when tracing i MAF-table it stops around 2,4-2,5V and don´t want to go any higher.
Did a testrun and measured voltage directly on MAF,it´s about 4,7-4,8V when boosting 0,7bar and 5000k rpm.
Is there something in the binfile blocking MAF-signal from reaching higher?
Downloaded stock 413Redlabel 623chip file...
I have been encountered the same problem. It's Tuner Pro issue. Split MAF table half and use upper or lower part at time. It will work. I don't know why this problem occurs, but i haven't considered it as a big problem.

MAF table is directly pointed by ADC value from MAF input. No "software" filtering.
JAS
Posts: 30
Joined: Tue Apr 23, 2019 9:13 pm

Re: M3.3.1 413

Post by JAS »

Yesterday we managed the car to run pretty well :D

Problem with MAF-table was related to limits in 023D and limit for rpm/vs airflow.
One of my knocksensors is probably broken and this is why it´s been so hard to map,no matter what ignition I adjusted the knockretarding took igntion down.
Been running with autocom diagnosis beside my Ostrich,in Autocom you can se realtime igntion and knock activities,without this you´re blind :wink:
You can even see Load-value in autocom!
Turned off knockretarding and adjusted car and it turned out really good!
Max airflow att 6000rpm is just above 900Kg/h and Load is peaking around 9-9,5

K-value is adjusted so Load is enough high to reach highest Load-value in tables,present K-value 0,76

It pulls really hard and smooth from 3krpm to redline without any problems.

Last problems to solve by now is my auto-tranny wont shift as it did before...It´s slipping when shifting between second and third gear,no matter throttle...
The 2 tables for AT-map in 413red XDF doesen´t effect anything,this needs more work in comparing my orginal binfile vs the adjusted one.The tranny worked 100% with stock bin.
olafu
Posts: 186
Joined: Tue Jul 26, 2016 12:35 pm
Location: Finland

Re: M3.3.1 413

Post by olafu »

Because adjusted K-value, it affects to load and maybe auto transmission gets engine load information from "load" or "injection time". I'm never check this, but maybe good thing to start... Did you rescale all tables what are load dependent or consists load values?
But about that tracing problem, can you trace now through whole MAF table? In my tests i can only trace first 128 cells and then rest of them by make other 128 cell table starting from half of MAF table... :roll:
Post Reply