frances

Hi all, I'm missing something about the EMS logic and i need some help.

Working with the emulator (OSTRICH + Tuner PRO) i've noticed that if i increase the minimum value of load (an ECU constant) the idle mixture gets richer while the AFM is given the same output, so i'm inclined to think that the AFM output is neglected @ idle.

But if i hand operate the AFM @ idle and open the flap, the engine gets rich and stalls. Why does this happen (due to the fact that the idle correction map depends only on rpms, not load)?

I trying to figure out the function of the minimum value of load on the overall injection logic.

Thank you.

Eric_Oz_S2

If you operate the AFM, you are instructing the ECU there is more air coming into the engine. The ECU makes the mixture more rich, but in reality there is no more air because you haven't opened the throttle.

What do you mean by "increase minimum value of load"? Are you referring to the idle fuel map or something else?

Eric

Rogue_Ant

Other than starting/cranking, the DME uses the AFM signal at all times...
Which table / cells are you adjusting?

frances

No i'm not touching the IT corrections at idle.
Yes the intuitive point is that if you manually open the AFM @ idle you flood the engine, but on the other hand why this happens also increasing the MiVL? (see after)

The cell that i'm adjusting is the 1174h value 11h translated into decimal and divided by 20 gives 0,85ms.
The previous cell 1173h, value C0 is the maximum value of load 9,6ms.
@ 1171h we have the rev limiter, value A2 which is 6480rpm. A2 into decimal is 162 then x40 = 6480.

How do i know that the 1174h is the minimum value of load?

Well i have here an old bosch manual that was sent to me via email by a user on another forum.
This manual specifically refers to the bmw M3 e30, 1988, which .bin (when i use a 2D graph editing SW) is really similar to the Porsche one (it's a Motronic design, i think a 1.3).
I own also an M3 e30.
The analogies between the two bins are really strong and it is easy to compare things cause, even if the values are different, the maps are stored with the same order.

Especially the rev limiter - maximum value of load (MaVL) - minimum value of load (let's call it MiVL) group.
But you can also find analogies in other parts, like the three parts of the AFM transfer function.

Now the adjusting of the MiVL cause the same effects on both cars.

The first effect is to enrich the mixture @ idle and the second is to shift the ECU reading on the injection timing scale ("load" PT scale) @ PT forward.

In other words suppose the minimum load scale value is 1.05ms, if you set the MiVL to 2ms, you won't read the 1,05ms point on the PT load scale anymore.

Now what i don't understand is why to change the MiVL affects the IT @ idle.

I'm attaching my .bin.
It is from a 951, 28 pin DME, obviously, i'm sorry, maybe i've chosen the wrong thread, i've seen now the 944 title.

Rogue_Ant

Ok, I think you mis-understand what the function are of the values at 0x1174h & 0x1173h...

These values simply determine the upper and lower limit of the airflow sensor. They do not directly define load.

The value at 0x1174h is the minimum airflow sensor limit. So if you raise this number enough, then the DME will ignore any airflow reading below this low-limit, and instead use the limit value. This is why you can enrich the idle AFR by raising the value @ 0x1174h.
The reverse is true for the value at 0x1173h - this value is the upper airflow limit. The same rules apply. The DME will ignore any airflow reading above this limit, and instead use the limit value.

frances

I wish I had misunderstood, unfortunately it seems I didn't get it at all .
Thank you very much, very very helpful.

I'd like to ask a couple of general questions:

1-Is it possible (does it make sense) to enrich the mixture lowering the injector constant (I'm confident that I'll find it sooner or later ) without replacing the injectors? Is the injector costant sometimes called "injector nozzle"?

2-Does any action which increase the injection timing possibly cause the ECU to overheat? (the maximum allowed MaVL is 12ms).

3-When i adjust the AFM transfer function (i found in this forum a fantastic excel file that allows to do it consciously) to work with different "load" sensors (often i have to translate all the function up) i can see the tracking light bouncing back to the beginning of the injection corrections table (@PT) just a few degrees before the WOT condition, while emulating.
Can it be that to translate up the transfer function cause the ECU to read excessively high Q values (IT=Q/nxK) and as a consequence, fools the ECU?
I have tried to avoid the issue by enlarging the IT (load) scale @PT to the limit of the MaVL and to lower the transfer function when passing 4V, but without success.
When the stock AFM is in, the emulation shows that, @ PT, the ECU reads correctly, even when approaching WOT, it doesn't bounce back.

Thanks again.

Rogue_Ant

Glad to help
And to answer your questions:
1 - You can enrich the mixture by adjusting the injector constant, as long as you are within the injector duty cycle limits. Injector nozzle could be another term for the injector constant.
2 - If you are using the stock impedance injectors (or equivalent), than the higher injector duty cycle should be within the temperature limits.
3 - I am glad you like my excel file . Most likely what you are seeing is the tracking light inconsistency. If the tracking light is true, than you might have made the transfer function too high/large. In this case, the 8bit load value could go over 255 decimal, and thus roll-over back to 0 decimal. Which is why the tracking goes back to the beginning of the PT table.

BTW, if you follow the link in my signature, I have an .xdf file for the 951. It might help you out.

frances

Well thank you again.
I have found a competence here, i haven't found anywhere else.
I've visited your link, i didn't know (i'm a new user) that you have done all of the job in disassembling the Motronic code, to make the ECU swallow new sensors, so, way to go.

I've also seen that you destroyed an engine "on purpose".
Rings failure after a triumphal 9 months run or something.

The .xdf looks the same i DLed from the TunerPro site: i've built (for the M3) the whole 8 pieces of the AFM transfer function with the volt reference scale (following your .xls), but, after having had an email conversation with Mark i realize that to put the pieces toghether and built the entire transfer function would have required me to hand input the 256 values of the volt scale (quite time wasting).
On this car (and on a 320is e30) i'm using a 0280120418 sensor which has two plugs, one for the idle/ WOT switches and another for the pot function.

The main issue is the 0,5 - 4,90 V output, while the AFM outputs 2,0 to 4,0V maximum (registered) @ PT. The A/F leaning effect (low voltage) is really effective (if one doesn't translate up the transfer function, you won't get things right).
So i have to play.

I'm considering a MAF or hot wire sensor, anything that has 4 pins, ATS integrated, analog output, 5V supply, 80-80 mm diameter maximum, 76 minimum.
Maybe the SW tuning will be eased by a MAF. It is also much close to the actual Airflow than a pot which isn't at all.

Rogue_Ant

Yep, the .xdf on TunerPro was done by me. And yes, I knew the NA motor would live a finite life - and I learned how much the NA motor could take.

You are correct the stock transfer function algorithm is very limiting. Rather than adjusting the transfer function, you could adjust the PT tables to achieve the AFR you desire...

You mention usig a pot, are you trying to use this in place of the AFM? I haven't tried an Alpha-N setup with the DME... I'm not sure if this is even possible without re-writing the transfer function algorithm. Adapting a MAF would probably be a simpler solution.

frances

My pot experiments has started on the 320is with a crappy (prototype) setup like what is shown in the first and second picture and end with a very clean install (third picture):



The stand alone pot setup (with the driving rod and hardware) has the advantage of being very adjustable.
In point of fact in a couple of minutes one can make the engine running by increasing the idle volt (while decreasing idle correction) in such a way that the PT corrections are read correctly. This requires a minimum action on the design of the PT corrections themselves.
Unfortunately the pot won't ever be centered correctly (accelerated wear of the pot itself).

The third picture setup is perfectly centered and it is a direct fit.
No additional HW needed. But there's no way of increasing the idle signal and currently i'm under the impression that starting from 0,5V (i can't get more volt adjusting the position of the pot without compromising the idle switching function) i can't simply act on the correction table, but i'll give it another try.
With 0,5V at idle, the engine doesn't even idle correctly after having set the idle corrections to the maximum possible.
That's why i'm looking and still can't find the injection constant on the map, just to have another parameter to "move".

What above has been done for an M3 e30 and 320is car and the map (relative to the stand alone pot setup) that i produced was installed on two s14 users cars (for free of course, I'm an amateur) without problems.

I'm currently working in my free time on the "second generation" (third picture). Which is way more clean to me. The physolofy is the same driving two generations of alpha-n professional tuners: the one who use the stand alone and the one who doesn't and choose the "clean install". In this last case the sensor used is the 0280120402 (Maxx Automotive). But the plug is different so one has to re-wire.

Now i'm studying and wondering about 951...But still tuning the s14.

frances

I'm going on with this thread cause i think it is interesting, due to the fact that it's always a MOTRONIC thing.
But if you think that it is a gigantic OT please, tell me.

Well the AFM transfer function is quite limiting while tuning (as previously stated) and the answer to the bouncing back of the PT fuel / advance table was that the calculated base injection timing via the Q/nxK is way too high when the AFM flapper output Volt is greater than 4/4.5Volts and so, even stretching the load scale of the PT map to the maximum possible value (12.75ms), we always get the bounce effect. To stretch the scale implies also to change the maximum load value, obviously.
This bouncing doesn't make the negine stalling so one can also decide to keep it as is and it is not possible to avoid it, working only on the AFM transfer function (and i don't know where to work either).

So, unfortunately, in my condition, one has to make a compromise between various things e.g. if you implement a maximum lenght load scale you can be sure that you limit the bouncing effect to the extremely low rpms, like 1000 / 1400, but you will be reading only a few map corrections at high rmps.
In fact that table (256 values) is only a recipient of corrections, choosen time by time by the ECU calculation of the IT = Q/nxK and the istantaneous rpm value.

I'd like now, if someone can tell me, to understand the differences between the AFM transfer map and the VQ table (if any).
As far as i know AFM transfer function translates the AFM flapper Volts in an experimental airflow value.
How is it related to the VQ table?
Does has a meaning to tune both the transfer function and the VQ table?
May anyone provide a bin and a starting address of the VQ table, so i can see what is the shape of it and make some comparison?