Technical Woj's turbo project

[cc1]

If you are using the lower impedance one, the wave form might distort due to too much gain which would effect higher rpm though or too much loading on the input creating a weaker signal? I would have thought the VR sensors would produce a sine wave and squared off in the ECU.
Worth a shot with the resistor but you could do with scoping the signal to compare.
On the Omex, you can set the switch from high to low gain at any rpm. It's normally set at 2k rpm. With the MPI sensor, I get the odd misfire at about 1.5k rpm but have yet to adjust the settings to see if this changes.

 

[woj]

This is very valuable info, you will get a thanks when I am with a browser that allows this

Now that you have mentioned it, I am sure this ECU has this switch too, I might have even seen it without knowing it, will look into this.

 

[thunderfox5]

cc1 is right, I forgot the stock sensor in these cars is VR and not hall effect

 

[cc1]

cc1 is right, I forgot the stock sensor in these cars is VR and not hall effect

The CAM sensor is hall on mine.
I have a setting to count pulses and is set to 3. This is due to the 3 lobes being on the Mpi CAM pulley and I'm not sure if this is right though. It makes starting take longer than I would like. Something else I have yet to play with.

 

[woj]

Looked into this thing with gains and all - in this ECU this is done by a separate chip, one that I do not have any documentation for whatsoever. I do not see any meaningful configuration of this chip going on though (but I still have to recheck, but there is a similar chip for the knock sensor and there I can see where it is configured) so I can only assume that the gain and filtering is done automatically. This also leads me to believe that it should not matter which exact sensor is used...

 

[woj]

Measured the resistance of my crank position sensor - 550 ohm when static. So I assume it is different from the one MPI Sei uses (according to specs 1350 ohm).

But, back to the drawning board. After 5 days of standing still with the new ECU connected the battery is dead flat - 0.5V. That means fault in the electrics, either bad construction or I got the wires wrong somewhere. In this case I cannot blame the ECU for not running well. So I have to figure this one out first, not to mention charging the battery - PITA because of the strut brace.

 

[woj]

:bang: It was not the wiring of the new ECU that killed the battery - I take guesses of what I was so sloppy about and forgot to switch off.

In the meantime, since I took the new ECU installation out of the car again, I made a bit more proper connection for the crank sensor signal shield. That is, I now have a connector to plug it into the ECU pin 77, as it is done in the orignal Seicento MPI installation, and not the body / ECU casing ground. Moreover, now I have the two shields for both position sensors connected as it is in the orignal loom, might be that it also does make a difference. If nothing changes I have at least one thing ruled out (crappy shielding). The next would be to try out a different crank position sensor, this is an investment of circa 30 Euros.

 

[blu73]

interior light?

 

[woj]

interior light?

 

[blu73]

:doh:

 

[thunderfox5]

It happens... Mine was stopped for two months in the garage and I forgot to disconnect the battery. After two months, it would ALMOST start. Guess what I left on? The radio clock/LCD.

 

[blu73]

One good thing about having an LED interior light, can leave them on for a couple of days and they use next to no battery.

 

[woj]

Developments Today I reconnected the ECU again with the factory TPS settings and the sensor shiled done somewhat better. Was a bit puzzeled when it hardly started and would not keep idle, on revs it was fine, but idle was very low, jerky, and it eventually died. On top of that the MIL would not go off. I did not have the laptop with me so I can only assume that either the ECU did not like my play with TPS calibration (and couldn't catch the right state of the engine), or because earlier I inadvertedly switched on the ECU on my desk with bare installation (= no sensors or actuators, just power), I think it might have not liked that and stay histeric about it for a while.

But, on the second attempt it started beatifully (if I wouldn't know better, I would have said the same as my 18F ECU does) and kept veru nice idle and reved very nicely. I could not drive it, will do it later, but stationary tests showed that it is OK.

And to the most interesting bit:

I'm not running cam sensors in either the Seicento (with an 8V) or the Punto (with the 16V swap). Therefore I would rule out cam signal as being the cause of it.

I had difficult time believing this, but indeed it seems that the cam sensor is more of a decoration on this engine. I know it is also used to check the timing belt accuracy, but for running the engine it seems it does not do much. I disconnected it while it was running - zero effect, engine runs the same, no MIL. I turned it off, restarted after letting the ECU finish the power latch phase, again, started nicely, run nicely, no MIL.

Will know more later when I hook up the laptop to it and drive it around the block.

 

[woj]

OK, I am very hesitant to use word ideal, but it is very good It seems that the whole problem was (partly, perhaps the fixed shielding helped a bit too) the TPS calibration indeed. In short, the ECU had problems gracefully transitting from idle to part throttle state and back, and that was giving horrible results on very light throttle presses.

The whole last ride was very smooth and the phenomenon occured only once. I will have to still investiage, because even though it was only once, it was also very heavy hesitation and hunting. Also, it seems that the fuel enrichment on sudden throttle presses is a bit weaker than in 18F and on taking off and gear shifting I need to press a bit more throttle not to experience a slight choke. It might be also the characteristics of full sequential injection - in full group there is always some "spare" fuel floating around the manifold, here not

 

[woj]

What it also might be - the factory ignition on this ECU is not very far apart between the idle and part throthle (idle / light load corresponding regions), so the transition between the two is smooth. I have my part throttle ignition (copied over from 18F) quite agressive comapred to the the idle ignition in the same region, my guess is that the hunting is because of the sudden change (and ECUs attempts to countermeasure it) between the two. I will try a lighter ignition map for part throttle.

 

[cc1]

Are you sure the injection cycle does not revert to batch firing if the ECU gets no reading from the CAM sensor as a default setting? Would expect a fault to be logged though.

 

[woj]

Are you sure the injection cycle does not revert to batch firing if the ECU gets no reading from the CAM sensor as a default setting?

It does not seem so, I tried experimenting earlier on on my test equipment and it never did any batch firing, it was always sequential. The only thing that bothered me was how does it know the right sequence.

(What I can also tell you is that there is a an earlier version of this ECU - 49F (unfortunately implemented on an undocumented CPU), this one has also sequential injection but does not use the cam sensor, it simply has an algorithm to guess the sequence, you can even see this guessing in the diagnostics readings).

Would expect a fault to be logged though.

Oh yes, the fauly is logged, but it is not considered serious enough for a MIL.

 

[woj]

OK, I think I am there. When warm it is running essentially as I would expect it to do. Fine changes in the VE table bring foreseeable AFR changes. I modded several tables: load delta enrichments, light load ignition, allowed for quicker transition from idle ignition to part throttle, and also moved away charcoal canister operation from idle regions (which in my case are 200 rpm higher than stock). All this seems did the trick, now it simply needs fine tuning of the fuel. Also the EBC is not as stable as it could be, but that's a matter of playing with the PID parameters.

What I still am not happy about is cold start and running way too rich during warmup. Don't get me wrong - it starts on the first turn of the key, but not as smooth as I am used too. And testing is so difficult because it warms up quick and then I have to wait serval hours to try again.

 

[blu73]

just an idea, but could you not fit a thermostat with innards removed just for messing about with cold running, they take ages to warm up like that - could make the rad fan run constantly too.

 

[woj]

Too much hassle to be honest, besides, starting it a couple of times takes it out of the proper cold state anyhow, no matter how cool the thermostat keeps the engine.

But no worries, I think I found the core of the problem. Below two CTS correction tables for VE (these are separate from the actual cold enrichment fuel correction tables), one from Sei program (that I kept using), the other one from a Panda 1.2 program. Interestingly enough, the Panda 1.1 program has still lower values than the 1.2, even though the engine is practically identical to the Sei one. So it turns out the correction values in Sei are exceedingly high, I wonder why In any case the values explain why I get AFR of 10.2 or so around 0 degress Celcius instead of 12.0-12.5. Add slight after start enrichment to this 10.2 and you get poor starting