Nearly stalling when starting from cold

[cliffb75]

Here is an example for you. Its from a now out of production 4 cylinder 2L 16v NA. Obviously the actual numbers will not be applicable to you, but the shape and trends of the curves are what you are after. This was at +10C ambient temp after an overnight soak.



LA3 lambda is lambda , ti is injection pulsewidth in ms, nmot is engine speed in rpm and ps is MAP in mbar

Important point to note is the shape of the ti curve. Notice how much fuel it is putting in for crank and afterstart, but how quickly it is removed. This also looks even more because of the load effect, but never the less you do need to add A LOT of fuel in the first few combustions compared to once the combustion chamber has some heat in it.

Compare that to your log with 13ms during cranking and then 6 ms when running, though you are running quite rich (0..

So as a guess, try upping the crank cold pulsewidth up to around the 30ms mark and see what happens. If it won't start you have definately gone too far! So take the fuel back out and get the car started, warmed through and tray agin next day with about 20ms. If it stumbles during the runup with 30ms but then is happy you know that is a little too much for the crank fuelling, but more is needed in the afterstart. Warm the engine through and leave it for another start next day with the revised numbers (less crank, more afterstart). If its starts and runs great, either leave it if you're happy, or try adding more until you go too far.

Its not rocket science or anything. Thats how we have to do it

[Tony C]

well changed to 20ms of CPW and the same thing tonight when i went round to start her,
fired up ran then revs died to near stall then picked up again and idles as it should 

Also i know this is a bit off topic, is there any chance between us all we can get an istruction manual done for what each setting in megatune is for and does, i know many of us have work comitments, BUT i think this would help new vems users by far. ( including me )

[cliffb75]

Need to see the log - did you actually achieve 20ms at start?

Remember that pic - 90ms at 10C. You're still a long way off that.

Am starting to wonder if its something else though. Using a dizzy may not be helping if its timed on the retarded side at low load.

Send us the log of the last start.

 

[Tony C]

will have to do a log tomorrow mate as never had laptop on me..

[Tony C]

how do you get the lambda to read from crank?
as went round today, left ignition on for at least a 1min, started car and the lambda was off, as if it had not warmed up. then after about a min of engine running as normal the red and green status bar started reading 
have i got to turn something off or on in the settings some where??

once we can get the lambda to read from the start i will do another log mate,

Pity there is no one down this kneck of the woods that is clued up on the vems 

[dnb]

Where abouts are you again?  (PM me if you don't want to write it in public)  If it's within easy reach when I'm next on the mainland I could pay you a visit...  Not sure when that'll be though.

There's a lambda warmup parameter lurking somewhere in megatune.  VEMS ignores the readings until this time is up (measured from after cranking IIRC). 

[Tony C]

Im near Newbury in Berkshire Dave

[dnb]

Not too bad for me then...  I need to visit Oxford in the next couple of weeks, so will be in the area.

[Tony C]

im about 2 miles from the A34 Dave
6 miles north J13 A34/M4 interchange

i will still do a data log in the morning for every one to look over.

also any one give me a step by step guide on calibrating my WBo2 want to make sure this is correct also, i read somewhere something about a terminal program 

[rob@vems.co.uk]

You can't calibrate with a used sensor.  I'm pretty sure that we had the sensor calibrated by Richard?

[Tony C]

ok Rob,
maybe im getting a bit paranoid 

[Tony C]

Hi,
have latest datalog and .msq file below .......

http://www.capriracing.co.uk/vems/vemsdata2.zip

[cliffb75]

Right then. Your data log



So it looks like you are still too short on the crank and afterstart pulsewidth (max 13, but you're running 5 once the lamda sensor has kicked in, to get 0.85) I would try increasing it lots more - like double it again. Having said that the runup and flare are really nice

But the more interesting thing is what happens at about 85secs, same as your first log. What is happening? Why does engine speed increase? IAC then ramps down to control the idle speed, and everything looks good.

Before this point, you can see that the engine is not actually achieveing target speed, its ramping up to it. The first log you sent did the same thing (ramping up, then overspeed for a bit, IAC ramps down and idle speed under control)

Does the engine have some sort of additional device for opening the throttle stop (the CA18DET has a wax capsule thingy in addition to the idle valve)? If so, is it operating properly? is it sticking until temp get up a bit or something?

Or, Is the power to the IAC delayed for some reason? (so it thinks it is active but the valve remains closed)

Or, Does the idle valve do something wacky (like close) when the IAC duty reaches 66% (66 was the max value on your last log too), or is the signal inverted (66 means closed)?

I'm wondering if the near stall after start and the flare later are linked somehow? e.g. why does the IAC duty drop when the engine nearly stalls at 42.4 sec?

Anybody else got any suggestions?

 

[Tony C]

im glad its got you baffled too 

[Jorgen]

I have a similar starting behavior on an Audi 4.2l engine even without idle control. When the engine starts the plenum is filled with air at atmospheric pressure, this is enough to let the engine overshoot on startup. The high rpm then pull a strong vaccum which will not even out quickly when the rpm falls back down again. This will cause the pulsewidth to be all over the place as well.

First we have the PID parameters where I'm pretty lost, but the integrator limits seem a bit high. Try setting ingegral decrease limit to 32 and integral increase limit to 64 as a start but after that you should stick to the "One setting at a time" rule.. The normal problem is "integral windup" and the limits usually help a bit.

You can possibly use "Extras->idle settings->Idle afterstart RPM, added value (rpm)" to allow a larger overshoot on startup, while keeping the "one setting at a time" rule in mind you can set it to 300 or 400 and see what happens. But we also risk an even larger overshoot, it depends a bit on the flow and speed of the stepper.

I'm sure that we can get the starting behavior better.

Jorgen Karlsson
Gothenburg, Sweden.
VEMS hardware development