Nearly stalling when starting from cold

[cliffb75]

Jorgen I hear what you are saying, and agree with the principle.

In this case though the IAC isn't doing much after start. In the first log where the engine showed the same engine speed profile, it just stayed at 66, so while what you are saying is correct in principle, I'm not sure it applies here.

Either way, that still doesn't address the issue thats bugging me - what the hell is happening at 85 secs? Something must cause the engine revs to suddenly increase. For those few seconds the speed is high and the MAP low, and then it all comes back under control, to similar values (MAP and injector pulsewidth) that it had before the speed increased. The only difference is that the IAC duty has ramped down. It suggests there is extra air coming from somewhere, but what would suddenly allow extra air after 85 secs? Either that or the ignition suddenly advances, but if you're still using a dizzy, I don't see how (unless that has some weird afterstart retard function?)

You're right Tony, it has got me rather baffled.

[Tony C]

 

im glad my car is still being a bitch after 10yrs

[Jorgen]

Jorgen I hear what you are saying, and agree with the principle.

In this case though the IAC isn't doing much after start. In the first log where the engine showed the same engine speed profile, it just stayed at 66, so while what you are saying is correct in principle, I'm not sure it applies here.

I think that it does, my car does the same and it has no idle control. My theory is  that the idle control seem to be too slow to handle the panic condition when it goes from start to run mode and find itself 5-600rpm to low. What it doesn't know is that the engine is allready shooting for 1500-2000rpm because of the high plenum pressure. If the idle control is quick enough to open up the valve a bit it will make the condition even worse.

The engine will shoot toward high rpm and will then run out of air as the plenum pressure has been equalized. This will make the engine rpm fall very quickly, my engines rpm will fall like a rock to 200-300rpm after the overshoot if there is no idle control. It doesn't help if the idle control has paniced and closed the idle air valve when it sees the overshoot.

We first need to try to set up the idle control to deal with this problem quickly to prevent the overshoot at the beginning as much as possible and then it need to quickly open up the idle air valve when the rpm start to fall after the overshoot. I don't know how to handle this as most of it is out of our control but increasing the afterstart rpm to 1400-1600rpm to let it try to stabilize at that rpm for 2-6 seconds before decreasing to the normal cold idle could be a workaround. Some idle PID tuning will most likely make it work better as well.

Can the rpm buildup at 85 second be reproduced? I ignored it as I assumed that the TPS calibration was out.

Jörgen

[Tony C]

We first need to try to set up the idle control to deal with this problem quickly to prevent the overshoot at the beginning as much as possible and then it need to quickly open up the idle air valve when the rpm start to fall after the overshoot. I don't know how to handle this as most of it is out of our control but increasing the afterstart rpm to 1400-1600rpm to let it try to stabilize at that rpm for 2-6 seconds before decreasing to the normal cold idle could be a workaround. Some idle PID tuning will most likely make it work better as well.

where do i set that
as every where i look there is afterstart this afterstart that 

thanks for all your help guys.

[dnb]

There's a setting in the "general idle controls" thingy in megatune for setting an addition to the idle speed for afterstart.  (I think - it's been some time since I used Megatune!)

As for PID tuning, I sent Rob a recipie for doing this that I used sucessfully on mine.  I will see if I can dig it out.

[rob@vems.co.uk]

I was going to post that this morning but I realised that its on my laptop at home rather than my PC at work
Plus its your work so you may as well take the credit/flak for it! 

[dnb]

LOL.  I have the same problem working on at least 4 different computers in a day...
I'll dig the words out when I get home.

[cliffb75]

Are you talking about a ziegler nichols oscillation method? http://www.jashaw.com/znclosed.htm

Works as a starting point, but generally still needs fine tuning.

Also for idle speed control, the PID terms generally need checking at a range of temperatures as you find that the optimum terms when its warm tend to be too aggressive when cold, since the IAC is a volume device not a mass device, so the load varies more when the air is dense.

In modern systems we use modelled engine drag torque to predict the required air mass, hence avoiding overshooting the target. We also tend to use a variable target to improve the return to idle. For the start and afterstart there are also various conditions on what controller parts turn on when (i.e. hold out the air path integrator until a certain condition) to avoid the problems Jorgen is describing with integrator windup due to beeing a long way from target.

However, I still reckon looking at Tony's log data we are not getting the valve closing causing the problem.

If Tony was running spark control we could use the 'ski ramp' ignition timing I described somewhere else to help this situation. Buts thats a sticky plaster really, not a cure for the problem.

[dnb]

A very slight variation on strategy to get to the ultimate gain, but yes, it's the same old ziegler nichols method in the end

I did find that most of my starting issues went away if I cracked the throttle a little while starting.  I always presumed this was down to the stepper being turned off, but I'm not so sure now.  I need the engine back in the car to play some more!!

[Tony C]

so what do i do at present?
not touch any thing and wait???
or adjust some settings??

Cheers

[cliffb75]

so what do i do at present?
not touch any thing and wait???
or adjust some settings??

Cheers

Play of course! 

For one thing, thats why you went for VEMS, so you could get stuck in and have a go.

And second, it ain't gonna fix itself.

The important bit is to be meticulous with your record keeping. Last time I did start calibration work I made a simple sheet with date, ambient temp, coolant temp, calibration name and log name on it, and filled it in for each start I logged. Later on (end of the day normally), I looked through the logs and added time to fire, peak flare (this is delta above target), lowest lambda (rich dip) and any notes worth mentioning (speed dip, lambda went lean etc etc). This allows you to quickly go through and see what is good and what is bad, and pick the calibration settings accordingly. You can then steadily home in on a good calibration. As you build up this databse of info, you will also start to get a 'feel' for what the engine wants, so if you do find a poor start in the future you wil have an idea how to sort it.

Go for it!

BTW have you got the lambda working right from the start yet? This will really help you out with knowing which way to go on the fuelling.

As long as you save the changes and note what is good and what is bad, you can always step back.

[Tony C]

no still not got the lambda working from the start yet mate.

[Jorgen]

I can't tell you what to do as I have very little experience with the idle control. But if I would start by lowering the integrator limits. (Settings->extras->idle settings advanced/PID.) Integrator increase limit is currently at 112 (or something like that) and integrator decrease limit is 64. Decrease these to 64 and 32 respectively and see if it changes anything.

I actually doubt that the above will make a big difference but I think that it's a change in the right direction and that it will work better together with future cold start tuning.

My next step would be to change the "Idle afterstart rpm added value" to 400 to make the target rpm 1400 or 1500rpm right after startup. This should allow the natural oscillation of the engine to stop before it drops the idle target to 1100rpm or so after 4 seconds. As I'm not a software guy I have no idea if the added rpm is just removed after 4 seconds or if it fades out over 4 seconds. But I guess that you can tell me in a few days If it's just removed after 4 seconds we will add this to the task list for the programmers.

Jörgen

[Tony C]

I can't tell you what to do as I have very little experience with the idle control. But if I would start by lowering the integrator limits. (Settings->extras->idle settings advanced/PID.) Integrator increase limit is currently at 112 (or something like that) and integrator decrease limit is 64. Decrease these to 64 and 32 respectively and see if it changes anything.

My next step would be to change the "Idle afterstart rpm added value" to 400 to make the target rpm 1400 or 1500rpm right after startup. This should allow the natural oscillation of the engine to stop before it drops the idle target to 1100rpm or so after 4 seconds. As I'm not a software guy I have no idea if the added rpm is just removed after 4 seconds or if it fades out over 4 seconds. But I guess that you can tell me in a few days  If it's just removed after 4 seconds we will add this to the task list for the programmers.

Did both the above,
and yup still the same, if anything it felt worse, as if it tried to die twice,
did a datalog of the start up also

[cliffb75]

Well lets see the log then