Crank to run taper not obeying!

Post by **muythaibxr** » Sat Jan 10, 2015 4:21 pm

Not sure I understand your question.

arran · Post by **arran** » Sat Jan 10, 2015 6:41 pm

Crank to run taper now set to 2, i'm going to the hardware store in a few minutes so can test then

I note you have fixed the typo in the name of that variable

piledriver · Post by **piledriver** » Sat Jan 10, 2015 10:07 pm

muythaibxr wrote:Not sure I understand your question.

I'm suggesting his closed loop idle initial value table settings may be a contributing factor, or perhaps the cure.

arran · Post by **arran** » Sun Jan 11, 2015 2:12 am

I have spent a LOT of time tuning the initial value table. With the current settings for fan idle up, ac idle up, dashpot and warm up idle speed adders together with the changes in intake air temperature as summer progresses here in Brisbane I have tweaked it to be fairly good.

I drove the car this afternoon with the 2 second value in Crank to run taper and the problem did NOT go away (I did need approx. 10 starts to make the problem appear though). However Ken specified the setting as "PID Wait timer" so I changed "Crank to run" back to 1 and changed "PID delay" to 2 and was NOT ABLE to make the problem return after another 12 starts.

So the theory is looking correct!

The longer PID delay however resulted in the car behaving undesirably every time it returned to idle when driving. If I cruise at 3000rpm and push in the clutch the rpms fall rapidly and the engine falls to 600rpm for a second before CL picks it up to target.

I'd like to understand more about the code around idle conditions.

Post by **muythaibxr** » Sun Jan 11, 2015 7:00 am

If the engine is falling back to 600 rpm then the initial value table is wrong or you have something going on with fuel or advance. The code is designed so that the engine is supposed to be stable above the target before PID engages. Can you log this behavior?

I fixed the timer so the error is up to .1 sec but based on your logs you would still have this issue with it set to 1 sec at least some of the time. Your engine sometimes takes around a second or a little longer to finish the initial RPM acceleration after crank. With the new code 2 sec would still work though.

The code is rather complicated around the idle conditions since as it has grown I have grafted in features as people have asked for them and it has become a bit unwieldy. I am in the process of simplifying it while maintaining functionality. I could use some testers in 1.5 to try the engine state feature. Right now it just logs data, and I want to see how the data matches the actual engine state once tuned. If that works all the features that require a TPS threshold and a MAP threshold and other parameters will switch to using the global engine state. You will get the engine states to match reality, and then all the other features will trigger consistently with each other.

arran · Post by **arran** » Sun Jan 11, 2015 1:50 pm

I can test some code.

I'll log the behaviour as it returns to idle and reply. With the 1 second delay it was working well, ac on or off. This is why I asked about the idle code logic (flow diagram?) ....does the initial value table + dashpot get consulted before or as Closed loop idle is entered? I am frequently reverse engineering the logic through trial and error. I think the idle speed of the warm engine on the throttle stopper with no idle valve correction is going to be around 600rpm. It has been a while since I checked that, which i'll do

Can you confirm the 0.1 second max error applied to the 1.3.4 b0?

piledriver · Post by **piledriver** » Sun Jan 11, 2015 6:44 pm

From what I have seen it sets the initial IAC values on entry to CLidle.
As noted they should be set just on the high side of the static values to prevent a stumble/stall.
idle-ve table and idle timing kicks in here, too, if active.
Cold advance is also useful.

I wonder if TS could be set up to show all the active idle features/tables in one view?
(might not work so hot on a small screen)

arran · Post by **arran** » Sun Jan 11, 2015 7:42 pm

piledriver wrote:From what I have seen it sets the initial IAC values on entry to CLidle.
As noted they should be set just on the high side of the static values to prevent a stumble/stal l.
idle-ve table and idle timing kicks in here, too, if active.

Here is where I'm not 100%. If as you say the initial value table is consulted on entry to closed loop idle how can this be any use at all if the ecu is waiting two seconds to engage closed loop? Same logic for dashpot adder.

Experience tells me, with a 1 second wait, that the drop to idle rpm is softened as initial value and dashpot adder are increased. As ken said i need to examine the exact response in a data log showing the difference between 1 sec and 2 sec. I really don't want to have to retune the initial value table as I have months of tweaks in the existing one.

Post by **muythaibxr** » Sun Jan 11, 2015 8:12 pm

The initial value table is used as soon as you lift the throttle (unless you tell the valve to stay closed above a certain RPM in which case it goes to the table as soon as you are below that RPM). The PID delay countdown starts as soon as all other conditions for entering PID are met (RPM not falling, MAP above the MAP threshold, TPS below the threshold, etc...). So think of PID delay as the amount of time after RPM becomes stable before PID will engage.

The initial values table should be set to give RPM that is at or slightly above the target, and the dashpot adder can be used to get you slightly higher still above the target on throttle lift.

If the RPM drops below the target before PID engages when set to 2 seconds, the initial value table or dashpot adder need increased, or you have a fueling or timing issue. You do have a fueling issue anyway; some of your logs show oscillation in AFR causing PID to react making the oscillation worse.

Also as I said before, a lot of the settings that are confusing are in the process of being simplified.

Ken

Post by **jsmcortina** » Mon Jan 12, 2015 4:39 am

Perhaps a flow-chart of the logic would help? That could be added to the manual.

James

arran · Post by **arran** » Mon Jan 12, 2015 7:22 pm

Thanks for the explanation of some of the idle variables Ken.

Took a data log this morning, zipped and attached.

The interesting thing is that there isn't much to see. Nothing changed from my last post when I mentioned that the crank to run taper was fixed but that the return to idle under normal driving was hitting 600 before recovering. I have not adjusted the throttle stopper. On this drive it behaved perfectly, everything behaved (not yet tested with AC though)! I have sort of seen that behaviour before, the ECU sometimes does not listen to changes immediately, even though I did a couple of dozen power cycles after applying the 1.3.4b0 code.

I'll continue to monitor and see if the situation changes.

James a flow chart would be splendid.
Also, Ken can you confirm that the 0.1 timer resolution was included in the 1.3.4b0 code?

2015-01-13_07.38.38.msl.zip

Post by **muythaibxr** » Mon Jan 12, 2015 9:06 pm

That code was not in the rev I gave you. It is checked in though so will be in the next release of any version (1.3.x, 1.4, 1.5)

arran · Post by **arran** » Tue Jan 13, 2015 3:11 pm

OK, I might revert the firmware back to 1.3.3.

Drove the car again this morning, over Mount Cootha for a bit of morning exercise on the push bike. The return to idle worked well on the forward trip, but played up when I jumped back in to drive home, dropping down to 600 briefly before CL activates (after 2 seconds). I took the laptop with me, just in case, so I fired it up. With the understanding of the idle process now improved thanks to this thread, as rpm is dropping through 2000 where the idle valve is armed I can see it loading the initial value from the table loading. Because the car was warm and heat soaked it was on a part of the table that needed more initial duty, quite a lot more in the 925 column.

I fixed that and it behaved on the way home. I'm not sure why changing the pid delay from 1 to 2 secs made that config problem evident.

I also checked the behavior with the AC on and it was great.

Arran

Post by **muythaibxr** » Tue Jan 13, 2015 5:39 pm

Glad it helped.

piledriver · Post by **piledriver** » Tue Jan 20, 2015 3:30 am

I just fixed my rough running immediately after cold start (only) issue...
needed to adjust my valves.

Had them set for zero lash (solid lifters/CrMo pushrods) and with the weather, it was a little past zero on cold mornings.
Would clear up ~about the time ASE kicked off, regardless of settings.

It's easy to overlook simple things like that, it will never show up in the logs except in extreme cases.

arran · Post by **arran** » Tue Jan 19, 2016 11:25 pm

Wanted to dig this thread up and comment on the crank to run function. Since the implemention of the 0.1 resolution the problem has almost gone away completely but it does still occur from time to time, I'm guessing that increasing resolution from 1 sec to 0.1 has just reduced the incidence, but not the behaviour.

Sometimes I can start the car and get no fast idle, and then simply turn off the car and start it again and it works fine. Basically same as before, just much less often.

Arran

piledriver · Post by **piledriver** » Wed Jan 20, 2016 3:09 am

I set the idle timing/correction and idle VE to use their own settings vs. closed loop trigger option.
seems to have helped a lot, although the air (pwm iac) or something still seems to wait for the PID to kick in.

A little "kick" from a few% of dashpot adder seems to aid the recovery but the rpms drop too fast for it to be smooth in every case. (RPMDOT limit too restrictive? max is 1000)
I have beaten on the IAC preset table until its black and blue.
The revs still drop below idle sometimes, occasionally it even dies, like when coming off the freeway at ~3k then coasting to a stoplight with clutch in.
Most of the time it works fine...

Lately running current release firmware, but I had been keeping up with the dev code for years previously.

I'm going to run through the ideas in this thread again, as its the ~only thing about my tune that's not quite as I'd like.

Post by **muythaibxr** » Wed Jan 20, 2016 5:43 am

piledriver wrote:I set the idle timing/correction and idle VE to use their own settings vs. closed loop trigger option.
seems to have helped a lot, although the air (pwm iac) or something still seems to wait for the PID to kick in.

A little "kick" from a few% of dashpot adder seems to aid the recovery but the rpms drop too fast for it to be smooth in every case. (RPMDOT limit too restrictive? max is 1000)
I have beaten on the IAC preset table until its black and blue.
The revs still drop below idle sometimes, occasionally it even dies, like when coming off the freeway at ~3k then coasting to a stoplight with clutch in.
Most of the time it works fine...

Lately running current release firmware, but I had been keeping up with the dev code for years previously.

I'm going to run through the ideas in this thread again, as its the ~only thing about my tune that's not quite as I'd like.

This behavior is almost definitely fuel or ignition related. If the valve is going to the same position every time, fuel is the most likely culprit. I had the same issue on my rx7 years ago whether in closed loop or warmup only. I added some bins on the bottom of the table and tuned the ones that get hit when recovering from overrun and the problem went away.

Post by **muythaibxr** » Wed Jan 20, 2016 5:46 am

arran wrote:Wanted to dig this thread up and comment on the crank to run function. Since the implemention of the 0.1 resolution the problem has almost gone away completely but it does still occur from time to time, I'm guessing that increasing resolution from 1 sec to 0.1 has just reduced the incidence, but not the behaviour.

Sometimes I can start the car and get no fast idle, and then simply turn off the car and start it again and it works fine. Basically same as before, just much less often.

Arran

Probably not the same cause. Datalog. Settings.

arran · Post by **arran** » Thu Jan 21, 2016 7:13 am

Could see that coming

Will do.

Megasquirt Support Forum (MSEXTRA)