Closed loop idle bug?

[teg]

Closed loop idle "stalled" today without any reason? PWM stalled to 90% (fully open) even rpm was over target? Using MS2 Extra beta8

[UnaClocker]

I haven't look at your stuff, but I'd imagine it's a TPS problem.. Is your TPS calibrated? Do you have the TPS settings correctly set so that the closed loop idle knows when your TPS is closed?

[teg]

You should look that log before guessing.

TPS is calibrated and it IS zero.

[Keithg]

That is weird. First things I would ask are:
is 90% truly a good value for 'open'. It is a good idea to have this value be slightly above a normal idle value on a hot day with all accessories running.
is 37% actually the closed duty? This is the value when the valve is closed all the way. I would suspect this should be either 0 or 25% for the 2 types I have seen and used. There may be others that I am unaware of, but AFAIK...
It appears that your idle can regulate at target RPM (end of the log) at ~40. You may want to play with RPM lag to make sure it always gets a good rpmdot value. The code defaults to 50. I ended up having to use 30 to get it to reliably get out of lockout. Also, your map at idle is pretty high. Is there some reason it is at 50? (cam, A/C running, etc) If this is a normal idle MAP value, you can bring your PID lockut map value closer to your idle MAP value. If it normally idles at 50 map, you may want to try 48 as a PID lockout value.

Maybe Ken will see something I did not, but closed loop idle is pretty exacting in the values you need to have for it to work well.

HTH,

KeithG

On 8/22/07, teg wrote:

Closed loop idle "stalled" today without any reason? PWM stalled to 90% (fully open) even rpm was over target? Using MS2 Extra beta8



Opel Kadett C City 1.2i Tbo MS-II PCB V3 squirt'n'spark

[teg]

Finally got megalogviewer working and foudn that closed loop idle started to control after Gwarm went to 100 (ASE ends?)

[MegaScott]

Is 37% really your closed value? that seems kind of high. On my setup i used 15% and that was really where the valve stopped moving at the low end. (use "Warm up only" to get this) I had to lower my idle screw at the lower end to get it to idle lower than 1000rpm because the stock Ford computer expects 1volt tps at idle.

I would tweak your PID a little better. I would try bumping up your proportional gain and lower your Intregal gain a little if it's hunting.

Also I see your Idle Spark advance is pretty low, you should run this up more to get more torque at idle to make it less likely to stall. You have around 9 degrees of advance at idle, I would run closer to 20-30degrees advance at idle and you'll see better idle quality.

Oh, also your VE table lower values drop off at idle a bit at lower Map, richen it up around the edges a bit and that will help.

[teg]

Is 37% really your closed value? that seems kind of high. On my setup i used 15% and that was really where the valve stopped moving at the low end. (use "Warm up only" to get this) I had to lower my idle screw at the lower end to get it to idle lower than 1000rpm because the stock Ford computer expects 1volt tps at idle.

It is actually 37%. Valve still moves if I go lower , BUT then it will open from other end. That is kind of fail safe position in wire break situation

I would tweak your PID a little better. I would try bumping up your proportional gain and lower your Intregal gain a little if it's hunting.

I was just starting to do PID tuning (with NZ-method), so those aren't final values.

Also I see your Idle Spark advance is pretty low, you should run this up more to get more torque at idle to make it less likely to stall. You have around 9 degrees of advance at idle, I would run closer to 20-30degrees advance at idle and you'll see better idle quality.

Stock engine has 4-6 degrees idle advance. My engine is old Opel OHV engine (1.2 litres) from 70's.

[MegaScott]

Back in the 70's there was no computerized ignitions, and the Vacuum advance left a little to be desired.

Bump up just the idle advance bins and you'll see a big difference. Also richen up those bottom VE values, when it leans out like that it will have a tendancy to oscillate and stall.

[teg]

Ok, thanks for your hints and comments. Hope emissions won't (HC) won't go too high..

BUT I still have problem with Idle valve control...

[MegaScott]

Ok, thanks for your hints and comments. Hope emissions won't (HC) won't go too high..

BUT I still have problem with Idle valve control...

All the newer OEM's use a scheme where the idle advance is upwards of 30degrees advanced, this serves two purposes, one it increases torque, so accessories such as A/C and alternator do not drag the engine down so much, plus it gives you more control when letting out the clutch.

I suspect as soon as you raise your idle advance you'll see the engine idle go up to a point where you will need ot lower your throttle blade opening, this will cause the Idle control valve to be forced open more, giving you more control over your idle.

Your Proportional gain is a little low, this could cause sluggish operation in closed loop, bump it up to 20 or higher.

Also your Integral might then be a little high, this can cause excess hunting, lower that until the hunting goes away, it will still over shoot a bit, but now that you have more air and more control over your idle it should only overshoot once a small amount, then settle down to the target speed.

[Keithg]

Actually, your HC may actually improve based on what Scott says. Basically, increase timing at idle until idle quality decreases. You will note this as MAP will continue to decrease then level out. Since you are mapping timing, you will not affect any of the rest of the map, only around idle. In carb days they used to have a slightly rich idle as a slightly rich idle is more stable than slightly lean. With EFI, you can get closer to a stoich idle which helps the 3 way catalyst (if so equipped) and idle quality.

KeithG

On 8/22/07, teg wrote:

Ok, thanks for your hints and comments. Hope emissions won't (HC) won't go too high..

BUT I still have problem with Idle valve control...

[teg]

It happend again. Idle control stalled or actually had wrong target. I had ~900rpm as target but it tried to adjust idle to 1300rpm.

So this is really a bug.

[teg]

I have a feeling that it will get first bin from rpm target curve and tries to keep in that value until CLT reaches last temp bin?

[muythaibxr]

I have been unable to get the behavior you are seeing to happen. I'll see if I can find anything this week.

Ken

[teg]

Hmm..

Interpolation of target rpm has done by this way:

Code: Select all

	    targ_rpm = intrp_1ditable(outpc.clt,8,(int *)pg5_ptr->pwmidle_clt_temps,
		    (unsigned int *)pg5_ptr->pwmidle_target_rpms);

And prototype of intrp_1ditable is:

Code: Select all

unsigned int intrp_1ditable(unsigned int x, unsigned char n, unsigned int * x_table, unsigned int * z_table)

What if pwmidle_clt_temps has negative values (as I have)?

[snillet]

All the newer OEM's use a scheme where the idle advance is upwards of 30degrees advanced, this serves two purposes, one it increases torque, so accessories such as A/C and alternator do not drag the engine down so much, plus it gives you more control when letting out the clutch.

You say that newer engine managements run at around 30btdc idle advance ?, hmm, thought i was seeing ghosts when i was getting into that area myself......, thought that something must be wrong for this to work so good. Car was really nice to drive in "jammed" traffic, no need to touch the throttle at all, just clutchwork

But hey, i´m not using MS2 extra, YET(getting there soon), so i´ll shut up now.

[muythaibxr]

Hmm..

Interpolation of target rpm has done by this way:

Code: Select all

	    targ_rpm = intrp_1ditable(outpc.clt,8,(int *)pg5_ptr->pwmidle_clt_temps,
		    (unsigned int *)pg5_ptr->pwmidle_target_rpms);

And prototype of intrp_1ditable is:

Code: Select all

unsigned int intrp_1ditable(unsigned int x, unsigned char n, unsigned int * x_table, unsigned int * z_table)

What if pwmidle_clt_temps has negative values (as I have)?

Could be the problem. I'll check into it.

Ken

[jsmcortina]

What if pwmidle_clt_temps has negative values (as I have)?

I knew we should have used Kelvin..

James

[teg]

It is not, yet, too late to start using Kelvins

[rb26dett]

What if pwmidle_clt_temps has negative values (as I have)?

I knew we should have used Kelvin..

James

woooooooo hoooooooooo

please metricise the code

that would be cool indeed. properly pure scientific grade code

i actually exclaimed "yes!!" when i read that. shows just how sad i am ;-)

fred.