Idle Control - methods?

AbeFM · Post by **AbeFM** » Wed Aug 28, 2013 12:46 pm

So, while there's still a fair amount of confusion on things like polarity, etc, I've got boost control somewhat working, and P, I, and D have effects at least related to the theory - i.e. a P-only controller actually works to some degree.

My question - does the current (release) idle code use the same PID code, or does it use code from some years ago.... I.E. if I use P-only control, will the idle valve open more as rpm dips?

Post by **muythaibxr** » Fri Aug 30, 2013 7:54 am

They all use variations of the PID listed here:

http://bestune.50megs.com/typeABC.htm

Boost is a type B controller, Idle is a type C loop.

With type C, the P term is decoupled from error, so yes if RPM drops, duty will go up.

Neither of those will allow you to hold a target with only a P term set. Type B's P operates on change in error, type C's operates on change in process variable.

Ken

AbeFM · Post by **AbeFM** » Fri Jun 27, 2014 7:38 pm

Hello everyone!
After many happy years of using PWM-Warmup, I've got a new motor, a new computer, and thought with some of the changes to the idle set up, I'd give things a whirl.

I'm happy to report that with next-to-no effort, the idle is basically working!

But I'm still having one issue, and I'm thinking that it's better fixed in software than settings. Of course, if it IS settings, I'd love some enlightenment.

I'm running an MS3Pro, (1.3.1 just installed today), and thought I'd try closed loop - the sensitivity slider seems to be a great addition. My problem is I have a ridiculous stereo, and when that, the cooling fans and headlights are all on at the same time, it can take some significant idle valve motion to keep things afloat - especially in response to those step-function loads.

My issue is this - due to all the accessories, the "last value" seems to be the best choice. However, I'm having this issue where at the last second before hitting the gas and leaving closed loop, the valve may be at 50 or 60%, verses the 22-32% it needs. When I have things set to close the valve after 2 seconds, but perhaps it is not? Anyway, as you can imagine, the motor "idles" at 3k and won't come down.

I've done this both in "simple" mode, and with PID=100/100/2 (I'm assuming simple mode is just 100/100/0? Neat trick that!).

What I was thinking was how great it would be if I could use the idle-table values as a maximum cap. But only if there isn't something I'm missing, like a max rpm or something.

Anyway, tell me where to look and I'll start looking. Most of the time everything is fine, but sometimes it'll just sit at high rpm. No reason to beat a dead horse over what the P-term "should" do, I just want it to work. Thanks!

Post by **muythaibxr** » Sat Jun 28, 2014 7:09 am

You can probably fix this by adjusting your lockout settings. RPMdot might be jittering too much up there, or MAP might be too low. Also, are you using the built in fan control to turn the fans on or something else?

AbeFM · Post by **AbeFM** » Sat Jun 28, 2014 2:56 pm

Which lockout settings would you recommend? Is there some portion of logs worth posting (presumably the events surrounding leaving CL idle the time it stuck)?

I thought I could bring it back down by reving over the idle-close limit (6500 on my set up), then it would "catch" on the way down, but it didn't.

What ROMdot/MAP influence are you talking about (i.e. which settings again)?

Thanks!

hardline · Post by **hardline** » Sat Jun 28, 2014 8:10 pm

idle.PNG

Shorten your PID Delay and Ramp to Target times. This is the settings I am running on my setup.

Also, Notice the typo?

Post by **muythaibxr** » Sat Jun 28, 2014 8:45 pm

Those are not the settings I was referring to. I was referring to the RPMdot setting and MAP setting. I will fix that typo.

AbeFM · Post by **AbeFM** » Sat Jun 28, 2014 10:23 pm

Yeah - I set that to around 3 as you did. I don't remember the typo, perhaps it is a different version. How would RPMdot and MAP help? When a big load hits, it would cut out? Seems reactive - i.e. too low, but I get it I think. You'd suggest lowering RPMdot and raising MAP?

I guess there is no way to get "RPM - target >> 0" to close the idle valve?

Post by **muythaibxr** » Sun Jun 29, 2014 5:56 am

Your problem is that PID is not engaging. So either there is too much jitter in actual RPMdot (so raising the RPMdot setting will fix that) or MAP is too low (lowering the MAP setting will fix that). In a future release I am going to rework the PID engagement code to make it easier to control when PID engages.

You don't necessarily want RPM - Target to engage PID... then PID will engage in overrun and during other odd situations.

AbeFM · Post by **AbeFM** » Sun Jun 29, 2014 5:27 pm

Is there an easy way to tell from the logs if PID is engaged or not?

Post by **muythaibxr** » Sun Jun 29, 2014 8:29 pm

You can look at status2. status2 & 0x80 will give it to you. In the log viewer, it'll be 128 when idle engages (unless you're using another feature that toggles a bit in status2, in which case you might have to create a custom field in the log viewer that does the bitwise and from above).

Ken

JAM · Post by **JAM** » Mon Jun 30, 2014 7:49 pm

I have just started to mess with the closed loop idle with MS3X (current release) and a PWM valve with and have some questions as well;

"Closed loop idle PID gains" has the slider with closed loop sensitivity (0-4000), and then advanced mode is all 3 variables open. I thought that the basic was just the slider but i see it also is active in advanced mode. i have messed with the slide by itself. Can someone explain how the 0-4000 slider works in conjunction with the PID settings in advanced mode? or in other words how do you tune these 2 sets of parameters together?

muythaibxr wrote:You can look at status2. status2 & 0x80 will give it to you. In the log viewer, it'll be 128 when idle engages (unless you're using another feature that toggles a bit in status2,

I am doing some tuning to attempt to make the rpm the same in and out of gear (auto trans) and it does so but with very long delay and on the way up it never quite gets there. I see status2 go to 128 right away, so it is working and i see the duty cycle on the valve slowly increasing so its doing but it may take 20 seconds to get there

Post by **muythaibxr** » Mon Jun 30, 2014 8:45 pm

The sensitivity slider controls the sensitivity of the whole algorithm. Think of it as a gain for all three PID values at once. The idea is to tune the slider to give you good behavior reaching the target, then if you need go to advanced mode and tune the PID values with the sensitivity you set in basic mode.

AbeFM · Post by **AbeFM** » Tue Jul 01, 2014 5:26 pm

I was just asking what is "basic mode" - is it just P=100, I=100, D=0?

The settings seemed to help a lot! I was able to bump up dashpot some to help, though I'm still getting stalls as the engine comes down - probably I'll have to go to the initial values table but I'm trying to avoid it - it is so hard to predict what the load on the motor is.

As an aside: I've always noticed braking hard makes the motor want to die - I can't imagine vacuum assist brakes could be doing this, but I'm out of other ideas?

JAM · Post by **JAM** » Wed Jul 02, 2014 4:38 am

muythaibxr wrote:The sensitivity slider controls the sensitivity of the whole algorithm. Think of it as a gain for all three PID values at once. The idea is to tune the slider to give you good behavior reaching the target, then if you need go to advanced mode and tune the PID values with the sensitivity you set in basic mode.

thanks, so do the advanced settings basically give the ratio (relative effect) between P,I, & D and the slider adjust the overall sensitivity of all 3 at once?

Post by **muythaibxr** » Wed Jul 02, 2014 1:51 pm

Yeah, you could look at it that way.

Post by **muythaibxr** » Wed Jul 02, 2014 1:53 pm

AbeFM wrote:I was just asking what is "basic mode" - is it just P=100, I=100, D=0?

Yep.

The settings seemed to help a lot! I was able to bump up dashpot some to help, though I'm still getting stalls as the engine comes down - probably I'll have to go to the initial values table but I'm trying to avoid it - it is so hard to predict what the load on the motor is.

Usually stalls are fuel or timing related. Do you have logs?

Ken

AbeFM · Post by **AbeFM** » Wed Jul 02, 2014 2:06 pm

Ok, I'll look there. I'd played pretty fast and loose with the tables around/under idle trying to get higher AFR's after I had CL idle in place.

It seems when it's cold, a higher dash-pot would solve some issues. Or mask them

richyvrlimited · Post by **richyvrlimited** » Fri Jul 04, 2014 3:56 am

AbeFM wrote:Ok, I'll look there. I'd played pretty fast and loose with the tables around/under idle trying to get higher AFR's after I had CL idle in place.

It seems when it's cold, a higher dash-pot would solve some issues. Or mask them

And yet when cold i want zero dashpot, (causes an unnecessarily high idle) but need it when warm.

arran · Post by **arran** » Fri Jul 04, 2014 7:46 pm

richyvrlimited wrote:
AbeFM wrote:Ok, I'll look there. I'd played pretty fast and loose with the tables around/under idle trying to get higher AFR's after I had CL idle in place.

It seems when it's cold, a higher dash-pot would solve some issues. Or mask them

And yet when cold i want zero dashpot, (causes an unnecessarily high idle) but need it when warm.

Dashpot is just going to have a transient affect on the idle, it adds duty for a short amount of time to make it a soft landing into the idle set point

Megasquirt Support Forum (MSEXTRA)