Siamese Code Trial - Take Five - R
Turbo Mini Forums

The "R" is to differentiate between mine and Paul's.....

First, 4 hours of total frustration trying to get the idle control to work....

The IACV just doesn't do what I expect in either closed loop or open loop control - It is NOT a faulty valve, I had it on the JimStim a few weeks ago and the actual position matched the readout in Megatune but the actual demand (as shown in Megatune) bears little resemblance to what I would expect....

The valve is out of the throttle body, attached to my "bench" (you may see it in some of the photos) and it does what Megatune says it should be doing but it just isn't right !!!!

Whilst I've most likely set something wrong, I did have a long search on the MS-Extra forum and found several interesting threads.....

Anyway, I taped up all the holes in the TB where the valve should be and set the throttle plate manually to get about 1K RPM.

I haven't bothered setting up proper logging yet, just photos of the wideband displays.

Bottom is inner, top is outer.
All this is just playing with the mixture, injectors set at 90, 90 with equal tables.



Mixture too rich obviously



trying to make it a bit weaker by adjusting "req'd fuel"



and trying a bit more....

But at this stage I was struggling to get the AFRs up just by changing the "req'd fuel" value so I went into the injector parameters and found I had the timing set at 2mS instead of 1mS (don't ask me how, I don't know...) but as soon as I set it to 1mS everything went seriously weak....



But it still ran although the idle was now a bit erratic.....

Upping "req'd fuel" to compensate


That is finally what I was expecting, the interesting point being that when rich, the AFRs are the wrong way around.

That is exactly the same as I saw on the carb when I first tested the wideband contollers out about 6 months ago.

Edited by Rod S on 4th Jun, 2009.

Schrödinger's cat - so which one am I ???

Exciting stuff.

Great stuff.

I would reduce your injection advance to see if that richens up the outer cylinders.

You should be able top get them close at idle with two pulses.

With the IACV, mine now only wants very little opening steps to get the right idle at this time of year. Is your CLT sensor giving ambient temperature at start?

Saul Bellow - "A great deal of intelligence can be invested in ignorance when the need for illusion is deep."
Stephen Hawking - "The greatest enemy of knowledge is not ignorance, it is the illusion of knowledge."

Schrödinger's cat - so which one am I ???

Any progress?

I would just leave the IACV control to open loop for now. One less thing to worry about.

I had to reduce the IACV table considerably to get a low enough idle with the latest code.

Saul Bellow - "A great deal of intelligence can be invested in ignorance when the need for illusion is deep."
Stephen Hawking - "The greatest enemy of knowledge is not ignorance, it is the illusion of knowledge."

Sorry Paul,

No progress on MS at all this weekend.

After the frustration with the IACV, I decided to leave the engine alone this weekend and start some of the work on the shell....

The frustration is I tried open loop control and it didn't work either...

Normal assumption would be I have the wiring to the stepper coils wrong but, if I power up/down (closed loop) the pintle runs up to the closed stop (a screw in my workbench) hesitates a second and winds back out to the start position.

This is exactly what I would expect... but the "start" position seems random and bears no resemblance to the "steps" in Megatune if I display the steps as a guage.....

And when I try other settings, the "guage" in Megatune doesn't even display what I would expect, especially on the JimStim......

I really don't want to run the engine again until I've got to the bottom of this (it's shot up to 3K RPM on start three tiimes now hence the IACV being moved to the bench and the ports taped over...) but I won't be happy about adjusting all the other settings until I understand why the IAVC doesn't do what I expect.

Just frustration at the moment - same as I've seen in some of your earlier posts - but I want this right before I continue.

So...... today, work on the shell only......

I'll try testing the IACV again on Tuesday.

Schrödinger's cat - so which one am I ???

The problem then is usually the "closed" position. If it is not working consistantly then it is usually due to it not returning correctly.

I would start by adjusting the "Start Value" in the Idle Control menu. This needs to be higher than the maximum number of steps required to close the valve. Mine is set at 360. Most valves need far less than that.

There is also the "cranking steps" and the "Idle Steps" to adjust. It may be that the cranking steps are too high for your valve.

Saul Bellow - "A great deal of intelligence can be invested in ignorance when the need for illusion is deep."
Stephen Hawking - "The greatest enemy of knowledge is not ignorance, it is the illusion of knowledge."

After much searching the MS-Extra forum and playing indoor with the JimStim with the IAVC sitting next to me, I think I've found most of the answers in here
http://www.msextra.com/viewtopic.php?f=91&t=30441

Although there is still a lot of setting to do on the engine, I have finally got it to do what I expect and want on the JimStim.

It probably didn't help that in all my previous reading of the standard code the 0-255 runs in one direction and in the Extra code it runs in the opposite direction.......

Schrödinger's cat - so which one am I ???

Yeah, things have changed quite a lot on that front in the last few releases of the base code and the documentation is all over the place. And I can see how it would be confusing to get it to do what you want if it's going the opposite way of what you're expecting.

Hopefully, it's going to be easier from now on. There are a few reports that the close loop is working very well once the PID parameters are correctly set. But you need a good tune around idle with a very flat response for it to regulate idle correctly. So that should be the first item on your list.

Jean

http://www.jbperf.com/

So I got the IAVC working the way I wanted on the JimStim but before I put it back on the engine I decided to look at the PWM for low z injectors to make sure I'd got that about right.



Lovely waveform, just what I was expecting !!!

But wind the RPM up to near 5k



Pulses very close to merging despite my calculations..

And, closer to 5K


Far top left and near middle bottom right, the part pulse lost.

We know the reasons, Jean has explained, but I thought I was working well away from this....

Schrödinger's cat - so which one am I ???

And when I ran it for real....

No problems getting it running, but the closed loop idle control.... !!!!

I now have the IACV correctly wired but I started with the classic unstable PID control loop.....

It reminds me of my apprenticeship - playing with PID loops on simulators.....

I started with a stable idle, then it started oscilating and the oscillations got bigger and bigger and then it was swinging between 500 and 3500 RPM with the poor IACV trying to follow !!!

More fun tuning to follow but at least I now have a working IACV.

Schrödinger's cat - so which one am I ???

Rod,

The disappearing pulse will be fixed in the next release but it will still not be good to run this close (as you know). And PWM looks to be doing what it should.

As for close loop idle, even if your idle is stable at the correct RPM, you also have to make sure it is stable when above and below this. What I mean is that if there is more load (or less air) and RPM is lower, it has to be stable there and not want to go up or down by itself (same thing with higher RPM). Otherwise, the PID control will try to fight with this self-correction and you'll have wild oscillations that will be very difficult or impossible to correct by tuning the PID parameters. You probably already knew that but I just wanted to possibly save you some time and frustration.

One more thing about the idle PID control. It might behave differently from what you expect from previous experience. I haven't played with it and have very little experience with tuning PID loops but this one requires you to start with tuning I then P and maybe D. From posts on the forum, anything else makes it quite difficult to tune.

Jean

http://www.jbperf.com/

Great stuuf. Let me know your closed loop IACV settings when yu have it sorted. I might try that at some time.

Saul Bellow - "A great deal of intelligence can be invested in ignorance when the need for illusion is deep."
Stephen Hawking - "The greatest enemy of knowledge is not ignorance, it is the illusion of knowledge."

And then I ran out of petrol.....

I'd just got the idle working fine and dropped the outer VE table to equal the AFRs and while I was getting my camera to photo the widebands.... the tank ran out...

I'll set up some proper data logging and try again tommorow...

Schrödinger's cat - so which one am I ???

Erm....

If your outers were running lean as the last photo above, then I would have expecte you to raise the VE for that pulse, not drop.

However, before adjusting independant VE tables and introducing another variable, I would have checked that the difference in AFRs was not due to the injection timing. 90 degrees is a bit high and some of the outer pulse may be getting into the innner cylinders.

But, it is good to have a different approach to me as I may have been doing it all wrong

Saul Bellow - "A great deal of intelligence can be invested in ignorance when the need for illusion is deep."
Stephen Hawking - "The greatest enemy of knowledge is not ignorance, it is the illusion of knowledge."

Mental block between doing it in the garage, finding my camera batteries were flat, running out of petrol, then eventually typing on the PC....

It was VE table 1 I dropped, ie the inners.

By the time I got to the PC I was back to thinking 1 is outer (as in 1 and 4....)

Anyway, pulsewidth 1 dropped and the inner AFR went up to match the outer. Unlike the earlier photo, I had already got the outers to ~14 and the inners just below 13 by adjusting "req'd fuel" so it was logic to try to weaken the inners.

I had tried moving the injection timing earlier but it made no noticeable difference, but when I thought about it at the time I decided that was logic in this scenario.... idleing at 900 RPM, nice high vacuum, pulse width was about 2 - 2.5 mS. At that speed a 1/2 engine revolution takes ~33mS so the pulses are nowhere near valve overlap.

Obviously once speed/load goes up, its a different scenario.

But in this scenario, nice short pulses pretty much when the respective valve is wide open, a change to the VE table to force a change in pulse width seemed the next logical test.

However, yesterday was a bit haphazard, no logging set up, just me experimenting to get a basic idea of how it all works. When the engine just stopped for no apparent reason (until I realised it was out of petrol) I even forgot to save the final msq when I had perfectly matched AFRs....

I must be more methodical in future - maybe I should write myself a "method statement" after all.....

Schrödinger's cat - so which one am I ???

Erm.....

If you are firing the injector at the 90 degree setting, that is 180 degrees before the valve is wide open.

But, as you say, probably a "real" VE issue.

Saul Bellow - "A great deal of intelligence can be invested in ignorance when the need for illusion is deep."
Stephen Hawking - "The greatest enemy of knowledge is not ignorance, it is the illusion of knowledge."

Schrödinger's cat - so which one am I ???

That's not how I understand it, but then I may be wrong.

The timing is in "advance" degrees, just like ignition timing.

It is realtive to TDC, but 90 degrees is before TDC.

You have to consider that the air speed in the port is very low at idle and it takes some time to get from the injector to the valve.

Saul Bellow - "A great deal of intelligence can be invested in ignorance when the need for illusion is deep."
Stephen Hawking - "The greatest enemy of knowledge is not ignorance, it is the illusion of knowledge."

Hmmmm.

As you watch the pointer on the Megatune display it shows 0 degrees at the top and plus 360 degrees to the right (clockwise) and minus 360 degrees to the left.

So I have assumed TDC is the zero position at the top and my setting (positive) to the right is ATDC.

If it is an advance table, I take your point, I have been reading it backwards.

Jean, can you clarify please......

EDIT - just so there is no confusion, this is how I'm running (simulated on JimStin inc. vacuum from a syringe...)



I have been assuming this means 90 degrees After TDC, am I right or is it an advance table ???

BTW, is there any way of keeping the alternate dials in Megatune (ie, like the MAP one I have here)..... every time I change them, they revert to the originals when I restart....

Edited by Rod S on 10th Jun, 2009.

Schrödinger's cat - so which one am I ???

The timing is for advance like it is for ignition timing. So a positive value is for BTDC and a negative is for ATDC. The best thing is to refer to this pic that I now have on my web page:


The Megatune dial is just that: a dial. So it will show numbers going higher in a clockwise way. The timing values are going in a counterclockwise way which may be confusing until you think of advance. It is more confusing than ignition advance because we're dealing with angles that cover the entire cycle instead of just a few degrees around TDC but it is consistent.

Jean

http://www.jbperf.com/

And for the Megatune dials, as miniminor says you can change the ini file to make that permanent. Or you can use TunerStudio and it will ask if you want to save your changes when you exit.

Jean

http://www.jbperf.com/

Phew, that's a relief.

My calcs actually show that it takes around 220 degrees of engine rotation for the air to get down the port at idle.

The fuel will take less time, at idle, because it exists the injector faster than the speed of the air.

Saul Bellow - "A great deal of intelligence can be invested in ignorance when the need for illusion is deep."
Stephen Hawking - "The greatest enemy of knowledge is not ignorance, it is the illusion of knowledge."

Oh well,

The last fundamental mistake I made was thinking VE table 1 was for cylinder 1 (and 4)....

At that time I said "at least I'm just running it on a scope, not an engine...."

I'd better revise that phrase to "at least I'm only running it on an engine on a trolley on the garage floor and not in a car....."

As they say, every day is a school day.... I've got to run wifey to the airport tommorow so I'll fill up a few cans of petrol on the way back.

Re the transit time, I'm assuming (or should that be hoping?) that mine is really low because the single injectors (per port) are as close to the valves as I can get them and are large....

We shall see tommorow.

However, the change in VE tables was far more effective than changing injector timing yesterday but that is probably down to my misunderstanding of what the injector timing is so was not making the right changes....

Schrödinger's cat - so which one am I ???