Basic Boost Control Advice

[AbeFM]

I just set up my car on MS3Pro, and I've gotten EBC working... Some of the time. About 1/3 of the time, I'll get the target I expect. Most of the time, I get wastegate pressure. What should I look for?

These three throttle openings happened at the exact same temperature.



I've got the boost control lower delta at 110% and the temp cut off at 130. The later isn't explained in the documentation that I can find, and I would like to know what it means.

Thanks all!

[AbeFM]

Verified this today over many, many attempts. I can even flip the throttle several times over the different boost levels in the table and get it to "catch" sometimes. As you can see from my log, it's not even TRYING to increase boost. It's pretty frustrating!! Please help!

[Matt Cramer]

Please post the full data log and MSQ in this thread.

[AbeFM]

Will attach files in second post. what I noticed is it'll work properly if initially boost builds quickly. at low rpms and/or slow tip in, it won't control boost. the higher the revs/ quicker the tip in the more it'll work. this makes me suspect the threshold setting.

will attach files, log is long, how can I edit out extra rows? I'll host on Google for now.

[AbeFM]

Cut out about 200 seconds in a text editor. It seems to load into MLV without issue.

I did ~10-15 throttle stabs in a row, rolling into it or tipping in more aggressively over a variety of road speeds and gears.

Let me know if you need more than this. I'm sure I can host it elsewhere or email you directly.

[franksidebike]

Hi AbeFM
Is your Boost Control Output polarity backwards?
Frank

[AbeFM]

I was, and still am, wondering that. Initially I had it flipped, it didn't work on two throttle presses, so I tried flipping it.

Then it worked a couple times, so I thought it was good. It DOES hold pressure just how I ask sometimes, would that happen if it were backwards?

I have the valve set up "fail safe" so if it turns off (0% duty or broken wire) the boost will return to wastegate/can pressure. I believe this is "normal" and I have it "inverted" but since it didn't work (only two times) normal I've been using inverted.

I guess I'll flip it again while I wait for a better idea. But if it's sitting at 100% all the time, why am I not getting boost?

[AbeFM]

Huh. So I tried "normal" settings, which I had high hopes for because it was always doing something - but it was oscillating and always around the wrong spot (like, 10 psi versus the 15 I command). I added more gains (from 40/20/20 to 60/20/10) and it was still terrible.

Then I put it back on Inverted, and moved gains to 60/20/15 and it seems to be much better? Then again, I'm in low gear, so faster boost, so I expect less of an issue anyway. I still think it didn't control a couple times, so still stumped.

[pit_celica]

First, you need to know if your boost solenoid is working correctly in open-loop. Do a run (not short WOT stab, try some slow 2000-4000 WOT runs in 4th gear) in open loop with your boost duty table filled with 100%. Do you have overboost or wastegate spring boost? If you have overboost, fill the table with all 0 and verify that you have wastegate spring boost.

If you are running the latest alpha firmware, the closed-loop code assume that more duty = more boost. Your plumbing/software parameters must match with this statement to be able to use closed-loop.

If you are running the current release firmware, the closed-loop code assume that less duty = more boost.

As for the Boost Lower Limit Delta, it works this way : the code begin to control the boost when the actual MAP is higher than [Actual_boost_target - Boost_Lower Limit_Delta]. From your MSQ, this means that you want the code to begin control boost when the actual MAP is higher than [185-110=75kPa]. I would suggest you to set your Boost Lower Limit Delta to something like 80. Right now, you are asking the code to control boost before there is even a positive MAP pressure.

I hope that it answer some of your questions.

Sam

[AbeFM]

oh, wow, I totally misunderstood the min pressure thing. I thought it meant 110 kpa. odd eat of expressing it, but I'll try.

Definitely the can is working, that's obvious, I get the same pressure all the time. when solenoid fires, I get consistent boost control. the issue is sometimes it doesn't engage at all.

so far, that is my issue, and I've seen nothing to explain it. I don't know if you viewed log, but it clearly isn't trying, just view picture I posted earlier - no ebc duty, no boost. I'm looking for lock out our engage conditions which aren't being met. Any ideas? mechanically, the system is fine. I'll worry about fast accurate build up when it consistently engages.

[AbeFM]

Ok, still in knots over this. Its very difficult with the PID loops being the transformed type to fix - since I can't just set a large P term and check that things are working how they should. I'm not trying to start a fight here, I just want to get this working, saying why I'm having trouble figuring it out myself.

The big issue is I get boost control both ways, valve inverted or non-inverted. I played with the PID settings and I get boost more of the time when I hit the gas, but still, its not always controlled.

As noted, I have the lower limit set to 110%, I'm unclear what this number means. Boost duty is always 100% until I get to a narrow range between 98-103 kpa, then it drops. Additionally, it'll stay solid for a while, then come down (not yet sure if it's at ~4700 rpm, or after a given amount of time, or what) but it just stops controlling.

This is all pretty confusing, I'm not sure why sometimes it controls, why it controls if valve is "inverted" or not, etc. There's no "it gets better when I do X, so do more of X till it gets worse".

Please help!

[pit_celica]

Ok, you need to verify some things. Answer these questions :

What firmware are you using?

In open-loop, when you are set to "Inverted", does less duty = more boost (100% duty = wastegate spring and 0% = overboost)?

In open-loop, when you are set to "Inverted", how much duty% is needed to achieve you target boost (185kPa if I'm looking at your MSQ)?

Answer these then we will have more hints about how your boost control should react.

Sam

[AbeFM]

ECU Signature: MS3 Format 0262.07P
ECU version: MS3 release 1.2.1 20130316 16:55 GMT

Heh - I went for a drive yesterday and flipped stuff around a couple times, and clearly got boost at 100% and wastegate pressure at 0%... And I forgot which way I had it set! Almost positive it was "normal" though. I will verify this before I implement any of your further advice - but I'll say that unpowered, the valve connects wastegate to boost pressure, and powered the solenoid vents the wastegate can to atmosphere - this I physically checked.

I'm still stuck on how, if the computer DID have the polarity of the valve backwards, it could ever choose to open it. Why would I get above can pressure? The controller should attempt to open valve, which would close it, and that's a steady state. Without understanding why the computer would attempt (in it's own head) to CLOSE the valve to get MORE boost.... I am at a loss.

[AbeFM]

Ok, so I checked the valve, when in normal mode, it opens with more duty cycle, therefore creates more boost when open. This is obvious from the logs as well, but now there is no doubt - I both blew through the valve with different open-loop numbers in there and drove it.

I set the Boost Control Lower CLT threshold to 100f.
I set the Boost Control Lower Limit Delta to 100% - because I can't find any documentation for what this does and I don't really understand it. Looking at the logs, it seems to start controlling at ~100 kpa, which honestly is what it was doing with the number set differently, so I'm just trying to turn that feature off until I understand it and can prove that it works.

I have the valve set to "normal".

I did a run, mostly at 50/0/0 but a bit at 30/0/0. Here's what I saw:
P/I/D at 50/0/0
Valve sits at 100% when out of boost. I'm not sure why it's not at 0, but ok. It is trying to increase boost, I guess.
As MAP crosses 100kpa, the valve begins to close. This is with no integral and no derivative gains. In my opinion, the valve should open here? Perhaps 100% somehow MEANS closed?
Boost builds until the valve hits 0% duty cycle (two data points) with a MAP of 135 kpa (boost target here should be 160 kpa) at which point the valve starts to increase to 100% duty cycle (and at this point, all bets are off as to what 0% actually means. On open loop, that would mean the wastegate is seeing manifold pressure, but here I don't know).

As duty cycle increases, boost level flattens out, then duty cycle increases and boost rises for two data points, and it oscillates back and forth with boost bouncing between 162 and 146 kpa while all else held constant (boost target of 185 throughout). The entire time the DC swings from 0 to 70% just about as fast as it can (a few log samples per cycle. about 40ms. FYI 30 ms is the control interval)


It seems to be controlling, but to the wrong value? Could this be one of those cases where an unselected option is having an effect? People see values in unused tables effecting things in other areas like VVT control... Perhaps the open loop or feed forward "initial duty table" is effecting something?

When I changed to 30/0/0 I get similar behavior, except: The swing of the duty cycle is smaller, ~51-20% dc, but the MAPs acheived are ~144-156 so not that different.

Nowhere in my log does the MAP ever get above 167KPA, well below the target, and never does it sit flat there.

So, basically, while I could make a lot of reasons why it's not perfect - perhaps the house routes are too long, etc, none of that explains why it's trying to drop the boost when it's WELL below the target. I'm not tuning, I'm trying to figure out why its broken.

[jsmcortina]

I set the Boost Control Lower Limit Delta to 100% - because I can't find any documentation for what this does and I don't really understand it.

Did you look at the Tooltip ?

James

[muythaibxr]

If you are only using a P term you will never converge on the target. You must use an I term to make it converge. The best thing to do is set P high enough to get you to reach the target with some overshoot, then add I and readjust P as needed. Then add D to get rid of overshoot and readjust I and P as necessary.

P responds to changes in error not error itself so just having a P term will keep you from reaching a target.

[AbeFM]

Oh, I remembered that from before. But I also remembered hearing of updated code some years back with a different PID loop. Does P actually do as described above (i.e. why does it start opening the wastegate BEFORE I get to my target? Why do I NEVER get to my target? A pure P controller should overshoot if too aggressive, or monotonically rise if not. I guess it could do what it is doing - oscillate around the lower point, but increasing the P gain should raise the spot where this happens. I'd started off with 20/20/5 but can't get any advice....


I went to 50/0/0 based on this text directly from the MS3Pro manual

The next step after setting up the target table and supporting settings is to tune the PID gains:

1. Set Integral and Differential Gains to 0% - This will let you focus on tuning the proportional gain.

2. Set Proportional gain to 100% and slowly lower - While tuning Proportional gain, higher numbers
mean slower boost climb and lower final boost. For safety, start with a very high gain (100%
should be sufficient). Find the RPM that typically spools quickly, and fully and quickly depress
the accelerator. Note how much boost is reached. If boost overshoots the target dramatically,
increase the Proportional gain. Otherwise, reduce the Proportional gain and try again. Do this
until boost reaches the target with a small amount of overshoot.

It also makes it seem like not only will P alone get you there, it also makes P seem inherently negative? That a larger P value is LESS control?

I just want someone to explain to me how to go about this, in what order, etc. I had PID loops working just dandy in college and since on other things, but somehow I can never make MS ones work - and I can't find instructions on how to make them work. Show me how to get overshoot and I'll be a lot more confident that someday I can hit my target.

[pit_celica]

Too much P make the control too aggressive in the way that it tells the wastegate to open as soon as boost is seen, not the way you are thinking. So, try with a large P and decrease until you start having overshoot. In my case, with P only, I wasn't able to get boost over the wastegate spring threshold when P was higher than 12. And, if P was lower than 8 or 7, it was infinite overboost. So, my conclusion is that the range where you have some control using P only is pretty narrow, but it is essential that you find it before even trying to add some I term.

I suggest you to keep lowering P until you have some kind of overshoot, then increase I, then you will need to increase P a little bit each time you increase I because you result in too much overshoot.

The P term is only for "how fast" is the control response. Too much P and you never go over the wastegate spring because the control "predict" that the boost will come and it has already opened the wastegate. Too little P and the control isn't reacting fast enough to a rise in boost and this result in a overboost.

Sam

[AbeFM]

Sam,
Thanks! That's straightforward advice. You're right - I was thinking of the controller as trying to raise the boost up, not trying to force it down.

Unfortunately, I'm out of town for a week. Will definately try that when I get back, I certainly had no idea that if I didn't see anything by 30 that it could still happen. That even explains the oscillations.

[baldur]

P responds to changes in error not error itself so just having a P term will keep you from reaching a target.

If P responds to the first derivative of the error, then what does the D respond to, changes in changes in error?