rwyatt365

In my ongoing battle with the Tuning Gods I've encountered another oddity...When I get into higher boost circumstances the Boost Duty "goes berserk". The duty will oscillate wildly between the min and max duty when nearing peak boost and won't stop until I back out of it (see log). This is something new that has cropped up since I swapped out (another) Churbo, which was supposed to be a twin to the one that I borked because of an oiling issue about a month ago. Here's the log;


I've done several things since I first encountered this about 2 weeks ago;
1) I've tweaked, and re-tweaked the CL Boost settings ad-infinitium. I've tried so many different PID and Gain slider settings that I'm pretty sure that none of that has made any appreciable difference,

2) I have swapped in two different wastegate actuators; a) the one on the new turbo had a 30psi spring (as "verified" by a bicycle pump with a crappy gauge from WalMart), b) the wastegate from the borked churbo that was working just fine (it has a 20psi spring, as measured by the same precision instrumentation) and c) the OG actuator from the CX Racing churbo that was also working just fine (with a 15psi spring). My thought-process was that the "a" and "b" wastegates were too strong and that the max psi from the turbo (230+ kPa) wasn't enough to open the WG enough for it to be effective, so the CL couldn't control it properly.

3) I checked the open/closed duty on the EBC (from DIYAutotune) using the Test Mode and set those values accordingly. One thing I did notice when doing that was that when I had the valve at max duty, the airflow "pulsed" at the solenoid frequency. I don't know if this an expected thing, but I'm wondering if this is contributing to the problem. Maybe my EBC is bad? I don't know. If I have to get another controller to verify if that is the source of my problems.

Bottom line...everything I've done has made no difference in the result. As soon as the controller tried to control, it goes bonkers.

Looking for input/suggestions. Tune attached for reference.

Skamba

If you have a 15 psi spring, and your target boost is 220 kPa at high rpm, that means you can gain about 17 kPa above pure wastegate boost pressure. At lower RPM's you have 195 or less target boost pressure, which you won't be able to reach with a 15 psi spring. That being said:
1) If you have the DIYautotune EBC, you can run it at 39hz.
2) I have the same EBC, and mine starts around 13% - how did you determine the 20% minimum value? Although it could be that it has different values at different frequencies.
3) Start the tuning again. Typically, these oscillations means that the sensitivity is too high. Although for the MS3, I find this thread to be useful.
4) Don't forget to set overboost protection. You're already 20 kPa above the highest value in your ignition table with the current amount of overboost.

Ted75zcar

iTrader: (2)

The tune and the posted image don't appear to match, the screen shot shows a max DC of 70%, the tune indicates 65%.

Not sure where you are, but you won't be able to regulate a 15psi spring to 210kpa at sea level. The minimum effective DC you will see out of these things is going to be around 30% or so. this means ... 1.3 * 15psi = 19.5 psi, or ~ 235kpa.

The tune has "342" which I assume means MS2. If you are on MS3, and have SD log capabilities, use them

the bias table has it pegging at 65% for a 210kpa target, this will result in ~ 30psi of boost. The controller is oscillating between the bias table and the CL value.

Get a lower psi spring for 210kpa
use 39Hz as previously suggested
go open loop, characterize and tune your bias table
configure CL primary parameters using "simple"
finish with tuning "advanced" PID

rwyatt365

I hear what everyone is saying, and I'll be getting a actuator with a lower spring rate, but the problem (in my mind, at least) is that I was able to control the WG that's on there now, with a similar (if not the same) DIYAutotune EBC on a "twin" of the current turbo, and was not getting these kinds of oscillations. This is what has me baffled - it was working successfully before, and now, not so much.

This is indeed for a MSPNP2 - so a MS2-based MS.

I'll change to 39Hz. first and see what happens. In the mean-time I'll order an actuator with a lower spring rate and try that. I even might get one of the new DIYAutoune EBC's and see what that does. If worse comes to worse, I'll punt and start back w/ open-loop and take the long uphill trek back to CL.

Tomorrow, we're expecting torrential rains, so I won't be getting any WOT pulls until the weekend.

Thanks for the inputs and more news later. PS - here's the "right" tune. I had copied a restore point from earlier thinking it was the right one.

Ted75zcar

iTrader: (2)

When you say EBC, do you mean the solenoid?

I suspect that you either had a lower psi spring on the other turbo, or that you were simply running on WG before and did not realize it. The screen shot you provided above is exactly what I would expect to see from your hardware with your tune.

Tune the open loop table, you will see what I mean. With that actuator setup, you are going to be well into the 230+ kpa range by the time you are into a usable dutycycle.

rwyatt365

Yeah...solenoid.

And the WG that I'm currently using is the exact same one that was on the other turbo! I swapped it out yesterday. That's what has me all confused.

I changed the solenoid frequency to 39Hz and re-calibrated the max & min duty cycles, and took a run this morning before the rains started...same f-ing thing! So...back to grade school. Open loop tuning it is.

PS - I'm also getting a WG actuator w/ an 8psi spring to remove ALL questions on that front.

rwyatt365

Just for closure (I hate threads that start out so good, and the leave you hangin with no resolution), here's an update...

At the suggestion of Ted75zcar, I've gone back to grade school and have been running Open Loop for the past week. What I've landed on is the following;


This will get me 230+ kPa max with the "alleged" 15psi wastegate. The "alleged" (EBay) 7psi wastegate just came in this afternoon, so I'll put that in tonight and get a log tomorrow to see what happens.

Just for ***** and giggles I ran what I had in Closed Loop and the plus/minus duty cycle oscillation was still there (no surprise), but not as violent - i.e. I think that the data in the datalog is being aliased because the sample rate is too low so what I'm seeing is a pale display of what is REALLY happening.

Anyway, I will post up a log and MSQ tomorrow of a WOT pull with the "7psi" wastegate.

Ted75zcar

iTrader: (2)

Looks like a 15psi actuator to me with that table. Bet if you targeted 235kpa the CL would work in basic mode with 200-400 sensitivity.

What turbo is that?

rwyatt365

It's an EMUSA (EBay churbo) "Universal GT28 IWG turbo" - basically a 2871 knock-off.

If the rains stop today (WOT on slick roads with cold tires is pretty dicey), I'll try OL with the 7psi WG, and then use CL and see what happens.

rwyatt365

Like I said before, I won't leave the thread unresolved. So, here is my latest update.

I've spent the last week reading everything I could about CL Boost tuning and Duty tables, ad infinitum. Then I came across this thread on msextra, "Boost control change testing (MS2 and MS3)", where the tuning process was essentially;
1 - Set PI&D to zero
2 - Tune the Initial Duty Table
3 - Set I to a value that will hold boost to the desired target
4 - Tweak P & D to adjust for over/under shooting

So, I did 1, 2 & 3, and then spent the last 4 days waiting for the streets to dry enough so I could go WOT without spinning the tires. Well, this morning was the test and this is the result...


It hit the 200kPa target and held there slick as snot! I'm stoked now. I'm going to save this tune and use it as a baseline for refinement. FWIW, My PID values are; P=20, I=100, D=0. I'm thinking that I might be underdamped and could possibly reduce the I and increase P to achieve the target faster (or maybe I'll screw everything up and have to revert to the baseline again).

I'm also thinking about replacing the old DIYAutotune solenoid with a new MAC solenoid, but that'll come later. For now, I'm liking that I can actually control the boost in CL mode.