TouringBubble

iTrader: (2)

Do you have any idea why there seems to be an offset of 10% as shown here? After you pointed me in the right direction today, I began to tune the BWGDC values and noticed that I nearly always have a 10% offset (as shown here) and it doesn't seem to be dependent on the error correction values.

Another thing, my error correction doesn't seem to be working. I'm overshooing target load in 3rd but am logging no correction at all in the WGDC. So, basically, I'm getting a random +10% deviation and no -xx% correction when I exceed the target load value.

Any thoughts?

mrfred

iTrader: (50)

In that particular tune, I'm pretty sure that I had the target load set to 240 across the rpm range. At about 2600 rpm, the WGDC jumps by about 10% during spool up as the ECU tries to meet the target load of 240. For that tune, I had my maximum "step" correction set to 10%. After the boost peak, the ECU compensates downward. Then the ECU compensates up and down as the actual load bounces above and below the target load. The bouncing isn't due to the corrections that the ECU is applying but rather the way my boost control system behaves as a function of WGDC. See the plot below for yesterday's log and how I've had to set my baseline WGDC. Anyhow, the final 10% increase occurs as the actual load drops into the last trough. After that the ECU makes no further corrections because the actual load starts dropping off, but the target load is still 240.

I've found that how *readily* the ECU compensates upward or downward depends greatly on the values used in the correction table. With some of the tables I tried, the ECU would readily correct upward if the actual load was just slightly below the target load, but then if the actual load exceeded the target load by a greater amount, the ECU would do nothing. I'm still sorting out this behavior. My general approach at the moment is to create a baseline WGDC curve that puts the actual load right onto the target load in 3rd gear runs, and then I am using a very small error correction (+/- 3 maximum) to allow for different gears and different weather.

RazorLab

iTrader: (8)

Boost is looking good now mrfred.

nj1266

iTrader: (6)

mrfred,

Have you tried to run less boost and advance the timing in the 3000-3700 rpm range to about 3-4* and see if you will make more power than with the current set-up? Your timing is going negative in this area and I suspect that it might be causing a loss of torque. Ofcourse, I do not know your AFR, but you might have to adjust it to the low 11s in that area if it is not currently set-up that way.

mrfred

iTrader: (50)

Yeah, looking much better than before. What a funky WGDC curve though. I'm scratching my head on that. We'll have to get back together to work on the tuning again soon. I'd like to look at what nj1266 suggested.

TouringBubble

iTrader: (2)

Mrfred, are you currently tuning boost with 0 correction and then planning to add correction after the WGDC is where you want it? I think that's the approach I'm about to take.

mrfred

iTrader: (50)

I'm doing 0 upward correction, but I do have downward correction enabled. Make sure to check your result in 4th or 5th gear.

nj1266

iTrader: (6)

I have a question: When you set your BDEL tables do you set them to the exact 2 byte load numbers that you are logging? Or do you set them slightly higher than that?

mrfred

iTrader: (50)

I set my target load (BDEL + offset) to what I would like to achieve, and then tweak baseline WGDC (with upward correction disabled) until I acheive my target load. If the motor starts consistently knocking before I acheive my target load, then I readjust the target load to a more realistic value and continue with baseline WGDC tweaking.

One thing I'm finding is that a few knock counts at sporadic locations in the powerband from run to run is not a sign of knock. It takes consistent knock counts of two or more at the same rpm range before I'll believe its real knock.

nj1266

iTrader: (6)

What I am trying to figure out is the relationship between BDEL tables and the Turbo Error Correction Table. There are no rpm rows for the correction table so how do those numbers relate to correcting the BDEL? Or are the numbers on the TECT not intended for the BDEL?

TouringBubble

iTrader: (2)

The error correction tables adjust the WGDC according to the difference of actual load from target load ...

If target load is 250, actual load is 260, error correction would adjust the WGDC the variable % amount (right column) according to the 10% load difference (left column).

I hope that makes sense.

mrfred

iTrader: (50)

The "Max WGDC" table is poorly named. It should be called "Baseline WGDC" table. When you are cruising, the car is at 0% WGDC (at least that's what EvoScan says), and when you stomp on the gas, the ECU looks at the engine RPM, then goes to the Baseline WGDC table, reads the WGDC number, and applies that value to the BCS. At that point, the ECU then starts comparing the actual load to the target load (BDEL + boost offset) about once every 0.4 seconds. If the actual load matches the target load, then the ECU continues to follow the WGDC value (vs RPM) in the baseline WGDC table. However, if the ECU sees a discrepancy between actual and target load, then it can/will apply a correction to the WGDC. The amount of correction depends on the values in the boost error correction table. The correction is simply to raise or lower the entire Baseline WGDC curve. So after the correction is applied, the ECU follows the raised/lowered curve. It checks again 0.4 seconds later, and if necessary, raises or lowers the curve again. This is why the correction table isn't a function of RPM. The graph shows the behavior well. The blue arrows show when corrections are made, and its apparent that the ECU is simply raising/lowering the baseline WGDC curve.

Jumperalex

Greetings. Been lurking for a while, and now finally deciding to start tuning (ahh to have the free time I had during my 2G DSM days) ...

Anyway I've been reading solid now for 2 days every thread I can find to come to an understanding of the underpinnings and methods for controlling boost with the ECU. I'm using MBC right now and ordered my tactrix cable 2 days ago and I've come up with a question and a comment I'd like to pose

Question: despite the results of datalogs run from turbo stall speed up through red line I'm still left wondering if having BWGDC set at less than 100% isn't losing some spool-speed capability when actually driving and say during corner exit when we aren't below turbo stall speed?

I actually do believe I understand, and would tend to agree (isn't that nice of the newbie to agree when he hasn't even flashed his first ROM ) with the hows and whys of the way mrfred seems to be doing things. In my early research I kept finding myself thinking along the same lines as he was. Of course I had the luxury or being an arm-chair tuner.

Still my thoughts are that Mitsu choose to use pills sized just right to get pretty much the boost they wanted with a BWGDC of 100% and the correct TBEC values to provide enough correction to adjust it down after the initial spike and then again back up as RPM increased. Of course they still sized the pill such that even at 100% WGDC and high rpm there would be taper. The benefit of this method would seem to be assurance that the WGA hadn't opened a single bit during spool (due to 100% WGDC) to make it happen as fast as possible.

So now comes my comment: I think once I get up and running I just might have to include a web cam mounted under the hood, looking at the WGA, to prove to me that the WGA isn't moving until it actually needs to.

Ok another question: has anyone already done this?

Because everything being said here and in a few other posts both makes sense, and seems to be producing the expected results (the hallmarks of a true functioning theory ... testable and predictive) but I still have that nagging feeling that if the ECU isn't being told to give 100% WGDC when BDEL /= Actual Load then we aren't going to get the best spool possible.

In fact it all makes me think that the TBEC table needs at least on of the positive correction values to be big enough to actually generate a 100% WGDC despite the BWGDC table.

So looking at the last mrfred chart (post #77 that is) I wonder if there is benefit to setting the -20 TBEC input value (left side) to as much as 60 (on right side) to account for his lowest BWGDC value (~40% @ 4000rpm). Because when I'm exiting a corner at 4000rpm and open her up, I want to know that the WGA isn't open even a crack until I get close to my desired boost. I hate corner exit turbo lag!!! hehe sorry

Thoughts? Comments? Well aimed flying carp?

TouringBubble

iTrader: (2)

Well, I'm not an expert on this, but I'll take a crack at it ...

You're partially correct and partially confused.

You are correct in that a 100% WGDC will help the turbo spool faster. I've got the low end of my map set at 100% for this reason and I have tested it. It's only really effective from like 3000 to 3500 RPM where the turbo really starts to pick up. At that point the WGDC is capped to limit boost levels.

Your confusion is based in the naming of the values we're discussing. The WGDC value isn't actually related to the wastegate. It's a measure of the duty cycle of the solenoid that controls flow to the wastegate actuator. So, 100% duty cycle on the solenoid does not really equal 100% duty for the physical wastegate. It's more like a measure of the flow capability of the pill and the solenoid's control of that flow.

When we change the pills, we are basically throwing off the scaling of these values. Think of it this way ...

Stock Pill WGDC @ 100% = 100% flow
Modified Pill WGDC @ 100% = 200% flow
Modified Pill WGDC @ 60% = 120% flow

So, with the modified pill we can use a lower WGDC% value to flow the same amount of air. To compare this to your question and observation, we are comparatively running more than 100%.

I really hope this isn't completely bogus ...

mrfred

iTrader: (50)

I think TB hit the key point which is that you appear to be equating the "WGDC" with the actual wastegate motion. Its completely understandable and is due to the sometimes less-than-precise naming given to the tables in ECUFlash. If we were to be more accurate, BWGDC -> baseline boost control solenoid duty cycle (BBCSDC!). The action of the BCS solenoid influences the WGA but there is not a 1:1 correlation between WGDC and WGA movement. Mitsubishi has designed the stock restrictor pills such that the BCS plays very little role in the control of boost. Most of the time, its just sitting at 100% DC. However, on a car with non-stock pills or a bleeder where the pressure at the WG is much lower when the BCS is at 100%, the BCS has to be tuned to set the desired load curve (just like TB's example).

There is also the point about making certain that the WGDC is closed until it needs to be open to prevent shooting past the target boost value. The BCS works in conjunction with restrictor pills and bleeders to regulate the pressure on the WGA. It is possible that under some circumstances, the WG may open slightly before reaching the target boost level. However, this is not unique to the Evo boost control system. I think that any passive restrictor or bleeder type boost control system can allow this to happen. And I think that you'll need to have the wastegate open a little early if you want to keep from shooting past your target boost level everytime you spool up in the high efficiency range.

Maybe one way you might be able to get away from this is with an EBC that uses a PID algorithm and has a very fast response time. Maybe jcsbanks can chime in here since I know that he's using an EBC.