Log of the TBEC -20 table cranking up WGDC. I had hard set WGDC for this RPM to be 95. But with TBEC going to -20 and it not being zero'd out, you can see TBEC working against WGDC.

Table is 2byte Load, Load Error, Boost, WGDC

 

Awesome work as usual mrfred - you should have a different title like "EcuFlash Boost Guru" . If we can get a control for target boost (using the JDM Map Sensor) using wgdc correct that will be phenomenal.

Anyways, I have seen two versions of the WGDC equation...
x*100/255
x/2

I know the difference is minor, but trying to get things sorted well I'm looking for the most accurate data. Thanks

 

If this is for EvoScan, its x/2.

 

Alternatively, you could use calculation based on the logs... (2byte Load)...

Target Load = (PSI you want / 0.145) + Baro.
Ex) 21 PSI to target load. (21/0.145) + 101 = 245.8 (2 byte Load)

I have dealt with same problem. Eventually, I lowered correction value on the - (minus area) so that it does not over work with WGDC to boost more than I wanted (Target Boost Engine Load). However, I left above 10+ percent over boost situation to be corrected like stock....

Finally, I had to reduce "Max WGDC" to fine-tune of the BOOST in 3K - 3.5K RPM zone...

Example on STOCK IX turbo I set with TBE and pill removed from the solenoid side only...

TBEL: (Load)
2250 229.4 (Remember to subtract your boost correction factor : Mine is @ 90 so 139.4 on the table)

RPM TBEL MWGDC (This is with pill removed from Solenoid side only)
2500 210.6 99.5
2750 226.9 97.5
3000 233.1 88
3500 245.6 93
4000 238.1 96.5
4500 231.9 96
5000 228.1 95
5500 225.6 96
6000 221.9 96.5
6500 221.9 97
7000 220.6 97

This gave me 21 PSI max gradually tapered to 17 PSI by 7000 RPM....

Correction
% diff % correction
-20 4.5
-17.5 4.5
-12.5 4.0
-10 2.5
-7.5 2.0
-5.0 1.5
-2.5 1
0 0
2.5 -1.0
5.0 -1.5
7.5 -2.0
10.0 -2.5
12.5 -4.0
15.0 -4.5
17.5 -8.0
19.8 -12.0



Eventually, I logged 2 byte load and calculated the difference from target to acutal... (Target load - Actual 2 byte load logged)/Target Load ... Based on this caluclation, I adjusted a Target Boost Engine Load again (which is what I came up with above)...

I honestly do not see the dead spot! It is within +- 3% of target boost engine load...

 

very interesting fine tuned setup
did you try with more boost ?
i'm running 24-25 peak tapered to around 20-21 and here is the 'slow' ECU reaction i'm experiencing : ECU is pulling down WGDC too late & too much at 4000 rpm and then up too late at 4800 to sustain around 240 250 load

 

No, I've never try to run that much as I only have TBE with stock turbo.....
I concurr with your statement. There are too much correction going on @ 4K RPM. Then, it is catching up with 4500 RPM where it is adding more % to catch up with Target Boost Engine Load. I would setup the TargetBoostEngine Load @ 270 for 4K PRM and then reduce the -20% correction to something smaller to +4.0% to examine how it handles... Then, just use the calculation to see the % difference in the acutal Load (2byte) versus what you intended through the "TBEL"... Eventually, you will see yourself getting closer to where you want it to minimize the boost drop..... This thread is espcially nice since it provides you with the difference without calculating. This is also more accurate method than how I used calculation as that would be what ECU is seeing.....

Have you also tried to reduce the WG reaction time to 3 from 10? I think "mrfred" listed this on the first page. Perhaps, this will accelerate the correction response time....

 

yeah , 'will try your ideas next week & post results
unfortunately I can't try mrfred stuff for the moment (no adress for EDM 88840016 ROM for the moment )
cheers

 

Many thanks to you MrFred for making this thread and sharing the info. I just started logging this WGDC error so I don't have much to input. Just that I didn't see any XML definition to log this in Mitsulogger so here is what I used:

 

Looks good, but its not WGDC Error. Its the Load Error (the difference between the actual load and the target load). The units are load.

 

Here is my explanation of the "slow reaction" in correcting the load value ...

Let's say that you are running at your exact BWGDC and all is fine. Suddenly, load is off and the error check decides to apply -4% WGDC correction. Well, on the next check load is where it needs to be, but the boost hasn't fully leveled out from the quick -4% correction. The ECU thinks that BWGDC - 4% is correct, but really it's more like BWGDC - 2%. The ECU can't realize that the boost is lagging behind the WGDC correction, so it can't stop the boost from dropping to the -4% it asked for. Once the boost catches up to the -4%, load is too low and the process repeats.

The solution is to slow down the error correction or lower the correction values. I've logged the same TBEC values against different correction latencies and the dips and spikes get worse with faster correction. Admittedly, I've only logged this on my own car so it may vary depending on the boost control method used.

 

I think I experienced this in my car for the first time trying this mod. I had set my correction timing to 3 and my error table was left unchanged in the overload area. As I got on the throttle and upped my load it suddenly felt like I hit fuel cut but my load didn't reach my boost limit value I think.

Today I'm trying smaller correction values and leaving the correction timing at 3.

 

Using the smaller correction values worked. My load curve is pretty smooth. There's just one anomaly I was hoping someone could help with.

So I'm wondering why I get the 228 load. My desired is 240 and my WGDC is 68 in my table. Could it be because my WGDC at 3000 in my table is 83? The reported is 69 so the ECU is correcting and maybe that 228 load is a result of the large WGDC correction?
And I'm wondering why the load error at 4000 is 10 when my target load there is 239.

 

could you post a screen of your WGDC, BDEL, losd offset, and TBEC tables? I'd be happy to take a look for you.

On another note ... I slowed down the error correction to "5" last night for testing. The wavy load/boost curve is mostly gone. I still get a fairly large initial spike, but I just need to make some fine adjustments to the TBEC table for that ... a little more correction for overboost and a little less correction for underboost.

I'm starting to think that Mitsu was on to something with the slower correction and larger correction values. This method gives the pressure time to "catch up" to the WGDC change. "10" is definitely too much time between corrections, but 5 seems very usable.

 

Thanks, TouringBubble

 

Everyone should try to tune with their TBEC table zero'd out. When you're as close as you can be, then re-introduce the TBEC.

Also, I've noticed that when logging Load Error, the error is very laggy. For example, if you have +10 Load Error at 4000RPM, chances are you were actually +10 at about 3750RPM and only now is it making changes.

This would explain roger smith's question as to why his load error is +10 at 4000 even though he is very close to his BDEL.