2009 Aussie RA - Tweaking and Tuning
Jan 2, 2011, 07:20 PM
#61
Hi Bryan.
Great - so my "High Gear" def is good! What's the story with the "27" value for the "Low Gear" value?
Will it be advantageous to do the following?...
- Reduce the "Reactive Solenoid WGDC Correction Interval (High Gear Range)" from 7 to 3.
- Halve all values in the "Reactive Solenoid Turbo Boost Error Correction" table... smaller adjustments because the ECU will be error-correcting the boost curve more frequently?
I want to get a smoother boost curve before I hit the AFR tuning. Right now, it hits 1.6 bar -> 1.4 bar -> 1.5 bar -> 1.6 bar, before starting its tail-off. This seems like a good place to start getting some experience in ECU boost control.
Rich
Great - so my "High Gear" def is good! What's the story with the "27" value for the "Low Gear" value?
Will it be advantageous to do the following?...
- Reduce the "Reactive Solenoid WGDC Correction Interval (High Gear Range)" from 7 to 3.
- Halve all values in the "Reactive Solenoid Turbo Boost Error Correction" table... smaller adjustments because the ECU will be error-correcting the boost curve more frequently?
I want to get a smoother boost curve before I hit the AFR tuning. Right now, it hits 1.6 bar -> 1.4 bar -> 1.5 bar -> 1.6 bar, before starting its tail-off. This seems like a good place to start getting some experience in ECU boost control.
Rich
Have you been logging your WGDC?
Jan 2, 2011, 08:25 PM
#62
Evolved Member
Thread Starter
I've dumped a load of adjacent addresses, and cross-referenced with other known-working EVO X ROM xml definitions, and I have found a likely candidate...
** AS YET UNVERIFIED **
** AS YET UNVERIFIED **
This has a value of 7.
This is the address used in ROM 53040010 for the aforementioned value. Sussed out by Tephra here...
https://www.evolutionm.net/forums/ev...-low-gear.html
I don't know for sure whether that's the correct address for 53610010 - not going to edit it - but at least I'm attempting to use some initiative!
I haven't been logging WGDC. Will add it.
Rich
** AS YET UNVERIFIED **
Code:
<table name="Reactive Solenoid WGDC Correction Interval (Low Gear Range)" category="Turbo" address="54688" type="1D" scaling="uint16"/>
This has a value of 7.
This is the address used in ROM 53040010 for the aforementioned value. Sussed out by Tephra here...
https://www.evolutionm.net/forums/ev...-low-gear.html
I don't know for sure whether that's the correct address for 53610010 - not going to edit it - but at least I'm attempting to use some initiative!
I haven't been logging WGDC. Will add it.
Rich
Jan 2, 2011, 11:46 PM
#63
Evolved Member
Thread Starter
Having spent the past I-don't-know-how-long reading up on the ECU's boost control style, I'm less inclined to go messing with it - the system seems to be doing its job reasonably well.
I didn't quite "get" that this was all really a Load based feedback system... and the load values I'm logging through the "boost peak and trough" areas are in fact very consistent...
"Ld1" is the load used for sparkmap lookup. "Ld2" is the one used for fuel lookup (I think!). I don't yet know how to log the special load value being used for boost-load targeting/correction.
Between the first boost peak @1.55 bar (3469rpm) to the second boost peak @1.54 bar (4719rpm), the load didn't vary by more than +/- 3. That's under 1.5% variance each way. Pretty good, given it's not direct boost control, but a load control.
Still, I reserve the right to tinker with it later!
Rich
I didn't quite "get" that this was all really a Load based feedback system... and the load values I'm logging through the "boost peak and trough" areas are in fact very consistent...
Code:
RPM Ld1 Ld2 Boost(BAR) 3000 161 158 0.90 3063 173 169 1.01 3125 183 180 1.14 3188 194 190 1.25 3250 206 202 1.37 3313 217 212 1.47 3375 226 221 1.53 3469 231 224 1.55 3563 232 225 1.52 3625 231 224 1.48 3688 229 222 1.45 3750 228 221 1.46 3844 229 223 1.45 3906 231 224 1.47 4000 232 225 1.48 4063 233 225 1.49 4125 233 226 1.51 4188 233 226 1.50 4250 233 226 1.50 4344 232 224 1.49 4406 231 223 1.50 4469 231 223 1.50 4563 231 224 1.51 4594 230 223 1.52 4688 231 223 1.53 4719 231 224 1.54 4813 231 223 1.54 4875 230 222 1.54 4938 229 222 1.53 5000 229 222 1.52
Between the first boost peak @1.55 bar (3469rpm) to the second boost peak @1.54 bar (4719rpm), the load didn't vary by more than +/- 3. That's under 1.5% variance each way. Pretty good, given it's not direct boost control, but a load control.
Still, I reserve the right to tinker with it later!
Rich
Jan 13, 2011, 06:25 AM
#64
Evolved Member
Thread Starter
Okay, here's a 3rd gear log, 2000rpm up to 5700rpm (at which point I changed up to 4th).
Selected raw data...
Code:
RPM Boost | Load| ActiveWGDC | | | | WideBandAF | | | | | 2250 76 0.2 83 14.2 2094 79 0.6 84 14.2 2063 82 0.8 83 12.4 2125 85 1.4 82 12.4 2156 87 1.7 81 13.0 2219 91 2.1 81 13.0 2250 94 2.5 80 12.9 2281 98 3.1 79 12.9 2344 102 3.6 79 13.2 2375 107 4.1 78 13.1 2438 111 4.8 77 13.1 2500 117 5.5 76 12.9 2531 123 6.3 76 12.9 2594 128 7.3 76 12.7 2656 135 8.4 75 12.7 2719 144 9.6 75 13.2 2781 152 10.9 74 13.3 2844 163 12.3 74 13.3 2938 174 14.2 73 12.8 3031 188 16.3 73 12.8 3094 203 18.5 73 12.3 3219 218 20.8 72 12.3 3281 230 22.5 72 12.1 3406 240 23.5 71 12.1 3469 243 22.9 72 12.1 3594 240 22.1 72 11.6 3688 236 21.2 73 11.6 3781 235 20.8 74 11.3 3875 236 21.1 74 11.3 4000 238 21.0 75 11.3 4063 240 21.6 77 11.3 4188 242 22.0 78 11.1 4281 242 22.2 79 11.1 4375 241 22.2 80 11.1 4500 239 22.2 81 11.0 4563 237 22.1 82 11.0 4656 235 22.2 83 11.0 4750 236 22.6 84 11.0 4844 237 22.6 85 11.0 4938 236 22.5 86 11.0 5000 234 22.5 87 11.1 5094 231 22.3 88 11.0 5188 228 21.9 89 11.0 5281 225 21.4 90 11.2 5375 222 21.1 90 11.2 5438 219 21.0 91 11.1 5531 216 20.5 92 11.1 5594 215 20.5 92 11.2 5688 212 20.0 93 11.2 5719 210 20.0 92 11.2 5313 217 21.7 86 11.4 4844 232 23.7 83 11.4 4688 242 24.4 82 11.3 4625 247 24.4 82 11.3
So, where to from here? After loads more reading this past week, I'm thinking...
(1) Decide on the exact boost curve I want. Not easy when you're a n00b!
(2) Temporarily zero out the first eight entries in "Reactive Solenoid Turbo Boost Error Correction" table, so unlink the upward feedback correction loop.
(3) Modify all three "Reactive Solenoid Base WGDC Map" progressively, to get the curve I want.
(4) Revert the TBEC table entries to original values.
(5) Modify all three "Boost Target Engine Load" tables to set appropriate new engine loads.
(6) Review ActiveWGDC logging to check for error correction, and review (5) if the curve is being raised or lowered
Feedback welcome...!
And yes, I did notice the boost touched 24.5psi on the upshift at 5700rpm WOT.
Rich
PS. I had a boost leak test performed the other day, and it came back all good. Everything seems to be running fine mechanically.
Jan 13, 2011, 09:42 AM
#65
Since your boost control is already setup because of the basemap here is what I would do to create your desired boost curve:
1. Turn off the error correction by zeroing out the right side of the "Reactive Solenoid Turbo Boost Error Correction" table.
2. Do a pull and log the car, espically WGDC and Load
3. Add or subtract WGDC depending on how you want your boost curve to look. Too little boost at a certain rpm? Add WGDC. Too much? Subtract WGDC.
4. Once you have your desired boost curve. Take the load curve values and fill in the two highest TPS rows in the "Boost Target Engine Load" table with just *slightly* higher load values then what you have logged. From there, fill in the lower TPS rows with smoothed out load values down to zero. See the current table to see how to do it
5. Turn back on error correction by putting the additive and negative values back in the "Reactive Solenoid Turbo Boost Error Correction" table
1. Turn off the error correction by zeroing out the right side of the "Reactive Solenoid Turbo Boost Error Correction" table.
2. Do a pull and log the car, espically WGDC and Load
3. Add or subtract WGDC depending on how you want your boost curve to look. Too little boost at a certain rpm? Add WGDC. Too much? Subtract WGDC.
4. Once you have your desired boost curve. Take the load curve values and fill in the two highest TPS rows in the "Boost Target Engine Load" table with just *slightly* higher load values then what you have logged. From there, fill in the lower TPS rows with smoothed out load values down to zero. See the current table to see how to do it
5. Turn back on error correction by putting the additive and negative values back in the "Reactive Solenoid Turbo Boost Error Correction" table
Jan 23, 2011, 05:50 PM
#67
Evolved Member
Thread Starter
What a pain in the **** it is, trying to stop the initial peak in boost levels! I've tried pretty much every trick in the book, but with load-based error correction enabled it still peaks, drops 2psi, rises 1psi, before stabilising.
With EC off, my revised BWGDC map gets me a lovely smooth boost curve in both 3rd and 4th. Switch EC back on, and it overshoots (then over-corrects) every time. Same old issue I've read about again and again... I've now experienced all the fun myself!
The BTEL map is a great way to manage boost levels once the t/c is spooled up. But it honestly just gets in the way during spool-up. This is a control system that uses "target" levels... but those levels are all over the place in the spool-up region.
The WOT engine loads at 2750rpm and 3000rpm
(a) vary a lot in the different gears.
(b) rise extremely fast, and
(c) vary depending on the "starting" rpm for your WOT run.
A moving target is a pointless target - hence the endless frustration with a target-based correction system. It has no hope of being "on target" at the very start of the boost curve... only once the boost levels start to level out will it become useful.
Ideas I still have to try:
- Try smaller downward EC values - at least try and get a smooth, consistent boost curve after the initial peak.
- Lower the BWGDC at 2500-3000rpm to "build in" the inevitable upward error correction factor.
- Changing the gear Xover to separate 3rd and 4th maps, and tune them differently... but I really wanted to tune "1st and 2nd" separately at some point. Plus, my XML address for the "rpm/mph Xover" field is currently wrong!
- (edit) Changes to the error correction scale as well as the values... to retain the large "way off-target" downward correction ability, but limit the smaller boost adjustments.
Or I could just put up with it and just move on to AFRs! Nah, I'm not giving up yet!
Rich
With EC off, my revised BWGDC map gets me a lovely smooth boost curve in both 3rd and 4th. Switch EC back on, and it overshoots (then over-corrects) every time. Same old issue I've read about again and again... I've now experienced all the fun myself!
The BTEL map is a great way to manage boost levels once the t/c is spooled up. But it honestly just gets in the way during spool-up. This is a control system that uses "target" levels... but those levels are all over the place in the spool-up region.
The WOT engine loads at 2750rpm and 3000rpm
(a) vary a lot in the different gears.
(b) rise extremely fast, and
(c) vary depending on the "starting" rpm for your WOT run.
A moving target is a pointless target - hence the endless frustration with a target-based correction system. It has no hope of being "on target" at the very start of the boost curve... only once the boost levels start to level out will it become useful.
Ideas I still have to try:
- Try smaller downward EC values - at least try and get a smooth, consistent boost curve after the initial peak.
- Lower the BWGDC at 2500-3000rpm to "build in" the inevitable upward error correction factor.
- Changing the gear Xover to separate 3rd and 4th maps, and tune them differently... but I really wanted to tune "1st and 2nd" separately at some point. Plus, my XML address for the "rpm/mph Xover" field is currently wrong!
- (edit) Changes to the error correction scale as well as the values... to retain the large "way off-target" downward correction ability, but limit the smaller boost adjustments.
Or I could just put up with it and just move on to AFRs! Nah, I'm not giving up yet!
Rich
Last edited by richardjh; Jan 23, 2011 at 08:25 PM.
Jan 24, 2011, 07:51 AM
#68
Evolved Member
Thread Starter
Okay, here's my current work-in-progress...
I'm focusing on 3rd gear, and working hardest on the 3000rpm to 4250rpm range - where all the action is.
The various boost maps have had their scales modified - I stole some pointless "6500" and "7500" rpm scales, and re-deployed them in the key area where most of the control is required. Worked pretty nicely!
Before starting this, I thought it was gonna look all nice, progressive and pretty. Instead, I've dialled in a steep BWGDC "low" of 67, and a BTEL that looks like Mt. Everest.
The BWGDC was sorted out with error correction disabled. Now it's on again, I'm having to spend more than a little time fine-tuning.
I can see why people switch to direct PSI based boost control - I just can't wait to see how this all looks when winter comes around!
This does seem to be getting results - currently touching 23psi at 3500rpm, holding 22psi to 5000rpm, then tapering to redline.
Will post up some actual results once I've put this whole boost tuning phase to bed.
Rich
I'm focusing on 3rd gear, and working hardest on the 3000rpm to 4250rpm range - where all the action is.
The various boost maps have had their scales modified - I stole some pointless "6500" and "7500" rpm scales, and re-deployed them in the key area where most of the control is required. Worked pretty nicely!
Before starting this, I thought it was gonna look all nice, progressive and pretty. Instead, I've dialled in a steep BWGDC "low" of 67, and a BTEL that looks like Mt. Everest.
The BWGDC was sorted out with error correction disabled. Now it's on again, I'm having to spend more than a little time fine-tuning.
I can see why people switch to direct PSI based boost control - I just can't wait to see how this all looks when winter comes around!
This does seem to be getting results - currently touching 23psi at 3500rpm, holding 22psi to 5000rpm, then tapering to redline.
Will post up some actual results once I've put this whole boost tuning phase to bed.
Rich
Jan 25, 2011, 08:09 AM
#69
Evolved Member
Thread Starter
Okay, so I've seen from subsequent testing that scooping out the BWGDC region to drop the initial boost "peak" has an unpleasant side-effect.
It means a gear upshift that lands squarely in the scooped out region means a quick and nasty downward WGDC adjustment. This is thanks to the SST's lightning-fast WOT shifts, I guess. The result: transient spike in the boost.
[edit - sorry, that's not making much sense! Too late at night, I guess! Maybe the stock system is slow to react, or whatever. Sleep beckons...]
So... that essentially means nearly 20 progressive tuning increments refining my boost curve get chucked in the big round filing cabinet!
There I was, trying to reduce the initial boost peak in my GST-based map from 23 psi to 22 psi, and I ended up with a transient 24psi. Score one for the beginner!
Still, I've got my hands dirty with boost control now, which was another aim of the exercise!
Looks like it's back to the old drawing board. Maybe I'll come at this from another angle, like getting the error correction working harder. Or voodoo.
Or maybe some kind soul will lever PSI Based Direct Boost Control into 53610010 ROM, so I won't have to kill myself with insane load targets!!!
Rich
It means a gear upshift that lands squarely in the scooped out region means a quick and nasty downward WGDC adjustment. This is thanks to the SST's lightning-fast WOT shifts, I guess. The result: transient spike in the boost.
[edit - sorry, that's not making much sense! Too late at night, I guess! Maybe the stock system is slow to react, or whatever. Sleep beckons...]
So... that essentially means nearly 20 progressive tuning increments refining my boost curve get chucked in the big round filing cabinet!
There I was, trying to reduce the initial boost peak in my GST-based map from 23 psi to 22 psi, and I ended up with a transient 24psi. Score one for the beginner!
Still, I've got my hands dirty with boost control now, which was another aim of the exercise!
Looks like it's back to the old drawing board. Maybe I'll come at this from another angle, like getting the error correction working harder. Or voodoo.
Or maybe some kind soul will lever PSI Based Direct Boost Control into 53610010 ROM, so I won't have to kill myself with insane load targets!!!
Rich
Last edited by richardjh; Jan 25, 2011 at 08:58 AM.
Jan 25, 2011, 04:13 PM
#71
Evolved Member
Thread Starter
Apparently not - mine seems to be a weird one that doesn't switch to the Spark Advance LBM during SST upshifts!
In the LBM, I have:
4000rpm / 240 load : -3
3500rpm / 240 load : -6
3000rpm / 240 load : -9
But during the actual shift from 4000 to 3000, timing used is 6* -> 5* -> 3* -> 0*.
It's a bit of a mystery...
Rich
Jan 25, 2011, 04:32 PM
#72
Evolved Member
Thread Starter
Don't know if you recall, but this odd "feature" of my car's engine management was something we saw previously from the logs, back in November:
https://www.evolutionm.net/forums/09...ml#post8844219
Sometimes it does use the LBM, and sometimes it doesn't. It was unresolved then... and now it's getting in my face as I try to get the boost right!
You mentioned at the time that I didn't need to worry about upshift boost spikes with this little RA turbo... but I must confess I'm a little rattled when it goes from 21.2 psi in one gear to 21.5 psi in the next... via 24 psi in the middle.
Rich
https://www.evolutionm.net/forums/09...ml#post8844219
Sometimes it does use the LBM, and sometimes it doesn't. It was unresolved then... and now it's getting in my face as I try to get the boost right!
You mentioned at the time that I didn't need to worry about upshift boost spikes with this little RA turbo... but I must confess I'm a little rattled when it goes from 21.2 psi in one gear to 21.5 psi in the next... via 24 psi in the middle.
Rich
Jan 29, 2011, 08:30 AM
#73
Evolved Member
Thread Starter
Using stock wastegate control, is it the norm to find that boost curves look like this at the start (initial peak -> drop-off -> recovery)...
I take it I can't really hope or expect to tune this dip out, even on stock 2-port?
By now, I must have tried most combinations of factory ECU boost control. Learnt a lot. But I didn't find a magic bullet for this one aspect.
Is this basically a "feature" of the stock 2-port solenoid system? Something about it not being able to react all that fast when the ECU is telling it what to do?
I've got my error correction such that it keeps boost at 21psi, plus or minus 1psi... from peak boost up to 5500rpm. Feels very smooth and progressive indeed.
Is a "+/- 1psi" range about the best I can reasonably expect?
Should I finally be happy now, and move on to the AFR tuning phase?
Rich
I take it I can't really hope or expect to tune this dip out, even on stock 2-port?
By now, I must have tried most combinations of factory ECU boost control. Learnt a lot. But I didn't find a magic bullet for this one aspect.
Is this basically a "feature" of the stock 2-port solenoid system? Something about it not being able to react all that fast when the ECU is telling it what to do?
I've got my error correction such that it keeps boost at 21psi, plus or minus 1psi... from peak boost up to 5500rpm. Feels very smooth and progressive indeed.
Is a "+/- 1psi" range about the best I can reasonably expect?
Should I finally be happy now, and move on to the AFR tuning phase?
Rich
Jan 29, 2011, 09:56 AM
#74
I wouldn't worry about the "ringing" in your boost plot too much. It is, after all, an inherent attribute of any closed loop feedback system. The goal is to find the best balance between gain and damping to minimize the overshoot, but still maintain a rapid spool-up. Changing to a 3-port will indeed give your boost control system a faster reaction time which will lessen the effect you are seeing, but not eliminate it.
You could try to use mivec to "absorb" the overshoot somewhat.
You could try to use mivec to "absorb" the overshoot somewhat.
Jan 29, 2011, 09:56 PM
#75
Evolved Member
Thread Starter
I've just had a bit of a breakthrough in my understanding of how this boost control is actually working.
My tuning foray used the GST base maps as a starting point. In Bryan's tune, it seems he has disabled upward correction via an all-zero setting in the Max Total Upward WGDC Correction table, eg.
Stock vs GST:
-> 
Consequently, it didn't matter what I put in the Turbo Boost Error Correction cells from -20 to 0... they weren't being used. Any time target load was higher than actual load, the ECU just carried on doing its thing with the BWGDC.
I couldn't see this until now, because my XML wasn't pointing to the right address for Max Total Upward WGDC Correction in the first place. For 53610010, it needs to be:
<table name="Reactive Solenoid Max Total Upward WGDC Correction vs TPS" category="Turbo" address="58d12" type="2D" scaling="OffsetWGDC"><table name="TPS" address="6265a" type="Y Axis" elements="9" scaling="Throttle %"/></table>
Now that I have this extra piece of information, the logs make a lot more sense. I couldn't see why error correction wasn't recovering following the post-peak drop...
I now get to play with the double-edged sword that is upward error correction. Too much, and it peaks too high at first. Too little, and it can't recover after the peak. Can't wait to try out more combinations...
Rich
My tuning foray used the GST base maps as a starting point. In Bryan's tune, it seems he has disabled upward correction via an all-zero setting in the Max Total Upward WGDC Correction table, eg.
Stock vs GST:
-> Consequently, it didn't matter what I put in the Turbo Boost Error Correction cells from -20 to 0... they weren't being used. Any time target load was higher than actual load, the ECU just carried on doing its thing with the BWGDC.
I couldn't see this until now, because my XML wasn't pointing to the right address for Max Total Upward WGDC Correction in the first place. For 53610010, it needs to be:
<table name="Reactive Solenoid Max Total Upward WGDC Correction vs TPS" category="Turbo" address="58d12" type="2D" scaling="OffsetWGDC"><table name="TPS" address="6265a" type="Y Axis" elements="9" scaling="Throttle %"/></table>
Now that I have this extra piece of information, the logs make a lot more sense. I couldn't see why error correction wasn't recovering following the post-peak drop...
I now get to play with the double-edged sword that is upward error correction. Too much, and it peaks too high at first. Too little, and it can't recover after the peak. Can't wait to try out more combinations...
Rich