Wastegate control, Boost spikes, PID control

[FreeFlyFreak]

EDIT: So you dont have to read the whole thread, I have summarized the only thing I learned for sure here:
The only thing I really learned after all my experimentation is that I can accurately set whatever Maximium wastegate duty cycle I want using the WGDC table.
38% in the table gives me max wastegate of 95% (with stock dynamics), this is the max Cobb recommends.
The formula is 2xWGDC table plus 10 (+/-1), the formula holds exactly with lower numbers and becomes off by 1 at higher numbers.
It is a useful thing to know, but not what I was looking for when I started.

If you want to read the whole thing go ahead, but some of the conclusions I initially drew turned out to be not correct after further experimentation, so bear that in mind when reading.

-----------------------------------------------------------------------------------------------------------------------------

Some of you guys may know this already, but I have not really seen it posted and I at least found Cobbs documentation confusing and lacking in detail about the wastegate tables and how they related to the whole system and the ECU logic that goes with them. They do mention everything but they don't really tell you everything you need to know clearly.

This post is a little long if you just want the Conclusions with out any explanation skip to the end of the Post.


First let me say I have done my first real tuning to this car with ATR over the past week or so.
Previous attempts with ATR over many many weeks were all just a failed attempt to get rid of high rpm knock in the 1-3 range.
After failing at all my attempts to kill knock, other than pulling load to below stock levels, I bought some Torco accelerator octane booster and used 16oz in my 14 gall tank of 91 oct giving an estimated 93 Oct, voila all knock gone!

So now with the knock under control I could try to tune for power.

I managed to turn up the load/boost a lot with no knock, but I was getting intermittent boost spikes. It was weird I couldn't figure it out, sometimes it would spike and other times it would track load perfectly with no spike. Same with the wastegate duty cycle (WGDC), sometimes it would do what I expected sometimes it would not. There seemed to be no rhyme or reason too it, but I knew there had to be both rhyme and reason tied up in the ECU logic.

I analyzed my logs over and over read different posts, most people were claiming to use the Boost throttle close tables to control boost but cld12pk2go in his ATR journey thread had not touched them.

Anyway to cut a long story short what I discovered was this.

If the car is at 0 WGDC and you floor the throttle you will more than likely spike, depending on what is in your Wastegate duty cycle table. examples would be: after a normal shift WOT to WOT, or if you lift off the throttle for slower traffic, then the traffic moves right and you floor it. Bam spike.
If however you had been driving in boost already, and floored it, the car would not spike, it would track load exactly. Also it would not spike if you floored it at low rpm and the turbo spooled gradually as the car built rpm into the spool region, again it would track load directly.

I got to thinking about it and last night had a eureka moment. PID feedback!

What I believe is happening is this, when the WGDC is closed there is no PID feedback action, if you floor the throttle from there, the car goes to the Wastegate Duty cycle table, looks up the value and that is what it sets whether it is right or wrong, for me at some rpms it was wrong.
Well actually what it sets in the important WOT area, is 50 plus what is in the table plus or minus a couple.

The boost and load then build to whatever that WGDC value achieves and very shortly after that the PID feedback (Boost dynamics/Load dynamics) makes corrections to the WGDC up or down to attempt to achieve either the targeted boost or targeted load. Whichever is lower in the tables.

I tested this out today.
What I did as an experiment was disable all nannies, zeroed the Boost/load dynamics to kill the pid feedback, and raised all boost/load targets to unreachable levels. Yes this is potentially dangerous, but I already knew on my car that even with the WG maxed (long story) I will only spike 22psi, high yes I wouldn't want to run it all the time but for experimental purposes I could live with it if it happened.

With all PIDS and targets disabled the car sets exactly what you program in the WG duty table. Fixed, not variable!
If you zero out the PID corrections and raise the boost/ load to unreachable levels you can see EXACTLY the WG duty cycle (+approx 50) being implemented straight from the table without any correction or limitation.
*WARNING* You do not need to do this. If you decide to try this for your own curiosity I recommend you experiment with lower values to minimize the risk of damage from overboost.
The car will still spike, but the important thing is you can control the max it spikes to by changing the values in the WGDC table.

If the WG duty table says 30 it sets 80 WGDC dead on, 40 in the table it sets around 90 WG, actually it sets 89 in my car. Things get sketchier the closer you get to 100%. In my car if I set 42.5 that gives me 95% WG the max Cobb recommends for durability of the WG system.

By the way, the car rips if you do 100% wastegate 100% of the time at WOT. LOL.... but it spikes like a mother. ;-p
Plus Cobb recommends sticking to 95% max to avoid the WGDC sticking.


So knowing this you can set more or less exactly the maximum of the initial spike you are comfortable with. It will still vary some with load on the engine and temperature because at this point it is a raw uncontrolled value. Very shortly after that the PID feedback will take over and modify the WGDC up or down to achieve the desired boost or load depending on what is being targeted (the lower value of boost or load).

The way I adjusted the WGDC values was to floor it from idle throttle (0 WG) just before each target rpm in the table, I used as high a gear as I was comfortable with to maximize load and therefore boost, also an up sloping road helped maximize spike.
To save a lot of work, cld12pk2go's table in his ATR journey thread is very close to what I ended up with. In fact once I had done my experimenting to confirm exactly what was going on I just took his table as a starting point and modified it for my car/mods.

The experiment I describe above is not necessary for you guys, I just did it for my own knowledge/interest and to confirm my theory of what was going on and why the car would sometimes spike and sometimes not.
You can do the same thing with the pid enabled, for safety, but you MUST start from 0 WGDC so the pid is not already active and the car will use the unchanged WGDC table value (+50 approx) initially. If the WGDC is already active the PID feedback loop may already be in progress and the spike would not occur in that case.

I think this is why if you just roll on the throttle a bit first to get the WG active before flooring it the car likes it better, because the PID has time to get control of things early on, before the spike region.


CONCLUSIONS AND REVIEW.
For those that dont want to read the whole thing, and a summary for those that I confused.

If flooring the accelerator with a current WGDC of 0 the car initially sets what is in the WG duty cycle table (+about50) for that rpm this could cause a spike if the WG Dutyclcle is too high, shortly after that (1/2sec or less) the PID feeback begins to modify that value using the Boost/load dynamics tables to achieve targets more accurately.

If flooring the accelerator when the WG is already alive and in pid feedback the car will track to the best of its ability the lower of boost or load specified in the tables. You can modify the boost/load dynamics values if it is not tracking it well enough.

If your car is spiking above what you want, the WG duty table is the correct way to address it. I do not recommend the Boost limits throttle close as primary boost spike control as some have recommended.
Closing the throttle with the waste gate still open and turbo spinning and building boost will protect the engine yes but the side effect is it would just back that pressure up to the turbo possibly causing surging and damage. It is preventing the symptom, not the cause of the problem.
The Boost limit throttle close is best used as a first line of defense against suck WGDC/boost creep/sensor failure, under the theory that it is better damage the turbo than the motor.

Some other useful rules of thumb I learned while doing this were:
1) To get a close estimate of the WGDC that will be initially set when flooring the accelerator from a 0 wastegate, no pid feedback situation. Take the values in the WG duty cycle table in the WOT area and mentally add 50.
2) Boost and Load are approximated in the important higher value area by a factor of 10. example: 15 psi approximates to a load of 150% on the AP.
3) This one fjames reported but I had forgotten. The values are "up to" and interpolated.
In other words if a table says 190 load 4000 rpm, 180 load 4500, 170 load 5000rpm, the car will attempt to acheive 190 UPTO 4K then 185 to 4.25K 180 to 4.5K, 175 to 4.75K, 170 to 5K
4) The car does indeed target the lower value of either load or boost I had heard various stories about this, so you can choose to tune for load or boost whatever you set as lower..... or even max WGDC if you want to be creative.

There were some others but I didn't write them down and they escape me now.

Hope this helps someone understand the logic of what is going with the ECU and Boost control.

Here is my current map, it still needs some work but it is definitely getting there.
It is intended to be my aggressive/track map for use with Higher Octane or Meth.
Once I get this map finalized I am going to go back to 91 Octane for a while and see if I can solve my knock woes there by using the VVT to lower the dynamic compression....
If not Ill just find out what is the minimum Octane booster I need to run to keep knock at bay.
Or go meth...

[jahman]

Thanks for your (possibly engine damaging) experiment to take one for the team, lol. I noticed my loads were lower than what I had in ATR, and now it makes sense that it must be targeting boost rather than load (resulting in lower load than whats commanded).

I also copied c12pk's wastegate table and it basically made spike disappear for good. The boost throttle close table was useless for me, didnt do a thing.

[cld12pk2go]

To clarify, I did end up tweaking the "boost limit - throttle close" table, but it was the last thing I changed and I really didn't have any spiking at that point.

It is kind of the last line of defense for me in case I am getting over my boost threasholds and my WGDC signal is a 0% (say for a failure mode where a vacuum line comes off/cracks/etc).

It is important to configure the WGDC table close to what you car actually needs to take a lot of error correction out of the control loop (sounds like yours now is).

I also altered the Boost/Load Dynamic tables to modify the action on the PID loop, which might have helped some, but I really didn't notice anything significant. I ended up being more aggressive on these with no over correction or instability issues so the tuning is probably pretty good.

One other thing to note is the WGDC required to meet your targets will greatly vary based upon ambient temperatures (at least with the stock turbo).

I tuned my WGDC when it was ~75°F out, and it actually is tweaked by half to one percent per °F in the range of 50-90°F via the control loop to achieve my targets.

This is strictly due to ambient air density effects and the stock snail being at its flow limits.



Oh if you want to keep the power up, plan to invest in a meth kit. It is way cheaper than racegas or octane boosters in the long run and much more effective.

[FreeFlyFreak]

Originally Posted by jahman I noticed my loads were lower than what I had in ATR, and now it makes sense that it must be targeting boost rather than load (resulting in lower load than whats commanded).

Try changing the bottom 2 rows in the WOT area of the Boost Targets table to higher values.
I used 1 psi over what I expect to see based on Gear based load.
I used a similar technique for Throttle requested Load A and C, I set the bottom two WOT rows to 1% higher than the Highest Gear based load table.

That way at WOT I know exactly which table is limiting and that is the Gear Based Load tables, every time.

I still have a little tweaking to do, but this is the first map that I have been relatively happy with, Ill attach it here if anyone wants to take a look at it.

The only thing I wish we had was a "Maximum wastegate duyt cycle" table.
In my map I target 95% WGDC for the max, using the gearbased load tables at lower rpms. However, I have a problem when accelerating downhill, less load/boost is built and the PID does its thing and maxes the WGDC in a failed attempt to achieve the target boost/load.
At the moment I am just accepting it. It is not like it happens on a regular basis just in certain situations.

Another technique I have thought about would be to design the wastegate tables to hit the boost/load you want with a max of 95% WG and then only allow the PID to correct down by zeroing out the corrections for underboost/underload
This would have the effect I want, but would have 2 undesirable side effects, it would spike more than a properly tuned PID map, additionally it wouldn't be able to correct the WGDC higher in case of an under boost.
Not sure which would be best.
The best would be an additional table.
"Max Wastegate Duty Cycle" where the pid can modify the WGDC but only up to whatever you set in Max WGDC, in my case 95%

I wonder if Cobb could add a table like that??
After all they added the LC FFS tables, and this would be more useful to me than those tables which I have never used.
Off to post in Cobb ATR forum......

Originally Posted by cld12pk2go I also altered the Boost/Load Dynamic tables to modify the action on the PID loop, which might have helped some, but I really didn't notice anything significant. I ended up being more aggressive on these with no over correction or instability issues so the tuning is probably pretty good.

Funny you should say that. I did that too.
I found that in the lower gears ie 3rd I needed a more aggressive correction on the underboost/load side. What was happening was that I have an initial WG duty low to avoid the spike but this left the PID with some catching up to do immediately, but the engine rpm was accelerating so fast and therefore the required WG increasing so rapidly that the load would not catch up to target until late in the rpms where basically requested load came down to meet actual.
The PID didn't correct fast enough because the rpm and hence the WG needed was accelerating faster than the PID would correct.
At least I think that is what was happening... LOL
Anyway, it is fixed now using the Load and Boost Dynamics table.

[fjames]

I've been playing with vvt at higher rpm (adding advance) to see what happens. I'm using small numbers so the changes are subtle, but it looks like it simply hits load targets with less boost and WGDC. Seems like a good thing.

I do virtually all my logging in 4th and 5th gear, so when I picture spiking, it's a WOT lift shift from 4->5. I've never been able eliminate that and I'm not sure it's possible with an otherwise driveable car lol. What I have done somehow is eliminate the WG spike that's always been associated with it. That really threw me, I'm so used to seeing the WG and boost spike together, but it seems to be under control (the WG.)

Interesting idea on throttle cut. I use it to get it to come down off the spike, otherwise it's left to the WG and that takes forever. I'm going to set mine a little higher in the shift-spike area and see if it stops it from overcorrection so much (it always does that.)

[FreeFlyFreak]

Originally Posted by fjames I've been playing with vvt at higher rpm (adding advance) to see what happens. I'm using small numbers so the changes are subtle, but it looks like it simply hits load targets with less boost and WGDC. Seems like a good thing.

What kind of numbers are you up to for vvt?

Originally Posted by fjames I do virtually all my logging in 4th and 5th gear, so when I picture spiking, it's a WOT lift shift from 4->5. I've never been able eliminate that

The map I am working on still spikes some on 4>5, 4>5 is always my worst.
Im going to lower the waste gate duty a little more at 5000 rpm that is always where I end up after a shift, that will lower the spike, and then let the PID increase the WG from there. But that wont be till next week.

Originally Posted by fjames Interesting idea on throttle cut. I use it to get it to come down off the spike, otherwise it's left to the WG and that takes forever. I'm going to set mine a little higher in the shift-spike area and see if it stops it from overcorrection so much (it always does that.)

i think the reason it takes forever for the WGDC to correct downward is because at that point you are targeting load, and the downward correction on load (right side load dynamics) is small when you compare it to the downward correction on boost. Just compare the load and boost dynamics and see what I am talking about. So you could make the numbers on the right hand side of the load dynamics table more negative to make the pid correct for the spike quicker buy reducing WGDC more quickly. But that is bassackwards to me and tackles the symptom not the root cause. It could also lead to unwanted side effects in other gear shifts, so instead what I would try is this:

Just forget the throttle cut, raise it to between targets and fuel cut and use it as a safety net , and use the WGDC and pid dynamics to address it this way:
Lower the wastegate duty cycle at the rpm where you end up after a shift, for me it is 5000ish.
Lower it enough to tame or kill the initial spike.
But then, in addition, increase the left side of the load dynamics (I presume you are targeting load at this point and not boost) so the pid increases the WG upwards more quickly from the initial "after the shift" table value to a value which maintains the appropriate load target.
In effect what you are doing is staging the WGDC.
Lets say for example you set 30 in the WG duty table, that would give you 80% duty cycle initially to avoid a spike above target, and then the PID would more quickly than now increase the WG up to the value required to maintain requested load. Addressing the cause is much better approach IMO than tackling the symptom. But as always it is a balance, if you make it perfect in 4>5 it may mess up the 2>3 so it is finding the happy medium where all shifts are acceptable, that is the key.

Right now I am using, for a load error of -0.19, a load compensation of 0.38, vs the stock 0.24
Try that and let me know if it works for you, should do.

[subparpunk03]

Originally Posted by FreeFlyFreak i think the reason it takes forever for the wastegate to correct downward is because at that point you are targeting load, and the downward correction on load (right side load dynamics) is small when you compare it to the downward correction on boost. Just compare the load and boost dynamics and see what I am talking about. So you could make the numbers on the right hand side of the load dynamics table more negative to make the pid correct for the spike quicker buy reducing wastegate more quickly. But that is bassackwards to me and tackles the symptom not the root cause. It could also lead to unwanted side effects in other gear shifts, so instead what I would try is this:

If I understand your post correctly you are saying that when you have an over-load condition you are on the right side of the load dynamics table. According to the cobb help file documentation, negative load error means you are over your load target:

Boost Error = (Desired Boost Target) – (Actual Boost)
Load Error = (Desired Load Target) – (Calculated Load)

So when you see either a boost or a load spike, you're on the LEFT side of the dynamics tables.

[FreeFlyFreak]

Hmmm..... it looks like I could have things flipped around in that post.......??
The basic principle should apply though.
I'll take another look at that part of it to make sure I understand that part correctly.

EDIT:
You know what. After looking at those tables again......
I think that the Cobb documentation you pasted is partially INCORRECT (the equation part, the rest looks correct).

Think about it. If you are overload or overboost you want to reduce the wastegate duty cycle.
All the negative correction factors are on the right side of the dynamics tables.

If you are underboost or underload you want to increase WGDC.
All the Positive correction factors are on the left side of the dynamics table.

The correction factors are the percentage change made to the Wastegate Duty Cycle value.
Negative correction to WG duty for overboost/load = RIGHT side.
Positive correction to WG duty for underboost/load = LEFT side of table

[cld12pk2go]

Originally Posted by FreeFlyFreak Think about it. If you are overload or overboost you want to reduce the wastegate duty cycle. All the negative correction factors are on the right side of the tables If you are underboost or underload you want to increase WGDC. All the Positive correction factors are on the left side of the dynamics table. The correction factors are the percentage change made to the Wastegate Duty Cycle value. Negative correction to WG duty for overboost/load = Right side. Positive correction to WG duty for underboost/load = LEFT side of table

This is exactly how I understand it to work.

[subparpunk03]

PID controller - Wikipedia, the free encyclopedia

I'm guessing that whats changing is one of the gain parameters, for either the proportional or integral term. It really depends on what the exact equation in the PCM is on how this stuff works. It might not be as intuitive as you think.


***EDIT***

OK after re-re-re-re-reading the PID doc, I'm guessing the value in the ecu tables is more like an overall multiplier for the equation, not one of the gain terms. Unfortunately, both explanations of the tables makes sense to me. For instance if the error is negative, by setting the compensation to be positive you seek to make the error less negative. Also, it could totally be as intuitive as you have stated. So who the hell knows. For now, I trust cobb.

Having said that, I've made the right side of the load dynamics table closer to 0 and I've noticed that it has caused the load to ramp up more slowly. From what little messing around I've done with the load dynamics table, it seems to behave as stated in the cobb docs.

[FreeFlyFreak]

Originally Posted by subparpunk03 So who the hell knows. For now, I trust cobb. From what little messing around I've done with the load dynamics table, it seems to behave as stated in the cobb docs.

Which part do you trust.
There is an inherent conflict in Cobbs documentation quoted below.

Originally Posted by cobb help file Boost Error = (Desired Boost Target) - (Actual Boost) This table represents a compensation (correction) necessary to counteract the Boost error that has occurred. The table is referenced by Boost Error (represented in mm Hg) on the x-axis. Table values are the percentage change made to the Wastegate Duty Cycle value.

The two highlighted parts conflict with each other. They cannot both be true when you look at the table.

Another thing I have noticed is that some of the stuff in their documentation is definitely incorrect. A lot of it is copied and pasted straight from a subaru document it references tables we dont have, mentions subaru a few times, and even talks about high and low wastegate tables, and logging turbo dynamics values.
I for one do not trust Cobbs documentation, I find it confusing, poorly edited, and often poorly worded.

[fjames]

Don't forget Cobb got the math exactly wrong (backward) on the worksheet for MAF/AFR correction.

Funny though, I've played around with the left side (assuming it's under boost/load) to try to limit spiking. My theory was exactly the opposite of your advice above- I was thinking that it was over correcting for the lack of load when shifting WOT. So by lowering the values it would correct less therefore lower spike. It did absolutely nothing as far as I can tell.

But, what if it was as subpar says. Then I was really lowering the over correction side. So, it should have not corrected for the spike so hard. But it does over-correct for the spike.

So in either case, it didn't do shit for me lol. It spikes, then overcorrects for the spike. One or the other should have been affected you'd think.

[FreeFlyFreak]

Originally Posted by fjames My theory was exactly the opposite of your advice above- I was thinking that it was over correcting for the lack of load when shifting WOT. So by lowering the values it would correct less therefore lower spike. It did absolutely nothing as far as I can tell.

Under my theory lowering the left side values would do absolutely nothing to prevent boost spiking, because you would be lowering the correction for underboosting/loading, and therefore would see no effect in an over boosting over loading situation.

Anyway, I try to will look into that part of it more next week if I find the time.

[cld12pk2go]

Originally Posted by subparpunk03 If I understand your post correctly you are saying that when you have an over-load condition you are on the right side of the load dynamics table. According to the cobb help file documentation, negative load error means you are over your load target: Boost Error = (Desired Boost Target) - (Actual Boost) Load Error = (Desired Load Target) - (Calculated Load) So when you see either a boost or a load spike, you're on the LEFT side of the dynamics tables.

It took me a while to see what you are saying...


Either the equations should be:

Boost Error = (Actual Boost) -(Desired Boost Target)
Load Error = (Calculated Load) - (Desired Load Target)

Or I am backwards in my previous understanding. Since I never zeroed out either side I wouldn't know. I only adjusted both to more or less aggressive.

From a purely mathematical point of view, it should be as I just wrote above to obtain the proper sign on the error; however, this is not a requirement for the logic to work.

[subparpunk03]

Dammit guys I had this all worked out in my head... Now i'm going to have to use science and empirical evidence.

and fwiw FreeFlyFreak, you can prevent load/boost spiking by making it ramp up slower, not just correct the spike more aggressively. I've done that successfully on my car... although maybe I just did just the opposite of what I intended... I'll try and science it up tonight.

[FreeFlyFreak]

Originally Posted by subparpunk03 and fwiw FreeFlyFreak, you can prevent load/boost spiking by making it ramp up slower, not just correct the spike more aggressively. I've done that successfully on my car... although maybe I just did just the opposite of what I intended... I'll try and science it up tonight.

I agree that if the WGDC table is set low enough, you can vary how it ramps up from the initial setting with the dynamics table thus preventing a spike. I do not agree that if your value in the WGDC is too high you can stop the spiking with the dynamics table, you can only control how quickly it corrects for the spike.
At least that is my interpretation of what I have seen so far.
And remember all my comments relate to starting from 0 wastegate.
If I start with the feedback active then I saw no spikes.

It is possible I only have a partial understanding, it is also possible I have part of this misunderstood.
I only did one day of experimenting on this, not much in the big picture, and I definitely didn't try all combination's.
Like I said the documentation is confusing, how else could we draw different conclusions from it and our own experiments.
With proper clear documentation there should be no need for all this shit, trying to figure out exactly what the ECU does with various tables. That is what the documentation is supposed to be for, but instead we have to experiment this stuff out for ourselves.

So I will do some more experimenting too, lets see if we cant get it 100% without doubt correct.
I believe the best way to do that will be to set the WOT portion of the WGDC tables to a constant value all the way across, lets say set 30 to give WG 80ish. At the same time set one side of the dynamics tables to zero..... or even set zero on one side and switch positive and negative on the other side, and see what happens, and then swap sides zero.

I will tell you this for sure so you dont have to try it.
If you set the WGDC table to hold 90 WG (set 40) all the way across, at WOT, and zero both the dynamics tables both sides. The car still boost spikes just as bad as normal and the WGDC appeared to open no slower or faster than with the dynamics tables stock.
In this configuration however it DOES set a constant WGDC of what ever you have set in the WGDC table + 50. There is no tampering or attempting to meet targets at all.

Also dont forget, unless you start from 0 WGDC everytime your results will vary because somethime it will have set up he feedback loop and you will be unable to draw conclusions .
If WGDC and boost/load feedback are already active when you floor the accelerator there will be no spike.

[fjames]

Originally Posted by cld12pk2go It took me a while to see what you are saying... Either the equations should be: Boost Error = (Actual Boost) -(Desired Boost Target) Load Error = (Calculated Load) - (Desired Load Target) Or I am backwards in my previous understanding. Since I never zeroed out either side I wouldn't know. I only adjusted both to more or less aggressive. From a purely mathematical point of view, it should be as I just wrote above to obtain the proper sign on the error; however, this is not a requirement for the logic to work.

That's why I pointed out that Cobb got the math wrong before with the MAF/AFR thing. I understand what subparpunk03 is saying, but I'm betting it's just them getting the explanation backward. It wouldn't make sense to make it so unintuitive (I'm assuming they can make it however they want.) It's good that people are following through on this though.

What kind of numbers are you up to for vvt?

I'm up to 15-10-5 for 5K-5.5K-6K. I never logged vvt in the past, so it's weird to realize that the increase in WGDC as rpms climb is directly correlated with vvt reducing. What I logged above was 10-7.5-5. I'm hoping to get to 15-15-5. Theory is reducing BAT, but not sure it will really have any effect yet.

[MS3Zoom24]

Originally Posted by fjames Theory is reducing BAT, but not sure it will really have any effect yet.

interesting cant wait to see what your findings are going to be

[fjames]

If you set the WGDC table to hold 90 WG (set 40) all the way across, at WOT, and zero both the dynamics tables both sides. The car still spikes just as bad as normal and the wastegate appeared to open no slower or faster than with the dynamics tables stock. In this configuration however it DOES set a constant wastegate of what ever you have set in the WGDC table + 50. There is no tampering or attempting to meet targets at all.

That's extremely interesting - it means that the higher than set WG numbers have nothing to do with the correction tables. I don't think everyone's "plus" numbers are the same - mine don't get that high. Of course maybe they would in another situation. I don't know how you would test to see the highest they could go. Maybe set WG at 5 everywhere.

What you're saying above fits into my thinking on WG though - it just doesn't fucking matter where you set it for steady driving. It's when the "unpredictable" happens, like a shift, that the correction maybe has an influence. Then, maybe the correction looks at the actual WG number as a reference. In other words, those numbers must have some meaning ...

[FreeFlyFreak]

Originally Posted by fjames That's extremely interesting - it means that the higher than set WG numbers have nothing to do with the correction tables. I don't think everyone's "plus" numbers are the same - mine don't get that high. Of course maybe they would in another situation. I don't know how you would test to see the highest they could go. Maybe set WG at 5 everywhere. What you're saying above fits into my thinking on WG though - it just doesn't fucking matter where you set it for steady driving. It's when the "unpredictable" happens, like a shift, that the correction maybe has an influence. Then, maybe the correction looks at the actual WG number as a reference. In other words, those numbers must have some meaning ...

Not sure what you are saying in the first paragraph, maybe if you rephrase it I will get what you are trying to say.
EDIT: It is possible you are misunderstanding my use of the term spike. By spike I am soley talking about a spike in boost/load. If I zero the dynamics, the logged WGDC does not rise above that set in the WGDC table (+50). The only time I see rises above what I have set in the WGDC table (+50) is when the dynamics are functioning. I hope I am making my self clear. It is hard to clearly put everything in words.

Regarding the second paragraph.
In normal driving it really doesnt seem to matter much what is in the WGDC table, the feedback system seems to take care of it all OK.
Where the problem comes in is if you floor the accelerator when the WGDC is 0, such as after a normal shift, or after slowing down with idle throttle. At this point there is no feedback and as far I as I can tell the car just sets the initial value from the WG duty cycle table, it is only after that happens and the spike has occurred that the feedback system takes over.
Hence if the initial value in the WGDC table is too high you get a boost/load spike over what you want, and then a correction down. If the initial value is set correctly the car will still spike, but you wont notice see it because the top of the spike is what you are targeting, then as load/boost starts to pull back from the peak of the boost, the feedback system increases the WGDC to correct and the maintain targets.

I keep trying to say the same thing in different ways, because I am not sure I explained it well initially, and it is a difficult concept to explain in words. It would be easier really with a diagram...........
Guess I could sketch something on a napkin...... take a photo..... and post it up LOL ;-)

[LBV]

I understand what you're saying fff and I thanked you as with the TurboXS BPV I'd get spikes of 2 psi above what my tune was set for (heck even spiked to 25 when I was on the 23 psi map for a couple of weeks).

I went back to my 21 psi map but still got upto 23 on the spike ... didn't like that and I reverted back to the stock BPV which I strongly believe is designed to leak boost at 21 psi. I'm ok for the winter as is but I'll definitley start lowering WGDC come spring time.

[fjames]

I understand what you're saying I think. The only problem would by your use of the term "set." I'm taking your use of that to mean actual WG value PLUS 50. So, if you're saying that, for instance, you made every WG value 25 in the table, zeroed all correction tables, then you'd see a maximum of 75 WGDC when logging, I understand you. If you're saying that you'd only log 25 WGDC, the I'm not understanding you.

If I really do understand you, then it means that you can't predict correction factors since the WG does whatever it wants. I've always assumed that the larger than written WG values were due to correction.

In normal driving it really doesn't seem to matter much what is in the WGDC table, the feedback seems to take care of it all OK. Where the problem comes in is if you floor the accelerator when the WG is 0 such as after a normal shift. At this point there is no feedback and as far I as I can tell the car just sets the initial value from the WG duty cycle table, it is only after that happens and the spike has occurred that the feedback takes over.

This is where we disagree. Look at the following slice of a log of mine - 4->5 shift point.



Line 4 is the critical one. The clutch is fully engaged (see rpm,) pedal is almost fully depressed, load is double digit, boost is in vacuum, but I've got 67% WGDC. This says to me that I'm asking for a lot more load than I've got, so the under load correction kicks in and tries to help. That 67% is higher than I'd see for steady-state.

If you look at line 3, I've already got 20% WG but only 3% pedal. It's like it "knows" you're flooring it (it definitely knows the speed you're moving the pedal) and activates a "hammer macro" - lines 3-6 are a pattern I see all the time for WG. Like it intentionally overshoots, immediately backs off a proscribed amount (lines 5,6) then stabilizes. The boost/load lags the WG a few lines, doesn't spike for another 150 rpm or so. As if accommodating the lag in the system is built into the ecu logic.

I just want it to overshoot a little less, but it may be the price you pay with a turbo system if you insist on hammering the gas. A rev matching upshift technique could also help, but would take a lot of concentration and time to program into you muscle memory.

[FreeFlyFreak]

Originally Posted by fjames I understand what you're saying I think. The only problem would by your use of the term "set." I'm taking your use of that to mean actual WG value PLUS 50. So, if you're saying that, for instance, you made every WG value 25 in the table, zeroed all correction tables, then you'd see a maximum of 75 WGDC when logging, I understand you. If you're saying that you'd only log 25 WGDC, the I'm not understanding you.

You understand that part.

Originally Posted by fjames If I really do understand you, then it means that you can't predict correction factors since the WG does whatever it wants. I've always assumed that the larger than written WG values were due to correction. This is where we disagree. Look at the following slice of a log of mine - 4->5 shift point. Line 4 is the critical one. The clutch is fully engaged (see rpm,) pedal is almost fully depressed, load is double digit, boost is in vacuum, but I've got 67% WGDC. This says to me that I'm asking for a lot more load than I've got, so the under load correction kicks in and tries to help. That 67% is higher than I'd see for steady-state.

My question would be...... what is in your WGDC table at 4500rpm?
67% is higher than you see steady state, because steady state is the corrected value, corrected by feedback.
The 67% indicates you have 17, or possibly more, in your WG dutycycle table for 4500 rpm. 17 is obviously too much for you at that rpm, because you end up with a 21.3psi spike because of it. Reduce your WGDC table in the 4500 rpm area by enough and that spike will lessen and hopefully disappear. (presuming that is where you normally end up after a 4-5 shift.)

Originally Posted by fjames If you look at line 3, I've already got 20% WG but only 3% pedal. It's like it "knows" you're flooring it (it definitely knows the speed you're moving the pedal) and activates a "hammer macro" - lines 3-6 are a pattern I see all the time for WG. Like it intentionally overshoots, immediately backs off a proscribed amount (lines 5,6) then stabilizes. The boost/load lags the WG a few lines, doesn't spike for another 150 rpm or so. As if accommodating the lag in the system is built into the ecu logic.

It is possible it sees the rate of change and anticipates, that would be part of a pid function.
But be careful reading too much into what happens where!
I use a DashDaq that samples way more frequently than the AP (AP approx 3 lines per sec, DD approx 46 lines per sec.) and it becomes even more obvious when you see that rapid sampling, that things don't line up on the logs.
I try to look at a bigger slice of the log at one time because of that.
What I am saying is don't assume that everything on one line occurred at exactly the same time, it most definitely didn't!
To show this, I have attached a log it includes a 3>4 shift, it is a log I made when I disabled all feedback and nannies and set my WGDC to go to 89% whenever I was at WOT.
You will see things don't necessarily line up, you will also see the results of a steady WGDC, there is an initial boost spike as the waste gate opens to 89% and then the steady state boost values of an 89% WGDC (on my car) after that.
The WGDC does not change, because on that map I disabled boost and load targeting, it holds 89% the entire time I am at WOT.
So what I am saying is set your WGDC so the top of your initial spike as the WGDC opens meets your target load/boost and then let the feedback catch it on the backside of the faux "spike" by increasing WGDC, so it then maintains your steady state load/boost.

In fact looking at your log again your steady state after the 67%WGDC spike is 42% WG.
As far as I know that may mean it is not possible to get rid of your spike entirely.
I haven't tried setting a WGDC that low, as far as I know 0 in the tables = 50% but I haven't tested that low..... so I don't know for sure. I am just making assumptions based on what I have seen up high.
I would imagine that to get rid of that spike all together you would have to set a WGDC less than 40%, I have no idea how to do that..... other than maybe tuning with the spring preload.

What I would try after looking at your log is set WGDC in the table at 4500 to 0 (or what ever you find gives you 50% actual WGDC), once you achieve an initial 50% WGDC it will still spike some but not as bad as now. You need mid 40s steady state and I don't know how to set lower than 50%....... In fact I have never tried to set lower than 75%(25 in table), by my theory you would need to set your WGDC to be LESS than 42% initially and I do not know how to do that.
So you may in fact be screwed in getting rid of that spike all together, unless you can figure out how to set something in the table to give you lower than 42%WG initially....
Or by tuning spring pretension.

Originally Posted by fjames I just want it to overshoot a little less, but it may be the price you pay with a turbo system if you insist on hammering the gas. A rev matching upshift technique could also help, but would take a lot of concentration and time to program into you muscle memory.

Sounds like your main problem is after the 4>5 shift.
If you generally end up around 4500 rpm after the shift, just change the WGDC table at 4500 to something lower, Im guessing much lower than it is right now. I would try 0 to start in the WGDC table at 4500 rpm and see what happens. If you need to move it up from there then do it. Im thinking your much lower WGDC values are due to your more extensive mods and better airflow/pumping efficiency through that autoexe intercooler, I am on SRI and AP only and need more WGDC to prevent the WG from opening.

[fjames]

That's a lot of data lol.

To answer your question, my WG is 30-35 in the 4K-5K area, and I have it at 25 in the highest three lines.

I've been thinking of zeroing out the WG in that area for so long I can't remember, but never have -it just seems so extreme. But the evidence seems to say it could work.

There's no doubt the IC has a lot to do with it. It's also why you get a large (50) increase over setting - that's much higher than I ever get.

Looking at your log, are you sure it's not trying to hit targets? I'm sure it will use throttle control to some extent to achieve targets. It's a fascinating log to ponder. My experience in general is that this ecu will surprise you - the internal logic has a strong personality and is hard to fool lol. I mean, it just doesn't make sense to me that it will use all that WGDC above the set values IF the correction is zeroed. The values are accurate in the table presumably, it can't go higher than 100. So, what if you set it to 90 everywhere. If it wants to add 50 on your car, would it just go to 100 and quit?

The logical answer too me is that there are more correction tables we can't access, or the behavior is hard-wired, and the correction tables we have only work in certain narrow situations.

As I think about it, if I'm going to zero something out, I think I'll try the left side of the boost/load correction tables. Until subpar proves us wrong I'm going with that being the "under" side. I've reduced it, but never actually zeroed it IIRC.

[FreeFlyFreak]

Originally Posted by fjames I've been thinking of zeroing out the WG in that area for so long I can't remember, but never have -it just seems so extreme. But the evidence seems to say it could work.

That is what I would do. The evidence says it will help. Signs point to yes........ see! even my magic 8 ball thinks so.

Originally Posted by fjames There's no doubt the IC has a lot to do with it. It's also why you get a large (50) increase over setting - that's much higher than I ever get.

Exactly, my stock IC causes so much back up I need to open my WGDC soleniod almost fully to build the same steady state boost as you do with 42%WG

Originally Posted by fjames Looking at your log, are you sure it's not trying to hit targets? I'm sure it will use throttle control to some extent to achieve targets.

I didnt log actual throttle so it is possible, but I would bet my last dollar that it isn't attempting to target....... EDIT in fact it cant be because I set my Boost targets and throttle close at like 24 and my load targets to 235, if it was trying to meet targets it would have made WGDC go higher than 89%

Originally Posted by fjames it can't go higher than 100. So, what if you set it to 90 everywhere. If it wants to add 50 on your car, would it just go to 100 and quit?

Funny you should ask that, the way I figured out that actual WGDC was WGDC in the table +50, I set I think it was 80 in the table.
Can you say Zoom Zoom!
100%WG every time I stepped on it. LOL
I mentioned it earlier.
That is where I began experiment and learned that for me 42.5 on the table = 100% WGDC initially, if flooring it from 0 WG and below, and 40 in the table was 89% WGDC, that was WGDC table + 50 close enough.

Originally Posted by fjames As I think about it, if I'm going to zero something out, I think I'll try the left side of the boost/load correction tables. Until subpar proves us wrong I'm going with that being the "under" side. I've reduced it, but never actually zeroed it IIRC.

It wont do anything for you IMO.
IMO left side is for underboost under load, zeroing it will not prevent your spike, you are overboosting that would be right side of the dynamics table if anything IMO, but try if it you want to satisfy your curiosity.
I mean mine was zeroed both sides and you saw the spike the 89%WGDC log I posted.
A spike of some sort is always going to occur when the waste gate suddenly opens from 0.
You just need to open it a smaller amount for your set up, so the boost peaks where you want it to, then the feedback will catch it to prevent it from falling off.

IMO what you need is 0 in the WGDC table around 4500 rpm.
My guess would be you also need 0 in the whole thing, above 2250 rpm.
Just try it....
It cant hurt anything.
The worst that can happen is you under boost and have to try again, but I am telling you, I don't think that will happen.


EDIT: One more thing. And I have no evidence for this at all it is pure speculation. But it is possible you may gain some control in this from the Boost Comp Gear tables for 5th gear.
I have never touched these tables and so it is pure speculation but, that would be the second thing I would play with after the WGDC.

[cld12pk2go]

I thought the actual WG% is just 2x the WGDC table value. Or at least that is the logic I used when I tuned my table.

I just took my actual logged WGDC% and divided by 2 and stuck those into my WGDC table. It seems to have worked great.

[fjames]

Originally Posted by cld12pk2go I thought the actual WG% is just 2x the WGDC table value. Or at least that is the logic I used when I tuned my table. I just took my actual logged WGDC% and divided by 2 and stuck those into my WGDC table. It seems to have worked great.

I've wondered if the actual logic is a multiple. But everyone's car is different - most people run stock WG and look at anyone's log with the stock IC ... WG is maxed all over the place, almost triple the values.

[FreeFlyFreak]

Originally Posted by cld12pk2go I thought the actual WG% is just 2x the WGDC table value. Or at least that is the logic I used when I tuned my table. I just took my actual logged WGDC% and divided by 2 and stuck those into my WGDC table. It seems to have worked great.

At the higher values 2x comes close, but I found WGDC plus 50 a better approximation in the values I checked.
I didn't check the really low values with the feedback disabled.....
It is possible that relationship doesn't work down low, but I have no need for those values and so didn't check them at all.
I'm guessing we'll find out if fjames zeros his out, will he get close to 0 or close to 50......

I have $5 on close to 50.


EDIT:

Originally Posted by fjames To answer your question, my WG is 30-35 in the 4K-5K area, and I have it at 25 in the highest three lines.

If you have WGDC 30-35 in the shift region that tells me the feedback prevented an even worse spike than you actually got!
30-35 in the table is 80-85% WG...... at least in my car.
You only hit 64%WG before the feedback dragged it down. That was because of the way you applied the throttle after the shift.
You didn't show the seconds but my guess is 1-1.25 secs from 0 to full throttle judging by the number of lines logged. That is enough time for the feedback to help stop something worse. My worst spikes are when I go from 0 to 100% throttle in less than 0.3 of a second after a rapid lift shift. The feedback doesn't do anything to an excessive wastegate setting in that case, until it is too late

[MS3Zoom24]

im with coldpack, on the WGDC being 2x the ATR tables,

i just checked my WOT logges for 2nd and 3rd, and they correspond within +7 on 3rd and 2nd,

which seems pretty close since there are so many other tables the affect WGDC!!!!

I have attached a log with the values in a blue box on the right corresponding to the WOT pulls

[FreeFlyFreak]

Originally Posted by MS3Zoom24 im with coldpack, on the WGDC being 2x the ATR tables, i just checked my WOT logges for 2nd and 3rd, and they correspond within +7 on 3rd and 2nd, which seems pretty close since there are so many other tables the affect WGDC!!!! I have attached a log with the values in a blue box on the right corresponding to the WOT pulls

Did you read the thread??

I am not talking about steady state values when they are being modified by feedback, which is what I see in your log.
I am talking about unmodified values.
To see the unmodified values, you have to step on the accelerator very quickly from 0 wastegate situation.

Ill tell you what.
Zero out your boost and load dynamics tables, and then tell me what your logs tell you.

EDIT:Actually, no need to do that....... look at your log line 657 that is right after you step on it. It tells you everything you need to know, your table is 29-28 if I read that right, your actual wastegate goes initially to 75% when you floor it, until the feedback corrects it down to where it should be ;-p Your 2x predicts 57, my 50+ predicts 78.5...... actual value 75 and that is possibly pid modified as dynamics were not zeroed.

EDIT AGAIN: Maybe you discovered a new rule of thumb, steady state values are approx 2x WGDC table, but really I think that is more related to what you have set as your boost or load target, not what is in the WGDC table.

[MS3Zoom24]

Originally Posted by FreeFlyFreak Did you read the thread?? I am not talking about steady state values when they are being modified by feedback, which is what I see in your log. I am talking about unmodified values. To see the unmodifed values, you have to step on the accelerator very quickly from 0 wastegate situation. Ill tell you what. Zero out your boost and load dynamics tables, and then tell me what your logs tell you. EDIT:Actually, no need to do that....... look at your log line 657 that is right after you step on it. It tells you everything you need to know, your table is 29-28 according to you, your wastegate goes initially to 75 when you floor it, until the feedback corrects it down to where it should be ;-p Your 2x predicts 57, my 50+ predicts 78.5...... actual value 75 and that is possibly pid modified as dynamics were not zeroed. EDIT AGAIN: Maybe you discovered a new rule of thumb, steady state values are approx 2x WGDC table, but really I think that is more related to what you have set as your boost or load target, not what is in the WGDC table.

ok so on line 657- im at 2600rpm observed WG at 75.36 and the ATR table is set at 32.7 (65.4 when doubled), which is close to the table. Then consider it is a underload situation, observed 78.1 load, 3rd hear trl at 2600rpm is set at 169 so load dynamic would want to ask for more WGDC.

This makes sense to me but this thread with PID confuses me a little bit, i have read it threw a couple times and i still need another read

it seems to me that WGDC table is doubled and then you have other values from boost or load dynamics to correct the changes necessary to achieve targeted boost or load depending on the ECUs target goal

[fjames]

Yeah, I try to shift the same all the time when logging, and also to avoid too much of a spike. But all I have to do to spike to 80 is raise load target .02, it all starts there.

Here's something maybe worth pondering - if you convert the correction values to percentages, it all works pretty much. In other words, a .10 correction would be 100% (doubling or halving the value.) If you look at the stock correction tables it falls into place pretty well.

Of course you've shown that zeroing the correction out still produces larger values. But what if it's using the same correction values internally for steady state, and what we have access to just applies in other situations?

Okay, here's what I did. I set the entire column for 4K and 4.5K at 20, and I have nothing larger than .10 for all left side correction (same as its been for awhile.) Something should change if it's the right track.

Originally Posted by MS3Zoom24 it seems to me that WGDC table is doubled and then you have other values from boost or load dynamics to correct the changes necessary to achieve targeted boost or load depending on the ECUs target goal

The ecu definitely has ways of adjusting for targets other than WG control. Make it a habit to always log Throttle Duty Cycle and you'll see it change sometimes in ways you can correlate to targeting.

[MS3Zoom24]

Originally Posted by fjames The ecu definitely has ways of adjusting for targets other than WG control. Make it a habit to always log Throttle Duty Cycle and you'll see it change sometimes in ways you can correlate to targeting.

What is the Throttle DC, and what does it do?

I always watch Throttle position but what does it do different??

edit: when you said .10 correction factor as percentage would be 100%, did you actually mean .10 = 10%

[fjames]

Good question, I've never logged position to compare. Duty cycle seems more "root" to me as it should relate to the DBW table. It's what the ecu is doing to your throttle, no matter what you say about it lol. 80 is max, I've never heard of anyone seeing more than 77 or so, and I usually see ~73.

[jahman]

fjames, what IC do you have?

My WGDC is maxed in a lot of places too and it seems the only way to prevent it is to put boost way down, or load way down. First I thought my BPV was leaky, replaced it, still maxes, and I know I dont have a boost leak. Stock IC is the next thing that makes any sense.

[FreeFlyFreak]

Originally Posted by fjames Of course you've shown that zeroing the correction out still produces larger values.

NO!!
That is not what I have shown or tried to explain. Exactly the opposite.
I guess I wasnt clear.

If I zero the dynamics correction I can set EXACTLY what I want as wastegate% by using the WGDC table.
There would be no larger (or smaller) values for WG than what I set.
That is what I have been trying to say the whole thread.

There is always a boost spike in higher gears, even with the dynamics zeroed but I can control where the PEAK of the boost spike ends up by changing the WGDC table values. If you set it correctly (would be a WGDC lower than steady state) and leave the dynamics inactive it will spike to your target, but then drop below target, not what you want, but wouldn't overboost.
BUT, if you have dynamics active, instead of the boost spiking to target and then dropping to below target, the dynamics will prevent that from happening, it would spike to target and then the dynamics would correct to maintain target.

I am thinking this is my last attempt at explaining verbally. LOL.

I need a diagram, but am useless at that stuff....... unless I try the Wenis approach and draw it out on a piece of paper, photo it and post it up.
Look at the log I posted..... at WOT wastegate is set for 89% with dynamics 0'd, 89% is what I get every time. No larger values, no smaller.

Originally Posted by fjames Okay, here's what I did. I set the entire column for 4K and 4.5K at 20, and I have nothing larger than .10 for all left side correction (same as its been for awhile.) Something should change if it's the right track.

That may reduce it some, IF you always ease on the throttle like that, because the Dynamics will have time to help you. I would not lower the left side correction, as IMO that is lessening your under boost correction, but hey do what you think will work. I have posted what I think.
BUT if you ever hammer it, with 20 in the WGDC, you will still spike, IMO.
That 20 in the table will try to set approx WG 70% and your car spikes with with 67%!
Zero that WGDC out then maybe you have a chance of seeing no spike or at least a smaller one every time, but if you do that you may want to raise the left side of those dynamics back up to where they started.

There is no ghost in the machine, or secret codes going on IMO, it is just ECU logic and feedback, that is all.
The differences between cars are in the Mods., Maps, and the way the driver drives, which makes all cars look like they are using different logic but they are not.

You just need to start zeroing stuff out to see exactly what is happening and why, otherwise the feedback just confuses you by masking stuff sometimes, and sometimes not, depending on how fast you mash the pedal and if the feedback is active when you do it.
This is what confused me to start. It makes it look like the ECU is using different logic, but it is not, it just can only compensate so fast, and if you floor it too fast from a no feedback situation it doesn't compensate till it is too late, if you ease the throttle in over 2 seconds, the feedback has time to make the needed corrections and it will track load/boost perfectly.

Here's something maybe worth pondering - if you convert the correction values to percentages, it all works pretty much. In other words, a .10 correction would be 100% (doubling or halving the value.) If you look at the stock correction tables it falls into place pretty well.

The Cobb documentation says it is a percentage.
I read it as 0.10 would be 10% correction to whatever wastegate you have currently active.
There is no way you would want a 100% increase over what is set.

Originally Posted by fjames The ecu definitely has ways of adjusting for targets other than WG control. Make it a habit to always log Throttle Duty Cycle and you'll see it change sometimes in ways you can correlate to targeting.

Yes it is the Boost Limits - Throttle close. In my experiment I increased it to 24psi so I could see what was going without it interfering. Yours may be having an effect if you have it set to 21psi or less. But as I said in my first post that is not a way I would recommend to control boost on a day to day basis.

Originally Posted by jahman fjames, what IC do you have? My WGDC is maxed in a lot of places too and it seems the only way to prevent it is to put boost way down, or load way down. First I thought my BPV was leaky, replaced it, still maxes, and I know I dont have a boost leak. Stock IC is the next thing that makes any sense.

He has autoexe TMIC I believe.

I am stock IC too. If you call for too much load you WILL max the WG especially high rpm high gear. The only way around that is to get something less restrictive or lower your targets, that will lower the WG required per psi achieved in the manifold.


EDIT:
I was looking at some logs I did which had forgotten about.
The map had a WGDC map very much like 12packs but with pids and load/boost targeting disabled, I found the 50+ rule does not estimate WG well down in the 25 table region but neither does 2x. The best estimate is actually midway between the two. So lets say for 25 in the table, 2x predicts 50, 50+ predicts 75, actual value was 61, and this is with targeting disabled. It could be those gear specific boost correction multiples that is making it difficult to come up with a rule of thumb that works for all gears and rpms....... That is what I am guessing. Anyway just though I would point it out. That the +50 does not hold at lower values.
I may try zeroing out the WGDC tables above 2250 rpm my self with the targeting disabled just so I can satisfy my curiosity.

[fjames]

Originally Posted by FreeFlyFreak If I zero the correction I can set EXACTLY what I want as wastegate by using the WGDC table. There would be no larger (or smaller) values for WG than what I set. That is what I have been trying to say the whole thread. Look at the log I posted..... at WOT it is set for 89% with dynamics 0'd, 89% is what I get every time. No larger values, no smaller.

I thought we had this clear, now you're confusing me again. I understood that on that log, to get the 89% WGDC, you had the map set to 40 WGDC, thus accommodating the +50 you always get. Is that incorrect? That's what I meant by my comment - your +50 isn't coming from the correction that we can set in ATR, it's coming from somewhere else - the WG isn't being limited to the 40 that you have in the map. If you had it set for 40 like I understood, that 89 could also be a safety limit the ecu is imposing because of the zeroed correction - in other words, without that safety it would just go to 100%.

Originally Posted by FreeFlyFreak BUT you ever hammer it, with 20 in the WGDC, you will still spike, IMO. That 20 in the table will try to set 70% and your car spikes with with 67%!

I usually get the ~doubling that others note - I don't get the larger increases over actual map settings that you see.

Originally Posted by FreeFlyFreak There is no ghost in the machine, or secret codes going on IMO, it is just ECU logic and feedback, that is all. The differences between cars are in the Mods., Maps, and the way the driver drives, which makes all cars look like they are using different logic but they are not.

Good point, but we don't know what the logic is, or what Cobb has decided to show us in ATR.

Originally Posted by FreeFlyFreak The Cobb documentation says it is a percentage. I read it as 0.10 would be 10% correction to whatever wastegate you have currently active. There is no way you would want a 100% increase over what is set.

Yeah, that's how you, me and everyone reads it, but that doesn't mean we're right. It's just funny that moving the decimal makes things work out right. 100% increase is smaller than what you see usually - it's just doubling the value. 150% increase of say 30, would be 75. Don't forget that it's tied to the level of error ... so the numbers still work whether it's a stock IC (missing load targets by a larger error) or upgraded.

Originally Posted by FreeFlyFreak Yes it is the Boost Limits - Throttle close. In my experiment I increased it to 24psi so I could see what was going without it interfering. Yours may be having an effect if you have it set to 21psi or less. But as I said in my first post that is not a way I would recommend to control boost on a day to day basis.

No, I'm not talking about that - I'm talking about the ecu having direct control of the throttle at all times. It will increase it as well as decrease it - blending that control into whatever else it does to meet targets. It's a good example of the ecu doing what it wants - everyone has the higher loads set to "80" in DBW, but no one ever actually sees 80 - it's always mid 70s. I always think of the values in ATR as "up to" that number because that's how the ecu seems to interpret them.

Originally Posted by FreeFlyFreak He has autoexe TMIC I believe. I am stock IC too. If you call for too much load you will max the WG. The only way around that is to get something less restrictive, that will lower the WG required per psi in the manifold.

Yeah, I've got the autoexe, but I doubt it makes any difference as long as you're lowering the pressure drop. The autoexe takes it to extremes, but anything would help.

jahman, did you see that thread on the guy having his stock core replaced by a competition radiator shop? That's what I'd probably do if I did it again. $400 all in, check it out.

[FreeFlyFreak]

Originally Posted by fjames I thought we had this clear, now you're confusing me again. I understood that on that log, to get the 89% WGDC, you had the map set to 40 WGDC, thus accommodating the +50 you always get. Is that incorrect? That's what I meant by my comment - your +50 isn't coming from the correction that we can set in ATR, it's coming from somewhere else - the WG isn't being limited to the 40 that you have in the map. If you had it set for 40 like I understood, that 89 could also be a safety limit the ecu is imposing because of the zeroed correction - in other words, without that safety it would just go to 100%.

LOL, No offense meant........ but you are too wrapped up in mystic secret stuff going on in the background IMO.
I can command wastegate 100% "at will" on that same map as I posted the log for (dynamics 0'd) by increasing the WGDC table to 50.

The way I think of that +50 is this:
The first 50% is WG spring preload, it is fixed unless you do something mechanical to adjust it, the other 50% is what you can control with the table.
That maybe why the relationship is not exactly linear, because the way the spring reacts is not exactly linear.
That may not be a correct analogy, but it is how I wrap my head around it.
I dont believe in the hidden stuff going on in the background.


I have tried to explain my previous explanations with a diagram instead of words.
I am not good with photoshop so forgive.....
Blue line is target boost/load
Black line, correct WGDC set, boost spikes to target and feedback keeps it there.
Red line WGDC too low, feedback then corrects up.
Green line WGDC too high, over boost/load, then spike dies and feedback brings it down the rest of the way.

I cant think of a way to explain it any better or clearer right now, and cant log either @ work, I will do some more experimental stuff when I can get to it. Time will tell if I am right or wrong... Hope that helps.

[fjames]

Originally Posted by FreeFlyFreak LOL, No offense meant........ but you are too wrapped up in mystic secret stuff going on in the background IMO.

That's funny. A lady friend of mine would say, "why are you (me) so obsessed with one little fact/detail?" In other words she'd be looking at my end of the conversation just the opposite of you lol.

The crux of it to me is that little fact - the WG table bears no resemblance to reality. I'm just thinking out loud looking for why. You've accommodated it to the point of auto-correcting for it in your head. Our thought processes are different so we have a hard time communicating, but my thought process isn't mystic, it's very grounded at the core - one little fact.

I remember when I got my '00 Si it was the first OBDII ecu for them, so I paid attention to stuff on that. There was an article that said OBDII standards called for something like 250 separate monitors (a huge increase from OBDI.) That was 10 model years ago. Even if a modern ecu is still just monitoring 250 variables, that's a boat-load of data to deal with, in real-time, in a very dynamic environment. So of course there's tons of stuff going on in the background that we can't access (most of which we don't need to.) ATR is just a gui for a small part of the ecu code that Cobb decided to let us access. There are probably parts they knew of but didn't let us have, parts that they didn't have yet, parts that they had but didn't understand themselves and on and on. It is what it is, and the more we play with it, the more holes we'll find. I think the WG table qualifies as a hole

I can't think of another table where both A and B are false.

A. Table "X" values are accurately logged within a reasonable margin of error.

B. If Table X values are significantly off in logs, the error can be explained by combining Table X with one or more other tables.

For WG, "B" seems to be true if you make my decimal point correction to the dynamics tables (oddly.) But "A" is always false.

Originally Posted by FreeFlyFreak I can command wastegate 100% "at will" on that same map as I posted the log for (dynamics 0'd) by increasing the WGDC table to 50.

Have you actually done that, or are you assuming. It would surprise me that the ecu would allow a full 100%. I see WG activity all the time in vacuum (monitoring real-time.) It's predicting (the need for boost) based on speed and amount of pedal movement. I see low values in under spring pressure boost all the time. I also can get into boost with no WG activity. In PT it depends on how "demanding" you are of it - how fast and far you move the go pedal.

Point is, this is PT. All the stuff we've been talking about is WOT, and I've long thought that something is up at pure WOT. I used to get 4+ KR in 4th gear and up starting at 4.5K. It was automatic at WOT. But at say 92% throttle, nothing. You had to go allll the way into it to get it to knock. I'll see if I can log a '90%" run and see what the WG does.

My point is that all bets are off at WOT. You said in your first post that it wouldn't load spike if there was active WG, but mine used to do it routinely. One of my chapters with ATR involved a lot of logging while easing into it as slowly as I could and still have a little left at WOT. So, 4th gear, start ~2.5K and reach WOT at ~5K. It would load spike every time right around WOT. Just a small spike, wrong to call it a "spike" really, but it would overshoot the target. At WOT you obviously trigger something in the ecu- this is old news, cars have had WOT circuits of some sort, electrical or mechanical, for a long time (like the secondaries on a quad carb or whatever.) When you go from 98% pedal to 100%, you're not just adding 2%, you have to be triggering something and it would be interesting to see if the WG is involved. Yeah I know, mystic ...

[FreeFlyFreak]

Originally Posted by FreeFlyFreak I can command wastegate 100% "at will" on that same map as I posted the log for (dynamics 0'd) by increasing the WGDC table to 50.

Originally Posted by fjames Have you actually done that, or are you assuming. It would surprise me that the ecu would allow a full 100%.

Ill just address the above part of your post at the bottom of this post...

The rest I am going to leave right now, as there is a communication problem, and I dont think I can explain it any better than I have.
I did go back over my previous posts and clarify/reword in some areas that I thought were unclear.
I would be be best to just finish my map and post the map and logs.
That will be all the info you, or anyone else needs, plus proof.

I will tell you this though, just 2 revisions after getting a grasp of the situation.
I drove home last night, trying to spike and got only one it was 4000 rpm high gear no feedback floor it.
When I say no spikes I mean no matter what rpm I floor it or what gear (3000-6000 3rd thru 6th), it not quite everywhere yet, but almost!
Just 2 revisions since I grasped what was going on.
I don't spike on shifts any more.
Plus I'm meeting all load targets =/- 2, some can be increased as feedback still has WGDC left to play with.
All this using the info I have posted here.
There is no secret handshake with this ECU.
You just need you know what is going on.
I don't claim to understand it all yet, but I am getting enough grasp that when I change things, what I expect to happen happens.

I no longer believe in the "all these cars are different" or the "ecu does things on its own"
The ecu does what you tell it, the documentation is poor and so we don't know what to tell it.
The feedback also makes it hard to to interpret what it is doing with the changes sometimes and therefore it can get confusing.
But it is not rocket science, The ECU is a computer, reading signals, and responding to what we change in the tables.

I personally believe the minimum amount of work went into the the Cobb documentation for 2 reasons.
1) ATR beta is free, are they really going to spend a lot of man hours for documentation.
2) If the documentation is not 100% clear it will "encourage" people to pay for a pro tune from a Cobb Authorized Tuner who has bought Access tuner pro.


You don't have to believe that what I have posted works.
Try the techniques I have posted/recommended and you will know it works.
I will post logs and the map when I am done.


To address your question in the quote above, here are logs of the experimental/accidental map I made that was 100% WG anytime I was WOT. This still had the PIDs working, but the WGDC table was set stupid high, and so I am guessng the feedback never got the number down to something that would decrease the WG from 100%.
.
Last 3 rows in WGDC were like this (WARNING! don't try this at home folks !WARNING):

50.00	50.00	50.00	50.00	50.00	32.70	46.00	57.00	68.00	78.00	90.00	90.00	90.00	90.00	90.00
50.00	50.00	50.00	50.00	50.00	32.70	46.00	57.00	68.00	78.00	90.00	90.00	90.00	90.00	90.00
50.00	50.00	50.00	50.00	50.00	35.00	46.00	57.00	68.00	78.00	90.00	90.00	90.00	90.00	90.00
50.00	50.00	50.00	50.00	50.00	35.00	46.00	57.00	68.00	78.00	90.00	90.00	90.00	90.00	90.00

I have to say..... that map was fun, an accident, but it ripped.
I have been yearning to put it back on.
LOL it really was fun.
But you can see from the logs I didn't wring it out, it was kind of a shock/surprise for me the way it worked and I was worried what I might do to the car.
But one day I'm gonna put that sucker back on and wring it out fully: