JB4 logging question: Target boost data unit convention

[i3igpete]

Sure didn't take much reading to figure out what causes throttle cuts on MT JB4 cars. Makes me wonder if some of my maps would work better in Normal mode lol. I have never logged in normal mode.

Pretty wacky find, at least it'll be easy to test and see if it works.

btw, did you ever figure out what the LOAD signal in the JB4 logs are? They're in the mid teens for both of our logs during pulls, so I don't think it's a DME output.

Sponsored

 

[ColonelAdama]

Pretty wacky find, at least it'll be easy to test and see if it works.

btw, did you ever figure out what the LOAD signal in the JB4 logs are? They're in the mid teens for both of our logs during pulls, so I don't think it's a DME output.

Don't quote me on this, but I'm pretty sure it's a meaningless variable for the JB4 on this platform. Thought I read that somewhere.

So with my new knowledge of the "on throttle overboosting", I started a new map from scratch today.

I did several logs on Map 0 and I could see that even the stock tune relies heavily on the throttle. I wondered if my random DME BT drops were because it is expecting to close the throttle when it hits target, and doesn't - instead lowering the target.

Now, I'm trying a new map to embrace the throttle closure on initial spool up. Seems to work well, pretty mild map but it feels stock smooth. Log is short but I've never had issues 5k+ anyway.

 

[razorlab]

I could see that even the stock tune relies heavily on the throttle.

Yup, the oem mapping uses throttle as part of the boost control, it does it well enough that you don't feel it.

Here is a stock 2020. You can see it does it based on pressure deviation.

 

[ColonelAdama]

Yup, the oem mapping uses throttle as part of the boost control, it does it well enough that you don't feel it.

Here is a stock 2020. You can see it does it based on pressure deviation.

I was aware that it did, but the throttle cut in map 0 was a lot more severe than I expected it to be. Not surprised if that's from the downpipe.

 

[i3igpete]

93 is perfectly safe for using Map 4. Your timing crash is not an octane issue. I think its because of underboosting. Let FF adapt up or set it to 58 yourself. Set boost safety to 23.

looks like ff was still sitting at the default of 50. changing it next time I go for a drive. looking at the latest log again, I do observe that Midway up the RPM range it does adjust up, but as soon as I lift off, the FF signal snaps back to 50. How many WOT cycles does it take to self adjust permanently?

I'm guessing the default boost safety of 31 psi might be to avoid triggering during lift off due to lack of BOV. unless the jb4 programming is smart enough to ignore those events.

 

[ColonelAdama]

looks like ff was still sitting at the default of 50. changing it next time I go for a drive. looking at the latest log again, I do observe that Midway up the RPM range it does adjust up, but as soon as I lift off, the FF signal snaps back to 50. How many WOT cycles does it take to self adjust permanently?

I'm guessing the default boost safety of 31 psi might be to avoid triggering during lift off due to lack of BOV. unless the jb4 programming is smart enough to ignore those events.

Boost safety only applies to sustained boost, not a little spike. When I am talking about FF adapting up, I am talking about FF value under user adjustment, not the variable in the log.

 

[i3igpete]

I am talking about FF value under user adjustment,

yep, that's the one I changed. But from what I can tell, the logged FF value snaps back to the user input FF value after you lift.

 

[ColonelAdama]

yep, that's the one I changed. But from what I can tell, the logged FF value snaps back to the user input FF value after you lift.

I'm a bit confused. If you mean the FF variable in the log goes to 50 when you lift, that's because 50 is basically a passthrough. JB4 not intervening wastegate control when not on the throttle.

Its probably just a coincidence your "user adjustable" ff value is also 50, because that the default.

 

[i3igpete]

I'm a bit confused. If you mean the FF variable in the log goes to 50 when you lift, that's because 50 is basically a passthrough. JB4 not intervening wastegate control when not on the throttle.

Its probably just a coincidence your "user adjustable" ff value is also 50, because that the default.

my mistake, I see what you're saying now. By changing FF to your recommend value of 58 and then 42, I can see the logged ff signal jumps from the passthru value to the input setting as soon as you get on it. lift, and it goes to 50 (regardless of user ff setting).

 

[ColonelAdama]

my mistake, I see what you're saying now. By changing FF to your recommend value of 58 and then 42, I can see the logged ff signal jumps from the passthru value to the input setting as soon as you get on it. lift, and it goes to 50 (regardless of user ff setting).

The FF value under user adjustable is just FF gain , it doesn't specify what the actual FF datapoints in your log are gonna be.

Those are a function of RPM, FF gain value, Duty Bias values, pedal input, boost targets, and a few other variables.

 

[i3igpete]

Had about an hour of running errands so I was able to adjust the bias table a bit. Started by flipping bit 1 to disable adaption, and then with each iteration I adjusted duty bias curve to align the shape of FF and WGDC. Each iteration I also reduced the PID gain from 20 -> 15 -> 10 -> 5 to weaken its response. This was the final log of the day where I switch the PID back to 20.

https://datazap.me/u/i3igpete/back-walmart?log=0&data=1-3-4-8-11-20&zoom=236-608&mark=260

It looks like there is a consistent closure at 5000RPM across multiple pulls, so I need to dial back the duty bias at that operating point. Otherwise, I tried to keep the logged FF as a consistent ~2 points below WGDC to ensure a positive PID response without any positive/negative swings. If I understand correctly, turning bit 1 back off will allow wastegate adaption to vertically shift my entire FF curve. so that it's hopefully line-on-line with WGDC.

Boost is within 1 psi above 3500. However, is my 14psi boost target below 3000 going to throw the PID controller off?

Timing looks better than before as well. seems to consistently approach 10 degrees at 6000.

Does anyone mess with the auto shift reduction? I checked a couple logs on jb4 forum and almost everyone leaves it at default.

 

[razorlab]

Timing looks better than before as well. seems to consistently approach 10 degrees at 6000.

Does anyone mess with the auto shift reduction? I checked a couple logs on jb4 forum and almost everyone leaves it at default.

Those timing curves still look so bad. They should be a smooth stair step increasing with revs. The fact that they are going up down like a saw means the car is unhappy.

 

[ColonelAdama]

Had about an hour of running errands so I was able to adjust the bias table a bit. Started by flipping bit 1 to disable adaption, and then with each iteration I adjusted duty bias curve to align the shape of FF and WGDC. Each iteration I also reduced the PID gain from 20 -> 15 -> 10 -> 5 to weaken its response. This was the final log of the day where I switch the PID back to 20.

https://datazap.me/u/i3igpete/back-walmart?log=0&data=1-3-4-8-11-20&zoom=236-608&mark=260

It looks like there is a consistent closure at 5000RPM across multiple pulls, so I need to dial back the duty bias at that operating point. Otherwise, I tried to keep the logged FF as a consistent ~2 points below WGDC to ensure a positive PID response without any positive/negative swings. If I understand correctly, turning bit 1 back off will allow wastegate adaption to vertically shift my entire FF curve. so that it's hopefully line-on-line with WGDC.

Boost is within 1 psi above 3500. However, is my 14psi boost target below 3000 going to throw the PID controller off?

Timing looks better than before as well. seems to consistently approach 10 degrees at 6000.

Does anyone mess with the auto shift reduction? I checked a couple logs on jb4 forum and almost everyone leaves it at default.

Yes, you are on the right track. Since you get consistent throttle cuts around 5k, consider dropping DB down a couple there and 4500 a little as well as you mentioned. You can keep PID at 20 for now.

@razorlab , agree timing still needs help. Should start climbing around 4k and peak over 10 deg redline. Lets get the boost control cleaned up more, then we can look into getting timing sorted.

I wonder if the car pulls some timing if it senses you may overshoot boost target -seems like it. These boost targets are totally fine for 93, unless its bad gas. That BM3 tuning guide says the DME only uses load targeting, no boost or timing targeting. It just uses both timing and boost to get desired load. It would make sense that too much boost would pull timing.

After the duty bias adjustments to finalize the FF curve, @i3igpete - you can try turning down FF gain to underboost on purpose. See if that cleans up timing at all. Don't mess with the duty bias any more after the adjustments at 4500 and 5000, we can just adjust FF gain from here on.

Why don't your boost targets in the log match your Map 6 settings? (shown below)

map6_15	map6_20	map6_25	map6_30	map6_35	map6_40	map6_45	map6_50	map6_55	map6_60	map6_65	map6_70
14​	14​	15.1​	19​	19​	19​	19​	19​	19​	19​	19​	19​

These won't work anyway. you need to taper boost targets on either end.

Try something like this:

map6_15	map6_20	map6_25	map6_30	map6_35	map6_40	map6_45	map6_50	map6_55	map6_60	map6_65	map6_70
13​	14​	15​	16​	17​	18​	19​	20​	20​	19​	18​	17​

Auto boost redux only needs to be changed if your gear changes aren't smooth and quick. higher number there = more boost dumped out the wastegate during upshift.

 

[i3igpete]

Those timing curves still look so bad. They should be a smooth stair step increasing with revs. The fact that they are going up down like a saw means the car is unhappy.

yep, totally agree. Want to see something wacky? On the old mitsubishi 6g72, the factory td04-9b turbos were horribly undersized and ran out of boost at 5000rpm, so the solution was to throw a TON of timing at the engine to make up the difference. People were fighting the stock timing map with piggybacks all the time.

 

[razorlab]

yep, totally agree. Want to see something wacky? On the old mitsubishi 6g72, the factory td04-9b turbos were horribly undersized and ran out of boost at 5000rpm, so the solution was to throw a TON of timing at the engine to make up the difference. People were fighting the stock timing map with piggybacks all the time.

Hahha, my roots are Apexi S-AFC's on DSMs, I know this all too well.

Sponsored