|
nyet
|
 |
« Reply #75 on: February 01, 2017, 02:29:47 PM »
|
|
|
Your request looks good... going to be hard to tune out the rest.
Might need to adjust Q2... can you flatten out IMX a bit (remove that dip between 2500 and 3000) so we can rule it out?
You'll get some overshoot, but that will tell us whether IMX should come up as a whole and if you need more Q2
You won't get ANY i max trim corrections at this level of fine tuning.
Don't worry, you are getting close now.
|
|
|
|
« Last Edit: February 01, 2017, 02:56:26 PM by nyet »
|
Logged
|
|
|
|
|
Dave9n3
|
|
« Reply #76 on: February 01, 2017, 02:51:13 PM »
|
|
|
Cheers, that makes me feel a lot better haha seriously felt like i was going round in circles!
Literally just got back in from doing some logging, by chance I just so happened to have been messing with that exact area and have flattened it to 70, prev was 68 probably giving that dip in imx between 2500 and 3000.
I presume I'm going to need to lift the values a bit more than they are set at at the moment though, if so i'll make some adjustments and log tomorrow.
|
|
|
|
|
Logged
|
|
|
|
|
nyet
|
|
« Reply #77 on: February 01, 2017, 02:58:36 PM »
|
|
|
Looking good, dont worry about small wiggles, you're looking to cure big over/undershoots.
Flatten/simplify IMX more and it will help tuning q2. Find a nice flat setting where over/undershoot is about equal, and we can move on from there.
Also, you're maxed out up top, you need to taper LDRXN more.
|
|
|
|
|
Logged
|
|
|
|
|
nyet
|
|
« Reply #78 on: February 01, 2017, 03:05:27 PM »
|
|
|
i'd start with drawing a straight line from 2250 to 4500 from 80 to 100, and flat 100 all the way to redline from there.
|
|
|
|
|
Logged
|
|
|
|
|
KasperH
|
|
« Reply #79 on: February 01, 2017, 09:42:38 PM »
|
|
|
Im on this topic like a hawk.  I'm dealing with the same PID headache at the moment  |
|
|
|
|
Logged
|
|
|
|
|
nyet
|
|
« Reply #80 on: February 01, 2017, 10:13:33 PM »
|
|
|
Bottom line: if your IMX has more than 2 inflection points you are doing it wrong.
|
|
|
|
|
Logged
|
|
|
|
4ringpieces
Full Member
Karma: +4/-4
 Offline
Posts: 232
|
|
« Reply #81 on: February 02, 2017, 06:55:12 AM »
|
|
|
Your obsessed with tuning out overshoot.
It's designed to overshoot, turbos have inertia, watergate ports can only control so much as they transistion from closed to open.
The reason your boost is a bit wappy, it because of pid.
It's trying to make it hit request, then as soon as it hits request it goes wappy. your not letting the boost go over requested for the pid to function correctly and control it down in a smooth manor.
IMO if you let it overshoot a little bit the pid will become happier and things will smooth out to factory like control
|
|
|
|
|
Logged
|
|
|
|
|
adam-
|
|
« Reply #82 on: February 02, 2017, 07:05:51 AM »
|
|
|
Yeah, I think you're expecting it to follow perfectly and that's just not how a PID works. You want it to overshoot.
|
|
|
|
|
Logged
|
|
|
|
4ringpieces
Full Member
Karma: +4/-4
Offline
Posts: 232
|
|
« Reply #83 on: February 02, 2017, 08:00:10 AM »
|
|
|
I'd recommend you watch motec closed loop pid control YouTube video it's very long tho. But you'll then understand why a slight overboost is desirable for pid function.
|
|
|
|
|
Logged
|
|
|
|
|
Dave9n3
|
|
« Reply #84 on: February 02, 2017, 08:06:20 AM »
|
|
|
I'm more bothered about the undershoot that seems to follow the overshoot on ramp, which I can feel when doing a 3rd gear pull. I'm going to flatten IMX as Nyet suggested then hopefully some Q2 changes will improve it a little more. I'm not after perfection, just something that feels better when driving  I'll watch that once I'm home from work for sure! |
|
|
|
|
Logged
|
|
|
|
|
Dave9n3
|
|
« Reply #85 on: February 02, 2017, 12:53:38 PM »
|
|
|
Just changed the file slightly and flatted out IMX a lot more, then had it increase to 95% beyond 4500rpm. Less points of inflection vs the revision before this one. Not got equal overshoot vs undershoot just yet though.
Altered LDRXN at the top end so req is closer to actual and it looks like it's about right now.
Should I raise IMX up, but keeping the same profile/shape to try and get an equal overshoot vs undershoot?
|
|
|
|
|
Logged
|
|
|
|
|
SB_GLI
|
|
« Reply #86 on: February 02, 2017, 01:04:11 PM »
|
|
|
Should I raise IMX up, but keeping the same profile/shape to try and get an equal overshoot vs undershoot?
Yes, you need slightly more IMX @3,500, and maybe a hair more Q2. Very close. |
|
|
|
« Last Edit: February 02, 2017, 01:07:42 PM by SB_GLI »
|
Logged
|
|
|
|
|
Dave9n3
|
|
« Reply #87 on: February 02, 2017, 01:10:30 PM »
|
|
|
Potential stupid question here but I set IMX flat at about 80%, this seems to then translate to an actual imax of say 85 on the logs. Is that down to some sort of correction made by the ecu? Thanks for the reply, it means I've got time to lift it, flash the file and take a few more logs before it gets too late I've not touched q2 just yet since I'm not sure in what increments I should be increasing it by so felt best to leave it until I knew, and or everyone was suggesting to start tweaking it haha |
|
|
|
|
Logged
|
|
|
|
|
nyet
|
|
« Reply #88 on: February 02, 2017, 02:42:37 PM »
|
|
|
LDIATA
|
|
|
|
|
Logged
|
|
|
|
|
|
|
