Pages: [1] 2 3 ... 5
Author Topic: Help with boost higher than expected  (Read 42801 times)
Tshirt2k
Full Member
***

Karma: +1/-0
Offline Offline

Posts: 78


« on: April 24, 2016, 10:13:48 AM »

Ive done an engine conversion with a K04 1.8T.
Engine hardware is as found when i purchased engine apart from rebuilding with rods and exhaust valves. 3" DP > 2.5" exhaust.
Wastegate is as i received engine. It was apparently running fine with an APR map but due to not being able to code out the various part that arent needed on that map, I decided to put a BAM 018AH flash onto it and have a go at tuning it myself.

Probem im having is, the boost is way over what is required. And i seem to hit a charge limit if i turn boost up.

I have adjusted fuel to a better profile but boost is always spiking high and staying there. Wastegate duty doesnt seem to be able to control it.

Can someone help diagnose whats going on? Boost leaks checks have been done STFT's are around -0.6%. I've taken some logs with standard LDRXN profile and with N75 disconnected.
You can see the boost just increases upto around 12psi. It seems to point to a stronger wastegate spring but the turbo seems to take longer to spool up. Unsure of what it should look like with the N75 dissed. I'd expect a k04 to spool earlier. Or could the map sensor be overreading? I will have to double check boost gauge to rule this out.

I havent tested wastegate pressure as yet but if it is stronger can i be pointed to some info on how to tune boost pid to enable N75 to control boost better.

Full log is too large to post it seems.
« Last Edit: April 24, 2016, 10:33:36 AM by Tshirt2k » Logged
Dave9n3
Sr. Member
****

Karma: +27/-0
Offline Offline

Posts: 289


« Reply #1 on: May 02, 2016, 08:52:43 AM »

Hi,

I'm finding similar after adjusting LDRXN, LAMFA etc making my own little map. I think it may be because we havent adjusted the boost PID? changing this may mean we get less overshoot - i however do not know how to adjust it just yet and or what maps need adjusting. I may be totally wrong though. seems to control nicely after the initial overshoot.

EDIT: grabbed this from the s4 wiki Smiley

KFLDRQ2 - LDR PID Q2 (differential term) - adjust this to compensate for overshoot when your boost ramp meets requested. Increase it if you see a lot of overshoot. If you have undershoot, try increasing KFLDIMX first. Only decrease Q2 as a last resort.
« Last Edit: May 02, 2016, 08:59:22 AM by Dave9n3 » Logged
Tshirt2k
Full Member
***

Karma: +1/-0
Offline Offline

Posts: 78


« Reply #2 on: May 02, 2016, 10:09:16 AM »

First I'll be checking operation of the wastegate as it seems to have some boost creep when on wastegate only.
Logged
nyet
Administrator
Hero Member
*****

Karma: +608/-168
Offline Offline

Posts: 12270


WWW
« Reply #3 on: May 02, 2016, 10:56:06 AM »

DO NOT TOUCH Q2 unless you are very intimately familiar with PIDs.

KFLDIMX is what you are looking for..

Either way, do not touch this until you at least do a little research on PIDs and how they work.
Logged

ME7.1 tuning guide
ECUx Plot
ME7Sum checksum
Trim heatmap tool

Please do not ask me for tunes. I'm here to help people make their own.

Do not PM me technical questions! Please, ask all questions on the forums! Doing so will ensure the next person with the same issue gets the opportunity to learn from your ex
nyet
Administrator
Hero Member
*****

Karma: +608/-168
Offline Offline

Posts: 12270


WWW
« Reply #4 on: May 02, 2016, 10:57:03 AM »

First I'll be checking operation of the wastegate as it seems to have some boost creep when on wastegate only.

Tune out boost creep with KFLDRL

For a constant output from the PID, KFLDRL should compensate such that the actual resulting boost is constant.
Logged

ME7.1 tuning guide
ECUx Plot
ME7Sum checksum
Trim heatmap tool

Please do not ask me for tunes. I'm here to help people make their own.

Do not PM me technical questions! Please, ask all questions on the forums! Doing so will ensure the next person with the same issue gets the opportunity to learn from your ex
Dave9n3
Sr. Member
****

Karma: +27/-0
Offline Offline

Posts: 289


« Reply #5 on: May 02, 2016, 01:54:19 PM »

Sorry guys, my bad! It's something I need to look into as well, every log I have has small overshoot and I'd like to make sure I don't have any Tongue
Logged
Tshirt2k
Full Member
***

Karma: +1/-0
Offline Offline

Posts: 78


« Reply #6 on: May 03, 2016, 11:52:30 AM »

Tune out boost creep with KFLDRL

For a constant output from the PID, KFLDRL should compensate such that the actual resulting boost is constant.

Is it just a case of adjusting dc on the last line?

Seems a bit crude to do that?
Logged
prj
Hero Member
*****

Karma: +1072/-481
Offline Offline

Posts: 6038


« Reply #7 on: May 04, 2016, 12:01:17 AM »

Tune out boost creep with KFLDRL
For a constant output from the PID, KFLDRL should compensate such that the actual resulting boost is constant.
That is not the function of KFLDRL in ME7.
The function of KFLDRL is to make sure that when the PID reacts by 10% it actually is 10% of the regulation range.
What you have suggested to do is completely wrong.

OP:
You need to linearize the boost pid, because your engine is nothing like your software anymore.
For that you need to set CWMDAPP to 16, and then control WGDC using LDRAPP.
Then make similar pulls with 10% increments in LDRAPP. 0 to as high as you can go without damaging anything (but ideally just a bit higher than the target boost you want to run).
After that the data has to be processed and KFLDIMX/KFLDRL generated from it. It is purely a mathematic process.
When that data is correct, some real world tuning will be needed at the lower RPM ranges in high gear with KFLDIMX. And only then the Q0/Q1/Q2 have to be adjusted as per PID mechanics.

It is not exactly a simple task.
« Last Edit: May 04, 2016, 12:04:49 AM by prj » Logged

PM's will not be answered, so don't even try.
Log your car properly - WinOLS database - Tools/patches
Tshirt2k
Full Member
***

Karma: +1/-0
Offline Offline

Posts: 78


« Reply #8 on: May 04, 2016, 07:09:43 AM »

Thanks for the explanation. I did wonder if DP and filters would affect this. Clearly it has.

I will look into this further. I did see a post where I said to start pid from scratch. Is that the way?
Logged
armageddon
Sr. Member
****

Karma: +20/-3
Offline Offline

Posts: 348


« Reply #9 on: May 04, 2016, 07:15:09 AM »

For that you need to set CWMDAPP to 16, and then control WGDC using LDRAPP.

should not be set to 8?
Logged
Tshirt2k
Full Member
***

Karma: +1/-0
Offline Offline

Posts: 78


« Reply #10 on: May 04, 2016, 07:19:20 AM »

should not be set to 8?

Read that here http://nefariousmotorsports.com/forum/index.php?topic=517.msg4013#msg4013
Logged
nyet
Administrator
Hero Member
*****

Karma: +608/-168
Offline Offline

Posts: 12270


WWW
« Reply #11 on: May 04, 2016, 11:19:26 AM »

The function of KFLDRL is to make sure that when the PID reacts by 10% it actually is 10% of the regulation range.

Which means that when the PID reacts by 0%, it is 0% of the regulation range.

Meaning if the output of the PID does not change, the boost should not change either.
Logged

ME7.1 tuning guide
ECUx Plot
ME7Sum checksum
Trim heatmap tool

Please do not ask me for tunes. I'm here to help people make their own.

Do not PM me technical questions! Please, ask all questions on the forums! Doing so will ensure the next person with the same issue gets the opportunity to learn from your ex
Tshirt2k
Full Member
***

Karma: +1/-0
Offline Offline

Posts: 78


« Reply #12 on: May 07, 2016, 12:04:26 AM »

In my XDF I have TLDOBAN.  I have read that this is zero 'd to help with overboost.

In my map it has 10.00 across the board. Would this help with initial overshoot?
Logged
nyet
Administrator
Hero Member
*****

Karma: +608/-168
Offline Offline

Posts: 12270


WWW
« Reply #13 on: May 07, 2016, 07:01:49 PM »

In my XDF I have TLDOBAN.  I have read that this is zero 'd to help with overboost.

In my map it has 10.00 across the board. Would this help with initial overshoot?

No.
Logged

ME7.1 tuning guide
ECUx Plot
ME7Sum checksum
Trim heatmap tool

Please do not ask me for tunes. I'm here to help people make their own.

Do not PM me technical questions! Please, ask all questions on the forums! Doing so will ensure the next person with the same issue gets the opportunity to learn from your ex
Tshirt2k
Full Member
***

Karma: +1/-0
Offline Offline

Posts: 78


« Reply #14 on: May 25, 2016, 10:42:50 PM »

That is not the function of KFLDRL in ME7.
The function of KFLDRL is to make sure that when the PID reacts by 10% it actually is 10% of the regulation range.
What you have suggested to do is completely wrong.

OP:
You need to linearize the boost pid, because your engine is nothing like your software anymore.
For that you need to set CWMDAPP to 16, and then control WGDC using LDRAPP.
Then make similar pulls with 10% increments in LDRAPP. 0 to as high as you can go without damaging anything (but ideally just a bit higher than the target boost you want to run).
After that the data has to be processed and KFLDIMX/KFLDRL generated from it. It is purely a mathematic process.
When that data is correct, some real world tuning will be needed at the lower RPM ranges in high gear with KFLDIMX. And only then the Q0/Q1/Q2 have to be adjusted as per PID mechanics.

It is not exactly a simple task.

I have actually been doing this. I had issues with this as no matter what duty I put I LDRAPP, the wastegate seemed to be similar to 0%.  Further investigation led to me realising that at 90%, wastegate duty was showing ~31% actual. The XDF I was using that I had found on here had no calculation for the map.  This was corrected and I could then see changes.

I have logs upto 70% but haven't yet been able to translate with spreadsheets I have

Although it seems from reading the 2.7T tune thread, I may only need to tweak DIMX as I have a stock wastegate and the larger DPs have an effect on the WG Duties required .  
« Last Edit: May 26, 2016, 02:42:36 AM by Tshirt2k » Logged
Pages: [1] 2 3 ... 5
  Print  
 
Jump to:  

Powered by SMF 1.1.21 | SMF © 2015, Simple Machines Page created in 0.044 seconds with 17 queries. (Pretty URLs adds 0.004s, 0q)