marantzvieta
Full Member
Karma: +3/-1
 Offline
Posts: 74
|
 |
« Reply #45 on: November 21, 2024, 06:50:17 AM »
|
|
|
THIS is why i wanted you to post it here. 140g/s for a 180hp tune is low. A healthy 150hp must flow about 130g/s at redline, a healthy 180hp should flow 150g/s at redline. You might have a vacuum or boost leak, maybe a defective component somwhere... this is why you are having underboost issues and not reaching requested loads. My car has reached 130g/s with 150hp tune, 152g/s with 180hp tune, and now I am flowing 164g/s at 5700rpm and 169g/s at 6300rpm. Which I believe to be the absolute reliable limit to stock k03 turbo. To estimate power it is said through the internet to use maf reading/0.8 but I consider this calculation too optimistic as 130g/s would translate to 162hp. I use mafreading*1,2 which translates 130g/s to 156hp (more accurate on my perspective). Same for 180hp, do the maths. 169g/s*1,2 is 203hp, which I belive to be accurate when running against similar power/weight cars. Smoke test your car and make a thorough check. Untill you don't get those values with stock tunes, do not tune any further. Best regards. |
|
|
|
|
Logged
|
|
|
|
doktor
Full Member
Karma: +0/-0
Offline
Posts: 50
|
|
« Reply #46 on: November 21, 2024, 01:22:14 PM »
|
|
|
THIS is why i wanted you to post it here. 140g/s for a 180hp tune is low. A healthy 150hp must flow about 130g/s at redline, a healthy 180hp should flow 150g/s at redline. You might have a vacuum or boost leak, maybe a defective component somwhere... this is why you are having underboost issues and not reaching requested loads.
My car has reached 130g/s with 150hp tune, 152g/s with 180hp tune, and now I am flowing 164g/s at 5700rpm and 169g/s at 6300rpm. Which I believe to be the absolute reliable limit to stock k03 turbo.
Sorry there was some misunderstanding, I was just wondering how any real figures above 180hp would be possible with a stock k03, but then Leonhard pointed out that my compressor map was wrong. I believe the car has the K03 with 2072 compressor which looks much better than the 1870. I did one more log and it seems it works acceptably well. The slight underboost happens between 2.5 and 3.5 krpm for 3rd gear pull. For 4th gear its better, and for 5th gear it is almost perfect. This looks like a job for PID tuning, but I don't even know any other maps except ZSOLLDRN. I think I will just leave it as is... Edit: About ZSOLLDRN... It says "Boost Control Filter Time Constant for Low Pass". I know what is a time constant of a low pass filter, but low pass where? The feedback loop can contain more lowpass filters... And the values in it (below 100ms) are still much shorter than the duration of the overboost (like a few seconds). -Jan |
|
|
|
« Last Edit: November 22, 2024, 02:21:40 AM by doktor »
|
Logged
|
|
|
|
doktor
Full Member
Karma: +0/-0
Offline
Posts: 50
|
|
« Reply #47 on: November 24, 2024, 06:45:29 AM »
|
|
|
A question about overrun fuelling (I am actually trying to do the opposite to most people - maximize engine braking capability), why do I see around 0.6ms load during overrun? Even several seconds after releasing the throttle, in a wide range of engine speed. Is this some kind of a bug or does it always fuel on overrun?
Also, I noticed the idle regulation starts at quite high rpm, like 1500 (other cars 1000-1200). I only see a relevant value called NDLSO (obere Drehzahlschwelle für digitale Leerlaufstabilisierung - Upper speed threshold for digital idle stabilization) which reads 1200 and does not seem to reflect reality. What are the relevant maps for overrun fuelling and idle regulation?
Thanks
|
|
|
|
|
Logged
|
|
|
|
marantzvieta
Full Member
Karma: +3/-1
Offline
Posts: 74
|
|
« Reply #48 on: November 26, 2024, 09:29:33 AM »
|
|
|
A question about overrun fuelling (I am actually trying to do the opposite to most people - maximize engine braking capability), why do I see around 0.6ms load during overrun? Even several seconds after releasing the throttle, in a wide range of engine speed. Is this some kind of a bug or does it always fuel on overrun?
Also, I noticed the idle regulation starts at quite high rpm, like 1500 (other cars 1000-1200). I only see a relevant value called NDLSO (obere Drehzahlschwelle für digitale Leerlaufstabilisierung - Upper speed threshold for digital idle stabilization) which reads 1200 and does not seem to reflect reality. What are the relevant maps for overrun fuelling and idle regulation?
Thanks
Hi mate, it's normal to have load on overrun, in the end there is air flowing through the maf... But this does not mean it is doing something with it. If you see TVSAM maps, they are (in theory) the time it stays in idle mode before fuel injection deactivation (overrun function). But with M3.8.3 I havent had much success with the ecu obeying the values in it, nor the idle reinsert speed you mention in your previous post... With 8D0907558E (chip, M3.8.2) all this maps have worked for me as epxected and always doing everything I modified. So I am a bit doubtful here on what might be different brom M3.8.2 to M3.8.3. I was conducting some investigation with it but had to postpone it a few weeks for work stuff. I found that 018R software is more responsive to changes than 018AQ, but still not toing 100% what i ask it to do... Lets post here if we do more discoveries. Best regards |
|
|
|
|
Logged
|
|
|
|
doktor
Full Member
Karma: +0/-0
Offline
Posts: 50
|
|
« Reply #49 on: November 26, 2024, 02:57:00 PM »
|
|
|
Thx for the info... I think I understand the load during overrun (but shouldnt the throttle plate be shut tight?), but I also see an injection period of about 1.2ms (which anyway mostly seems to be about double the load), during the whole overrun. It is never zero. I guess this is just some calculated value while real is 0? This is vag-com group 2, "inj. period". However, back on topic, does anyone understand why "load" is in miliseconds? I would understand if it was related to actual injection time, like a theoretical value or something, but the actual injection period mostly seems to be about double that value. What was the idea behind this? And how to log injection time above 16.32ms?  One more thing, the ignition table goes only up to 10ms "load", but the ecu allows up to 12.7ms without scaling anything. How is ignition timing calculated beyond 10ms? Does it take the last line or extrapolate? |
|
|
|
« Last Edit: November 27, 2024, 05:38:02 AM by doktor »
|
Logged
|
|
|
|
marantzvieta
Full Member
Karma: +3/-1
Offline
Posts: 74
|
|
« Reply #50 on: November 27, 2024, 04:14:27 PM »
|
|
|
Thx for the info... I think I understand the load during overrun (but shouldnt the throttle plate be shut tight?), but I also see an injection period of about 1.2ms (which anyway mostly seems to be about double the load), during the whole overrun. It is never zero. I guess this is just some calculated value while real is 0? This is vag-com group 2, "inj. period".
Yes, besides the ecu calculates an injection period, if function Overrun is active, it will not execute it. (I imagine, as there is no funktionsrahmen for it this ecu) However, back on topic, does anyone understand why "load" is in miliseconds? I would understand if it was related to actual injection time, like a theoretical value or something, but the actual injection period mostly seems to be about double that value. What was the idea behind this? And how to log injection time above 16.32ms? Load is defined as "theoretical injection time to obtain lambda 1". Afterwards there are many other calculations wich make up the effective time and even a load threshold (TLRAN.0, .1) where it does not calculate towards lambda 1 and uses preprogramed / specified fuel injection quantity when in open loop. One more thing, the ignition table goes only up to 10ms "load", but the ecu allows up to 12.7ms without scaling anything. How is ignition timing calculated beyond 10ms? Does it take the last line or extrapolate?
It does take the last line available, yes. Happens something similar when MAF is saturated when turbo is changed and the ecu sees more rpm but not more air. It does not calculate any further thus making the car run lean. MAF saturation is a very bad thing, but you can live with 10ms as the last line in KFZW  Best regards |
|
|
|
« Last Edit: November 28, 2024, 03:08:21 AM by marantzvieta »
|
Logged
|
|
|
|
doktor
Full Member
Karma: +0/-0
Offline
Posts: 50
|
|
« Reply #51 on: December 19, 2024, 07:22:03 AM »
|
|
|
I will revive this thread with a few questions:
1. I still don't get the point of KFTLWS. As the load is calculated from MAF anyway, with the target set in KFLDS, at which point does the ECU use KFTLWS? Or is it just a minimum threshold under which a limp mode is activated?
2. Is there a way to adjust idle speed, other than through vag-com/vcds?
3. Turbo PID control. I know this is a thin ice, but which are the relevant maps? It seems the turbo takes a bit longer to spool at kickdown than the one in AUM which I can compare it to, even though it is smaller (k03 vs k03s). I think the regulation is slow, I mostly see a boost undershoot, with no hint of oscillations. What does ZSOLLDRN do? I understand what is a time constant of a low pass filter, but of which filter (where in the loop)?
Edit: I think I found the answer to #2. NLLM...
-Jan
|
|
|
|
« Last Edit: December 20, 2024, 09:59:35 AM by doktor »
|
Logged
|
|
|
|
marantzvieta
Full Member
Karma: +3/-1
Offline
Posts: 74
|
|
« Reply #52 on: December 24, 2024, 03:27:00 AM »
|
|
|
I will revive this thread with a few questions:
1. I still don't get the point of KFTLWS. As the load is calculated from MAF anyway, with the target set in KFLDS, at which point does the ECU use KFTLWS? Or is it just a minimum threshold under which a limp mode is activated?
2. Is there a way to adjust idle speed, other than through vag-com/vcds?
3. Turbo PID control. I know this is a thin ice, but which are the relevant maps? It seems the turbo takes a bit longer to spool at kickdown than the one in AUM which I can compare it to, even though it is smaller (k03 vs k03s). I think the regulation is slow, I mostly see a boost undershoot, with no hint of oscillations. What does ZSOLLDRN do? I understand what is a time constant of a low pass filter, but of which filter (where in the loop)?
Edit: I think I found the answer to #2. NLLM...
-Jan
Hi mate, 1. I would personally not bother with KFTLWS, i have not touched it for any tune I've made and results have always been perfect. I belive it may use it for limp mode home and maybe other calculations, but it just not influence normal operation (or haven't felt it does). 2. No idea on my side but sould not need to touch it. If you are having idle issues maybe your throttle body is bad? 3. PID on these is a mystery for me too. I did overcome the second it takes to spool by making a modification to my tune in summer, as the hotter air makes the turbo have to work more, and with the run in summer logging channel 027 (N75 DC) and adapting KFLTV to it, I haven't had spool delay nor in summer or winter. Besides this, I have a hint on how to overcome this, as I got my hands on a tune with several maps related to air temps and turbo actuation modified and I need the time to have a thorough look at them (includes PID maps modifications. Will let you know what I find. Wish you have a merry christmas. Best regards, |
|
|
|
|
Logged
|
|
|
|
doktor
Full Member
Karma: +0/-0
Offline
Posts: 50
|
|
« Reply #53 on: December 24, 2024, 05:08:57 AM »
|
|
|
Hi marantz, I did overcome the second it takes to spool by making a modification to my tune in summer, as the hotter air makes the turbo have to work more, and with the run in summer logging channel 027 (N75 DC) and adapting KFLTV to it, I haven't had spool delay nor in summer or winter.
How did you "overcome" it? Besides this, I have a hint on how to overcome this, as I got my hands on a tune with several maps related to air temps and turbo actuation modified and I need the time to have a thorough look at them (includes PID maps modifications. Will let you know what I find.
This is very cool to hear! I am not really looking for step by step instructions and I know something about control theory as we deal with it in my work (power electronics). Just would like to have a basic idea how its implemented and what are the relevant maps... Have a nice holiday Jan |
|
|
|
|
Logged
|
|
|
|
doktor
Full Member
Karma: +0/-0
Offline
Posts: 50
|
|
« Reply #54 on: March 14, 2025, 04:24:03 AM »
|
|
|
I just had this idea, KFTLWS is almost certainly used with defective or unplugged MAF, basically running open-loop (at least for high loads) with only the throttle body angle as the input. This should imply that this table takes into account the original wastegate cracking pressure and resulting boost level. Shouldn't this table be increased for turbos with higher wg cracking pressure to protect the engine with defective maf? Or is it really not important?
|
|
|
|
« Last Edit: March 14, 2025, 04:26:13 AM by doktor »
|
Logged
|
|
|
|
error1997
Newbie
Karma: +2/-0
Offline
Posts: 15
|
|
« Reply #55 on: March 27, 2025, 02:36:16 AM »
|
|
|
I just had this idea, KFTLWS is almost certainly used with defective or unplugged MAF, basically running open-loop (at least for high loads) with only the throttle body angle as the input. This should imply that this table takes into account the original wastegate cracking pressure and resulting boost level. Shouldn't this table be increased for turbos with higher wg cracking pressure to protect the engine with defective maf? Or is it really not important?
KFTLWS is used as a replacement of the load signal in the ecu, dependent on rpm and opening of the TB when there is a maf sensor dtc, whether it being too high airflow or too low. In that scenario KFTVLDRE is used for n75 feedforward control and the boost pid is turned off. To tune it I simply unplug n75 valve and log rpm, tb angle and engine load. 3. Turbo PID control. I know this is a thin ice, but which are the relevant maps? It seems the turbo takes a bit longer to spool at kickdown than the one in AUM which I can compare it to, even though it is smaller (k03 vs k03s). I think the regulation is slow, I mostly see a boost undershoot, with no hint of oscillations. What does ZSOLLDRN do? I understand what is a time constant of a low pass filter, but of which filter (where in the loop)?
The PID in this ecu is as simple as it can get. 1.The boost pid is activated when TBA is above DKLDSPN. 2. The deviation is formed by Actual load - specified load ; specified load taken from (KFLDS*KFLDSAK*KFLDSHK) and limited by LDSMXN. 3. The deviation is fed into the different PID components. 3a. P part - KFLDP 3b. I part - KFLDI 3c. D part - DLDRN and ZSOLLDRN(the way I understand it as a filter value for the D part, lower numbers -> more aggressive D) 4. The n75 DC is calculated by summing all of the PID control outputs and adding (KFLDTV*KFLDTVAK*KFLDTVHK) In some ecus the underboost deviation is disabled in others it can throw a dtc and disable boost control replacing the n75dc values with KFTVLDRE. If there is overboost condition ( actual load is higher than KFLUL) a fuel cut is introduced. From experience the original mafs could be very inaccurate (to get good fueling on stg2, stg 3 cars KFLF looked like the surface of the moon) , thus for anything above simple stg1 I wire a ME7 maf and calibrate with it. |
|
|
|
|
Logged
|
|
|
|
doktor
Full Member
Karma: +0/-0
Offline
Posts: 50
|
|
« Reply #56 on: March 29, 2025, 01:45:02 PM »
|
|
|
Thanks for the reply error1997, appreciate it.
I might have a more serious question. The ECU is able to process a "load" of up to 12.7 ms without difficult rescaling. However, some tables, especially the fuel enrichment - KFLF (and also timing KFZW) - have axes only up to 10 ms. What happens when running long term at like 11-12ms and high rpm? Does the ecu simply use the last line of KFLF or does it extrapolate? In the first case, you will most likely get a too lean condition for these operating conditions, with unwelcome consequences...
|
|
|
|
|
Logged
|
|
|
|
|
