Tim D

Could you include all knock data and FCut and ICut in the log please? It looks like the knock threshold is being exceeded in places so I expect that knock control is pulling timing?

Thanks, that's great, I will give it a go...

Hi all, I set up Launch Control on my Subaru Impreza STI and configured it to disable at 25 mph. This works perfectly. However, my Vehicle Speed Sensor appears to have an intermittent fault, meaning that the reported speed sometimes drops to 0 mph. If this happens during a launch, the car gets stuck in launch mode. Is there an alternative way to disable launch mode (as a quick fix until I can fix the speed sensor)?

Excellent, thanks for the speedy reply Adam, much appreciated.

Hi all, I've been trying to fine tune a weird problem with idle control on my Subaru Impreza STI (EJ257) and noticed that the vehicle speed (LF Wheel Speed) as seen by the ECU doesn't return to 0 mph very nicely. It sort of hangs around 4-6 mph, then drops to 0 mph at the last minute, usually after coming to a standstill. I think this is detrimental to the transition to idle control. I looked at datalogs from 4+ years ago and these also show similar effects when coming to a standstill, so I am wondering how precise the indicated speed should be? I've attached a screenshot of a datalog showing this behaviour (I came to a standstill very smoothly, not as depicted in log): Happy to post logs etc, but wanted to know how good it should be, I'm assuming what I am seeing is sub standard? Any info/tips would be greatly appreciated.

Out of interest, did your new cable have a ferrite bead (the bulge in the cable!) and your old cable not have this? I'm interested in how significant this is?

It looks like your are reaching the MAP Limit which is causing Fuel Cut to kick in (to protect the engine). IMO, you should either reduce the boost target so that you don't reach the Map Limit or If the engine can handle more boost, increase the MAP Limit.

Thanks for your help, that worked perfectly.

Thanks Wanderer, thanks Adam, The engine has recently been rebuilt, so I guess this is pointing towards Adam's response. I had a quick look in PCLink, should I use Cam Angle Test to measure and fix this?

Hi all, Please could someone have a look at this datalog of my 2007 Subaru Impreza (EJ257), it shows the LH inlet cam following the target position nicely, but the RH inlet cam seems to stick when returning to low RPM. Does this look like a mechanical problem? Here's my tune and log files https://drive.google.com/file/d/1_2xjmL4tqKDDjRRaazbYCyisHcrCuHOQ/view?usp=sharing https://drive.google.com/file/d/18IJacWlOX1kb1Xx30rFBh6uxh54VSj6q/view?usp=sharing Thanks in advance

I had a couple of ideas on this, should be pretty quick to prove/disprove: You mentioned "logging a lot of parameters at high frequency", could you try removing most of the parameters just to see if it has an effect? My logic being, if there is a lot of interference, USB might cope better with less data being transmitted, hence pointing towards comms probs? Is your laptop graphics driver up to date, or has it recently updated itself and maybe introduced a problem? I know that graphics card/drivers have the potential to crash a laptop...

I noticed that AP is around 0.8% at idle and AP/TP lockout is 0.5%, meaning that ignition idle control is never activating, you could try setting this to 1%. Also, the Idle Ignition Table should have a value of 25 degrees at 0 error to match the ignition timing at idle. It currently has 10 degrees. You could try 0 degrees at -500 rpm error, 25 at 0 rpm error and 30 at +500 rpm error. I think this will help...

Thanks Adam, that has completely fixed it!

Hi techies... Something that has been posted by many people in the past... My laptop (Asus GL552V) takes a long time (>1 minute) to connect to the ECU. It was ok some time back, not sure what has changed! I've tried: Updating display drivers Display scaling set to 100% Updating all other drivers Power plan set to High Performance Run CCleaner Pro Defrag Run PCLink.exe as administrator Always have laptop plugged into charger See YouTube clip showing 1 minute from ignition on/start before showing as Online. I usually have to wait for the connection speed to ramp up too. Please can someone recommend some ideas that I might have overlooked?

Pete, great tip, thank you. Like you, I have also tried using conditional filters for overrun fuel cut but found it much better to turn it off like you mentioned. Maybe Lambda takes a while to stabilise after the overrun fuel cut has re-enbled? The strange thing is, I've had great results from using mixture map in the past, so I think I need to do it again and scrutinise the datalog before accepting the corrections. I have seen 'flat lines' in my datalogs in the past where USB comms appear to momentarily freeze, maybe that was happening and I didn't spot it?

Thanks Adam, glad my approach wasn't flawed at least! I'll have a closer look at the datalog next time.

Hi, I successfully used Mixture Map in the past, but more recently, I used it but I set "CLL Pos Trim Limit" and "CLL Neg Trim Limit" all to zero to effectively prevent Closed Loop Lambda corrections taking place (and upsetting the tuning process). However, the results produced a pretty bad fuel table, so I reverted to setting CLL Lambda Mode to OFF. This worked perfectly. I'm puzzled why my first approach did not work if no corrections were taking place? Any clues? Thanks

Good point.... However, to elaborate on my setup, I configured the dash to flash some lights at me if the 'log memory use' is less than 5% (streamed from ECU). So, under normal conditions, the memory is 100% full and recycling, until such time that I modify what's being logged or the log gets erased. This means I get notified when the log has been erased, even if I'm not trying to download it. The dash flashes until the log reaches 5% full, giving me several minutes to observe the event. I think that all makes sense!

I wondered if an erroneous Chip Erase command could have been the cause, but the datasheet quotes 45 s typ so I guess that rules that out. Is it possible that the datalog is actually intact (when it appears erased), but it’s being reported as erased by what ever method is used to monitor the data being written to Flash? A counter for example? As I mentioned before, my dash highlights when the datalog is empty, so I know straight away when this happens, it’s always on power up. However, I guess it could be on power-down as the dash switches off and I wouldn’t know until the next power-up! It’s never happened during normal engine running though, that’s for sure.

Thanks Adam, currently, my ECU logging conditions are: IO ON = Not Used RPM Above = 2500 MAP Above = 30 TPS Above = 0 Off Delay = 25 Re "When you download logs do you "delete all" or just leave the old logs in there?" I wasn't aware you could delete old logs (apart from changing what gets logged to force it to start a new ECU log)? I'm interested in which FLASH chip is used in the G4+, if you're happy to divulge?

I've seen this many times on my G4+. I have identified that the ECU always records data properly, but on occasion, it will get erased. I believe it's when starting the engine, perhaps a dip in voltage or a voltage spike? This gives the impresson that the ECU hasn't recoded any data. To expose this, I display 'log memory use' on my custom dash. Normally, the memory is 100% full (recycling), but when it gets erased, it's 0% full. I'm not sure if the ECU needs a permanent 12 V supply to keep the datalog or whether it's stored in non-volatile memory? I expect Adam will know?

Thanks Koracing, great advice. I will certainly try your idea out. Much appreciated.

I would never have worked that out - thanks again for the speedy response Adam! I had also misinterpreted how timing is re-introduced. So for an Advance Delay of 1.0 s and an Advance Rate of 0.5 degrees/sec, 5 degrees of timing is restored after 11 s providing you stay in that cell. My datalog was from a 10 minute track session and the 9 knock events were mostly around 5500 rpm which is around peak torque, suggesting it was probably real knock? This is a great forum!!

Thank you for the speedy reply, I'm still confused though. I should have included my tune and log to help explain: https://drive.google.com/file/d/1CtbSuLON4Rc5y7WZJKokeaEF8vvMiYRf/view?usp=sharing https://drive.google.com/file/d/18GwVXLmhthmeB2o6udd0pL4r_3_TkMFi/view?usp=sharing At time 2:41:900 Knk Level Cyl 1 exceeded Knock Threshold resulting in Cyl1 Knk I-Trim removing 5 degrees of timing and Knock Count Global incrementing to 1. This is what I expected. But, at time 5:33:845 Knk Level Cyl 1 did not appear to exceeded Knock Threshold but Cyl1 Knk I-Trim removed 5 degrees of timing (indicating that Knock Threshold had been exceeded - perhaps a Knock event was too quick for the logging rate?). Knock Count Global did not increment. There are a fw occurences of this in the log file. I don't think there's any other trims that are influencing this. Why doesn't Knock Count Global increment on these other Cyl Knk I-Trim events? Please can you help?

I thought I understood what Knock Count Global meant but after reviewing my datalog, I don't understand it. In my log file, I have Knock Threshold which I believe is the same as Knock Target in the tune. The help file says: Knock Count Global - The total number of times knock level has exceeded the knock target for all cylinders. Does this mean, Knock Count Global increments whenever ALL cylinders simultaneously exceed Knock Threshold? or maybe, Knock Count Global increments whenever ANY cylinder exceeds Knock Threshold? I understand that Knock is only interpreted (and Knk I Trim activated) if system is active (i.e. not inhibited by any lockouts) Please can someone explain what events have to take place for Knock Count Global to increment? Thanks in advance.