cj

In simple terms, the FP control system is set up so that the ECU tells a separate fuel pump controller whether to run the pumps fast or slow. It does this by sending a voltage (or ground) signal down the aux2 pin when certain cirteria are met. Relevant parameters to your situation are: RPM threshold: the RPM at which point the signal is sent down this wire to the FP controller DC%: injector duty percentage - works as an OR operation with RPM. If you go over this injector duty cycle or this RPM then the control voltage is sent to the controller polarity: whether the "switch to high mode" signal is to send ground or to send "nothing" (ie the "go slower" command is to send ground). The expected state of the FP controller are: off (when the fuel relay is turned off - not relevant here) slow: some pwm'd low output mode fast full flow rate. What i've been trying to explain is that you need to identify how the voltages on that FP controller input (aux 2 output) affect what the controller does with the pump speeds. It sounds like the "dont do anything" signal from the ECU as it stands is "fast mode". This will work fine if you just set both thresholds to really high numbers (10k rpm & 99%DC), but this means you dont have a "slow mode" option for low load times. It just means your pump is noisy-er and wears quicker. If you want to reconfigure this to the point where idle-ish loads swap to low flow mode, you would first need to identify what the signal is that makes it swap to high flow mode, then configure the ECU output to provide that signal. Note that some outputs can send 12v or ground, and some can only send ground or nothing.

Most limits in the link ECU's have a "soft kick in" range. The way this works is if you set a limit value of 7000rpm, and a threshold/soft limit/starting point of 500rpm, the limiter will start to kick in at 7000-500 = 6500rpm. Depending on how much change the engine needs to start "limiting", this may mean the limit is effectively 6500rpm, or it may mean you just start losing power, and the point where it actually "limits" is somewhere between 6500 & 7000rpm. Same thing applies to MAP limits etc. If you run everything in non-advanced limit mode I think it uses something like 300rpm and 10kpa less than your defined limits as the soft starting points. if you swap to advanced limiting settings you can tune this.

sounds like progress. Next thing then is to figure out what FP control settings will make it behave properly. Try changing the polarity of aux2 to high and see if the problem get better/worse. can you hear the fuel pump at idle? Maybe have someone put their head in the boot and check? I'd be tempted to disconnect aux 2 from the ecu temporarily (cut the wire, un-pin it from a connector anywhere between ECU and FP controller), then at idle (so pump is running), connect it to ground, then floating, then 12v and see which connection makes the pumps jump up into high flow mode. From there you can work out what to configure in the ECU, or if you need to add a relay to deliver the right voltage.

unless you've got oil flow issues to the vvt solenoid (blocked screen, old oil, watery oil), then that looks like PID overshoot on the VVT control solenoid. Swap to custom PID mode and increase the D value by 1-2, or reduce I by 0.01-0.02. Try again after each change as you need to be very careful with these settings to prevent advance when you really dont want it. The ECU will warn you "here there be dragons" when you tell it to switch to custom PID mode for VVT - take it seriously.

most link's dont have the option to change it as the only support saturated (high impedance) mode. If it exists in your ecu it is under fuel > injector setup. Because you have been running with a ballast resistor in place, the ECU has always been running in high impedance mode so there is nothing you need to change. Low impedance injectors + resistor ~= high impedance injectors. It sound like you already know this, but you need to remove the ballast resistor & bridge the wires that used to go to it, otherwise you have no power to your injectors. Just change the injector dead times, flow rate etc and you should be good to go. You will likely need to re-tune some of your fuel table after this.

Worst case you need to move a coupe base idle table numbers up or down, but your closed loop idle will hide most of that anyway. The rest of your tune (spark/fuel/boost) is unaffected.

Only thing I can think of is because its a stepper motor you need to power cycle the whole car after changing settings as the whole system is based on "X steps from start position" The big assumption here is that start position is always 0. Looking at your log where you said it idled low, weirdly the idle position is 222 @ just under 900rpm with coolant at 87ish and mid 20's IAT. this is quite a ways from the ~230 steps it was showing on your previous log. I wonder if your stepper isnt fully resetting at key off. your hold power value is 0 seconds, try setting it to 3 or 4 seconds and see if things become more consistent.

Heres one I prepared earlier on the ethrottle PID tuning process

that normal warmup log shows that after closed loop idle does its thing, the resulting idle positions at the various temps were: 20 30 40 50 60 70 80 188 198 206 210 216 225 228 You should put these values into the base table and then interpolate up from there - which is pretty flat anyway so 90+ should probably all be 230-235 ish. This will mean closed loop is doing a whole lot less real-time modification to the idle table, and the open loop values that are used before CL can kick in will be much closer to the idle you want. Using these plus the IAT fuel trim changes should help. If idle is still bit unstable try turning on ignition idle control, and if it still hangs just above 1200rpm when it should be idling, try bumping up the "RPM lockout" in idle speed control menu to 500.

The highlighted cells below are the base idle position at a given coolant temp. The cell from 70-100 ish are normal operating temp. You can see in your last log that your coolant temp @ startup was 83deg C so the 80 & 90 cells would be interpolated for this. Try adding 10-20 points to these values and try again. This isnt a 1 step fix, but is a good way to check if we are on the right track - ie we should see some improvement. This is the table you will re-build/re-check in open loop mode (change in "idle speed control") if we cant fix it with small tweaks. FYI you can upload files to google drive/dropbox/onedrive and link them here if the logs are too big. The slow start when warm issue might be the IAT trim you mentioned. Try zeroing out the hot 0% tps cells highlighted below, or maybe even change the 50+ cells (just in the 0% row) to + 3-5%. Heat soak will cause the IAT to read really high which can cause insufficent fuel for warm start. This is likely not the cause of your high idle though.

Do you mean the G4 management software? If its a windows (non RT) tablet you just install it like any other app - download, run, etc. If your tablet runs android or windows RT/ARM then it cant be installed.

The really short answer is you are having overrun fuel cut kick in, which is causing the rpm to drop every couple seconds. This is happening because your idle RPM is so high that the ECU thinks that you are coasting to a stop. When it drops below ~1800rpm in your config its not allowed to cut fuel (so you dont stall rolling up to a stop sign), and it turns the injectors back on, pushing RPM back up. How to fix it... When its idling normally, does it idle this high or more like ~850rpm like it says its configured to? Can you post another log of it idling normally as I suspect it is leaning really heavily on the close loop adjustments. You need to get the idle down enough so that the ECU actually considers it "at idle" and the idle control systems kick on. idling at 1700rpm you are about 600rpm above this point and so most the corrections to hold an idle RPM of 850 arent even applied. you could try increasing the "warm" numbers in your base idle table (higher numbers = lower idle on stepper motors normally). If that doesnt help you could run through the idle setup process by swapping it back to open loop, running it through a warm up cycle while adjusting the base idle numbers at each temperatue point, then swap back to closed loop once open loop is behaving well. There are also things you can do with idle ignition to make idle more stable, but you need to get the idle within ballpark first or it will have the same problem of not kicking in at all if your RPM is this far above target.

a couple of things you should change 1) ethrottle relay in the top right needs its trigger side positive to be fed from after the fuse rather than the output of the relay itself - I suspect this is just a typo though. 2) I wouldnt run sensor grounds to your coils, its much more likely to leak electrical noise onto your signal ground wiring which causes all kinds of issues, and coils typically work fine from an engine ground. I know the AEM guide says to do this, and I can only assume its so they have a cleaner comparision between trigger signals and ground to prevent dwell/timing issues on the coils themselves, but i'd run engine ground to both coil ground wires. I havent used these coils myself so cant give real world feedback but in general this seems like a bad idea. 3) The shield ground pin should only connect to the shielding wire around the trigger wiring (assuming it is shielded wire). It should not be grounded at the engine end or to the sensor itself. The actual trigger pins should receive sensor ground (ground out).

no, if the module is configured properly, it will receive RPM signal from the ECU and therefore not start up until the engine is above 400rpm. Once it sees good RPM, it will then start heating the sensor, and 5-10 seconds later it will start showing a lambda value. If you look at the value of "Lambda 1 status" it should show something like "low rpm" or "heating" and when it switches to "ok" then you should see a lambda value logged as "lambda 1" see the guide here https://linkecu.com/documentation/canlambda.pdf

@Adamw beat me to it while I was busy drawing graphs and eating dinner, but my graphs have more lines on them so i'm going to post it anyway Neither number is probably right because of the interpolation issues. The real number at the point you mention is probably around 105F. Red lines show approx actual sensor response. Green are the values I calculated that you would have seen at the time the screenshots were taken. Yellow lines show the approx value if the sensor calibration had been more complete, vs where it crosses the white line of the calibration as entered.

You can build one using these chips for a few $ https://www.jaycar.co.nz/4024-7-stage-ripple-carry-counter-divider-cmos-ic/p/ZC4024 Depending on which output you take the signal from it will divide by 2/4/8/16/32/64/128 I've used these in the past to adapt high frequency gearbox outputs down to something that can be fed to a DI at lower frequency and they have worked fine. Only thing to check is whether you need a pullup on the input & output circuits - input you need to add a pullup resistor to 5v or 12v, output side (ECU input side) you do in the ECU config for the DI.

can you grab a trigger scope at idle and about 5-6k? Those dwell spikes are a similar symptom as what Richard Hill was talking about with trigger or timing issues.

Having re-read your original post I think i got slightly mixed up by your mention of ethrottle TB swaps & PID. There are 2x set of PID's relevant here, 1 in the Ethrottle config itself, which is what I suggested you test, and it seems to be fine. This is used to translate TP target into TP motor voltages to move the throttle plate The second completely independent PID's are for the cruise control system itself - ie these are the settings the ECU uses to determine what TP target should be in order to add/remove 5kph or whatever it needs to hit speed target. The numbers you posted look like the cruise control PID's, and so you should change those to fix constant speedup/slowdown. Technically this would be considered "overshoot" on the PID algorithm, and you can either reduce I or increase D to stop this happening. Small changes at a time are a good idea.

lambda going lean would line up with fuel delivery problems. Because you have it working with a factory ECU we can likely rule out mechanical issues - blocked air or fuel filter, faulty fuel pump, bad regulator, etc. Based on this, this only this I can see in the config that would impact fuel pressure is if the fuel pump isnt being controlled properly by the link ECU. I'm assuming all the pinouts etc are correct, which just leaves the fuel pump control being wrong - maybe its not sending the right signal to run the pump at full power, maybe the PWM signal needs to go to 12v instead of ground or vice versa. Looking at the wiring diagram, these dont run a simple on/off relay, but a separate fuel pump controller, and there is a pin that is PWM'd to tell the fuel pump controller how hard to drive the pump. My money is on the config for this pwm signal being wrong (or maybe on the wrong pin, so it stays on low power at all times) - hence my suggestion that if you dont see any improvement from ecu config changes, put ground or 12v straight to the controller so it runs full power, then you know what the voltage needs to be and can work backwards from there to make it happen.

I think its something to do with your fuel pump control causing the pump to shut off or go into low output mode. The bit where your foot is flat, but your RPM is dropping is the key point for me. Can you get your wideband running so we can see if its going rich/lean while the revs drop at full throttle? Alternately, get a fuel pressure gauge/sensor installed? Maybe try setting the fuel control cutover to something like 4k & 50% DC as a test, and if that makes no difference, try disconnecting pin aux2 from the ECU and applying straight 12v to it (which should force the pump to high throughput mode)

Can you do a log of this happening + post your map? it sounds like you are exceeding one of the parameters that needs to be met to let one of the idle systems kick in. Maybe ignition idle control never kicks in and your timing values in the main igntition table are too high for the revs to drop.

what does the wiring to those 2 solenoids look like? Is the ECU grounding an output to turn them on, and what does the wiring on the "other" side of them look like - ie is it permanent 12v, switched, via some other relay? Have you got anything else connected to the ecu that is permanently fed 12? injectors, coils etc?

A log will be helpful, but as a starting point, your ignition table being 0 or negative just off idle is going to cause weirdness Try setting these cells to say 15 in 500 & 1000 columns, and 20 in the 1500 column. This will make it idle higher and mean you need to drop the idle valve numbers or adjust your idle screw, but it will stop the savage jump in power output between idle & 2k rpm.

It doesnt do this. If you have configured spark as "direct", its always direct.

The wiring diagram i'm looking at for 2008 WRX's shows the temp gauge on the dash as canbus. http://jdmfsm.info/Auto/Japan/Subaru/Impreza/2008/Impreza 2008/ The diagram you have matches what i've got for a 2007 model. As long as you have the 2007 version of the ECU, the CAN pins are free (plug C 27 & 35) so you could de-pin them from the main connector and wire them up to the canbus pins on the ECU. I'd be hesitant to just bridge onto the wires as even though they are unused, only the link guys could tell you if they are still physically wired to something. If you have the 2006 spec ECU, one of those pins is ANVolt5 so you would lose this input by doing this. Theres a pretty good chance these use the same device ID's as the legacy's of similar years, and so if you wire up the canbus pins and transmit a user defined stream on ID 600, @ 20hz with bit 3 set to ECT +40 it should make the temp gauge work There are a other parameters transmitted via CAN to get the whole dash to work however. This guy has a good starting point but if you want everything to work you will need to do some testing & captures with your own vehicle with the factory ECU installed. https://subdiesel.wordpress.com/ecu-analysis/can-messages I suspect you wont see vehicle speed registered in the ECU out of the box either - the abs unit feeds the dash, but expects the ECM to receive this via CAN. Either set the ECU to receive this on the right ID, or jumper into the Pink/Green wire (pin31 on the dash or pin5 in the OP connector) to get a traditional VSS signal on whatever DI the ECU used to receive VSS on with the older models. Receive VSS is probably 16 bits on ID512 on bytes 3 & 2 (byte 3 is the bigger half of the number), and the value sent is 16x the actual value, so a received value of 1000 = 62.5kph. FYI the reason it goes full hot is because the temp gauge wires to the ECU (so this should read right), then ECU sends CAN to the body computer, and the body computer sends CAN (on a different bus) to the gauges. Error state for that second temp message is 0xFF (ie as hot as it can read), and because the body computer isnt getting a signal from the ECU, it's in error state. I suspect your ABS warning is because of a similar lack of canbus comms from the ECU to the ABS unit, but i'm not sure which value you will need to send to keep it happy.