Currently running a slightly tuned nistune (type 2 board on R32 ECU) with load scale adjusted and some tweaks to fuel and timing using standard injectors and MAFs. Only mods being exhaust and boost controller.
I've got some purple top HKS 600cc injectors and a pair of GTST MAFs.
Can I check what the correct process should be here?
1. In current condition do a few road pulls and log MAF voltages
2. Turn off o2 feedback and check and note down AFR measured with wideband at idle (with engine hot ie not in cold start)
3. Fit injectors and use the nistune injector resize from 440cc to 600cc to automatically adjust K and get it running correctly
4. Set lag time to new value and tweak to get the same AFR at idle.
5. Do a few pulls on the road while logging and check AFRs are sensible.
6. Fit new MAFs and use nistune to change the VQ tables, tick "halve values" and then manually double K value.
7. Do a few pulls on the road while logging and check MAF voltages are sensible and the cells that are now being read have similar fuel and timing settings as before even though they will be in different places on the map.
Does that sequence sound correct and are steps 6 and 7 right? I've seen some conflicting stuff about whether to hcange injectors or MAFs first, and some conflicting stuff about halve values and doubling K.
Advice on MAF change and Injector resize for RB26DETT
Moderator: Matt
Re: Advice on MAF change and Injector resize for RB26DETT
Good idea to get a log of your AFRS in addition to RPM/TP (load) so you can see this in the AFR tracer on playback to see what your current AFRs are looking like
Clear out your O2 feedback trims (Operations > Active test > Clear self lean) as those will affect your adjustments. Prefer not to disconnect the O2 sensor, but instead watch the O2 alpha feedback trim gauges to see if it is trimming before/after doing the resize operation.
AF alpha is short term trim and is reset to 100% (no trim) each startup. AF alpha lern is long term trim and adjustments are stored in the ECU and used on each power on. Disabling O2 sensor by removing it does not stop the ECU using this previously saved trim value. These trims are multiplied against the K and fuel map calculations for final injection amount
Do the injector resize firstly. 440CC > 600CC and then fine tune injection multiplier until your off idle mixutres are back to normal and then fix your latency
After this fix the new MAFs and halve the values. Latest Nistune version also automatically doubles K value. Check your AFRs again and then fine tune. Also monitor your O2 alpha/alpha learn trims as these should be around 100% if they are lower/higher then it will affect your K
Clear out your O2 feedback trims (Operations > Active test > Clear self lean) as those will affect your adjustments. Prefer not to disconnect the O2 sensor, but instead watch the O2 alpha feedback trim gauges to see if it is trimming before/after doing the resize operation.
AF alpha is short term trim and is reset to 100% (no trim) each startup. AF alpha lern is long term trim and adjustments are stored in the ECU and used on each power on. Disabling O2 sensor by removing it does not stop the ECU using this previously saved trim value. These trims are multiplied against the K and fuel map calculations for final injection amount
Do the injector resize firstly. 440CC > 600CC and then fine tune injection multiplier until your off idle mixutres are back to normal and then fix your latency
After this fix the new MAFs and halve the values. Latest Nistune version also automatically doubles K value. Check your AFRs again and then fine tune. Also monitor your O2 alpha/alpha learn trims as these should be around 100% if they are lower/higher then it will affect your K
Re: Advice on MAF change and Injector resize for RB26DETT
That's smashing thanks Matt.
I re-read the manual(s), I last played in depth with nistune many years ago and the manuals have all been updated since I last looked.
Logging the "pre changes" was well worth while.
The lambda feedback had been disabled by the last person to map the car, by setting the feedback temp to 121c. As it turns out one of the lambda sensors was faulty (csv logs of the trace show one the two sensors occasionally going completely out of synch, one reading 1v while the other reads 0.2v etc). So I've left it like that.
As you suggest I cleared the AFR trims (with closed loop disabled new trims aren't being recorded as you'd expect). Interestingly the "untrimmed" idle afr was 13.9, nice and stable though. 14.7 on gentle load, which I'm assuming is equivalent to gently rising revs/speed on constant throttle or holding constant rpm on a slight gradient (cross checking with the estimated arfs to make sure it was in a "0 adjustment" 14.7 region of the fuel map). And healthy afrs on a WOT sweep.
I got a bit turned around setting k and the latency, stupidly I started trying to set the idle first. The suggested K was 170 but I found 156 gave me the best 14.7 on gentle load (does that mean my injectors are more like 660cc?), and a 0.8ms lag time gave a good 14.7 idle. And sensible afrs on a WOT sweep. The "off idle" tripped me up a bit as I was getting slightly unstable and generally rich readings just revving the engine sitting static - maybe that was just being fuzzy headed about the process initially.
I did a couple of hours adjusting and logging and started to figure out which bits in the log csv file to filter and sort in excel so as to get true data for idle vs cruise vs WOT sweep; speed, rpm, tps, tp etc to find the type of run and temp, aac to throw out the warm up or aac affected data. Only then looking at the AFRs.
I also tried two approaches; with and without ticking the box to rescale the TP scale.
If you don't tick the box to rescale the TP scale you end up reading of a different part of the map for the same MAF voltage (because K is a variable in the TP calculation). This means I'm a bit leaner and more advanced than before (setting up a bigger MAF at the same time I think would counter this a bit), but luckily the current map is super safe and a few degrees retarded and half a point richer than an earlier tune which wasn't knocking.
When I did tick the box the TP scale came up too short and I was needing an extra couple of columns- I guess because the auto rescale was using the estimated 170 not 158 that I ended up with. So I stuck with not ticking the box and will manually rescale TP scales by the adjustment factor 158/233.
The reason the map is so conservative is that when the car was being mapped it was randomly and unexpectedly switching to the knock map, I suspect faulty wiring or a faulty sensor. So a few degrees were taken out of the main map and the knock and standard map were set the same. I can now see knock fault DTC code and the flag in the logs. Anyway I've got some independent knock sensing to setup and a resister box built to wire-in in place of the standard system. It's a Phormula KS4 which has a warning circuit which I thought I could wire in to the resistor box to make it go open circuit when the warning fires, which would trigger the knock fault and switch to a proper knock map - obviously a bit clumsy compared to the way the ecu tries to manage knock (I understand it tries temporarily pulling some timing before reverting to the knock map if the knock continues).
The knock counter and display in nistune were haywire by the way, so I couldn't tell if the sensor was basically doing it's job or completely poked, the board is on firmware rev14, would 15 be better?
I'm thinking of unplugging the lambda sensors and putting a bung in the front elbow and fitting a EGT probe in the rear elbow (I'll be running a wideband all the time). Am I correct that the ECU will work fine and not make any untoward adjustments with the sensors unplugged and the feedback disabled (in this case by the high feedback temp setting)?
I've updated the adr file for the new injectors, I'm not sure what impact if any that has in nistune itself but seemed to be the tidy thing to do, I have some memory of that being a requirement in the old days - maybe that was when I was using liveedit ....
I re-read the manual(s), I last played in depth with nistune many years ago and the manuals have all been updated since I last looked.
Logging the "pre changes" was well worth while.
The lambda feedback had been disabled by the last person to map the car, by setting the feedback temp to 121c. As it turns out one of the lambda sensors was faulty (csv logs of the trace show one the two sensors occasionally going completely out of synch, one reading 1v while the other reads 0.2v etc). So I've left it like that.
As you suggest I cleared the AFR trims (with closed loop disabled new trims aren't being recorded as you'd expect). Interestingly the "untrimmed" idle afr was 13.9, nice and stable though. 14.7 on gentle load, which I'm assuming is equivalent to gently rising revs/speed on constant throttle or holding constant rpm on a slight gradient (cross checking with the estimated arfs to make sure it was in a "0 adjustment" 14.7 region of the fuel map). And healthy afrs on a WOT sweep.
I got a bit turned around setting k and the latency, stupidly I started trying to set the idle first. The suggested K was 170 but I found 156 gave me the best 14.7 on gentle load (does that mean my injectors are more like 660cc?), and a 0.8ms lag time gave a good 14.7 idle. And sensible afrs on a WOT sweep. The "off idle" tripped me up a bit as I was getting slightly unstable and generally rich readings just revving the engine sitting static - maybe that was just being fuzzy headed about the process initially.
I did a couple of hours adjusting and logging and started to figure out which bits in the log csv file to filter and sort in excel so as to get true data for idle vs cruise vs WOT sweep; speed, rpm, tps, tp etc to find the type of run and temp, aac to throw out the warm up or aac affected data. Only then looking at the AFRs.
I also tried two approaches; with and without ticking the box to rescale the TP scale.
If you don't tick the box to rescale the TP scale you end up reading of a different part of the map for the same MAF voltage (because K is a variable in the TP calculation). This means I'm a bit leaner and more advanced than before (setting up a bigger MAF at the same time I think would counter this a bit), but luckily the current map is super safe and a few degrees retarded and half a point richer than an earlier tune which wasn't knocking.
When I did tick the box the TP scale came up too short and I was needing an extra couple of columns- I guess because the auto rescale was using the estimated 170 not 158 that I ended up with. So I stuck with not ticking the box and will manually rescale TP scales by the adjustment factor 158/233.
The reason the map is so conservative is that when the car was being mapped it was randomly and unexpectedly switching to the knock map, I suspect faulty wiring or a faulty sensor. So a few degrees were taken out of the main map and the knock and standard map were set the same. I can now see knock fault DTC code and the flag in the logs. Anyway I've got some independent knock sensing to setup and a resister box built to wire-in in place of the standard system. It's a Phormula KS4 which has a warning circuit which I thought I could wire in to the resistor box to make it go open circuit when the warning fires, which would trigger the knock fault and switch to a proper knock map - obviously a bit clumsy compared to the way the ecu tries to manage knock (I understand it tries temporarily pulling some timing before reverting to the knock map if the knock continues).
The knock counter and display in nistune were haywire by the way, so I couldn't tell if the sensor was basically doing it's job or completely poked, the board is on firmware rev14, would 15 be better?
I'm thinking of unplugging the lambda sensors and putting a bung in the front elbow and fitting a EGT probe in the rear elbow (I'll be running a wideband all the time). Am I correct that the ECU will work fine and not make any untoward adjustments with the sensors unplugged and the feedback disabled (in this case by the high feedback temp setting)?
I've updated the adr file for the new injectors, I'm not sure what impact if any that has in nistune itself but seemed to be the tidy thing to do, I have some memory of that being a requirement in the old days - maybe that was when I was using liveedit ....
Re: Advice on MAF change and Injector resize for RB26DETT
Firstly you need to resolve the knock sensor issue. There are two knock sensors in the R32 and if either of these has a problem it will make the knock count keep incrementing and you will have knock DTCs. Either fix the wiring /replace the sensor or in worst case stub out the knock sensor with 470k ohm resistor and use the knock monitor very closely during tuning since you will no longer have any factory ECU knock protection if the sensors are stubbed out
Bad knock sensors will pull timing in the maps and may also affect mixtures so it is important to fix this firstly before proceeding
With regards to injectors, just get the basic AFRs right with the injectors and then do the MAF swap afterwards. As TP gets changed it will access a different point on the maps so just getting it roughly right at this stage is all that is required
As you will notice, moving K constant (injection time) lower affects the resulting TP (load) reading in the maps. Once you do the MAF swap afterwards it will move K back higher again and result in a more normal TP value. After this you can work on properly tuning the vehicle
As a note: We prefer to have K constant as close to the factory settings as possible due to its affect on TP in these fuel and timing tables, as well as other recently identified tables which use TP also
Bad knock sensors will pull timing in the maps and may also affect mixtures so it is important to fix this firstly before proceeding
With regards to injectors, just get the basic AFRs right with the injectors and then do the MAF swap afterwards. As TP gets changed it will access a different point on the maps so just getting it roughly right at this stage is all that is required
As you will notice, moving K constant (injection time) lower affects the resulting TP (load) reading in the maps. Once you do the MAF swap afterwards it will move K back higher again and result in a more normal TP value. After this you can work on properly tuning the vehicle
As a note: We prefer to have K constant as close to the factory settings as possible due to its affect on TP in these fuel and timing tables, as well as other recently identified tables which use TP also
Re: Advice on MAF change and Injector resize for RB26DETT
So I;m still waiting for filters for the RB20DET MAFs so I finished the job using the standard MAFs. I ended up with pretty much the automatically scaled values for K. The latency was where I went wrong I think. Along with worrying too much about getting the idle AFR right, that seems to be fairly variable depending on conditions anyway.
I also did some diagnostics on the knock sensors; the manual say they should measure 0.3v at the ecu at idle, and I was getting 2.5v so that doesn't seem right. The resistance was 250kohm on both sensors which means they are not open circuit. Anyway I thought I plug a headphone amp into the pins on the ecu for the sensor and ground and get lots of interference and buzz on both - which I think would explain the voltage and the fault code. My guess is the shielding is gone somewhere or some other fault in the loom.
Anyway plan is to either run new shielded wire to the sensors or just stub it out and fit the single sensor for the KS4 and run with that.
I thought of using the failsafe output on the AEM gauge I have (which is a switch to ground when the AFRs fault out of your set range for a given boost level) and use that to break the knock sensor stub resistor circuit and trigger a switch to a knock map, or some other limp mode on the ecu.
Can you think of any other sensor inputs that would trigger a limp type mode if they went open circuit?
I also did some diagnostics on the knock sensors; the manual say they should measure 0.3v at the ecu at idle, and I was getting 2.5v so that doesn't seem right. The resistance was 250kohm on both sensors which means they are not open circuit. Anyway I thought I plug a headphone amp into the pins on the ecu for the sensor and ground and get lots of interference and buzz on both - which I think would explain the voltage and the fault code. My guess is the shielding is gone somewhere or some other fault in the loom.
Anyway plan is to either run new shielded wire to the sensors or just stub it out and fit the single sensor for the KS4 and run with that.
I thought of using the failsafe output on the AEM gauge I have (which is a switch to ground when the AFRs fault out of your set range for a given boost level) and use that to break the knock sensor stub resistor circuit and trigger a switch to a knock map, or some other limp mode on the ecu.
Can you think of any other sensor inputs that would trigger a limp type mode if they went open circuit?