SD - first test success
#301
Evolved Member
Thread Starter
Thinking about the EGR more, if the mixing of the exhaust gases with the inlet gases is good, then the composite temperature recorded by the MAT should be accurate and along with the pressure allow accurate calculation of the air mass entering the engine. The only issue is if this EGR gas is oxygen depleted in which case there might be a richening effect? The MAF needs compensation for EGR because it cannot see the recirculation. I think though that this is likely to be a minor effect, let's not talk ourselves into addressing problems before we know whether they are present? If they are do we have an OBD pass EGR removal?
#303
EvoM Guru
iTrader: (50)
mrfred, it has SAS but I have it disabled. I thought most of the high power guys (who would want SD) have EGR disabled? It maybe that you do need some accel enrichment, especially with E85, but the golden rule is usually to get the base map right before considering enrichment.
I don't know the Evo's EGR system, but if you had to leave it enabled, someone would have to test the effects, it works during closed loop doesn't it?
I still believe the delta doesn't matter (and not having it WOULD matter) because we're accounting for it in our calculations.
I don't know the Evo's EGR system, but if you had to leave it enabled, someone would have to test the effects, it works during closed loop doesn't it?
I still believe the delta doesn't matter (and not having it WOULD matter) because we're accounting for it in our calculations.
The range of RPM and load where EGR operates on the Evo is pretty narrow. It would have a richening effect, but only occur rpms and load where almost no one drives. The bigger issue is that the EGR system heats up the IM pretty good, even when its not operating.
#304
Evolved Member
Thread Starter
It can be a bit soggy if too rich. Can you do a comparison between the MAP and MAF from a big log (helps because you get averages) like Eric showed for Tephra's log?
The IM heating effect should be compensated for as long as the MAT reading reflects the air temperature fairly well rather than the metal temperature of the inlet manifold. Depends on how well the sensor is in the air stream, what its thermal mass is etc. I know that I see a quick response on the stock sensor just from the cooler air hitting the MAT a second or two after opening the throttle.
The IM heating effect should be compensated for as long as the MAT reading reflects the air temperature fairly well rather than the metal temperature of the inlet manifold. Depends on how well the sensor is in the air stream, what its thermal mass is etc. I know that I see a quick response on the stock sensor just from the cooler air hitting the MAT a second or two after opening the throttle.
#305
EvoM Guru
iTrader: (50)
I'll see if I can get some more logs today.
The business part of the GM IAT sensor sits pretty far in the IM. Should be plenty far to be an airtemp only reading. From what I have read, the GM sensor is supposed to be very fast.
You can see what the GM IAT is doing in tephra's log that he posted. I know from my own setup with the GM IAT in the UICP that the difference between air temp at the MAF and at the UICP is minimal (at most 2C) during cruise, even when the outside air temp is 30C. There must be a pretty substantial heating effect in the IM when the airflow is pretty low, probably from both EGR and throttle body heating.
The business part of the GM IAT sensor sits pretty far in the IM. Should be plenty far to be an airtemp only reading. From what I have read, the GM sensor is supposed to be very fast.
You can see what the GM IAT is doing in tephra's log that he posted. I know from my own setup with the GM IAT in the UICP that the difference between air temp at the MAF and at the UICP is minimal (at most 2C) during cruise, even when the outside air temp is 30C. There must be a pretty substantial heating effect in the IM when the airflow is pretty low, probably from both EGR and throttle body heating.
Last edited by mrfred; Feb 13, 2009 at 10:25 AM.
#306
Evolved Member
Thread Starter
Looks fast. I have the TB heating disconnected, did it before I started logging MAT. Not had trouble down to -5C, that is mild compared to USA temperatures though.
#307
EvoM Guru
iTrader: (50)
I drove the car around for about an hour this morning with the SD patch. Except for it being a little soft, it drove great for everything I tried. Perfectly well mannered, and I'm very picky. The softness is more than likely from the car running pig rich. Even in closed loop, the AFRs were oscillating between 12.5-15:1. For some reason the LTFTs are not building up as quickly as I would expect to correct the rich conditions. Another thing I noticed is that when I came to a stop a few times, the AFRs when super rich for about 10 sec. The STFT was pegged at -25% during that time. I got a huge log that I may post later. I may try tuning the VE tables this weekend, but lots going on.
#308
Evolved Member
Thread Starter
My LTFT were quite slow to adjust too (and tephra also noticed this), until I got the STFT really tight in all situations, then they would adjust as you were looking at them, but only +- 1 % or so, as that is all it needs now. Kind of opposite of what you'd expect - you need the trims to move when it is all wrong - or perhaps the code is conservative so that it doesn't adjust if it is too chaotic so you don't compensate too aggressively for a failing O2 sensor? Now all the VE is dialled in it behaves like stock though - to drive and in the logs. My 2500 RPM 60C coolant 70 load knock has gone with 2 degrees taken out of the four cells and a change to the coolant based ignition retard. I never felt it, just saw it on logs, and it did it on the MAF sensor too.
cossie1 has just applied the patch to 90550001 (as I posted - so the code works - although he hasn't unplugged the MAF yet) and reports similar to you - a bit rich and fluffy in cruise - I'll post up his Evoscan maptrace when I find the best MAF based log to do it with. On his MAF sensor his load is about 15 lower than his MAP(kPa) - my stock values are based around them being equal, so he saw a little richening in the few full throttle bits he did.
cossie1 has just applied the patch to 90550001 (as I posted - so the code works - although he hasn't unplugged the MAF yet) and reports similar to you - a bit rich and fluffy in cruise - I'll post up his Evoscan maptrace when I find the best MAF based log to do it with. On his MAF sensor his load is about 15 lower than his MAP(kPa) - my stock values are based around them being equal, so he saw a little richening in the few full throttle bits he did.
Last edited by jcsbanks; Feb 13, 2009 at 10:39 AM.
#310
Evolved Member
iTrader: (8)
I don't know if this would work or not but when I'm setting up a map on MAP based system, I do my tuning at a constant RPM and vary load.
Drive down a road at say 3000 RPM in 5th. While logging, use the gas and the brakes to slowly increase load while keeping engine speed constant. In the log, if you find the engine is running rich (or lean) in all the load sites, you need to adjust the Engine Speed VE. If the engine is running rich (or lean) only under certain loads, then adjust the MAP VE.
JCSBanks, while your method makes perfect sense and seems to be effective, I'm wondering how difficult it would be to put in a seperate MAP calibration table and MAP VE table? Seperating out MAP calibration from engine VE might make it easier to implement different map sensors and also make it easier to tune for the VE of the engine.
Drive down a road at say 3000 RPM in 5th. While logging, use the gas and the brakes to slowly increase load while keeping engine speed constant. In the log, if you find the engine is running rich (or lean) in all the load sites, you need to adjust the Engine Speed VE. If the engine is running rich (or lean) only under certain loads, then adjust the MAP VE.
JCSBanks, while your method makes perfect sense and seems to be effective, I'm wondering how difficult it would be to put in a seperate MAP calibration table and MAP VE table? Seperating out MAP calibration from engine VE might make it easier to implement different map sensors and also make it easier to tune for the VE of the engine.
#311
Evolved Member
Thread Starter
mrfred, no it is the same as SD3.
03whitegsr, sounds a good method. Another table would add more complexity to the programming. To change map sensor: copy the axis of the table to the clipboard, edit the axis scaling to suit your new sensor, paste the old axis values again. The map calibration can then be forgotten forever. If you have multiple MAP sensors for the same ROM then you're a tuner and you're cute enough to cope?
03whitegsr, sounds a good method. Another table would add more complexity to the programming. To change map sensor: copy the axis of the table to the clipboard, edit the axis scaling to suit your new sensor, paste the old axis values again. The map calibration can then be forgotten forever. If you have multiple MAP sensors for the same ROM then you're a tuner and you're cute enough to cope?
![Smilie](https://www.evolutionm.net/forums/images/smilies/smile.gif)
#312
Evolved Member
Thread Starter
Count shows the number of times cossie1 hit each zone in this log. Average shows Load/MAP as a percentage. The load scaling is actually MAP(kPa) as this is our input to our VE tables, and load is the output.
The present Evoscan map tracer load and RPM zones are not centred so 30 load and 2500 RPM is really 30-39.9 load and 2500-2999 RPM, whereas in the ECU our tables are centred.
So from this log we can see that we need a bigger log with more VE zones hit, but we can still try to sculpture our VE tables from what we can see.
We can for any cell with big counts choose to display average RPM and load, and from that take a note of the % VE we need. Be careful if you look at a 10 load and 500 RPM, all the info may be from 19 load and 975 RPM.
A few from the 2000 RPM area which is commonly used and important to get right... in this log we have good counts (a good number of samples) from 20 to 80 load (I have rounded a bit):
25 kPa (displayed as 20) 93% of 25 is 23
35 kPa (displayed as 30) 87% of 35 is 30
45 kPa (displayed as 40) 87% of 45 is 39
55 kPa (displayed as 50) 85% of 55 is 47
70 kPa (displayed as 60) 85% of 70 is 60
90 kPa (displayed as 80) 85% of 90 is 77
So if this log is representative (I would do another with more samples), I would make MAP bins at 15,25,35,45,55,90,120,340 kPa and put the load in the second for row for each of these at 13,23,30,39,47,77, and I'd work out the 120 and 340 kPa sections from suitable logs as below. I'd set the RPM VE at 2000 to 100%.
Now onto idle, the important cell is 40kPa and 500 RPM. Average RPM in this cell for this log is actually 909 RPM, average MAP is 47kPa, bit higher than mine idles at, but they vary. So really we want to adjust our 45kPa and 1000 RPM cell to get the idle right - to give 73% overall. For 2000 RPM we set 45kPa to 87%, so we need RPM VE at 1000 RPM (and probably 500 RPM) to be 100/87*73=84%.
I only have a snippet of full throttle from cossie1, but it is displayed as WOT. From this I would say the 120 and 340 bins should be set to 91% - ie 109 and 309 giving a final table for cossie1 (although it could be better with bigger logs):
15,25,35,45,55,90,120,340 kPa
13,23,30,39,47,77,109,309 load
You need the pressures at the extremes of what the sensor is ever likely to see on your engine, otherwise it wll go flat beyond the ends of your table!
Sounds complex, but I did this once and the car drove very well, I revised the idle based on actual STFT and have left the VE table at that since with no issues.
If you've been spoilt by MAF sensors all this may seem complex, but on an SD ECU you either have to complete a good VE chart for every engine, or some VE tables, or just go through every cell and set an IPW. It can be laborious. I found setting up the VE on mine easier than mapping every cell on a MoTeC, and the result is just as good, with much better starting and idle control than the MoTeC.
The present Evoscan map tracer load and RPM zones are not centred so 30 load and 2500 RPM is really 30-39.9 load and 2500-2999 RPM, whereas in the ECU our tables are centred.
So from this log we can see that we need a bigger log with more VE zones hit, but we can still try to sculpture our VE tables from what we can see.
We can for any cell with big counts choose to display average RPM and load, and from that take a note of the % VE we need. Be careful if you look at a 10 load and 500 RPM, all the info may be from 19 load and 975 RPM.
A few from the 2000 RPM area which is commonly used and important to get right... in this log we have good counts (a good number of samples) from 20 to 80 load (I have rounded a bit):
25 kPa (displayed as 20) 93% of 25 is 23
35 kPa (displayed as 30) 87% of 35 is 30
45 kPa (displayed as 40) 87% of 45 is 39
55 kPa (displayed as 50) 85% of 55 is 47
70 kPa (displayed as 60) 85% of 70 is 60
90 kPa (displayed as 80) 85% of 90 is 77
So if this log is representative (I would do another with more samples), I would make MAP bins at 15,25,35,45,55,90,120,340 kPa and put the load in the second for row for each of these at 13,23,30,39,47,77, and I'd work out the 120 and 340 kPa sections from suitable logs as below. I'd set the RPM VE at 2000 to 100%.
Now onto idle, the important cell is 40kPa and 500 RPM. Average RPM in this cell for this log is actually 909 RPM, average MAP is 47kPa, bit higher than mine idles at, but they vary. So really we want to adjust our 45kPa and 1000 RPM cell to get the idle right - to give 73% overall. For 2000 RPM we set 45kPa to 87%, so we need RPM VE at 1000 RPM (and probably 500 RPM) to be 100/87*73=84%.
I only have a snippet of full throttle from cossie1, but it is displayed as WOT. From this I would say the 120 and 340 bins should be set to 91% - ie 109 and 309 giving a final table for cossie1 (although it could be better with bigger logs):
15,25,35,45,55,90,120,340 kPa
13,23,30,39,47,77,109,309 load
You need the pressures at the extremes of what the sensor is ever likely to see on your engine, otherwise it wll go flat beyond the ends of your table!
Sounds complex, but I did this once and the car drove very well, I revised the idle based on actual STFT and have left the VE table at that since with no issues.
If you've been spoilt by MAF sensors all this may seem complex, but on an SD ECU you either have to complete a good VE chart for every engine, or some VE tables, or just go through every cell and set an IPW. It can be laborious. I found setting up the VE on mine easier than mapping every cell on a MoTeC, and the result is just as good, with much better starting and idle control than the MoTeC.
#314
Evolved Member
Thread Starter
Suggestion for cossie1 attached.
It will be up to 9% leaner than before which may be too much as I only based it on the single decent full throttle run I had from your logs. It now may well go too lean at the top with baro still plugged in.
This thread and the last one are not a substitute for empirical testing and tuning, it just gives a starting point
It will be up to 9% leaner than before which may be too much as I only based it on the single decent full throttle run I had from your logs. It now may well go too lean at the top with baro still plugged in.
This thread and the last one are not a substitute for empirical testing and tuning, it just gives a starting point
![Wink](https://www.evolutionm.net/forums/images/smilies/wink.gif)