Oh, cool, I'll look into the bbc code thing... I have seen that.
I've tried saving with freeze set heaps of times but everytime it comes unstuck when i reopen. I'll try another version and see if it works.
The O2 readings etc.... It can happen that one cylinder can run short of fuel(blocked injector say), and make the others run rich, so the O2 readings stay more or less the same. Other times one cylinder can have a missfire so the other cylinders are compensating by going rich, which means there is a lot of fuel coming out of the faulty cylinder and the O2 will show that. It could be that one cylinder can not get fuel and the others will pour fuel in to compensate and make the O2 show a rich mixture. There are heaps of things that can happen.
We should wait and see what Mike says but I reckon the ignition timing is behaving as subaru would have it behave.
I'd log more fuel parameters myself and study them.
I'll have a look at the revscan, which I just grabbed, and see what they are measuring as best I can without an actual ecu here to look into. It might help me make more sense of the figures.
ECU learning issues
Moderators: Helpinators, Moderators
log1call: I've thought about getting the injectors serviced/checked. But it does pose a problem since I can't really easily go back to the stock components (maf, ecu, injectors), while the larger 440cc are out of the car. Hmm, wonder if there is anyone in the Seattle area who can do them in-shop the same day...
ScottyS: Yup, going to swap out the ECU tomorrow I think. Normal mode... as in no black/green connectors connected? Correct.
ScottyS: Yup, going to swap out the ECU tomorrow I think. Normal mode... as in no black/green connectors connected? Correct.
'11 WRX Limited
'94 SS | 3" TBE, 07 TMIC, TD05H-16G, Revtronix Stage 2, Walbro -- Sold
'94 TW | R.I.P.
I must have some sort of wiring/hardware problem...
I swapped out the TW ECU and installed the SS ECU + rev chip. It's been sitting out of a car for a good 2 months, looks like it finally reset itself? Turned the engine on and started to data log. Spark Learn = 75, Spark Base = 0, Spark Advance = 15, Fuel Trim (Long Term) = 0. Put the car into clear mode (black/green connectors connected), backed out of the driveway and drove down a hill. As before, a few seconds of driving, the steady CEL was flashing. Once at the bottom, disabled clear mode, fired up the engine and started logging again. Spark Learn = 0, Spark Base = 0, Spark Advance = 15, Fuel Trim (Long Term) = 0. BLAH!
On the drive up the same/short hill, I logged only RPM, Load (Stock/Stage 1), Knock Retard (Short Term), which got me a granularity of 4-5 data points a second. Highest load logged was 1.4 g/rev, yet there was 0 knock! What the heck is going on here?
I swapped out the TW ECU and installed the SS ECU + rev chip. It's been sitting out of a car for a good 2 months, looks like it finally reset itself? Turned the engine on and started to data log. Spark Learn = 75, Spark Base = 0, Spark Advance = 15, Fuel Trim (Long Term) = 0. Put the car into clear mode (black/green connectors connected), backed out of the driveway and drove down a hill. As before, a few seconds of driving, the steady CEL was flashing. Once at the bottom, disabled clear mode, fired up the engine and started logging again. Spark Learn = 0, Spark Base = 0, Spark Advance = 15, Fuel Trim (Long Term) = 0. BLAH!
On the drive up the same/short hill, I logged only RPM, Load (Stock/Stage 1), Knock Retard (Short Term), which got me a granularity of 4-5 data points a second. Highest load logged was 1.4 g/rev, yet there was 0 knock! What the heck is going on here?
'11 WRX Limited
'94 SS | 3" TBE, 07 TMIC, TD05H-16G, Revtronix Stage 2, Walbro -- Sold
'94 TW | R.I.P.
Did even more data logging after swapping the SS ECU back in. Took two ~25mi drives, which mostly consisted of the highway, plus some city street stuff.
Datalog-2009-09-09.xls todays log.
From the logs, I've noticed that the first time I accelerate after turning the engine on, the Knock Retard (Short Term) is registering 4 - 1 degrees of knock retardation for ~4 seconds. It did this both times I started up the car today after a long sit. Yet, both of the ~50mi driven today, there was ZERO knock logged.
Oh yea, before I started the second ~25mi run tonight, right after firing the engine up the rpms were bouncing low so I revved it a little bit (1.5-2K rpm). ~15 seconds after that, the rpms got so low the car died/stalled out. Remainder of the drive home was relatively uneventful, besides the BCS disabling itself again when I tried to go 1/2 throttle in 5th. Twice the Knock Retard (Short Term) logged crazy high values of 128 and 64, which I'm thinking is a logging glitch since there was gaps in other columns around that same time.
Comparing the various parameters of the ECU from just today and how it's learned...
* Spark Learn Factor went from 75 to 0%
* Spark (Base) started at 0, to 36, 31 and now 40. Good or bad?
* Fuel Trim (Long Term) went from 0 to 5.85%
* Boost DC Learned 1/2/3/4 went from 0, to 254 and back to 0 again.
1) What's the deal with the Knock Retardation only the first time I start driving?
2) With no Knock Retardation logged during ~50mi of highway/city driving, do you really think it's a wiring/sensor problem now?
3) I've got a laptop for a week or more now. What granular data do you want logged to help troubleshoot wtf is going wrong?
Datalog-2009-09-09.xls todays log.
From the logs, I've noticed that the first time I accelerate after turning the engine on, the Knock Retard (Short Term) is registering 4 - 1 degrees of knock retardation for ~4 seconds. It did this both times I started up the car today after a long sit. Yet, both of the ~50mi driven today, there was ZERO knock logged.
Oh yea, before I started the second ~25mi run tonight, right after firing the engine up the rpms were bouncing low so I revved it a little bit (1.5-2K rpm). ~15 seconds after that, the rpms got so low the car died/stalled out. Remainder of the drive home was relatively uneventful, besides the BCS disabling itself again when I tried to go 1/2 throttle in 5th. Twice the Knock Retard (Short Term) logged crazy high values of 128 and 64, which I'm thinking is a logging glitch since there was gaps in other columns around that same time.
Comparing the various parameters of the ECU from just today and how it's learned...
* Spark Learn Factor went from 75 to 0%
* Spark (Base) started at 0, to 36, 31 and now 40. Good or bad?
* Fuel Trim (Long Term) went from 0 to 5.85%
* Boost DC Learned 1/2/3/4 went from 0, to 254 and back to 0 again.
1) What's the deal with the Knock Retardation only the first time I start driving?
2) With no Knock Retardation logged during ~50mi of highway/city driving, do you really think it's a wiring/sensor problem now?
3) I've got a laptop for a week or more now. What granular data do you want logged to help troubleshoot wtf is going wrong?
'11 WRX Limited
'94 SS | 3" TBE, 07 TMIC, TD05H-16G, Revtronix Stage 2, Walbro -- Sold
'94 TW | R.I.P.
Yup well I don't know. A lot of those figures are contradicting one other and not even figures that subaru use as far as I know. All those boost Dc figures must be internal figures being used for calculating and not the ones normally read by a subaru scan tool. There are engine revs and engine revs that are different on the same line. there are O2 readings of zero. Timing "learn" figures that must be nominal internal figures used fr calculation because no engine has seventy five degrees of advance. There are all sorts of screwy figures there that must be used in internal calculations which we do not need.
I notice that you never opened the throttle up and you don't mention whether the motor actually played up. If you are only driving lightly then there should not be any knock, nor will there be any short term fuel trim or long term fuel trim if the ecu has been reset.
If you want to find a problem then you need to drive in such a way that the problem manifests itself. You also need to log data that shows the problem, and no other data which will slow down the logging, confuse the issue and make it hard to read.
I would concentrate on one aspet at a time and if it was me doing the job I would concentrate on fuel first.
I would log TPI, MAF, IPW O2, then I would drive in one gear, third if possible, I'd drive up a big hill from low revs to high revs first at half throttle, then again at fuller throttle, as much throttle as possible without getting too shorter run but hopefully enough throttle to make the motor misbehave. If the motor started to do something bad I'd prolong that condition as long as possible to see where it went.
If nothng showed up during those two runs then I'd either go harder at the hill to try and make the problem manifest or I'd log other data related to ignition and knock. The main thing though is to get data while the problem is manifesting.
It's the relationship between driving and figures and between one figures and another that tell, not the actual figures themselves.
I notice that you never opened the throttle up and you don't mention whether the motor actually played up. If you are only driving lightly then there should not be any knock, nor will there be any short term fuel trim or long term fuel trim if the ecu has been reset.
If you want to find a problem then you need to drive in such a way that the problem manifests itself. You also need to log data that shows the problem, and no other data which will slow down the logging, confuse the issue and make it hard to read.
I would concentrate on one aspet at a time and if it was me doing the job I would concentrate on fuel first.
I would log TPI, MAF, IPW O2, then I would drive in one gear, third if possible, I'd drive up a big hill from low revs to high revs first at half throttle, then again at fuller throttle, as much throttle as possible without getting too shorter run but hopefully enough throttle to make the motor misbehave. If the motor started to do something bad I'd prolong that condition as long as possible to see where it went.
If nothng showed up during those two runs then I'd either go harder at the hill to try and make the problem manifest or I'd log other data related to ignition and knock. The main thing though is to get data while the problem is manifesting.
It's the relationship between driving and figures and between one figures and another that tell, not the actual figures themselves.
More data logging today, this time with EvoScan so I could get AFR data from the LC-1. I also have the chip in normal mode with the switch (and think before it was too, except just the jumper). Several times, romping in WOT the BCS would oscillate for a bit, but there wasn't any cut or stumbling. Things are feeling good for the most part. After idling a bit, the rpms start to bounce, which I recall happened before with the TW so it's fine. Spark learn is still 0 though. I also logged the FA1 parameter (0x15A9h), which contains the KS "pinging" binary indicator, for the drive.
Short version: Things are behaving better today. I also have a new tank of gas as of Wednesday night... maybe the last tank wasn't really 92 octane?
Long version: The KS indicator NEVER went off the whole drive; normally driving or while on various levels of boost. Here is the log file from today: EvoScanDataLog_2009.09.11_13.23.39.xlsx. I'm noticing that while up to ~6 PSI, the AFR's are still staying in the 14-16 range. Isn't that too lean for that much boost? The BCS still oscillates quite a bit, but slowly settles down after several seconds. This will get better as the ECU learns?
Examples of boost from the log:
1814
1841 - 11 PSI, yet still ~15 AFR
1858 - ~11.5 PSI average, 11.2 AFR average
1895 - 2.2 PSI average, 14.8 AFR average
1907 - 3 PSI average, 15.3 AFR average
1979 - 10.8 PSI average, 12 AFR average
2175 - 3.8 PSI average, 15.4 AFR average
2312 - 7.4 PSI average, 11.9 AFR average
2545 - 3.4 PSI average, 15.3 AFR average
2598 - 2.2 PSI average, 14.7 AFR average
2870 - 6.7 PSI average, 13.1 AFR average
3090 - 3.9 PSI average, 14.5 AFR average
As things stand now, I'm going to keep the chip in normal mode for the next few days. If the Spark Learn is still at 0, then time for something more. If it DOES increase, then next week I'll change it back to power mode and see how things progress. Thoughts?
Short version: Things are behaving better today. I also have a new tank of gas as of Wednesday night... maybe the last tank wasn't really 92 octane?
Long version: The KS indicator NEVER went off the whole drive; normally driving or while on various levels of boost. Here is the log file from today: EvoScanDataLog_2009.09.11_13.23.39.xlsx. I'm noticing that while up to ~6 PSI, the AFR's are still staying in the 14-16 range. Isn't that too lean for that much boost? The BCS still oscillates quite a bit, but slowly settles down after several seconds. This will get better as the ECU learns?
Examples of boost from the log:
1814
1841 - 11 PSI, yet still ~15 AFR
1858 - ~11.5 PSI average, 11.2 AFR average
1895 - 2.2 PSI average, 14.8 AFR average
1907 - 3 PSI average, 15.3 AFR average
1979 - 10.8 PSI average, 12 AFR average
2175 - 3.8 PSI average, 15.4 AFR average
2312 - 7.4 PSI average, 11.9 AFR average
2545 - 3.4 PSI average, 15.3 AFR average
2598 - 2.2 PSI average, 14.7 AFR average
2870 - 6.7 PSI average, 13.1 AFR average
3090 - 3.9 PSI average, 14.5 AFR average
As things stand now, I'm going to keep the chip in normal mode for the next few days. If the Spark Learn is still at 0, then time for something more. If it DOES increase, then next week I'll change it back to power mode and see how things progress. Thoughts?
'11 WRX Limited
'94 SS | 3" TBE, 07 TMIC, TD05H-16G, Revtronix Stage 2, Walbro -- Sold
'94 TW | R.I.P.
*cough*ericem wrote:your in power mode eh. Ya try normal mode for a while and see how things progress.
BSOD2600 wrote:As things stand now, I'm going to keep the chip in normal mode for the next few days....
'11 WRX Limited
'94 SS | 3" TBE, 07 TMIC, TD05H-16G, Revtronix Stage 2, Walbro -- Sold
'94 TW | R.I.P.
Well I'm glad to hear it seems to be going better.
I couldn't figure out how to open that latest file.
I have been looking at all those figures from yesterday and I still don't understand where they all come from or what half of them mean. Some look like standard subaru figures, but they don't always behave as I'd expect. They are not always within acceptable ranges either, which makes me think they may be either from the wrong addresses or they are being calculated/adjusted wrongly. As far as I know, subaru doesn't differentiate between models when it comes to reading the SSM, there are different addresses for different models but the data we collect is always the same. Perhaps someone else out there will/can throw some light on them.
If your O2 sensor detects a rich mixture that keeps it struggling to keep the mixture within range, it "learns: the ecu by applying a correction factor to the fuel map. That correction is long term adjustment, all the other adjustments are short term fuel adjustments.
If the knock sensor detects knock it retards the timing but that is very temporary and any other timing adjustments are all prearranged in the timing map and we mechanics don't call them "learn", just correction, for temperature for instance.
I don't know whether you had a look at that log file I posted but if you did I think you would probably agree that the data was behaving as it should, and that it was clear and easy to see what was going on both with regards how I was driving the car and what was happening under the bonnet. Too much data confuses things and not having enough data to show how the car was being driven at any particular moment makes it hard to understand.
Anyway, it's all very interesting. I hope it stays going well for you and I'll keep an eye out in case it doesn't. There is always somehing new to learn.
I couldn't figure out how to open that latest file.
I have been looking at all those figures from yesterday and I still don't understand where they all come from or what half of them mean. Some look like standard subaru figures, but they don't always behave as I'd expect. They are not always within acceptable ranges either, which makes me think they may be either from the wrong addresses or they are being calculated/adjusted wrongly. As far as I know, subaru doesn't differentiate between models when it comes to reading the SSM, there are different addresses for different models but the data we collect is always the same. Perhaps someone else out there will/can throw some light on them.
If your O2 sensor detects a rich mixture that keeps it struggling to keep the mixture within range, it "learns: the ecu by applying a correction factor to the fuel map. That correction is long term adjustment, all the other adjustments are short term fuel adjustments.
If the knock sensor detects knock it retards the timing but that is very temporary and any other timing adjustments are all prearranged in the timing map and we mechanics don't call them "learn", just correction, for temperature for instance.
I don't know whether you had a look at that log file I posted but if you did I think you would probably agree that the data was behaving as it should, and that it was clear and easy to see what was going on both with regards how I was driving the car and what was happening under the bonnet. Too much data confuses things and not having enough data to show how the car was being driven at any particular moment makes it hard to understand.
Anyway, it's all very interesting. I hope it stays going well for you and I'll keep an eye out in case it doesn't. There is always somehing new to learn.
I've updated the data log from a few hours ago with the logging data from the drive home. This time, fewer parameters so I have 3-4 entries a second. Seems to be pretty solid now. Noticed that when I got home, Spark Learn Factor is finally at 50%!
Log entry - notes:
2819 - 13.3 PSI avg, 10.5 AFR avg (10 second 5th gear second pull)
4011 - 13.1 PSI avg, 10.6 AFR avg (7 sec 4th gear pull)
4973 - 3.6 PSI avg, 14.9 AFR avg
5090 - 10.7 PSI avg, 12.3 AFR avg (4 sec 3rd gear pull)
5258
log1call: You need the Microsoft Office Compatibility Pack to open the XLSX file.
Which parameters are out of spec? I also posted my updated Data.xml file to the EvoScan thread. I've gotten my ECU parameters/conversions from the b10 source code, VWRX Subaru Select Monitor Scan Tool ini file, and RevScan.dat binary file. Which ROMID are you using? Mine is 723622. Also, mind posting your data.xml file so I can see what conversions you've got in there?
Log entry - notes:
2819 - 13.3 PSI avg, 10.5 AFR avg (10 second 5th gear second pull)
4011 - 13.1 PSI avg, 10.6 AFR avg (7 sec 4th gear pull)
4973 - 3.6 PSI avg, 14.9 AFR avg
5090 - 10.7 PSI avg, 12.3 AFR avg (4 sec 3rd gear pull)
5258
log1call: You need the Microsoft Office Compatibility Pack to open the XLSX file.
Which parameters are out of spec? I also posted my updated Data.xml file to the EvoScan thread. I've gotten my ECU parameters/conversions from the b10 source code, VWRX Subaru Select Monitor Scan Tool ini file, and RevScan.dat binary file. Which ROMID are you using? Mine is 723622. Also, mind posting your data.xml file so I can see what conversions you've got in there?
'11 WRX Limited
'94 SS | 3" TBE, 07 TMIC, TD05H-16G, Revtronix Stage 2, Walbro -- Sold
'94 TW | R.I.P.
Hi, just downloaded that compatability pack and I will have a look at your second to last and latest logs tomorrow morning and let you know which ones I reckon are not correct as best I can, and, perhaps more usefully, which ones are trustworthy. I doubt it is the conversion factors that are wrong because I look at the relationship between figures and between figures and the motors behaviour... I can generally spot the data that is behaving correctly even if the figures are way out, especially if I apply scaling of my own.
My ecu parameter file is here... http://cid-4ca3c3459aaa7f7f.skydrive.li ... blic703315 which I made myself. and my ecu is 703315. That ecu file and those addresses are ones I made myself. I discovered them using ssm data dump. First I found the maps, then the addresses where data changed with changes in the motor, then I logged strings of live data from the address while I did things to the motor designed to cause predicable behaviour which I was looking for. It's not very scientific but it works. I actually suspect it's a lot faster and finds more usefull data that the experts way of dissasembling program... which is beyond me anyway to be truthfull.
I have been a mechanic for thirty-seven years, from before they even had computers in cars. I have seen all the developments, all the systems, pretty much all the combinations of systems as they are used in the various makes. My understanding of how they work, my theory, my practic is sound. I worked for toyota for a year and a half and when i moved there I was the boy from the country, within three months I was the highest paid toyota mechanic in new Zealand. Toyota N.Z. used to ring me and ask if they could send problem cars from all over to me to trouble shoot because they kept records of productivity and results and they knew I could diagnose and fix anything. Logic and reason are how I do it, along with sound basic principals of course. All the manufacturers use the same data, in the same ways, to achieve the same results.
There are only a very few principles involved in any particular problem, the trick is to isolate which ones they are, repeat them till you can cause the problem at will, then reason what is happening. If you can isolate the one and only thing that will cause it, all the better. By the time you have the things or thing that causes a problem you mostly have the cause and the verification it is cured once the work is done. If the cure can't be verified then there is more work to be done. That's how I achieved the 1oo% success rate and a 110% productivity rate that toyota liked so much.
Anyway, hope that doesn't sound like boasting, it isn't. You don't get to be the best by having an ego, quite the opposite. I like to help people, and I'm not talking about you here Bsod, but the watchers, I like to help them understand that motors are logical, they are very simple piles of components that operate in planned and predictable ways, and that we can understand what is happening if we take the time to learn the basic principles involved really well.
My ecu parameter file is here... http://cid-4ca3c3459aaa7f7f.skydrive.li ... blic703315 which I made myself. and my ecu is 703315. That ecu file and those addresses are ones I made myself. I discovered them using ssm data dump. First I found the maps, then the addresses where data changed with changes in the motor, then I logged strings of live data from the address while I did things to the motor designed to cause predicable behaviour which I was looking for. It's not very scientific but it works. I actually suspect it's a lot faster and finds more usefull data that the experts way of dissasembling program... which is beyond me anyway to be truthfull.
I have been a mechanic for thirty-seven years, from before they even had computers in cars. I have seen all the developments, all the systems, pretty much all the combinations of systems as they are used in the various makes. My understanding of how they work, my theory, my practic is sound. I worked for toyota for a year and a half and when i moved there I was the boy from the country, within three months I was the highest paid toyota mechanic in new Zealand. Toyota N.Z. used to ring me and ask if they could send problem cars from all over to me to trouble shoot because they kept records of productivity and results and they knew I could diagnose and fix anything. Logic and reason are how I do it, along with sound basic principals of course. All the manufacturers use the same data, in the same ways, to achieve the same results.
There are only a very few principles involved in any particular problem, the trick is to isolate which ones they are, repeat them till you can cause the problem at will, then reason what is happening. If you can isolate the one and only thing that will cause it, all the better. By the time you have the things or thing that causes a problem you mostly have the cause and the verification it is cured once the work is done. If the cure can't be verified then there is more work to be done. That's how I achieved the 1oo% success rate and a 110% productivity rate that toyota liked so much.
Anyway, hope that doesn't sound like boasting, it isn't. You don't get to be the best by having an ego, quite the opposite. I like to help people, and I'm not talking about you here Bsod, but the watchers, I like to help them understand that motors are logical, they are very simple piles of components that operate in planned and predictable ways, and that we can understand what is happening if we take the time to learn the basic principles involved really well.
Hi,
I have been looking at your evoscan file and seeing what address and conversion factors you are using in there because I don't know what addresses you used for the logs posted so far. Assuming the logs I'm looking at are using those evo addresses and conversion factors where possible...
Looking in the log from thread post dated Tue Sep 08, 2009 5:24 pm, and looking at "coast three"...
Engine speed. looks right to me.
coolant temp. what it says and looks right.
TPS. Right address but needs multiplying by .02 to get voltage I think. Note that my throttle works the reverse of this if you are examining my logs.
Load.I think you are using the right address but you should use the raw data without any modifying to get a more understandable figure.
pulse width. you are using x*0.128, try x*0.256.
Spark learn... ignore it. It's something Revtronics use but it's not very useful to mechanics.
Idle speed... Revtronics again, only useful if you are remapping their chip.
IAC base steps... probably right but not very responsive so not very useful.
EQ target. Not sure what that is... if you can tell me I will see what I can figure, otherwise ignore it.
Fueltrim short term. Not sure. It's not following O2, I's suspect it's a revtronics calculation which we could ignore.
Airflow Hitachi. I don't see that in evofile, It's figures do seem to follow revs/throttle though so it may well be relevent.
spark base, something for revtronics? That won't be the idling timing. I'd ignore that one.
Spark advance, well that should be the right address to give you actual live timing but you are meant to (x-128)/2 which would give a wrong figure I'd imagine, unless your log already has some factor applied?
fuel trim long. Revtronics map adjusting figure. Not very relevent.
TPS scaled. Revtronics? doesn't seem to track TPS to me.
boost D.C. looks like the right address and conversion factor, the behaviour is strange though.
IAC raw. revtronics, not useful and doesn't seem to be related to anything except scaled throttle. i'd ignore it.
Barometric pressure. Looks like the right address in evofile but the wrong address in these logs.
MAP. looks like the right address in evofile, figures are wrong in this log though. Well they are not Kpa anyway, not really familiar with torr.
O2. right address in evofile, wrong conversion I think. In this log the figures are not regular and don't seem to be following IPW. What address did you use in this log?
Knock retard short. Revtronics I suspect. I'd ignore it.
Maf voltage. Seems like the right address in evofile but the conversion factor looks wrong. the figures look right in this log though.
The four boosts... well I'm not sure which is which and most seem to do nothing, that one that learned4 looks like it might be right though.
IAC. evofile says you are using an address and conversion factor which should give a D.C.% of around 45% at idle. Your spreadsheet says your figure is a raw figure, the raw figure is meant to be read from different address than the D.C figure. You might have to have a hunt if you really want to know the raw data that the ecu uses to do it's calculations.
When we are looking in ECUs, there are lots of places we can find data that effects ignition timing. There is a base setting that's modified by temperature, throttle, air flow, revs and the knock sensor. None of those things though permanantly alter the base setting. We could read all those factors compensating figures and try and figure if they are all exactly as they should be or if there was something wrong with one figure, but mostly as mechanics we don't need to. If we can log the timing as it is live, check it changes with changes in all the different factors that effect it, then we really don't need to log every bit of data as long as it's responding. Most of those other factors are all ready being logged. When the motor revs up the timing should advance. If the motor gets a load on it, the timing should retard. If the motor gets too cold or too hot it should advance. If the revs get too high and the IAC can't control the revs the timing should retard. The only thing we need to log that is extra is the knock to see the timing is responding to that. A lot of those figures you were logging are internal calculations, how much to advance it if the motor gets too hot for instance. Until we are sure we need to know that figure it's best to ignore it in the interests of clarity.
The same can be said for fuel and boost. All we need to check initially is that they are all responding in the correct ways to the motor and the main factors that effect it. If we read the A/F mixture and check it responds to IPW and that they both respond to O2, throttle, air, etc, then we don't need to know the exact factor they are adjusting by. All we need to know is that they are responding. We can deduce things far quicker than we can interperate screeds of figures in spreadsheets or even graphs.
If anyone is logging data I'd recommend evoscan because it's the best on the market. I'd recommend just logging the bare essentials, time, revs, throttle, air flow, temp, IPW, MAP, O2, IAC, base timing, knock correction, boost D.C. Most other things can be deduced or it will become obvious that they are required with just those basic figures. Often only some of those figures will be needed, often only two figureds are needed to confirm a hunch or deduction. That revscan program would probably have a lot of useful information in it if you were remapping, but for diagnosis, it's too much.
I have been looking at your evoscan file and seeing what address and conversion factors you are using in there because I don't know what addresses you used for the logs posted so far. Assuming the logs I'm looking at are using those evo addresses and conversion factors where possible...
Looking in the log from thread post dated Tue Sep 08, 2009 5:24 pm, and looking at "coast three"...
Engine speed. looks right to me.
coolant temp. what it says and looks right.
TPS. Right address but needs multiplying by .02 to get voltage I think. Note that my throttle works the reverse of this if you are examining my logs.
Load.I think you are using the right address but you should use the raw data without any modifying to get a more understandable figure.
pulse width. you are using x*0.128, try x*0.256.
Spark learn... ignore it. It's something Revtronics use but it's not very useful to mechanics.
Idle speed... Revtronics again, only useful if you are remapping their chip.
IAC base steps... probably right but not very responsive so not very useful.
EQ target. Not sure what that is... if you can tell me I will see what I can figure, otherwise ignore it.
Fueltrim short term. Not sure. It's not following O2, I's suspect it's a revtronics calculation which we could ignore.
Airflow Hitachi. I don't see that in evofile, It's figures do seem to follow revs/throttle though so it may well be relevent.
spark base, something for revtronics? That won't be the idling timing. I'd ignore that one.
Spark advance, well that should be the right address to give you actual live timing but you are meant to (x-128)/2 which would give a wrong figure I'd imagine, unless your log already has some factor applied?
fuel trim long. Revtronics map adjusting figure. Not very relevent.
TPS scaled. Revtronics? doesn't seem to track TPS to me.
boost D.C. looks like the right address and conversion factor, the behaviour is strange though.
IAC raw. revtronics, not useful and doesn't seem to be related to anything except scaled throttle. i'd ignore it.
Barometric pressure. Looks like the right address in evofile but the wrong address in these logs.
MAP. looks like the right address in evofile, figures are wrong in this log though. Well they are not Kpa anyway, not really familiar with torr.
O2. right address in evofile, wrong conversion I think. In this log the figures are not regular and don't seem to be following IPW. What address did you use in this log?
Knock retard short. Revtronics I suspect. I'd ignore it.
Maf voltage. Seems like the right address in evofile but the conversion factor looks wrong. the figures look right in this log though.
The four boosts... well I'm not sure which is which and most seem to do nothing, that one that learned4 looks like it might be right though.
IAC. evofile says you are using an address and conversion factor which should give a D.C.% of around 45% at idle. Your spreadsheet says your figure is a raw figure, the raw figure is meant to be read from different address than the D.C figure. You might have to have a hunt if you really want to know the raw data that the ecu uses to do it's calculations.
When we are looking in ECUs, there are lots of places we can find data that effects ignition timing. There is a base setting that's modified by temperature, throttle, air flow, revs and the knock sensor. None of those things though permanantly alter the base setting. We could read all those factors compensating figures and try and figure if they are all exactly as they should be or if there was something wrong with one figure, but mostly as mechanics we don't need to. If we can log the timing as it is live, check it changes with changes in all the different factors that effect it, then we really don't need to log every bit of data as long as it's responding. Most of those other factors are all ready being logged. When the motor revs up the timing should advance. If the motor gets a load on it, the timing should retard. If the motor gets too cold or too hot it should advance. If the revs get too high and the IAC can't control the revs the timing should retard. The only thing we need to log that is extra is the knock to see the timing is responding to that. A lot of those figures you were logging are internal calculations, how much to advance it if the motor gets too hot for instance. Until we are sure we need to know that figure it's best to ignore it in the interests of clarity.
The same can be said for fuel and boost. All we need to check initially is that they are all responding in the correct ways to the motor and the main factors that effect it. If we read the A/F mixture and check it responds to IPW and that they both respond to O2, throttle, air, etc, then we don't need to know the exact factor they are adjusting by. All we need to know is that they are responding. We can deduce things far quicker than we can interperate screeds of figures in spreadsheets or even graphs.
If anyone is logging data I'd recommend evoscan because it's the best on the market. I'd recommend just logging the bare essentials, time, revs, throttle, air flow, temp, IPW, MAP, O2, IAC, base timing, knock correction, boost D.C. Most other things can be deduced or it will become obvious that they are required with just those basic figures. Often only some of those figures will be needed, often only two figureds are needed to confirm a hunch or deduction. That revscan program would probably have a lot of useful information in it if you were remapping, but for diagnosis, it's too much.
Spark learn factor was at 100 Saturday morning, but I take it all back that things are going better...
Saturday evening drive home, I was logging: knock retard long slow, boost DC, MAP, wideband and FA1, which meant I have a logging resolution of 3-4 entries a second. During a 5th gear, WOT pull, after 6 seconds on boost (it stabilized around 13.2 PSI), FA1 changed from 8 (binary 1000) to 10 (binary 1010) for a single entry. During that same time period, 'knock retard long slow still' remained -1 (what it always sits at). I believe FA1 of 10 means the FP (fuel pump) and KS (pinging) binary parameters are on, right? Anyways, once I got home, I turned back on all logging as I idled in the driveway. Spark Learn Factor is back to 0.
Sunday, drove over Snoqualmie pass and had the cruise control set at ~75 mph. Near the top of the pass for the last few miles, the boost was on a solid ~5-6 PSI for a long time (due to cruise control). From time to time, I could feel the car have little misses, shudders, etc. I've driven ~180mi since Saturday and the Spark Learn is back to 18.75%.
I'm starting to run out of ideas WHY this is happening, because data logging with either EvoScan or Revscan, is not really revealing a massive amount of knocking, which I've been told would cause this to happen.
Saturday evening drive home, I was logging: knock retard long slow, boost DC, MAP, wideband and FA1, which meant I have a logging resolution of 3-4 entries a second. During a 5th gear, WOT pull, after 6 seconds on boost (it stabilized around 13.2 PSI), FA1 changed from 8 (binary 1000) to 10 (binary 1010) for a single entry. During that same time period, 'knock retard long slow still' remained -1 (what it always sits at). I believe FA1 of 10 means the FP (fuel pump) and KS (pinging) binary parameters are on, right? Anyways, once I got home, I turned back on all logging as I idled in the driveway. Spark Learn Factor is back to 0.
Sunday, drove over Snoqualmie pass and had the cruise control set at ~75 mph. Near the top of the pass for the last few miles, the boost was on a solid ~5-6 PSI for a long time (due to cruise control). From time to time, I could feel the car have little misses, shudders, etc. I've driven ~180mi since Saturday and the Spark Learn is back to 18.75%.
I'm starting to run out of ideas WHY this is happening, because data logging with either EvoScan or Revscan, is not really revealing a massive amount of knocking, which I've been told would cause this to happen.
'11 WRX Limited
'94 SS | 3" TBE, 07 TMIC, TD05H-16G, Revtronix Stage 2, Walbro -- Sold
'94 TW | R.I.P.
The shuddering, etc, could be a lot of things. I know that I went through everthing I could think of on my car to get rid of stuff like that.
Assuming that everything in the block and valve train are OK......optimize the fuel system and electrical system.
I have:
-Diamond coil pack
-fixed/resoldered all suspect connector leads in the engine bay (ignitor chip, knock sensor, MAF, etc)
-got a newer alternator, the old one was worn out
-grounding harness to multiple points in the engine bay
-new Walbro fuel pump
-new fuel lines/filter
-new plug wires
-tightened plug gap
-all new gaskets for the intake manifold
-all new vacuum hoses in the engine bay
You get the idea. When I started with this project, the car had a LOT of little hesitations, and the combination of all of this has made it pretty darn smooth. There was no one thing that I can point to and say that it was the culprit.
If you really suspect the knock sensor is giving false information, then SERIOUSLY re-solder the connection on the harness side, and maybe even run a new wire (of the correct type) like the other poster talked about before. I also remember someone saying that they put a rubber gasket between the knock sensor and the block, as it was picking up piston slap or something.
Assuming that everything in the block and valve train are OK......optimize the fuel system and electrical system.
I have:
-Diamond coil pack
-fixed/resoldered all suspect connector leads in the engine bay (ignitor chip, knock sensor, MAF, etc)
-got a newer alternator, the old one was worn out
-grounding harness to multiple points in the engine bay
-new Walbro fuel pump
-new fuel lines/filter
-new plug wires
-tightened plug gap
-all new gaskets for the intake manifold
-all new vacuum hoses in the engine bay
You get the idea. When I started with this project, the car had a LOT of little hesitations, and the combination of all of this has made it pretty darn smooth. There was no one thing that I can point to and say that it was the culprit.
If you really suspect the knock sensor is giving false information, then SERIOUSLY re-solder the connection on the harness side, and maybe even run a new wire (of the correct type) like the other poster talked about before. I also remember someone saying that they put a rubber gasket between the knock sensor and the block, as it was picking up piston slap or something.
1991 SS build thread: http://bbs.legacycentral.org/viewtopic.php?t=40430
Ignore the spark learn factor. Find the live ignition timing figure which gives you a reading of around seventeen to twenty-two at idle and which goes up with revs and down with load and tops out at about fifty at revs when you ease your foot off the throttle... that will be your actual ignition timing that the motor is running on.
Find the knock correction figure that goes up to very temporarily when you floor the throttle and.or when the boost and load come on hard, it should only go to about twelve and it should cause the ignition timing figure to retard when all other factors remain the same,ie, throttle and revs dont change. That is your knock correction.
Find youe boost cut, the one that comes on and causes the MAP figure to drop.
Find an O2 reading that follows IPW and swings through it's range at idle regularly.
If the knock correction is working, and there aren't any trouble codes for knock detection, then I don't think that is your trouble. If the circuit breaks the code will set. If you have too much knock correction the code will set. If you have neither of those scenerios then it won't be the problem.
Check your IPW against O2 and A/F correction.
You could have water in your fuel, a blocked filter, faulty pump.
Find the knock correction figure that goes up to very temporarily when you floor the throttle and.or when the boost and load come on hard, it should only go to about twelve and it should cause the ignition timing figure to retard when all other factors remain the same,ie, throttle and revs dont change. That is your knock correction.
Find youe boost cut, the one that comes on and causes the MAP figure to drop.
Find an O2 reading that follows IPW and swings through it's range at idle regularly.
If the knock correction is working, and there aren't any trouble codes for knock detection, then I don't think that is your trouble. If the circuit breaks the code will set. If you have too much knock correction the code will set. If you have neither of those scenerios then it won't be the problem.
Check your IPW against O2 and A/F correction.
You could have water in your fuel, a blocked filter, faulty pump.
Mike@Rev states that the spark learn factor is an indicator of what the ECU is doing with the whole learning process. If that's not at 100%, then spark is retarded a lot along with reverting back to fuel maps.
Well on my coil pack hunch, went out and measured it. Both the FSM and Haynes manual both have the same specs for things, so either they're both wrong and reversed... or mine is majorly bad. lol. anyways, installing a new coil pack and fuel filter this afternoon...
Primary side:
Between 1 and 2: 13.7K ohms (should be 0.6-0.7 ohm)
Between 3 and 3: 13.2K ohms " "
Secondary side:
Between terminal No 1 and No 2: 0.5 ohm (should be 10.4-12.5K ohm)
Between terminal No 2 and No 3: 0.7 ohm. " "
Primary terminal to case: must be greater than 10M ohm, since the digital ohm meter wouldn't read anything.
I'll do more data logging with those parameters to in the upcoming day(s).
I've got some of the things which ScottyS has listed off:
-coil pack (installing new one this afternoon)
-walbro fuel pump
-new gaskets for rocker panel covers
-new vacuum lines + zip ties
-new alternator (well as of ~30K miles ago)
Well on my coil pack hunch, went out and measured it. Both the FSM and Haynes manual both have the same specs for things, so either they're both wrong and reversed... or mine is majorly bad. lol. anyways, installing a new coil pack and fuel filter this afternoon...
Primary side:
Between 1 and 2: 13.7K ohms (should be 0.6-0.7 ohm)
Between 3 and 3: 13.2K ohms " "
Secondary side:
Between terminal No 1 and No 2: 0.5 ohm (should be 10.4-12.5K ohm)
Between terminal No 2 and No 3: 0.7 ohm. " "
Primary terminal to case: must be greater than 10M ohm, since the digital ohm meter wouldn't read anything.
I'll do more data logging with those parameters to in the upcoming day(s).
I've got some of the things which ScottyS has listed off:
-coil pack (installing new one this afternoon)
-walbro fuel pump
-new gaskets for rocker panel covers
-new vacuum lines + zip ties
-new alternator (well as of ~30K miles ago)
'11 WRX Limited
'94 SS | 3" TBE, 07 TMIC, TD05H-16G, Revtronix Stage 2, Walbro -- Sold
'94 TW | R.I.P.
bah, disregard the coil pack stuff. The FSM has the primary/secondary numbers reversed
'11 WRX Limited
'94 SS | 3" TBE, 07 TMIC, TD05H-16G, Revtronix Stage 2, Walbro -- Sold
'94 TW | R.I.P.
Crikey, that's a worry. I just checked another diagnostic manual and they also had the ohms readings back to front.
That ignition learn then isn't a very accurate name for what ever it is. If it's also effecting the fuel maps then it's some internal calculation that's being used to decide when to revert to safe mode by the sound of it.
I'd just find the actual live ignition timing figure and use that. That's the only thing that the motor cares about. It doesn't care whether there is a debate going on in the background about what orders it should get, it just follows the orders it does receive..
That ignition learn then isn't a very accurate name for what ever it is. If it's also effecting the fuel maps then it's some internal calculation that's being used to decide when to revert to safe mode by the sound of it.
I'd just find the actual live ignition timing figure and use that. That's the only thing that the motor cares about. It doesn't care whether there is a debate going on in the background about what orders it should get, it just follows the orders it does receive..
This problem still hasn't been resolved! Ended up having a leak in the silicone TMIC tubing for a few weeks. Got a new one bought / installed 200+ miles ago.
On the suggestion from Mike @ Rev, I threw in bottle of octane booster this last time I filled up and did more data logging. Before the octane booster, I was getting knock retardation only in the ~2.1 g/rev engine load range. Now after the booster, it's logging knock retardation in the ~1.5 g/rev range. Not good and now progress is going backwards.
I printed out the knock sensor diagnostics and went over everything this afternoon.
1. I'm seeing ~4.87V at the knock sensor connector in the engine bay (B38).
2. knock sensor has 561K Ω
3a. B127 No 5 -> B38 No 1 has continuity. As I hold it, the auto-adjusting ohm meter drops from 1 to 0 Ω.
3b. B38 -> engine or body. No continuity or ohm reading, so I assume it's above a few MΩ.
I unwrapped the KS wiring in the engine bay back to the main harness bundle -- all looks stock and decent condition. I also unwrapped the KS wires to B127 and it took looks stock / good condition. I believe this would be sufficient proof that the knock sensor wiring is up to spec as it should be.
I'm about to take the chip / MAF out and go back to stock and see how things go, because this shit is starting to get really annoying.
On the suggestion from Mike @ Rev, I threw in bottle of octane booster this last time I filled up and did more data logging. Before the octane booster, I was getting knock retardation only in the ~2.1 g/rev engine load range. Now after the booster, it's logging knock retardation in the ~1.5 g/rev range. Not good and now progress is going backwards.
I printed out the knock sensor diagnostics and went over everything this afternoon.
1. I'm seeing ~4.87V at the knock sensor connector in the engine bay (B38).
2. knock sensor has 561K Ω
3a. B127 No 5 -> B38 No 1 has continuity. As I hold it, the auto-adjusting ohm meter drops from 1 to 0 Ω.
3b. B38 -> engine or body. No continuity or ohm reading, so I assume it's above a few MΩ.
I unwrapped the KS wiring in the engine bay back to the main harness bundle -- all looks stock and decent condition. I also unwrapped the KS wires to B127 and it took looks stock / good condition. I believe this would be sufficient proof that the knock sensor wiring is up to spec as it should be.
I'm about to take the chip / MAF out and go back to stock and see how things go, because this shit is starting to get really annoying.
'11 WRX Limited
'94 SS | 3" TBE, 07 TMIC, TD05H-16G, Revtronix Stage 2, Walbro -- Sold
'94 TW | R.I.P.
I'm having some ECU learning issues as well. I don't get what's wrong with the chip, seems like it's working for everyone else. That said, I just recently swapped back in my Hitachi MAF, but left in the chip. Everything seems to be running PERFECTLY now, except it just dumps fuel at WOT 
Try swapping out the MAF just to see what happens. It worked for me. You and I have the exact same setup, except you have a TD05 and I have a VF39.
Try swapping out the MAF just to see what happens. It worked for me. You and I have the exact same setup, except you have a TD05 and I have a VF39.
-Aaron
2000 Audi S4 - 2.7L Twin-turbo, 6 Speed
[quote="evolutionmovement"]It was me. And those are my balls. Happy Sunday![/quote]
2000 Audi S4 - 2.7L Twin-turbo, 6 Speed
[quote="evolutionmovement"]It was me. And those are my balls. Happy Sunday![/quote]