OFT Basic tuning procedure

[Kodename47]

Closed loop MAF scaling:
http://forums.nasioc.com/forums/show....php?t=1427448


Open Loop MAF Scaling:
http://www.ft86club.com/forums/showp...&postcount=995


As far as adding timing, I would advise only adding timing over 2500 and 1 load. Then you're not likely to add tip in knock. The rest of the table has no influence over WOT so IMO there's no need. Unless of course you think partial throttle is lacking....

[shif7i7down]

Originally Posted by Tromatic

Really interested in this thread. The OFT is going to make it possible for people to grenade their engine, and I want to make sure I'm not one of them!

Everyone that is currently a "pro tuner" at one time or another was at the same stage the op is currently in. If the guy wants to explore,great. Just do in a safe manner. And what better way to explore? enlist the help and knowledge of a forum!

[mid_life_crisis]

Originally Posted by HSayaovong

That's good to hear, since I'm on 93oct and i log for knocks i should be fairly safe to play with my timing"in small increments of course". Without to much fear of blowing something up right? And also by leaning base timing B alone for fail safe.

Also @jamesm , you don't happen to know of any tutorials on how to adjust MAF scaling do you? From the sound of everyone's response it seem that the wild swings in my stft and ltft are due to the maf scaling right?

This is the second time you've mentioned leaving "B" alone, but at least twice posters have said that the one to leave alone is "A". Can someone clarify?

[HSayaovong]

Originally Posted by mid_life_crisis

This is the second time you've mentioned leaving "B" alone, but at least twice posters have said that the one to leave alone is "A". Can someone clarify?

sorry, I meant leave A alone and adjust B., you can check for this in Romraider when you compare the different tunes, it's B that's adjusted for different tunes, not A.
Thanks for pointing that out.. didn't mean to confuse anyone.

[steve99]

Originally Posted by Toyota John

I am interested in the flash tuning process too. I have done a little tuning on my MR2 with a Apexi PFC and have taken the advanced EFI 101 class as well as read some books on tuning. But much of them deal with tuning live being able to see what cell you are on, making changes and watching the differences. With the flash system that we are using this procedure won't work. I am assuming that to tune with a flash system you would make a logged run on a dyno, look at the logged data, flash and repeat. Anyone want to share their procedure that works for them?

Yes correct for "flash" tuning sequence is, log, Analise results , change,flash ROM, log, Analise change flash ect.

To do true live tuning you would need a different ECU or emulator where some or all of the map tables are stored in RAM so they can be changed on the fly . I believe hydra http://www.hydraems.com/ can do this to some extent but it costs $$$$$ not really worth it for average person. Probably Motec and other as well.

Saw somewhere here where some Ecutek guys has a pseudo live tuning setup where they hijacked a couple on inputs to the ecu to switch/offset some custom maps on the fly for a sort of live tuning effect.

[jamesm]

With ecutek you have 4 map modes, so you only have to flash once every 4 runs if you use them right. Combined with a ~1 minute flash time, it makes it far less of a pita.

[HSayaovong]

hey @jamesm and @Kodename47 , i've updated the first post to reflect what you guys been saying and alot of stuff from the links that were provided. can you guys check it out and advise on revisions?
I would be great if we can get a thread with accumulated information to help people that are new to tuning and want to learn.

thankx

[Tromatic]

Originally Posted by HSayaovong

hey @jamesm and @Kodename47
I would be great if we can get a thread with accumulated information to help people that are new to tuning and want to learn.

thankx

I asked about that a while ago, and the answer seemed to be "ain't nobody got time fo' dat." Basic tutorials and even written information is sorely lacking. Half of these threads would go away given even basic written documentation and stone simple "your first tune edit" guide.

[Kodename47]

Originally Posted by Tromatic

I asked about that a while ago, and the answer seemed to be "ain't nobody got time fo' dat." Basic tutorials and even written information is sorely lacking. Half of these threads would go away given even basic written documentation and stone simple "your first tune edit" guide.

The thing is that there is a lot of information already on the web. The Romraider lot have been doing Subaru ECUs for years now and most of the way our ECU works is on their site. There's a few minor changes, namely the fact we use PI and DI and that info is already posted here. I could create a thread, but it would just be links to RR or NASIOC.

Very few people will give away their ideal fuel and ignition maps it seems at the moment and would you risk running someone else's maps anyway?

The way I see it, Google is your friend and have a deep look around what's around already with Subaru ECU information. If you have any questions then ask.... That's exactly how I learnt. Read info, compare maps which will inevitably lead to more questions which you then research. I found it quite satisfying.

[steve99]

Originally Posted by Kodename47

The thing is that there is a lot of information already on the web. The Romraider lot have been doing Subaru ECUs for years now and most of the way our ECU works is on their site. There's a few minor changes, namely the fact we use PI and DI and that info is already posted here. I could create a thread, but it would just be links to RR or NASIOC.

Very few people will give away their ideal fuel and ignition maps it seems at the moment and would you risk running someone else's maps anyway?

The way I see it, Google is your friend and have a deep look around what's around already with Subaru ECU information. If you have any questions then ask.... That's exactly how I learnt. Read info, compare maps which will inevitably lead to more questions which you then research. I found it quite satisfying.

Is the stock o2 sensor good enough to use to OL MAF Scaling or is an aftermarket wide-band O2 a necessity ?

[Kodename47]

Originally Posted by steve99

Is the stock o2 sensor good enough to use to OL MAF Scaling or is an aftermarket wide-band O2 a necessity ?

I keep on seeing/reading varying reports. My opinion on the matter is that you can use it for N/A tuning so long as the sensor has been scaled correctly. I have no idea what Shiv's O2 sensor scaling is like in comparison to the given ECUtek scale. It seems that the sensor is more accurate the closer to stoich (14.7) but loses it's consistency further away from that based on pressure and temperatures. Therefore I wouldn't do repeated runs without allowing a cool down period if you're doing OL fueling, this doesn't necessarily mean stopping just don't do WOT constant runs back to back.

To make OL scaling easier, you could set the map to a flat AFR of loads 1 and above - eg all cells 12.5/12.6 etc. This makes it easy to analyse and won't be an issue for a couple of runs that are needed for scaling. Due to the sensor accuracy, you want it as lean as possible BUT remember if your MAF scaling is already off then you could run leaner than the commanded AFR.

How I've been looking at it just for MAF scaling on an NA car:
- 1st runs with a map of 12.0/12.1 - If you've already got logs, you have an idea whether your scale makes you run lean/rich. Do a few pulls to get good data - I like 3 or 4 pulls from 2.5/3k to reline.
- 2nd runs around 12.5
- 3rd runs - fine tune with as lean as you're happy to go, but reduce timing a little to prevent knock. With the DI capability I personally would go as high as 12.75 but only for a couple of pulls.

That should be enough. If in doubt, always make the MAF scale a touch higher. I'd rather run 0.05 richer than leaner.

If you can visualise the data in graphs etc, you can see if the stock sensor starts to read inconsistently and then discard that data.

Once you can do this, I would seriously look into latency and PI/DI scaling as it's just an evolution of this technique.

Also worth noting on CL MAF scaling, as per @jamesm's screencast, don't go mental and randomly adjust each point. You want to adjust everything together.

[steve99]

Originally Posted by Kodename47

I keep on seeing/reading varying reports. My opinion on the matter is that you can use it for N/A tuning so long as the sensor has been scaled correctly. I have no idea what Shiv's O2 sensor scaling is like in comparison to the given ECUtek scale. It seems that the sensor is more accurate the closer to stoich (14.7) but loses it's consistency further away from that based on pressure and temperatures. Therefore I wouldn't do repeated runs without allowing a cool down period if you're doing OL fueling, this doesn't necessarily mean stopping just don't do WOT constant runs back to back.

To make OL scaling easier, you could set the map to a flat AFR of loads 1 and above - eg all cells 12.5/12.6 etc. This makes it easy to analyse and won't be an issue for a couple of runs that are needed for scaling. Due to the sensor accuracy, you want it as lean as possible BUT remember if your MAF scaling is already off then you could run leaner than the commanded AFR.

How I've been looking at it just for MAF scaling on an NA car:
- 1st runs with a map of 12.0/12.1 - If you've already got logs, you have an idea whether your scale makes you run lean/rich. Do a few pulls to get good data - I like 3 or 4 pulls from 2.5/3k to reline.
- 2nd runs around 12.5
- 3rd runs - fine tune with as lean as you're happy to go, but reduce timing a little to prevent knock. With the DI capability I personally would go as high as 12.75 but only for a couple of pulls.

That should be enough. If in doubt, always make the MAF scale a touch higher. I'd rather run 0.05 richer than leaner.

If you can visualise the data in graphs etc, you can see if the stock sensor starts to read inconsistently and then discard that data.

Once you can do this, I would seriously look into latency and PI/DI scaling as it's just an evolution of this technique.

Also worth noting on CL MAF scaling, as per @jamesm's screencast, don't go mental and randomly adjust each point. You want to adjust everything together.

After logging for an hour or so whacked data into CL Maf scale spread sheet see below. Look like CL area is pretty good only a couple of minir tweeks required.

Looking at the WOT runs I did for OL scaling similar to stuuf I sent you a while back it was off a fair bit Measured AFR 11.3-11.5, commanded AFR 12.3-12.5 (STFT=0, LTFT =+2%). Weird all the error in top end?

Unless I am doing something wrong ?

[Kodename47]

I wouldn't change it point by point, make sure you change it as a curve. If the lowest end of the MAFv scale get's increasingly distant from 0% error at as you near 1v then you need to look at latency. Random errors on the chart could be due to where the PI/DI systems change. What I would add to the chart above is % change. I have a table setup that compares my new table to OEM scaling. You can have the percentage of change increase/decrease as you go up the scale, but ideally you want that to be a smooth transition. Looking at the table above, there's nothing I'd change below 2v and notice how you have + error around 2.5v and then - error after.

For example, on one I've done recently the % change from the stock scale at 2.5v is 1% but I'm seeing 6% at the top end. I look at the data and work out at which points I should start to smooth the transition to keep the scale a nice curve.

Remember there will be no STFT in OL.

[HSayaovong]

Originally Posted by Kodename47

I wouldn't change it point by point, make sure you change it as a curve. If the lowest end of the MAFv scale get's increasingly distant from 0% error at as you near 1v then you need to look at latency. Random errors on the chart could be due to where the PI/DI systems change. What I would add to the chart above is % change. I have a table setup that compares my new table to OEM scaling. You can have the percentage of change increase/decrease as you go up the scale, but ideally you want that to be a smooth transition. Looking at the table above, there's nothing I'd change below 2v and notice how you have + error around 2.5v and then - error after.

For example, on one I've done recently the % change from the stock scale at 2.5v is 1% but I'm seeing 6% at the top end. I look at the data and work out at which points I should start to smooth the transition to keep the scale a nice curve.

Remember there will be no STFT in OL.

Wait, I can't seem grasp this... by apply change evenly across a section of the maf scaling table..wouldn't this create skewed and inaccurate result?
I.e reducing from 1.16v to 1.7v down .03 would have more of an impact on the lower voltage then on the lager one wouldn't it?
So by tackling each given voltage column and adjust each one to the ltft readout should yield greater accuracy shouldn't it?

Please advise. Thanks