WTF Tuning Part 1: Reading, Data Logging, and Analyzing Logs

[Heide264]

General Disclaimer to Start Off
You shouldn't blow your car up doing this. Don't make map changes unless you know what you are doing. If you do, don't blame me. You typed the number in. You flashed it to the car. That being said, I've never heard of anybody doing damage by logging data. Bottom line, as a blanket statement, I'm not responsible for your car. Just putting it out there.


Introduction
This is the first entry in a string of threads that I plan to write. I have been road tuning my own car now for over a year now (with help from others), have learned a good bit along the way, and wish to pass some useful knowledge on to others. I can’t supply a “conclusive tuning guide” through forum posts – nor am I qualified to do so.

I’d recommend picking up the following books to get a head start if you want:
• Designing and Tuning High-Performance Fuel Injection Systems by Greg Banish. Short book. Great initial overview.
• Turbo: Real-World High-Performance Turbocharger Systems by Jay K. Miller. This book goes in depth on JUST turbochargers. Given that most other books are general tuning books, this is a nice supplement. Lots of pictures and a surprisingly interesting read
• Engine Management: Advanced Tuning by Greg Banish. Short read, but a bit more in depth than the earlier Banish book. Good read though.
• Forced Induction Performance Tuning A Practical Guide to Supercharging and Turbocharging by A. Graham Bell. This book is not for the timid. It’s long, somewhat boring, and old. That said, I learned most of the fine details that I know from this book. If you are serious about it, this will take you back to the “classics” of tuning – an important aspect commonly overlooked on forums I feel. You’ll learn a ton though if you can bite the bullet and make it through.
• Maximum Boost: Designing, Testing, and Installing Turbocharger Systems by Corky Bell. I haven't read this one yet, so I can't offer an opinion on it. It did receive good ratings on amazon, which is what I based my book choice off of.
• Performance Fuel Injection Systems: How to Design, Build, Modify, and Tune EFI and ECU Systems. by Matt Cramer & Jerry Hoffmann. I just finished this book. It was a good read, but focuses more on using standalone ECU's or retrofitting an older carb'd car with an EFI system. You'll go down into the fine details - most of which is magically done for us guys using the stock ECU. Good book, but I wouldn't put it as a priority.

There are some good links that Matt has posted below. They contain some great information. I am only posting books above due to the ease of understanding the material, the assurance that they are written by a credible person, and the logical order the material is presented. That does not mean that an internet thread is bad, by any means. I just wanted to use this opportunity to give credit to the authors above, as well as provide a nice concrete starting point.

However, I am writing these entries assuming that you have not read those books. I know not everybody has enough time to read all of the above. My posts will focus on newer 32-bit ECUs with open source ECU definitions/parameters. I can offer some suggestions for the older ones on request. I will also assume you have some version of excel that isn’t from the Stone Age.

[Heide264]

Data logging is something that anybody with a Subaru should be able to do in my opinion. It will show you many issues with your car that can occur over time or when doing modifications – be it if they show up as detonation events, air/fuel errors, or boost errors. There are several devices which you can use, but I only plan on focusing on two of them: the OpenPort/Tactrix cable (OP2) and the Cobb AccessPort (AP).

OpenPort 2.0 (OP2.0): This is an ‘open source’ cable that you can buy premade from Tactrix. It is around $200 new currently, and I’d suggest just buying it instead of downloading the hardware design. There are a few older versions (and knock off versions) which will work with ‘older’ models (pre-CANbus – around 05 to 06ish). It allows you to read data or ‘re-flash’ your cars ECU with a new ‘map’ or tune. The newer 2.0 version brings with it a few perks such as standalone logging to an SD card (no laptop required! Woo!). It also takes advantage of the newer CAN bus found on modern vehicles… which is fast.

Cobb AccessPort (AP): This is a ‘plug-n-play’ tool for Subarus. It comes preloaded with some Off-The-Shelf (OTS) maps that are good for stock cars, cars with intakes, downpipes, or some common mods. They also provide a means for professional tunes. They are similar in function to an OP2.0, with some slight differences. I don’t personally use one of these, so my definitions may be slightly different than theirs.

You need to decide on a purpose of your log before you start. You may be investigating your cars boost control, fueling, or just a general check for any detonation. Each one will require specific parameters. You cannot usefully log every parameter at once – you lose resolution (datapoints/time) with every added parameter. The OP2.0 will give you a much better resolution on a CAN bus equipped vehicle… where as the AP you will have to give your chosen parameter list a bit more thought.

If you are using an AP, jump to the next reply. I’m going to use the rest of this reply to go over configuring an OP2.0 for standalone logging.

To configure your OP2 for standalone logging, some tech savy-ness is in order.

First, you should get the most recent version of ECUFlash so you have the most recent drivers/firmware for your OP2. If you go to Help -> About EcuFlash... I have Version 1.43.3252 showing. It works with my LC-1, which was my primary issue with the version on the Tactrix website. I got this one off of an Evo forum actually. After installing it, connecting your OP2 to your laptop and going to Help -> Licensing should apply the new firmware I think.

Next, open up your tune in EcuFlash. If you don't have one, hook it up to your car and read it. If you have definition problems - you are outside the scope of this write up. Either go to RR forums or shoot me a message and we'll get you up and going. Open up the 'ROM Info' section and write down your ECU ID (mine is 7412597007). Next look at the bottom of the entire scroll box for a "Patches" menu - check that and reflash your tune to your car. That will enable you to read various parameters much faster using the CAN bus.

Alright, now you need your microSD card and a bit of patience. You'll need to move a logconfig file onto the root directory of your nice new microSD card. I'll attach mine for reference. Don't rename it. Don't put it in a directory. The catch here is that mine is made for my ECU ID. All of the values read from the ECUs RAM (xFF or most floating values) are specific to an ECU. You will need to go through your EcuFlash or logger definitions and find yours. I can help with this, but it is a good practice to learn how these files work. Ctrl + F will be your friend.

Set your trigger at the bottom. I either use the defroster or RPM > 0. You'll see that I include both in my file and comment out ( the one I don't want.

One last interesting note - you'll notice the scaling in the log config file is in Reverse Polish Notation (RPN). I'd recommend giving it a wikipedia read. I am a bit partial to RPN calculators after getting the hang of them.

Any more detail about the OP2 config file is a bit out of the scope. Give it a go, and I can help you out if you post up your ECU ID and your current file. It takes some work and was fairly frustrating.

[Heide264]

A good place to start is to do a ‘general’ log that includes any detonation events, as well as a brief summary of what may have caused them.
• RPM
• Engine Load
• Manifold Pressure or Boost Pressure
• MAF_V
• A/F Learning #1 (Long Term Fuel Trim)
• A/F Correction #1 (Short Term Fuel Trim)
• IAM (Dynamic Advance Multiplier)...NOT DA
• FBKC (Knock Correction)
• FLKC (Knock Learning)
• Ignition Advance

Note that with that many parameters on an AP, you will not get ideal resolution – will not be too useful for ‘pulls’. It is more useful for general driving to see if you are pulling any timing due to detonation or if you have any long term trims that need addressed.

Alright. So now that you've gone for your drive and (hopefully) have a large .csv file, pop it in excel.

There are three commands in excel I use constantly when dealing with log files - filter, freeze top row, and insert scatter plot. To make my life easy, on the newer excel versions, you can customize the top bar where it normally has undo/redo/save/etc. Just go into excel options -> customize, and go to town. The buttons then become hot keys binded to Alt+#. Just press the alt key if you aren't sure for the overlays to pop up.

I have mine configured as such:


Whenever I open a log, I immediately hold alt and hit 2, 3, 'r' (freeze frame submenu option). This gets me set and going to filter for any knock events.

After everything is sorted, I just check all three knock parameter columns for anything odd. Pay attention to make sure your IAM/DAM is always maxed... you aren't getting FLKC in any rpm/load zones... and you aren't getting a lot of FBKC. A few events of FBKC here and there happen. I'd be concerned if you get any that are above 1.5 engine load or about 4k rpm. Also if you see any more than 5 for a half hour drive, I'd start to look into it.

Knock can be caused by several different factors, but that's beyond the scope of this first write up. The primary purpose for now is just to see if you have any issues.


Here is an example of a random bit of FLKC I picked up on my way to work. This was over a 40 minute drive on back roads, and it turned out to occur around 1.2-1.4 engine load and around 3k RPM. My car is noisy, and I have a feeling the knock sensor got confused by the 3 mile long stretch of (very) fresh tar and chips I went through.

[Heide264]

This is the fun part. To start off, I use the rear defroster to trigger pulls.

For just about every pull, I analyze:

• Fueling
• Any Detonation (obviously)
• Boost/WGDC/Corrections
• Ignition Advance

Fueling:

When under boost, you need a wideband o2 sensor to accurately measure fuel. Sorry. If you don't have one, you will be guessing. Again, do not trust the stock o2 sensor under boost. I'll have to take a plot of mine vs. lc-1 output sometime and add it in here to prove a point.

To analyze your fueling on a Cobb, You'll want the following parameters in a pull:

• RPM
• MAF_Voltage
• Final Fueling Base
• Wideband o2 output
• Injector Duty Cycle (Pulsewith calculation)
• If you have the resolution needed - add knock parameters

Anyhow, you'll want your target fueling and your actual fueling over the RPM range, as well as any correction to apply over the MAF voltage range. I log Final Fueling Base (FFB) and use that. Others have interpolated their open loop fueling table using a spreadsheet. I prefer to use FFB, because it shows what the ECU is actually aiming for. In the newer ECUs, you see some rather large 'undefined' compensations here and there.

I'll leave my formulas in text so you can see them. My target fueling is equal to 14.7/FFB (From the definition file), and the correction I apply is equal to actual/target. It may be better to use a percent error formula, but this seems to work for me as is. I do round down my corrections. EDIT: I used (lc1-Target)/lc1 this time around, and it appears to be a better way to go. Open for suggestions on this one.


...And drag 'em down.

Click your RPM column header to select the entire column, then ctrl click both your lc-1 and TarFuel colums... Scatter plot 'em. I suggest just moving the chart to a worksheet to make it easier to see. A lot of the detail is lost unless you format your graph correctly. Size both axis to show as much resolution as possible. Add major grid lines to the horizontal axis. And lastly, I change the page size to 8.5"x14" to get a bit more viewing space.

This is a well formatted (slightly sloppy fueling, but not too bad) graph. See how much detail you can pull off of it?


Next, plot your MAF Voltage against your Correction factor. Format appropriately. You should be scaling the vertical axis from about .8 or .9 to 1.1 or 1.2 Mines unusually sloppy here:

^That should be the change you would apply to your MAF scaling. It does fluctuate a lot when the wastegate opens... I doubt yours will be smooth either. I will probably apply the changes from 3.3V and over though.


Finally, browse down your Injector Duty Cycle column. You shouldn't see any numbers past around 85% most say. If so, you are pushing your injectors pretty hard. It's something to keep an eye on - I'm pushing 90% in the 4-6k region.

Just about covers it for fueling.


Boost Stuff

I am assuming you are using an ECU controlled EBCS here. If you aren't, I will address that in a later post or you can always shoot me questions.

To analyze boost control on a Cobb, You'll want the following parameters:

• RPM
• Primary Waste Gate Duty Cycle (WGDC)
• Target Boost
• Manifold Pressure/Boost pressure (Experiment to see what works. Use the 2-byte parameters when possible for better resolution. I'd also recommend relative parameters over absolute personally. My Relative Manifold pressure maxed out at 18.5psi, so I have to use Boost Pressure Absolute. 04-07 32 byte ECUs -STis/Legacys look to be okay with Manifold Pressure Relative)
• Turbo Dynamics - Integral
• Turbo Dynamics - Proportional
• Wideband o2 output (Just to check to make sure your fueling isn't way off)
• If you have the resolution needed - add knock parameters

The main thing you should be concerned with is your target boost Vs. actual boost. The rest is for diagnostic purposes. So, go ahead and plot that.
Two examples to examine here:

During this pull, the power came on smoothly and it was a overall a strong pull. This is my lower boost map, so it wasn't overly torquey. Just a nice smooth & strong pull. There was a slight hole right around where the wastegate opened up. This was noticeable while driving and felt almost as if the engine was pulling a bit of power. It was due to a lower than ideal initial wgdc.





This one was from after I put my equal length headers on. The pull felt very weak and I almost thought my engine wasn't running quite right. Turns out this was due to poorly adjusted turbo dynamics as you'll see later.

The important part here for now is to see if you are hitting your target boost, and that it is a smooth process. You may not notice that your boost settings are out of whack or not quite right for your car. Correcting these makes a huge difference in the feeling of the car. Commonly you will see dips like my issue above, periodic boost 'spikes'/bumps, or boost creep depending on your individual car.



Now for the analysis. I feel that even if your boost curve looks good, you should start getting in the habit of examining how your boost control is working. It will familiarize you with how a PID (Proportional/Integral/Differential) control system works, as well as showing you if you need to adjust your initial wgdc tables.

I graph RPM Vs. WGDC, Integral correction, and Proportional correction. I'll post up the graphs for both of the above scenarios:

Here is the better of the two pulls. Notice how my WGDC was a bit too low, causing the wastegate to open a bit early and cause my dip in boost.




Here is the crappy pull. This is due to an improperly tuned control system. Since I was 'overboosting' according to the ECU, it corrected my wgdc so it would allow more boost to vent. When the overboost cleared, it did not clear the negative correction fast enough. This was mainly due to me not properly tuning my turbo dynamics when I added my EWG. The issue became much more apparent when I added the equal length headers and the boost control system had to be more active in a smaller amount of time.


That about ties it up. I will add in corrections and suggestions from here out, but this encompasses all of the info I've planned on covering for in this thread. Please chime in on any discussion points and ask questions. If you have suggestions or recommendations for the next thread, toss them my way.

Brandon

[EJ257]

Nice start so far, bud

A few other threads that may be useful to link to:
OPENSOURCE TUNING: Your questions answered here!!! RomRaider + Tactrix
Datalogging with Romraider:
RomRaider • View topic - How To: Get started with tuning and logging
RomRaider • View topic - How To: Make useful data logs
RomRaider • View topic - How To: Read Data Logs
RomRaider • View topic - Subaru's knock control strategy explained
RomRaider • View topic - How To: Use Learning View to verify closed-loop MAF scaling
RomRaider • View topic - How To: Understand And Use Learning View's Knock Information
How Subaru’s Factory Boost Control System Works
A Complete Tuning Guide - NASIOC

[xcEXrdr]

Damn...never knew my OP2.0 had the SD card slot. Gonna have to try that out.

I'll have to follow your posts. Maybe your suggestions will help me get more involved in the tuning process. All I've done so far is flash my old car and log it to look for problems. Should log the new car just to see if looks OK.

[116Fanatic]

Good read so far...cant wait!

[EJ257]

WTF Tuning would like to welcome Donkey on board, as another resident tuner to the team

[Heide264]

Just as a followup.. I have been on vacation this week. Been reading and beaching it up. Be jealous.

Will get this finished up over the next few days. Be sure to fire any suggestions or questions concerning this my way and I will get them worked in.

[90crxfreak]

in for later

[Heide264]

Alright guys. This should have all the material in. Start picking at it and let me know anything you want me to add in.

[Alexmartynyuk]

This is the first entry in a string of threads that I plan to write. I have been road tuning my own car now for over a year now (with help from others), have learned a good bit along the way, and wish to pass some useful knowledge on to others. I can’t supply a “conclusive tuning guide” through forum posts – nor am I qualified to do so.

I’d recommend picking up the following books to get a head start if you want:
• Designing and Tuning High-Performance Fuel Injection Systems by Greg Banish. Short book. Great initial overview.
• Turbo: Real-World High-Performance Turbocharger Systems by Jay K. Miller. This book goes in depth on JUST turbochargers. Given that most other books are general tuning books, this is a nice supplement. Lots of pictures and a surprisingly interesting read
• Engine Management: Advanced Tuning by Greg Banish. Short read, but a bit more in depth than the earlier Banish book. Good read though.
• Forced Induction Performance Tuning A Practical Guide to Supercharging and Turbocharging by A. Graham Bell. This book is not for the timid. It’s long, somewhat boring, and old. That said, I learned most of the fine details that I know from this book. If you are serious about it, this will take you back to the “classics” of tuning – an important aspect commonly overlooked on forums I feel. You’ll learn a ton though if you can bite the bullet and make it through.

However, I am writing these entries assuming that you have not read those books. I know not everybody has enough time to read all of the above. My posts will focus on newer 32-bit ECUs with open source ECU definitions/parameters. I can offer some suggestions for the older ones on request. I will also assume you have some version of excel that isn’t from the Stone Age.

Might want to add Maximum boost by corky bell and something on how to read compressor maps.

[Heide264]

Might want to add Maximum boost by corky bell and something on how to read compressor maps.

I haven't personally read that one. I'll put it in there, but I'll add a note that I haven't personally read it. The name rings a bell though, and I may have to pick it up sometime.

Thanks

EDIT: Updated with the above and also included another book I finished up on vacation last week.

[Donkey]

and something on how to read compressor maps.

Stealth 316 - Turbocharger Compressor Flow Maps

A compressor map isn't the end all. Unfortunately it doesn't take into account turbine wheel flow,turbine housing A/R or compressor housing A/R. All of those things will dictate how a turbo will perform,not just the compressor map alone.

[Heide264]

Stealth 316 - Turbocharger Compressor Flow Maps

A compressor map isn't the end all. Unfortunately it doesn't take into account turbine wheel flow,turbine housing A/R or compressor housing A/R. All of those things will dictate how a turbo will perform,not just the compressor map alone.

I'd rather keep the specifics of that stuff out of this thread a bit. The turbos book I posted goes into greater detail on compressor maps than most would need.

Agreed on Donkey's part though. A compressor map is only a small piece of the puzzle. A thread three times the size of this could possibly cover the basics.

I'm looking to start measuring my turbos efficiency and I/C efficiency experimentally as opposed to theoretically. I will make sure to post up my results after I (finally) get around to it. I do not expect to see things that closely agree with the shady compressor maps that I've found. That'll be down the line a bit though, and I want to keep this thread along the train of the basics for now.