Thoughts On New Automated Tuning Software

[Bliksem]

Been reading up on this software being developed by an RX8 owner on another forum. Heres the link: http://www.rx8club.com/showthread.php?t=133797

Baseline v 1.01 BETA

PPO2 – The science of tuning

EXECUTIVE SUMMARY:

In order to maximize performance and simplify engine modification and management; Baseline will compile an almost infinite amount of ongoing engine performance data and scientifically determine the optimum fueling and timing information for your unique vehicle power plant; regardless of the level of modification without endless dyno tuning sessions; or cumbersome mathematical equations. If ragged edge performance is what you crave; a dyno can be used in order to specify your Rich and Lean Best Torque Range for the Three Dimensional Air Fuel Map (or you can use commonly accepted Air Fuel Ratios). The result is the most powerful street friendly tune possible given your driving style. This will DRASTICALLY simplify the tuning process without undue risk to your engine and wallet; for less that the cost of two tanks of gas.

METHOD:

The method is simple once you get someone else (namely Baseline) to do the heavy lifting (IE a BUNCH of math) for you. The standard equations used by the “Zen Masters of Electronic Fuel Injection” require specific engine data as no two engines are alike. By using this software, your engine data gets collected en masse, and grouped into the table structure that a host Engine Management Systems use. Plug in your EMS of choice, drive around for a few hours/days/weeks/months while happily logging info. Then dump the enormous log file; or files into Baseline; push the “Easy Button” and viola, the FUEL and TIMING MATH get done for you.

RESULT:

The result is an average value based on the Throttle Position, PSI of Boost or Vacuum, Engine Speed (RPM) grouped by the map cells you specify; (you can use any scale of your choice, so if you need a 3x3 or 30000x30000 resolution; Baseline will do it). Baseline notes the difference from your target values and the actual sensor data; and gives you a new map. Input those numbers “Tune1” and drive on. The more data you collect, the better your average values will be. You can also use the standard deviation to de-tune for a safer starting point. The level of accuracy of the “average” is fairly staggering +/-.05% deviation seems to be the norm. This is below the threshold of most EMS’s to even use…

LEARNING CURVE:

Once the log data is grouped and the “Tune1” is generated; Baseline will back up your old logs and collect the next set of logs based on your “Tune1” parameters; those logs are then re-run against “Tune1” and any deviations are corrected to create “Tune2” and so on.

For example your Target AFR may be 12; and Actual AFR is 10. “Tune1” pulls a bit of fuel and re-logs for an Actual of 12.0000001; “Tune2” will then add in a bit of fuel and end up with 11.9999999; “Tune3” will take some out and so on… similar to Closed Loop driving but with an engine history that a stock PCM/Engine Control Unit could never hope to retain.

Additionally, any modifications on the vehicle would be adjusted to over time; so you should be able to mod away without worry. There will also be a separate space in Baseline to store what engine mods you have done; so you can see what if any effect these have on your performance. So if after adding that intake you love and watching your Volumetric Efficiency go down the tubes; you should be able to asses the effectiveness of your purchase; or lack thereof.

This last piece I am particularly excited about because we turbo guys can now swap out snails and compare the results fairly easily and with a high level of precision.

As an added note – the following fringe benefits can also be realized:

• Virtual Blueprinting of your engine: Based on having a huge amount of sensor feedback logged; the system will “learn” your engine characteristics. For example; if your injector is rated at .6 lb/min of flow; and you set your duty cycle to 100% but only flow .599999 lb/min; Baseline will catch it on subsequent runs and re-task your other injectors to make up the difference. As of now it will not change the injector size to compensate; but the end fueling will be adjusted to hit your target AFR.

• Engine Reporting: With millions of sensor data logs; you can view your engines performance history like never before; see Volumetric Efficiency Graphs, Fueling Maps, Timing Changes, a Virtual Engine Dyno, Most Often Accessed Load Cells, Temperature vs. Airflow, Decode the MAF sensor, Tune vs. Power, Turbo / Supercharger Boost Profiles; you are only limited by your imagination. We will be gathering a list of pre-formatted reports in addition to offering “ad-hoc” reporting in the future.

• You will also be able to organize and save every tune you ever produce from Baseline on your engine. Going back to an older tune will be a snap along with comparing tunes to each other.

• Diagnosing engine problems should be a bit easier; if deviation increases without a corresponding engine change or other environmental variable – something is up.

Baseline will also premiere the Online Tune Warehouse. With users permission; Baseline will be able to access the Warehouse via the internet and upload your engine information and tune into a global database. This will allow for aggregate comparisons for research purposes and also for new Baseline users to download an initial tune from the web in the event that they have no Basemap to work with. Since this feature is all driven from the same model with no human factors; all the numbers can be directly compared to each other.

* No personal information would ever be collected; and you would have to turn on this feature; but hopefully the idea of accessing aggregate information will appeal to most people.

A SPECIAL NOTE ABOUT TIMING

All tuners know that timing is a KEY component of engine tuning; it will make or break an engine. Baseline is developing a custom mathematical timing model based off of industry leaders in engine physics. The goal is to determine flame speed propagation and crank angle in pursuit of maximum cylinder combustion pressure (12 degrees after Top Dead Center.) This model will incorporate your Boost Pressure, Octane, Temperature, Cylinder Volume and Air Fuel Ratio.

OEM timing maps are built by manufactures using advanced sensor information and thousands of hours of experimentation. No aftermarket company or individual tuner has the same level of resources at their disposal.

The same “safety” factors come into play on an OEM timing map as on their fuel maps. Manufactures have to incorporate all tolerances in the manufacturing process and must take into account the wide variations in fuel octane, changing operating conditions and other mechanical issues.

We hope to provide a 3D Timing Map where our users can input the Degrees After TDC that they want maximum cylinder pressure to occur; and Baseline will do the rest!

I dont know enough about the science of tuning but I thought the thread was an interesting read and it would be interesting to see what some of our forum members have to say.

btw Apparently he's now at the stage where it can be used for PowerFC

Edited October 14, 2008 by MintR33

[rob82]

Good in theory but the more modified an engine becomes the more likely it is that this system wont work. Most highly modified engines are too dynamic for the system to learn once and try to reuse that value again at that same point for the amount of accuracy your talking about. And it kind of allready exist - have we not heard of LTFT and STFT's?

[Guilt-Toy]

Pass

[CEF11E]

hmmmm.

you are still relying on someone's calculation and then the computers estimation of a calculation.

I am sure it works but it will never be as good as a good tuner that can spot things going wrong.

tuning is not a black art like some make it out to be. still best off taking it to a tuner you have heard about and has a good rep.

[VB-]

i'm no tuner but like to think i know a thing or two, i only skimmed it, but the biggest problem i see is ignition timing, and i wasn't convinced by that extra note at the end. they are working purely on theory by the sounds of it, theory doesnt allow for shit batches of fuel etc. for most things i think a self tuning ecu could work, with a wideband input etc,

but where all the power is, is in ignition timing, knocking is probly the main thing to look for when tuning a motor. there is nothing mainstream atm that can MEASURE knock, there are sensors that can 'hear' it, and although they send out a signal, i dont think there is anything that can convert knock into a fixed value.

now when ionization current technology becomes more mainstream, i think this could work better (if my understanding of ionization current is correct)

[saliya]

Well, I'm going to go out on a limb here and say that with appropriate logging you should be

able to get a better AFR tune with software like this and appropriate sensors than from <insert your favourite tuner here>.

No tuner is going to spend a week running your car at ALL load points and under all conditions to get perfect AFRs.

No reason for the logging/software not to, though. Properly-calibrated wideband sensor and off you go. Keep the

sensor and logging ability in your car, if things change you can update the fuelling. But that's AFRs.

The article mentions using calculated timing points; don't know how this would compare with the 'advance until no power gain'

or 'advance until knock' theories. I think you have to do timing individually, per-engine, using appropriate equipment -

perhaps a fatal flaw in the concept.

Regards,

Saliya

[700HP-GTR33]

OMG.. ..Tuners will become extinct

The car will run itself to work and back.

make me a cup of coffee and do my work for me....

Ok on a serious note...self tuning cars in theory sounds good.

It gets its own tune by the conditions you and your engine are in at any one time.

Sounds good.

But how is something like this going to be able to tune correctly.....not all tuners use a knock as part of their tuning

requirements.

Handing over complete control to a computer may just be the fatal flaw as said.

It cannot smell, hear or see if there is a major problem about to happen. The thing may just cut off

some things and starve the motor of fuel or timing....both of which can be real bad.

Has every one forgotten already the Qantas flight last week

[Scooby]

Check this out re Vipec Quicktune http://www.vi-pec.com/page_files/QuickTune.html particularly the words:

The QuickTune software is much faster then a good operator. It is not possible to manually tune as quick as the QuickTune . A typical engine will take less than 20 minutes to fully map on an engine dyno.

Doesn't do knock or ign, but pretty impressive if it does what it claims.

[GTR1993]

autronic auto tune?

Using a target AFR table speeds things up, but you still have to go over it manually and do ignition tables.

[Scooby]

not sure how it compares to the autronic (there's been much discussion on vipec vs link vs autronic elsewhere).

sure the tables need to be set up but presumably it should cut the cost / time for tuning quite a bit at 20 mins for a full fuel map? the vipec literature also suggests that Quicktune is more accurate than most tuners:

The accuracy of the QuickTune can be setup in the software from 0.5% to 5.0% accuracy. The default value is 2% accuracy, this is generally a lot more accuracy then most tuners manual tune engines.

one tuner suggested to me that the Quicktune won't become popular because it reduces the time/ cost of tuning and that means less $ for what many see as a good cash-cow. most good workshops have too much work anyway so i'm not sure that would become a reality, they'd just be doing something else.

Edited December 7, 2008 by Scooby

[SECURITY]

im currently trialing the software and working closely with the developer of the PPo2 system, it was designed around rotary engines as he drives an rx8t but works very well with piston motors. i was a bit skeptical at first but ive gotten some fair good gains from it, im the only skyline/turbo trailer at the moment but so far we've been picking out the bugs from it and it should be very skyline-friendly once the final version comes out.

it works like this:

1. you log all of your data via a data logging tool that has the ability to export to excel. you can run 5mins+ of data or consecutive bits of data. just get as many logs as you can! but it doesn't matter - you can come back and re-do your logs once you've completed a tune and it will fine tune it further.

2. run your logs through the software and then it will output a injector map that you can add to your current figures.

thats pretty much the run down of it, there are alot of steps involved in setting up your car - very detailed information you have to input before you can start tuning.

the preferred afr's were around 14.7 off boost then 12-12.1 on boost, then 11.8 above 17psi. I ran 3 maps into the software (they were 3 runs @ 10.5afr's in 2nd gear - probably around 20secs of logs each) the results were 11.1 - 11.5afr's to redline.

at the moment this is all run through microsoft access but once the bugs are all sorted it will be moving to a stand alone program. we have talked about making it an 'add on' to data logging software but i think that will happen after the initial product launches and much much much more time is spent in development.

Edited December 7, 2008 by SECURITY

[PPO2 Kane]

Joe (Security) - just hit me up about this thread. Funny how things spread on the internet.

My name is Kane - I am "that guy" doing this software.

The RX8 thread is long and off topic in some ways - typical. But if you want the up to date information without the BS; go to www.ppo2performance.com

Keep in mind; we are developing now - so not everything is "proven" yet - then again; the software is free right now....so not all bad news.

I have been running it on my RX8 for a year (with one motor still); and have tuned an FD RX7; Acura Integra RSX; Toyota MR2 SPyder; Joe's Skyline; an Evo; and a half dozen or so other RX8's with it. Right now we are working on cleaning up the fueling models - and we have a timing model done but need to be verified. We are using the University of Idaho's graduate engineering department to help us out with the timing.

There were a lot of good points brought up in this thread - and I am excited to get some good feedback and try to answer any more you guys have. So fire away (read the site first please - it answers a-lot already).

Last note; the old "tuner can do it better" argument. A good tuner can; I'll never dispute that; there are too many variables. But on a street driven car... for the money difference; is it competitive? And that tune is only good for the motor at that VE% level (with the exception of some MAF cars)... so new cam profile, rising rate fuel regulator, intake (for MAF cars), climate (MAP), means you'll be back on the dyno again.

IMO - the extra 20-30 HP I have to sacrifice to safety using the software is worth the few G's I'll sepnd on dyno time. But I don't race much. Everything I have done is in the context of street driven car.... but the science is universal so could I do a race / more aggresive tune? Sure if your willing to risk it...same with a real tuner.

EDIT - oh yeah the timing thing the reason why it is such a big deal is because while all detonation is bad; pistons can take a bit of it (trial and error tuning); Rotaries cannot take much if any without grenading the engine. So the idea of hitting peak cylinder pressure without ping is attractive to me especially. And I think a bonus to Piston folks too.

Edited December 7, 2008 by PPO2 Kane

[STATUS]

No tuner is going to spend a week running your car at ALL load points and under all conditions to get perfect AFRs.

No reason for the logging/software not to, though. Properly-calibrated wideband sensor and off you go. Keep the

sensor and logging ability in your car, if things change you can update the fuelling. But that's AFRs.

most good tuners use start of the art wide bands (changed 3 weekly in my case due to the amount of leaded fuel i see on mine.) and tune all the above load points as part of a normal tune... well at least they should be.

Edited December 8, 2008 by URAS

[PPO2 Kane]

most good tuners use start of the art wide bands (changed 3 weekly in my case due to the amount of leaded fuel i see on mine.) and tune all the above load points as part of a normal tune... well at least they should be.

I think he is speaking more about the law of large numbers.

If you strap it to the dyno on Day 1 - With Gas 1; and Temp 1 - then you get a fuel map that covers 3000 data points. Cool; that is a reasonable statistic for a tune.

Now - if you log 200,000 datapoints over a few days of driving; then your tune is a better sample of the engine; alot better.

Now if you continually log/tune - then that is the best infinite right? So as you approach infinity - you get a closer and closer model of the actual system you are measuring.

The difference between STFT, LTFT and Baseline; is that I run a LTFT for every load cell in the motor going back forever... and I have no "limits" to the adjustments I can make - so big VE changes are not going to peg the PCM. But the concept is the same yes. And my LTFT Targets are set by the user - not the factory.

[VB-]

infinite is nice, and essentially your trying to get the most accurate VE map, but problem is, your ECU will only have one VE map at the end of the day, and your motors actual VE is constantly changing anyway. you'll end up with a tune the same as if an actual tuner dyno'd it on a day with moderate temp/humidity wouldnt it?

every tune should have a slight margin of error anyway, allowing for hot/humid/crap fuel, i cant see this thing giving a tune thats THAT much better than a guy running it up on the dyno and playing with the numbers himself.

i'm not a tuner, the above is pretty much what i think i know please let me know if i've missinterpreted something

[PPO2 Kane]

infinite is nice, and essentially your trying to get the most accurate VE map, but problem is, your ECU will only have one VE map at the end of the day, and your motors actual VE is constantly changing anyway. you'll end up with a tune the same as if an actual tuner dyno'd it on a day with moderate temp/humidity wouldnt it?

every tune should have a slight margin of error anyway, allowing for hot/humid/crap fuel, i cant see this thing giving a tune thats THAT much better than a guy running it up on the dyno and playing with the numbers himself.

i'm not a tuner, the above is pretty much what i think i know please let me know if i've missinterpreted something

It is a bit of a "yes, no, and maybe".

A good tuner; I would say we would be about equal... if I had to guess; for a street tune. Now - every "tuned" car that I have run on Baseline; the software has found holes in... so take that as marketing BS or lack of really good tuners; your call.

But price vs what you get. $100 for some software you can use forever vs a few hundred every tuning session...

For me personally; I like being able to re-tune everytime I mod the car; since I am always in the garage tinkering with crap.

[VB-]

yeah definately makese sense now.

one question, i've only quickly skimmed over it, but i'm still missing something, how does the computer determine ignition timing? obviously when tuning for power, knock detection is probly the most critical thing to making the motor last, but from what i've seen, there is very little on the market that is reliable enough at detecting knock (nothing that can do it without human input) , and what is reliable enough, i havent seen mention of being used with this setup

[rob82]

dOes it take into account transient fueling?

[PPO2 Kane]

dOes it take into account transient fueling?

Do you mean acceleration enrichment or transition from vacuum to boost?

Transition is actually that is the one area where I find the most holes.... Since it happens so fast; and most people are not accounting for latency of the AFR readings; it tends to have bad lean or rich spots depending on the tuner.

With Baseline; law of large numbers again; we measure the transition a bunch of times and can snapshot it pretty well; all the FI MAP guys that have used it comment on how much smoother that part is.

As for acceleration enrichment; the beta has not been activated yet - but I have used it and we use a gamma mode to calculate it. The difference is we calculate the base fuel amount to ""gamma-ize" off of the most common basemap cell in use at that load point. So if you gamma is 1.2 = 120% of base fueling; and at 4000 RPM your most commonly at 2 PSI - then we use 120% of your fueling at 2 PSI. Motec is a good source for more information about Gamma stuff. Some of that still needs to be fleshed out before I activate it in the Beta.

[PPO2 Kane]

yeah definately makese sense now.

one question, i've only quickly skimmed over it, but i'm still missing something, how does the computer determine ignition timing? obviously when tuning for power, knock detection is probly the most critical thing to making the motor last, but from what i've seen, there is very little on the market that is reliable enough at detecting knock (nothing that can do it without human input) , and what is reliable enough, i havent seen mention of being used with this setup

Since I hate clicking links too.... HEH

Here is the copy from my blog.

Timing is the single most important element of a properly tuned engine. Since the hot rod days; performance enthusiasts and automakers alike have been in pursuit of the perfect instant to initiate the spark event in order to achieve the maximum cylinder pressure at a known and specified point to make the most torque - aka Mechanical Advantage (typically 12 degrees After Top Dead Center for a piston motor) or to be a little safer/later in order to avoid detonation (retarding timing). This is of paramount importance to safe and effective power.

The proposed method of today’s dyno tuning (think Jeff Hartmen - How to Tune and Modify EFI Engines) is to over-retard timing and slowly advance it until you begin to see small signs of detonation; and then back it off a small bit for safety. In simple terms think of it this way; the MOST advanced timing (max cylinder pressure closest to TDC) that you can obtain without detonation will be the most amount of power that can be obtained with that particular fuel. Obviously, pre-ignition and detonation are serious concerns with drastic consequences but the idea is to be as advanced as you can.

Another method is to use a pyrometer and work the timing until you see the coolest EGT’s while still making good power. If you start out over retarded the EGT will be very high as most of the burn happens outside of the motor; as you advance it will cool the EGT’s until you start to get some detonation - which will super heat the motor and cause the EGT’s to rise again.

In both of these options; you would then likely want to scale back your timing in order to cover the possibility of a tank of “bad gas”; the amount of safety would be a personal choice.

Here is how the science works

In order to determine the proper timing for an engine; you have to know how fast the fuel propagates in the combustion chamber. In Explosives; this is referred to as Brisance; and the value is a known entity - so TNT always burns at the same rate. In the automotive realm; our little controlled fuel-air explosion is often measured by referring to the flame speed, and there are two separate values; turbulent and laminar. Once the spark ignites there is a flame front that moves very quickly and burns the skin of the fuel drop (I think of it as a shockwave) this is the Laminar flame speed. Then the molecules that are unburned inside the Laminar bubble begin to be consumed; this slower but more powerful burn is the Turbulent flame (and the one more responsible for power). So, lets simplify and think of it as two separate events (two different Brisance numbers).

In theory, if you were to know these two flame speeds, and you knew the volume in the chamber that was available along with the cylinder geometry (rotory engine chambers are not round after all) and spark plug location; you can determine quite readily how long from ignition until max cylinder pressure is achieved - then subtract that time from your goal of say… 12 Degrees ATDC and bang; you’re done.

In reality, the volume of the cylinder is always changing; so we have to think of it as a phased event:

Phase 1: After spark, the mixture is still being compressed (since ignition is often initiated BTDC. So flame speeds are X1 and Y1. This speed actually changes for each change in volume or each rotation of the crank. Remember, the engine is moving all the time.

Phase 2: At TDC; the flame speed is X2 and Y2 based on our maximum mechanical compression.

Phase 3: After TDC; the flame speed is X3 and Y3; as the turbulent flame increases exponentially; and the volume increases for each rotation of the crank.

The combination of these three sets of numbers are computed into derivatives in order see the effect of our time-lapse on our flame propagation / heat / volume changes. Based on this information we can calculate the number of milliseconds from spark to maximum cylinder pressure. With the RPM and our time-lapse information we can then calculate the number of degrees of rotation required as Z; subtract that from your goal timing of (in our example) 12 Degrees ATDC and therefore initiate timing 12-Z degrees.

This is ROUGHLY the same models that are used in closed loop driving on your OEM engine; coupled with a knock sensor check - in order to verify that the math is meeting real world expectations.

The feedback (knock sensor) is important for a lot of reasons; one being fuel makeup, Air Fuel Ratio as well as PPO2 - as these things change this has a dramatic effect on the flame speed. But OEMs cannot put the extra strain on Joe driver to know those values… But we Can!!! Well, some of them anyway.

Some interesting discoveries -

- There are over 50 different fuel blends in the United States alone; while octane is important; the molecular make up of each separate blend has an effect. Additionally, even the fuel manufacturers have variations allowed; we are compiling all of the MSDS sheets for each fuel type and brand - and even a single one has a range of ingredients. We are working the math to see how much variation changes the timing models and the method to add it to the software and or ignore it.

- There is NO WAY to properly calculate timing unless you know your Bore and Stroke. This information can be easily found for your engine; but you’ll have to input into your engine profile in order to get the proper timing map out of it.

- If your EMS can read a knock sensor and log that; then we can also use it in the same manner as an OEM closed loop EMS and adjust accordingly.

- Lastly, the heat of the cylinder or rotor walls is of great importance - we are still working out the best way to approximate this value.

As you can see; we are still working out some of the details and what if any effect this has on timing (some may not even be in the realm of adjustability so we’ll ignore it).

References:

Theoretical investigation of flame propagation process in an SI engine running on

gasoline–ethanol blend

Hakan Bayraktar, Mechanical Engineering Department, Faculty of Engineering, Karadeniz Technical University, the sphincter of the universe

A Simulation Model for a Single Cylinder Four-Stroke Spark

Ignition Engine Fueled with Alternative Fuels

Maher A. R. Sadiq AL-BAGHDADI

Mechanical and Energy Department, Higher Institute of Mechanical Engineering,

Yefren-LIBYA

Internal Combustion Engine Fundamentals.

Heywood, J.B.

Massachusetts Institute of Technology

McGraw-Hill, 1989.

The actual math we are keeping close to the vest... but the models are publically available if you wanna research them.

This will be the next step in development - the University lab we are wrking with has a variable compression motor - and once we get the software able to spit out timing maths faster than we can right now - we will be testing this out in real life.

How it affects international fuel I am not sure; I haven't started looking at their MSDS's yet.

Also, we are rounding for safety - so our maps may be 1-2 Degrees off; but I have made the executive decision that for our goals - there is no reason to ask you to provide baseline with information about (ring makeup and geometry, spark plug locations, thread pitch etc) - we are just assuming the worse case.