How to tune your engine (EFI) - Alfa Romeo Bulletin Board & Forums
Reply
 
LinkBack Thread Tools Display Modes
post #1 of 45 (permalink) Old 03-12-2009, 07:55 AM Thread Starter
Senior Member
Platinum Subscriber
 
danyl's Avatar
 
Join Date: Jun 2007
Location: Savannah, GA
Posts: 305
How to tune your engine (EFI)

I'm now knee deep in the learning curve for tuning my Megasquirt and have gained some very valuable knowledge.

Now that I'm actually tuning my Megasquirt myself, I've had many revelations about "tuning theory" which I'd like to share.

First, anyone who states: "Just set your AFR for stoichiometric" has clearly NEVER tuned a car.

I've seen numerous people who believe this and write about it on the various web sites, but let me share some details.

I'll appeal to the carburetor guys first. You all know that once you set your idle mixture screws to approximately 2 turns out and get the car to temperature you then adjust the idle mixture for maximum vaccum (or minimum MAP). And of course the end result is a RICH idle! Not an AFR of 14.7!

Second, there's the problem of heat build up. Exhaust Gas Temperatures (EGT) should be in the range of 1300 to 1500 degrees. At WOT and AFR (Air/Fuel Ratio) can easily exceed this, and therefore requires richer ratios (more like 12.5)

At low RPM and WOT (TPS 100%) one wants various mixtures for better performance and this requires non-stoichiometric mixtures as well.

After looking at well over two dozen ".msq" files for Megasquirt from "success" stories, NONE of them had more than 40% of their fuel maps in the stoichiometric range.

Which brings me to a point I once made in a prior thread: I predict that the future of EFI OEM and aftermarket systems will include built-in accelerometers, EGT sensors as well as the standard Wide Band Oxygen sensor. This will allow the computer to go well beyond closed-loop mode! It will allow the computer to actually adjust the AFR table!

To make this thread interesting I'll explain VE and AFR tables for those less familiar.

The computer (ECU) needs some sort of recipe (fuel map) to tell it how much fuel to add for EVERY combination of RPM, vacuum (or MAP), and throttle position encountered. Since this would be an incredibly large three dimensional speadsheet, the computer usually has an abridged map and extrapolates the amount of fuel to inject when the engine is in-between particular values.

The table (spreadsheet) is in the computer's memory and runs instantly upon power-up of the car.

This table (which tells how much fuel to squirt) is referred to as the VE table which is short for Volumetric Effiency. I've dumbed this down a little, since really the VE table is then run through an algorithm that calculates the amount of fuel to inject, but for practical purposes think of the VE table as the "amount" of fuel to inject.

As your engine ages, gets dirty, bad gas, different altitudes, and just changes with driving habits, it requires adjustments to the VE table to run at perfection. This is where the Oxygen Sensor and AFR table come into play.

The O2 sensor and AFR table are not necessary for engine function. Many early fuel injected cars did not have them. Their computers were programmed with a VE table from the factory and used this table for all driving during it's life time.

The O2 sensor and AFR are critical for a do-it-yourself mechanic to program his own EFI as well as any one interested in fine tuning or performance tuning their cars.

The O2 sensor and AFR (air/fuel ratio) table run in a "closed-loop" mode. The readings from the oxygen sensor tell the computer what the ratio of oxygen to unburnt fuel is within the exhaust. This feedback is then used in a calculation by the computer (ECU) to make on-the-fly adjustments to the VE table. The VE table isn't really reprogramed in a factory unit, rather the amount of fuel is modified on a squirt-by-squirt basis.

It is this AFR table which I am arguing about. Many seemingly experienced individuals seem to think that the AFR table (which is ALSO pre-programed by the factory in the computer's ROM) should be set to an air/fuel ratio known as stoichiometric. This is the perfect ratio of air to fuel and for normal gasoline, and is 14.7 molecules of air to one molecule of fuel.

Only about 50%-60% of the cells (or bins) within the AFR speadsheet on a properly tuned car will end up at 14.7 however. The rest will be leaner or richer based upon the needs of that particular engine.

How does the do-it-yourself tuner know what these values should be? Indeed, he may start with a perfect 14.7 ratio for all the values, but then he/she needs some way to know how much to change each target value.

Factories have extensive test equipment like dynometers and test tracks to fine tune thier AFR table before the computers are programed and shipped to the customers, but without a more extensive "closed-loop" system built into the automobile, at 100,000 miles, the original AFR table may no longer apply to the needs of the engine.

I expect many luxurary cars over the next few years will start advertising artificial intelligence tuning systems that uses data from built-in acceleromters (G-force meters), and engine temperature sensors, as well as pitch and lateral roll meters, to fine tune the driving experience.

Danyl
1987 Spider Veloce
danyl is offline  
Sponsored Links
Advertisement
 
post #2 of 45 (permalink) Old 03-12-2009, 01:05 PM
Duk
Registered User
 
Duk's Avatar
 
Join Date: Aug 2007
Location: South Australia
Posts: 1,733
Garage
Exclamation

Quote:
Originally Posted by danyl View Post
First, anyone who states: "Just set your AFR for stoichiometric" has clearly NEVER tuned a car.
I've never heard or read that. Anybody who says such a thing should be avoided . Having said that, there are/have been factory produced cars that do run constant 14.7:1 AFRs, even turbo'd 1s.

Quote:
Originally Posted by danyl View Post
I'll appeal to the carburetor guys first. You all know that once you set your idle mixture screws to approximately 2 turns out and get the car to temperature you then adjust the idle mixture for maximum vaccum (or minimum MAP). And of course the end result is a RICH idle! Not an AFR of 14.7!
Lots of engines prefer to idle with a rich mixture, and is largely effected by camshaft overlap. Sequential injection systems typically allow leaner idle mixtures as there is less tendancy for the fuel to fall out of the minor amounts of air flowing past the valves.

Quote:
Originally Posted by danyl View Post
Second, there's the problem of heat build up. Exhaust Gas Temperatures (EGT) should be in the range of 1300 to 1500 degrees. At WOT and AFR (Air/Fuel Ratio) can easily exceed this, and therefore requires richer ratios (more like 12.5)
EGTs are effected by more than just AFRs. Ignition timing has a huge effect, less advance = higher EGTs.

Quote:
Originally Posted by danyl View Post
Which brings me to a point I once made in a prior thread: I predict that the future of EFI OEM and aftermarket systems will include built-in accelerometers, EGT sensors as well as the standard Wide Band Oxygen sensor. This will allow the computer to go well beyond closed-loop mode! It will allow the computer to actually adjust the AFR table!
Autotuning abillities have in use by the various aftermarket computer makers here in Australia for some years now. Some of them are apparently very, very good.

Factory systems of the last 15-20 years are far more advanced than a lot of people realise and comparing a simple system like the megasquirt to just about any factory system is very ambitious.
Simply quoting (and I know you haven't done this) maximum power an ENGINE made with brand xxx computer is really stupid. Tuning for max power is the easy part, and, as you have alluded to, it's all the parts between an idling engine and max power that is the fun part.
Please, don't get me wrong, I'm not having a go at you, and I'm deffinately not some 'expert', but I have studied (via the 'net) the workings of the Nissan ECCS type engine management systems and read as much as I can from as many creditable sources as I can find.
Some exstra reading:
Stuff about tuning
More stuff about tuning
ECU Tuning ('tho about Nissan ECUs, the complexity for an old system is quite amazing)
Innovate Motorsports Forum - Powered by vBulletin
http://www.motorator.com/videos/299 When your really know what you're doing
I also reccomend Greg Gordons book on Supercharging.

Slowly Progressing Vortech Supercharged 1990 Alfa Romeo 75 Potenziata. Out of Action Twin-Charged 1988 AW11 MR2. Current Daily Driver, The Glorified Taxi 2006 BF FPV F6 Typhoon.

Last edited by Duk; 03-12-2009 at 01:08 PM.
Duk is offline  
post #3 of 45 (permalink) Old 03-12-2009, 01:13 PM
Registered User
 
turbolarespider's Avatar
 
Join Date: Apr 2003
Posts: 1,837
Interesting theory.

Couple of notes- the prediction of hardware? Go out on a limb sometime. Well, sort of. We've (F) have had WB sensors in production cars for a couple of years now, and probably will be across all lines by the end of the decade. But not for obvious reasons. We've had NB O2 sensors in production for 30 years now- for the most part, all for emissions. Accelerometers are part of most cars for dynamic control, although they do virtually nothing to how the engine runs directly. For fuel, you really only need a O2 sensor and some way to measure speed and load.

For the map of a/f vs. speed/load: Except for the start, all points are actually stoich. But, we are allowed a few exceptions- one for a sustained, high, demand from the driver; second- more commonly for component protection. There's a HUGE nebulous that covers what componenets can be protected, BTW- any are ok, as long as you proove it. (valves, ports, manifolds, catalysts, sensors, etc)

Still, since there are no EGT sensors that are reliable enough, you won't see them on gas cars anytime soon. No real point to them.

Back to the idea of the VE tables- don't get the idea that it's the grand way that OEM's calculate the desired fuel. Not even close. It's a lot more to it than that, since not only is fuel calculated (and it has to be very close all the time), but spark also needs to be calculated. And you need to factor in EGR and/or cam timing accurately. And purge flow.

And when/how to shift.

All without the driver ever knowing it.

Needless to say, MS is a very simplistic system, that actually can be even more simple, but it would almost require access to a dyno. It's all the other details that makes it very, very hard. That's why the claims of 1M test miles (equivallent) are always made....

But interesting point of view, I must say. entertaining read.
turbolarespider is offline  
Sponsored Links
Advertisement
 
post #4 of 45 (permalink) Old 03-12-2009, 01:15 PM
Registered User
 
turbolarespider's Avatar
 
Join Date: Apr 2003
Posts: 1,837
Quote:
Originally Posted by Duk View Post
. Tuning for max power is the easy part
Oh, so very true.

And Oh so easy to manipulate to make people happy....

E-
turbolarespider is offline  
post #5 of 45 (permalink) Old 03-12-2009, 02:37 PM
Registered User
 
Join Date: May 2004
Posts: 4,075
For what it's worth, I generally tune for an air fuel ratio of 12.5:1 for full throttle or high manifold pressure settings. I go for 14.7:1 when the engine is under more than 3" of vacuum. The idle mixture usually has to be quite a bit richer than stoichiometric. I am really not sure why, but all engines I have worked with seem to like that. I am sure Eric could explain it. Then once all that's done I set up the acceleration enrichment (like the accelerator pump on a carb) and ignition timing.

The fuel seems to be the easy part. With an NGK wide band sensor and data logging it's pretty easy. The ignition however requires a dyno to tune for max power because I just can't seem to feel 10hp one way or the other.

Greg Gordon,
HI Performance Store, Inc
Greg Gordon is offline  
post #6 of 45 (permalink) Old 03-12-2009, 11:44 PM
Duk
Registered User
 
Duk's Avatar
 
Join Date: Aug 2007
Location: South Australia
Posts: 1,733
Garage
Quote:
Originally Posted by Greg Gordon View Post
The fuel seems to be the easy part. With an NGK wide band sensor and data logging it's pretty easy. The ignition however requires a dyno to tune for max power because I just can't seem to feel 10hp one way or the other.

Greg Gordon,
HI Performance Store, Inc
I look forward to the day when there is some sort of feedback system that is available to the enthusiest that can genuinely measure (and log) the actual output of an engine in real world conditions.

I did think about a load cell attached to the engine as either an actual engine mount or as a torque arm. Amplify, filter and log against load and engine speed.
The other idea was to use a pneumatic type of engine steady strut with a pressure sensor attatched.
Either method should (IMO ) show changes in torque output in real world conditions and be able to show small enough changes to provide meaningful feedback for tuning, especially tuning ignition maps.

Slowly Progressing Vortech Supercharged 1990 Alfa Romeo 75 Potenziata. Out of Action Twin-Charged 1988 AW11 MR2. Current Daily Driver, The Glorified Taxi 2006 BF FPV F6 Typhoon.
Duk is offline  
post #7 of 45 (permalink) Old 03-13-2009, 04:23 AM Thread Starter
Senior Member
Platinum Subscriber
 
danyl's Avatar
 
Join Date: Jun 2007
Location: Savannah, GA
Posts: 305
My test system

My first stab at this:

My Megasquirt, using the Megatune Generated Table
Adjust the Idle portion of the graph using the MAP sensor reading
Set upper RPM/WOT at 12.5

Use the LMA-3 (Aux-Box) with the following components added:
RPM input from inductive clamp
EGT sensor input
Speed sensor mounted on the driveshaft
MAP (redundant, but necessary for LogWorks)
Forward Acceleration

The LogWorks software can use the speed sensor input with car weight, and accel force, to calculate HP and torque

I'll run at steady speeds in all gears and RPM ranges on a long straight road.
I'll log each run.
I'll then make minor adjustments to the VE table in each range and repeat the runs.
I'll shoot for maximum HP, acceleration, torque in each range, without exceeding EGT limits or extremes of AFR.

This however will be a considerable amount of manual labor and time.

What if there was a computer program that could do the same over time!

Megasquirt and many other aftermarket systems autotune the VE table only to achieve the AFR values. What if autotune would periodically make minor "TEST" adjustments and then review the effect they had on HP and torque, keeping those that showed a positive improvement!?!

Thoughts?

Danyl
1987 Spider Veloce
danyl is offline  
post #8 of 45 (permalink) Old 03-13-2009, 05:23 AM
Duk
Registered User
 
Duk's Avatar
 
Join Date: Aug 2007
Location: South Australia
Posts: 1,733
Garage
Quote:
Originally Posted by danyl View Post
What if there was a computer program that could do the same over time!

Megasquirt and many other aftermarket systems autotune the VE table only to achieve the AFR values. What if autotune would periodically make minor "TEST" adjustments and then review the effect they had on HP and torque, keeping those that showed a positive improvement!?!

Thoughts?
Nissan and more than likely all the others have their short and long term fuel trims that are learned from the feedback from the EGO sensor and applied as a correction percentage to the fuel maps. This isn't learned from full throttle sensor readings, where the narrow band EGO sensor is innacurate, but learned in around the stoich. areas. I'm referring to the older Nissan ECU, I'm sure current day set ups are similar in princilpe but much more effective.

For an engine computer to be able to applied corrections to be constantly achieving best performance in any driving condition, a means of engine torque output feedback would be required.
Until a means of effectively measuring the very end result (engine generated torque) can be implimented into whole shooting match, then I believe things will pretty much stay the way they are.

Maybe someone like Innovate are working on something similar in concept.
Mybe I should paten the idea and sell it all to you guys and retire next week

Slowly Progressing Vortech Supercharged 1990 Alfa Romeo 75 Potenziata. Out of Action Twin-Charged 1988 AW11 MR2. Current Daily Driver, The Glorified Taxi 2006 BF FPV F6 Typhoon.
Duk is offline  
post #9 of 45 (permalink) Old 03-13-2009, 06:12 AM
Registered User
 
turbolarespider's Avatar
 
Join Date: Apr 2003
Posts: 1,837
Quote:
Originally Posted by Duk View Post
For an engine computer to be able to applied corrections to be constantly achieving best performance in any driving condition, a means of engine torque output feedback would be required.
Until a means of effectively measuring the very end result (engine generated torque) can be implimented into whole shooting match, then I believe things will pretty much stay the way they are.
Realistically, for road drivers, and the calibrations for said road drivers from OEM's, it's 99% never about optimum performance.

1st choice- optimum emissions- so that you can keep the emissions systems cheap.
2nd choice- optimum druability- you'll always find that the vehicle is calibrated rich to keep the components happy.

Optimum performance is only important for a few seconds on a fresh acceleration, after that, component protection takes over.

Also, for the most part, the knock sensor does a good job keeping the spark more optimum than not. So much time is spent finding the best spark values, and then finding where the car will run on real fuel... sometimes the spark is the same, most of the time it's not- but a knock sensor will keep it closer to the optimum values. For performance, spark trim (2-5 deg) is far more important than getting the fuel perfect (12-13:1 afr on real gasoline).
turbolarespider is offline  
post #10 of 45 (permalink) Old 03-13-2009, 09:14 AM Thread Starter
Senior Member
Platinum Subscriber
 
danyl's Avatar
 
Join Date: Jun 2007
Location: Savannah, GA
Posts: 305
Duk,

I think you missed part of my setup. The LMA-3 uses a speed sensor on the driveshaft, an accelerometer, and the known weight of the car and driver to calculate HP and torque. That's my exact point.

Essentially it is NOW within financial reasonability to have cars with built in dynometers. The computers can now go BEYOND EGO to control the engine.

There is also a product called G-Tech which mounts on your dash which, uses the weight of the car and an RPM signal to calculate Torque and HP. I don't think the RPM signal is as reliable however and it doesn't sense O2 or MAP.

The technology IS out there in pieces and parts. I want to see it all together in one nice user friendly and affordable package that Monitors & Tunes within the SAME system.

I'll go as far as saying that there might even be a person or two on this BBS that has the programming skills to program Megasquirt to query the LMA-3 device by Innovate and import the data and make the calculations to AUTOTUNE based upon torque and HP!

Danyl
1987 Spider Veloce
danyl is offline  
post #11 of 45 (permalink) Old 03-13-2009, 07:01 PM
Registered User
 
Join Date: Mar 2005
Location: Milpitas CA
Posts: 2,984
I run closed loop on mine(VEMS). I have target lambda setting for each cell, so at idle and low load I run lambda of 1 or even 1.1. Under load you need to run richer to cool it down some to keep it from detting. so I run more like .85
At for tuning it can go very fast once you get it to run just let it run closed loop and log it and go for a drive.
then just adjust the VE by how much the closed loop had to in the log for that cell.
one drive and it is 85% tuned a few more and you get a very good tune. Now getting the right spark table it a lot harder as it is not closed loop. The knock can help a tiny bit. But a in cylinder PSI is needed to tune it right.
If you have a CAT, running any where outside of lambda will melt it if prolonged.
WOT is OK at you will only be there for a short time. But runing rich at idle is a good recipe for a melted CAT. Too lean and the cat will not work ether but it is not clear if that will kill it.
getting acc-enrichment can be a bit tricky and if that is wrong it will drive very bad. like having a busted powerpump in a carb.

1987 black Milano Verde, 1987 (made in 1986)blue $200 gold with ABS and LSD
1972 White spider 2000 Veloce
slyalfa is offline  
post #12 of 45 (permalink) Old 03-14-2009, 02:35 AM
Duk
Registered User
 
Duk's Avatar
 
Join Date: Aug 2007
Location: South Australia
Posts: 1,733
Garage
Quote:
Originally Posted by danyl View Post
Duk,

I think you missed part of my setup. The LMA-3 uses a speed sensor on the driveshaft, an accelerometer, and the known weight of the car and driver to calculate HP and torque. That's my exact point.
I did read it and I too have an LM1 but only the RPM converter, not the Aux. box. The reason why I see more use in a load cell added to an engine to measure torque is so that every area within the map can be tuned to get the absolute best results. Tuning a cruise ignition timing map points can return decent fuel ecconomy impruvements and throttle response but it's easy to go past the point of peak torque for that rev/load point and actually loose torque without reaching the point of detonation, especially with an engine with lowish static compression ratio.
Kinda hard to do an acceleration test at part throttle and get consistant results. You could do it, but..........

Danyl, you'd also be familiar with Innovates AFR plotting and target AFR software. I liked the idea of instead of filling the grid with measured AFRs, the grid is filled with the output of a load cell. You could literally achieve a rev/load plot of the torque output of your engine in the same scale as the rev/load plots of your tuning maps (well I think it's )
That way you should be able to see, cell by cell, if you are going forward or backward with your tuning, and not just at full throttle.

Slowly Progressing Vortech Supercharged 1990 Alfa Romeo 75 Potenziata. Out of Action Twin-Charged 1988 AW11 MR2. Current Daily Driver, The Glorified Taxi 2006 BF FPV F6 Typhoon.
Duk is offline  
post #13 of 45 (permalink) Old 03-14-2009, 05:18 PM Thread Starter
Senior Member
Platinum Subscriber
 
danyl's Avatar
 
Join Date: Jun 2007
Location: Savannah, GA
Posts: 305
Load Cell

Very good point about the load cell (Duk). I've been thinking about it since you mentioned it. It might be possible to build some sort of lever arm that could be added to the T-bar or differential itself, that could measure the instant torque upon changes of power transfer to the rear wheels.

It would of course, be limited to just the inital transfer of power and there would be no real means of getting a true ft-lb equivalent, but the raw data could be compared against itself with each adjustment in the VE table. Any increase in the load cells value would be considered an improvement in torque.

Danyl
1987 Spider Veloce
danyl is offline  
post #14 of 45 (permalink) Old 03-16-2009, 04:38 PM
Registered User
 
msiert's Avatar
 
Join Date: Aug 2004
Location: Nebraska
Posts: 3,908
Danyl,

I have been going through a learning curve on this stuff too and after seeing the tables that were set up in my ecu I admire you for doing this from scratch ..........your own self. I had my Autronic ECU installed by an Autronic dealer, the engine builder and then tuned on a dyno.

Wanted to say...............your dead on about needing to adjust the tune for engine wear. For me the Autronic SM2 has "Autotuning" which really simplifys this type of tuning.

I just purchased a Innovate LM-1 with an Autronic interface cable so I can tie the A/F meter into the Autotuning function. For now I just want to use the LM-1 to data log the A/F and compare it to the Target A/F Table. If I see that I'm getting off then I can run the Autotune function to bring it back in line................the autotuning function does all the math and makes all the necessary changes to bring the tune back in line.

If you think it would be helpful, down load the Autronic SMC/SMD/SM2 software in the link below and I will email you my Cal file. It has my A/F target table, base fuel delievery, Ignition timing ect, ect.

Autronic - Fuel Injection Systems, ECU

If this would be helpful or just interesting to anyone else just email me [email protected] and I'll send the Cal file to you.

Darkside Member

Last edited by msiert; 03-16-2009 at 09:55 PM.
msiert is offline  
post #15 of 45 (permalink) Old 03-16-2009, 11:32 PM
Registered User
 
bcal's Avatar
 
Join Date: Jun 2003
Location: sydney australia
Posts: 272
If you have a WBo2 sensor and you are realistic about your AFR tables you can get a pretty decent tune using the VE calculator in Megalogviewer.

I tried using autotune in megatune but found that I couldn't get decent results.

It may take a couple of passes but you can save yourself a lot of time and fuel.
BTW I found that aiming for stoichiometric anywhere under 2000 rpm on my nord engine was next to useless. And don't expect much less that 12.5:1 at idle.

These really are dirty little engines, which is probably why they're no longer made.

Bye for now.....
Brett.
bcal is offline  
Reply

Quick Reply
Message:
Options

Register Now



In order to be able to post messages on the Alfa Romeo Bulletin Board & Forums forums, you must first register.
Please enter your desired user name, your email address and other required details in the form below.

User Name:
Password
Please enter a password for your user account. Note that passwords are case-sensitive.

Password:


Confirm Password:
Email Address
Please enter a valid email address for yourself.

Email Address:
OR

Log-in










Thread Tools
Show Printable Version Show Printable Version
Email this Page Email this Page
Display Modes
Linear Mode Linear Mode



Posting Rules  
You may post new threads
You may post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off

 
For the best viewing experience please update your browser to Google Chrome