Self learning EFI - Design Thoughts

[ToonRboy]

Interesting discussion. Went 'round on a subject like this once on a different post. I was suggesting the "killer-app" of an aftermarket programmable ECU would have a sort of "wizard" that would prompt for various key variables that would then result in a very good base map. I was promptly told that I knew nothing about tuning, and that all the system really needed was the normal inputs (O2, AIT, coolant temp, rpm, MAP, etc).

'can't argue with that. Just wondering why so many guys with their aftermarket programmable ECU's have vehicles that drive as if they were carburetted (with the wrong jets & no choke).

[slyalfa]

mine drives much much better then the L-jet and that was my first time installing one.
I did the hardware but others do the installs so it was my first install. I found tuning easy to do. but I know the timming could be better. if I had some way to get the PSI in the cylnder I bet I could get a few more MPG. but with a WBO²geting the mix is easy. just tell it what I want get it to start the let the WBO²do the corrections log it while driving and change the MAP to what ever the delta of the WBO²was. in one drive you are 90% there and allready runing much better then the L-jet
do it a few more times and the WBO²will be almost dead on to start with and you have a perfect VE MAP. now for the spark that is the hard one as there is no (cheap)good way to tune that.

[Duk]

I found using my Innovate LM1's data logging (and it's software LogWorks)the most useful way to get good AFRs when tuning my Trust Emanage Ultimate on my Toyota MR2. There is simply too much happening at any given time to make good decisions as to what's going on. I know that on a chassis dyno that can hold a givven engine speed helps to slow it down, but no dyno fan can pump the air a car moving in the real world experiances. Road tuning also costs less. Lots less.
There is still the issue of achieving the best ignition timing to achieve the best power when detonation isn't the problem (torque output decreases with more igntion advance but detonation isn't happening) and the best power without detonation (when detonation is the determining factor).
I had thought of trying to use a load cell on the engine (in the same way that a load cell is used on a dyno brake) by either using a compressive cell to replace an engine mount (on the compression side, the side that the engine places more load onto when it torques over) or use a simple strain gauge that acts as an engine steady.
Get the output from the load cell, amplify it, possibley filter it (to clean up engine vibration) and send that output into a datalogging device like the Innovate. An increase in torque from an improved state of tune should result in an increase in load cell output. Obviously a change of tune that results in a decreasse in torque would show a lower output.
The best part (if it worked) is that using the LogWorks software, you could actually show the load cell output against engine speed and load in the same way that most programable computers show injector open time and ignition advance in a rev/load manner. The other advantage is that it's all done in the real world at much reduced costs than paying for dyno time.

[ToonRboy]

Quote:

mine drives much much better then the L-jet and that was my first time installing one.

[slyalfa]

The cams in the software sim does show that and they also do match the runners. it is as alfa used the same software. I never found out what is different in the old cams and why they do not work as well as the sim shows. it must be somthing in the head? Or after reading some more the published specs for the stock cam might not match the real cam. but I have to say just driving around the S motor does seem to have a lot more down low power.
a real plus with the tall 1st of the Verde.

I think the L-jet works well for what it is. as simple as it is you would think it would run like crap. and it run kinda OK but it is day and night to somthing with smarts. The later motronic on the other hand is a very good system for a batch system. using the flap on the motronic system was a wast when it can work with the hotwire.
I think the L-jet is a good example how you can just blast some gas in anywhere close to the mix and the motor will run. but I would say in openloop mode a step backwards from a carb. but a set forward when in idel or cruze when the NBO² kicks in. but on the flip side the super simple nature of it makes for a reliable system.

[ToonRboy]

Software. In programming, the first thing you learn is "garbage-in garbage-out".

Automobile engines ran on carbs for almost 100years (still dominate certain racing). So, yeah you can go out and spend a grand on a programmable system and then that+ getting it tuned. And if they (or you) you do a good job, you'll make more power than L-Jet. But how will it run on 10* mornings, 100* afternoons, hot starts, or when it's raining cats and dogs? That's where it's hard to beat a good factory system that has tens of 1000's of man-hours of development work to make sure it performs as expected in all conditions.

People take for granted their modern cars and how well behaved they are. Turn key, drive off - doesn't matter if you just unburied it from a snow berm, walked off a golf course at 5200 feet, or a rest stop in the desert (after 10 hours of driving).

There have been loads of fellas out there blaming the flapper L-Jet for choking their engine's stock potential. Then after considerable time, effort, and money, gain a couple of HP at 5k rpm by replacing it.

For instance a stock Subaru Legacy GT 2.5 (4-cam) makes ~177HP today! It's very expensive and difficult to make more usable power with an engine design like ours. Without VVT, & variable intakes, you just can't have both worlds at this displacement. If peak HP and bragging rights is what you want (or a pure race car), then radical cams and aftermarket programmable injection (and engine work) will get you there - just don't expect it to behave well on the steet.

[Tifosi]

Mabe, mabe not.

I've got a relatively mild build on mine (all top end stuff, but stock bottom end, no compression boost) and a programmable EFI system using individual throttle bodies in there to replace the L-jet.

I'm within a couple mpg of what the L-jet had, (and likely could meet or exceed that if I wanted to fart around with it enough) but the power difference is brutally obvious, yet it'll still do closed loop operation and run in stop-n-go traffic all day if the need arises without acting stupid, loading plugs, or requiring any other 'tricks' to keep it going smoothly. (at most there's a tiny 'dead spot' between 1200 and 1500 rpm when the throttle gets whacked quickly due to the massive change in manifold pressure that the ITBs create at that RPM range)

Granted it'll give a little huff of smoke out the pipe the first time or two you run the revs up after puttering around for a length of time, but beyond that, quite streeable and a very noticable performance upgrade.

Murray/msiert's is decently well behaved on the street also from what I've gathered, and I know for sure he's got way more going on than I do.

I've found that it's not so much what you put in for programming, but what you program.

EG: the system I have reccomends tinkering RPM fuel values and leaving MAP values alone to do the dial in.

I found it was MUCH more effective for my purposes to work with the RPM fuel only a very little and do most of the tuning with MAP values.

But then as you said, garbage in, garbage out, and that's absolutely true.

One can screw up the tuning with any number of induction systems, not just the programmable ones. (I've seen some pretty horrific carb tuning in my day)

[Duk]

Quote:

Originally Posted by ToonRboy Automobile engines ran on carbs for almost 100years (still dominate certain racing).

[slyalfa]

I have been very happy with the hot/cold starts. it is rare for it not to start on the first hit. hot cold in the ice etc. I can not say the same for the old l-jet. there were windows where it could get in a no start. unless uning floodclear. and one of the windows was start go 2 blocks to get gas. and it would never start unless I held it down to flood clear. run a bit less or a bit more and it would start but just that distance and no start.

The L-jet has no maps for anything other then the base wich is only a few taps. (aka the trimmed resistors in the flap box ) somthing like 16 or less. I did have to mess with the cold enrichment a few time to get it right but I did it here where it does not get cold. so I only could do in late at night. but it seemed to work fine when I went real cold as it started right up first try. I only tuned the cold start at night above frezzing.

and I drive my car every day so I know how much better it works in the hot in trafic etc then the L-jet did.

I did have to detune it bit on the low just off idle. for the bumper to bumber traffic where you seem to have to slide the clutch for about 50min. I had too much advance at first so it would want to take off. giving it some retared timming for a bit longer made a big difference in the slower then walking driving.
and I know the timming could be improved but it works so well I have not messed with it. but that is the only place to improve would be to find the best timming. I know there are some bad tune out there that only tune WOT and have the rest of the map way off. but with a closed loop system it just comes together.
anything that is off in the map will be correced anyways. but a log will show if there had to be any

[bcal]

I believe that the Ford ECU as used on their V8s is a bit of a learner. Some of the hotrod blokes replace all the ugly manifolding with something nicer, disconnect/reconnect the battery, fire her up, take it for a drive and it adjusts itself to the new intake dynamics.

And there is another good reason for playing around with aftermarket EFI systems that people never seem to mention.....ITS FUN.

[ToonRboy]

There is no doubt in my mind that better/smarter aftermarket programmable fuel injection could be developed. Not in terms of features or ability's - which many already have, but in the programming/tuning interfaces, fuel (spark) table self analysis tools etc... To make it easier for the DIY tuner.

Here's the issue: Many of the actual tuners - know precisely what has to be done to tune the system, within the user interface of the system they are using. But don't necessarily know what's going on behind that serial connection. As for the programmers, they can build an excellent system - but the user interface (for the layman) is often left wanting. Mostly because engineers are like programmers are like scientists. Not like everyone else.

If I may draw a parallel: We all know how to use our computer, but that doesn't mean that we know, or need to know the inner workings or are able to affect change to it (other than hosing it up). So essentially, what I am talking about would be similar to the leap from Windows 3.11 to Windows '95 in terms of Plug-n-Play. -If anyone recalls, prior to Windows '95 every device had to be manually configured to work properly...

[Duk]

Another example would have to be electronic boost controllers. For something as relatively simple as boost control of a turbo charged engine. It took the engineers who designed and programmed the micro-controllers, quite some time to make them as smooth and as consistant as a purely mechanical set up while actually reaping the benefits of closed loop electronic control.

[slyalfa]

I have yet to try the electronic boost controller as I have not put a turbo on anything yet. but I did know that the firmware guys seemed to be messing with it for a while. but also a lot of the stuff is also drag race stuff that I do not think I would ever use on my street car. the anti-lag , the reduced power for low gears etc.
but the boost controller is easy in a way as all the info needed is there it is just geting the code right.
but the spark control is hard as the info needed is not there. so guessing based on other things can be hard. if you have a good way to hear a det you. you can back off but you can not just keep on going untill det. there are many time where you will not det and you can advance too far. there was some talk about just using the spinup time as you can see each power stroke. There are some ECU that use that to tell when a missfire happens. but that seem to not be a good way after a lot of talk. As far as I know only way is to get the PSI of the cylnder. And anything other then that will be a bad guess. There might be a time where every motor has one future? No one would have guessed that we would have a O² in every car back in the 70' or even the part of the 80'. so some sort of Peak PSI system that is cheap might pop up and be common. ION is kinda that and it might improve? but I think that is the last puzzle part. everything else is there. get that last part and I think everything will fall into place.

[Duk]

I'll give this thread a little kick up the bum .

Here in Australia, about half the after market ECU producers allow you to download their tuninig software for free. I highly reccomend that anybody looking to instal an aftermarket ECU get their hands on the tuning software of the intended ECU. This will allow familiarisation and lets you see both potential and limitations of your chosen ECU.
Correction maps and data logging are the big thing (in my opinion) when it comes to completing your modified car that gets used in the real world, rather than a race track special, or (worse