The L-Jet (FEDERAL) chronicles and AFM hacking

[a1fa]

I've posted this on another forum as well, but at this point, I am just doing research and running feasibility studies -- bear with me.

You grow to love L-Jet not because it is simple, but because it was hard work that made it possible to tick and tock when it needs to. Frankly, my understanding of L-Jet is that sole purpose in life is to drive fuel injectors, and it is good ad its job, but not as good as a modern FI system. My only pet peeve with L-Jet is lack of instant diagnostic information, and knowing what that thing is doing, and why is it doing it, and all of this from comfort of the drivers seat. Granted, this information is not needed once the motor is tuned, and frankly, I don't really care for it in a modern car.. If you are a fiddler, but like me, don't like to literally screw things up, L-Jet leaves a lot to be desired. Electronics these days overcome all challenges.

With several days of downtime, while waiting for alternator rebuild, I thought on how to improve some tunebaility on the L-Jet for all under $20.
Believe it or not -- L-Jet, and D-Jet perhaps, could easily be "re-programmed" on the fly -- literally! You could easily build your own tables for certain conditions and make L-Jet respond accordingly -- start, idle, cruising, wot... Imagine being able to set your own desired AFR as you are cruising down the interstate, or going up a hill....

The L-Jet has a very magic backdoor that could be exploited and hacked to achieve desired tune.
_________________
Learn to ride. Ride to learn.

[a1fa]

Sure, besides complete replacement (*Squirt) there are other commercial solutions (Greddy and SplitSec), and a lot of BMW and Miata guys have gone down this path, but there are even easier solutions. L-Jet only takes three (or four) inputs to operate: Engine Speed, Temp Sensor II, AFM, and Temp Sensor I. All of these variables are adjustable on the fly, except the engine speed.
_________________
Learn to ride. Ride to learn.

[a1fa]

I was really hoping to build some excitement before I spill the beans; but here it goes... I am planning on developing an AFM emulator based on the ARDUINO platform specifically for the L-Jet. This could easily be converted to other "digi" systems. The code and all the parts will be released as open source, and my goal is to keep this under $20 in parts.

The L-Jet produces variable output from 1.5 to 7.5v, and it is not linear output; instead it is logarithmic. My understanding is that "digi" systems produce 0-5v in a linear fashion which could make it even easier to convert.

The AFM emulator I am proposing is going to work in a piggy back fashion to the existing AFM hookup.

My plan will be in 5+1 stages:
Stage 1:
Build Hardware
Read AFM Angles and map stock output voltage to build a curve/resistance formula.
The "digi" crowd will probably not have to build a curve, since I believe the variable voltage is linear. Good for you.

Stage 2:
Emulate AFM output based on AFM input on a stock map; no input adjustments
For all of you familiar with Arduino you know that maximum voltage output is 5v, this will be overcome by using boosters, and producing a strong and solid 1.5v-7.5v volts. The digi crowd will not have to use boosters.


Stage 3:
Add o2 readings into the algorithm to adjust and target specific AFR
Exclude o2 readings during AFM positive or negative rate of angle change.

Stage 4:
Add coolant temperature sensor to seek desired AFR at specific engine temperatures

Stage 5:
Remove AAR
Remove CSV
Build cold-start enrichment logic.
Build idle logic and hardware to support it.

Stage 6:
Build barometric fuel adjustment ; if anyone is interested to compensate for altitude adjustment.
_________________
Learn to ride. Ride to learn.

[Spinal Tap]

Sounds cool. You're going down the right road, as the ECU only has a few different inputs it uses to calculate it's "output" of injector duty-cycle. Keep us posted on your progress.

You probably don't need to worry about Stage 6 though, if you're still using the AFM. The AFM will "auto adjust" to altitude due to the way it measures airflow, and any difference/error is compensated for by the ECU via the O2 sensor anyway. Good carburetors do this "auto adjustment" too...

The "Digi" series of Bosch ECU's are based on a very simple Intel MCS-48 series, 8048H CPU. It has 40 pins, 1K (x 8-bits!) of ROM, and 64-bytes of internal memory. This is the same controller that most keyboards have used since the IBM PC came on the market. You can even get the IP for it to embed the MCS-48 in your favorite FPGA. Most computer BIOS's have it built in (I forget why). For a good tutorial, google "grokking mcs-48", and enjoy the Heinlein reference.

It seems like it'd be easier to me, to just read out the ROM from an ECU, reverse engineer it (there only a few dozen instructions, quick PERL script work), and "rebuild" it with a PIC processor and use the extra power of the PIC to run your Check-Engine-Light, or any other instrument you care to think up. Me, I'd add a Knight Rider light to the front of the van!

At least that's what I've been thinking of doing...

Best of luck!
_________________
'85 Westy
'74 MG-B

[Spinal Tap]

Also, there's an old thread, that I don't think went anywhere, about hacking the Digi's.

Read all about it here: http://www.thesamba.com/vw/forum/viewtopic.php?t=382053&postdays=0&postorder=asc&start=0
_________________
'85 Westy
'74 MG-B

[a1fa]

I'm having a hardest time getting the position sensor MLX90324. No one is carrying it in stock, and the factory is refusing to let me buy one directly from them. There is a 24week period wait time on obtaining the sensor from distributors. The hunt continues for the chip. The chip can be replaced with a traditional potentiometer, but I want to build a frictionless system.
_________________
Learn to ride. Ride to learn.

[Spinal Tap]

What you're really looking for is called a rotary position sensor. They come in several flavors: potentiometer, Hall Effect, flux-density, and then there there are rotary encoders, and so on.

What you're looking for should be non-contact, virtually friction-less, and built for the automotive environment, have a 0.5-4.5V output, 5V input, etc...

Something like this:

That is a rotary position sensor, in this case an AN101101, made up in Wisconsin by a company called Cherry. It's designed for "Throttle and valve position sensing", which is exactly what you're doing.

Read all about it here:
http://www.cherrycorp.com/english/sensors/Rotary_Position/an1011.htm

And if you're nice, they'll send you free samples. But make sure you ask for the non-spring version, of you want to spin them with very low torque. the default spring in them is probably good for throttle pedal feel, but is too strong for other purposes.

Good luck!
_________________
'85 Westy
'74 MG-B

[a1fa]

Spoke to soon.. Found a sensor I was looking for. It will tuck nicely under the output pin, while magnet will sit on the flap shaft. I will be giving cherry guys a call too. I'd like to experiment with that too.
_________________
Learn to ride. Ride to learn.

[reluctantartist]

I am curious to what you come up with. I have a California westy with electronic ignition but also a 412 with the same setup as yours which I would be open to modifying so it handles the crappy ethanol we are stuck with better.
_________________
1982 Westy, 1974 412 Variant... Yes, Aircooled's are great! Oh and I do have modern computer controlled vehicles too, but I just don't care about them.

[a1fa]

I need to find the L-Jet schematic. I'd like to experiment with providing L-Jet with 5v reference, and 0-5 variable. I want to see how it will compensate for lack of voltage. Does anyone have a spare L-Jet we could use for bench testing?
_________________
Learn to ride. Ride to learn.

[GoWestyYoungMan]

http://www.thelotusforums.com/forums/topic/41947-b...80-esprit/
The OP researched the hell out of his topic so you may find what reference info you need.