Idiot questions from lazy newby

Jon V. jon at
Tue Aug 24 04:55:07 GMT 1999


I've been trying to find the answers to these questions myself scanning
through the list archive, but frankly there is just too much there for me
to read, and if there is a FAQ I'm too wasted to find it... so I'm going
to kill everyone's time with my silly questions. Please, feel free to
flame me... but if you could throw in at least a token factoid. I 
appreciate it! :-)

What I'm doing...

I have an old "performance" car (early '70s, it is really an on POS, but 
hey, you work with what you've got..) that I play around with. The car
has an OK engine, but it is early '70s technology... the most advanced
piece of electronics on the thing is... well, the HET point replacement I
installed in a BFE parking lot so I could make it home after the 
distributor started filling with oil every 20 miles.

It is a 2L engine and they are pulling about 140HP out of it, but
it is capable of a lot more. (Re-carbing alone is said by the engine
manufacturer to add about 20HP.)

So I am going to "performance enhance" the engine.

Actually, the engine has some life left in it and I like driving the car,
so I went ahead and bought a second engine (same type) and I'm going to
screw *it* up... then when I finally break a timing belt or whatever I'll
have a new engine to throw in.

Which brings me to the list-relevant parts. I think it'd be pretty snazzy
to implement a fuel injection and ignition controller for the old
rattletrap... but before I go and get myself sold on the idea, I need a
reality check.

I had intended to ask a bunch of leading questions at this point, filled
with all sorts of "gems" of knowledge to show that I'm really in with the 
EFI scene, but I decided that since I don't know a my ASS from my EGO and
I get weird shrinking feelings when people say things like 'Lambda' I
figure I'll take a different approach.

Assume that I have an IQ at least 20% higher than this email indicates...
no, that's too open... assume that I have an IQ of at least 100. Assume
that I know how to do something besides waste everyone's time... say maybe
some software development (high level languages, some assembly) experience
and a little bit of HW (enough to run screaming), and that I know which
end of a wrench the little spinny things hook up to.

Assume that I have two FI VW engines just sitting around waiting to
sacrifice their existence to the gods of experimentation. A selection of
spare computers ranging from 386/33Mhz to Cyrix 6x86 or Alpha 166Mhz
that can be reassigned. Familiarity+ with Linux. A selection of
basic electronic tools (VOM, Oscilliscope, device programmer, scientific
calculator) and an occasional window of access to even better tools (aka
real electrical engineers that owe me favors). 

Assume a fairly modest performance envelope.  Redline ~7K, maybe a little
higher, naturally aspirated, <9:1 compression. I expect 160HP, if I'm
lucky 180HP. I'll need to do a distributorless ignition at the same time.

Assume that I am what I sound like, a dirt poor crazyperson.

Oh, and if it is no bother, assume that I have the ultimate source of PIC
16C65A OTP PLCC parts... people I work with chew through them like popcorn
and the company policy is "engineering begets engineering." IOW they
encourage projects.... up to and including the time the one of the 
EEs fitted flashing LEDs on all the snails he could find around the
building... but I digress.

Given all of those assumptions, what is the slickest way to get started?

I don't like duplicating effort (but I don't want to buy... logically
inconsistent) unless I need to... I certainly don't want to need to
duplicate basic research.

Should I find a junkyard GM computer and work on reprogramming it? Should
I jimmy up some IO for a PC motherboard and start with that? My OS
preference is Linux these days, but I have worked long enough to know that
preference don't mean jack when there is a job to do.

What *I'd* like to do is set up a system that uses the PICs as IO
controllers... say one PIC per sensor, with all the code to refine the
sensor data and bounce it (on a serial bus) to a central processor. IOW
there would be a HET sensor on a cam pulley connected to a PIC which
abstracts the data into RPM and bounce it out each time Cly #1 is TDC...
Set up a sensor/PIC pair for everything (EGO, MAF, CP, TP, RPM, MAT,
EGT, ECT, TLA, TLA, LFLA, ETC, ETC.) which just transmits its data based
on some sensor specific criteria (the RPM might be sent each time the
engine is TDC, the EGO every 10th of a second, TPS whenever it changes) to
a central computer which can log the data, and control the injectors,
ignition, etc.

The extended idea is that because everything is bussed together you could
have actuator/PIC pairs (e.g. an injector) that act a lot like the
sensors. For example, the CPU configures each injector with timing
information (MS from TDC to open, MS from TDC to close) and a common
sync pulse (engine TDC) is constantly issued. At that point if the CPU got
behind for a few revs the engine would just sort of "hold" until things
were caught up. 

The stretched idea is that everything is really bussed together, and the
sensors just share their data, and the actuators just take the data and
act on it... and the CPU just oversees the process and makes sure that
everybody is talking.

See, now that's exactly what I said I wasn't going to do... damn these 
fingers. All of that is, as you are saying to yourself, sky high
dreaming, and I agree, I should pull my head out, but it has been lodged
in there for so long that my spine has set. 

Whatever I do, I'd like it to be educational, but not in the bad way that 
people mean when they say "it was educational". I want to be educational
AND work. It doesn't have to work the first time, but it needs to do
better than the carbs someday. I don't really want to buy a ready built
solution... both because it seems like a waste of a good hobby project,
and because if this works I've got several other engines ranging from a
700+ CID industrial/truck engine to 1100CC motorcycle engine that I'd like
to look at too.

So, can anyone politely ignore the utopian "all the little computers 
working together" part and tell me how much a project like this is going
to cost, and will it work better than carbs? FWIW, recarbing with the
"performance" carbs would probably cost about $900.


The eagle may soar, but the weasel never gets sucked into a jet engine.

More information about the Diy_efi mailing list