Introduction from Perth, Western Australia

[Mike (Perth, WA)]

At 04:03 PM 6/03/2000 +0800, you wrote:
>Goodness - that makes at least three of us within shouting range.

Yeah hi Bernd,  surprised to found that out after I posted :)
Saw the first ref back some time back but never had enough time to follow
it up - I'm committed (?) to getting something bit mroe serious done to my
car this year... Hence building a twin T3 exh manifold, need more MIG
practice though,

>Could this be growing out of desparation due to isolation? :-)

hehe - Possibly, Perth is after all the most isolated capital city in the
world (At first I thought you were from Perth Scotland, been there once).

>>My main interest is to install my own EFI to replace the existing bosch
>>unit but in the same housing so its a drop in replacement - except for a
>
>That's the reason I want to do data acquisition first.

mmm Yeah - should be possible to 'hang off' a simple 4 channel analog and
perhaps 4 channel digital with time ref for a few hundred. Trouble is the
damn connectors can only be sourced from wreckers :( So I'd use those tiny
probe clips (open ECU, attach, close, run cables out to pluggable block),
that means minimal upset of existing circuit.

I remember the days I'd be in the back seat with my HP analyser, 12v inverter,
buddy in front seat - I'd say 'gun it' and me hastily pressing buttons on
the woeful user interface on the HP whilst seeing the trace jump around ~`:o

>>At moment I'm curious of best conversion resolution for hot-wire AFM, I've
>>seen the 8051 projects with 8 bit resolution, I thought 10 bits would be
>>more appropriate for multi-point as a minimum - can use an SPI 12bit and
>>wonder if there are any comments on best AFM resolution. Oh incidentally,
>>the output from my bosch (exisiting) AFM is log. ie. Quiescent at around
>>1.6v then rises logarithmically to around 5v at maximum load - anyone
>>have any observations on this type of AFM and practicality of using log
>>to linear i/p conditioning with their EFI front end ?
>
>Logarithmic is good because it gives you "better resolution" at, and
>near idle speeds. You detect the significant changes more accurately.
>Don't fight it - go with the flow. :-)

Oh yes - understand that - especially at light cruise I'm on somewhere
like 10-15% throttle and quite happy with the bosch cpu for smoothness
but can't do anything to it to improve injectors etc. I'm interested
in the signal conditioning electronics, the Bosch ECU has a large PCB
area (and a proprietary hybrid) devoted to the AFM output before it goes
into a 6803 8bit micro. If a regular a/d were used then this hybrid could
be redundant - I'm therefore curious as to why boscj did it that way and
hence my query about the 8bit - if that shed some light on the rationale ?

>You can set up a table to aid in the conversion of AFM output to
>massflow. Interpolation can be done fairly quickly so not all
>possible values need to be stored... otherwise with 10 bits you'd
>need 1024 pairs of values -- do a hash lookup, dividing by say 32
>will give you only 32 pairs of values. The log-linear table will by
>its nature provide better quality numbers at the low end of the
>scale.

Yes understood, perhaps I should have been more specific in my query,
I'm curious if 8bits were sufficient resolution for the throttle body
system as against using say a 12bit for multipoint (with perhaps a
little more flow and precision) ?

A lookup table with 12 bits would be OK but I suppose I'm curious what
the conventional wisdom is here - do most late 80's,early 90's ecu's use
10, 12 bits and/or did they use opamps to convert to linear then a/d or
simply a/d the log o/p of the AFM ?

>Note that the airflow is unsteady in practice anyway; the meter
>output should be taken with a grain of salt. Use the feedback from
>the O2 sensor to close the control loop.

Yeah sure, I was thinking some correction with MAP too. Just how unsteady
is that output - as a result of positive feedback effects on normal turbulence
at the inlet piping - on my VL its woeful - theres a real bad sprung pipe
behind the AFM and huge pipe discontinuity in front :( ?

It would be ideal to have it run in feedback loop with a derivable signal
from the dyno - that would be nice - I guess its been done before though,
ie. Dyno feedback to ECU 'locks in' range for max power etc :) ?

>I assume that you've read all the available Bosch documentation on
>the metering process. Do you have (access to) a flow-bench?

Yes, have docs from 1982 but don't have a flow bench - not too much involved
though, was going to put one together in next few weeks in my sleepout, hope
it won't wake the neighbours, now looking for a single phase motor and
centrifugal air pump <sigh>

>The Bosch successor to L-Jetronic (flap-type) was effectively
>Digifant (aka L2-Jetronic). It uses the "same" flow meter. ADC is
>8-bit as far as I can tell. Interestingly, the flow meter output is
>supposedly linear with flow-rate to simplify the task of the ECU.

mmm OK - I thought the 'L' in 'L-jetronic' generally referred to air metering,
hence this included platinum hot wire - true mas air flow etc but with general
log output ?

>I'm not planning to reverse-engineer the puppy because I'm
>restricting my exposure to published materials in the public domain.

Fair enough, what-ever's the most practical I suppose.

Although 8bit conversion across whole range is real elegant, I like the idea
of precision and have some experience with kalman filters, these might end
up with a 10bit corrected AFM from a 12bit raw etc... ie Might drop the low
2 bits into a correction formula whilst using top 8 bits as raw table
lookup...

You see, to fit in the existing ECU enclosure I'd like to keep the hardware
count as low as possible - if I can use the onchip flash instead of external
bus devices then I have a smaller (simpler) PCB, more IO lines etc etc...

At some point I'd like to run CNG - given its only $12 per 200K as against
$60 per 400K for petrol and I could fill up at home with a compressor off
the gas line - see Sunday times Auto section - 5th March... Interesting.

Anyone done gas EFI on this group or know of any links to that ?


Rgds ~`:o)

Mike Massen         Trading as "Network Power Systems" and "Network Computers"
Perth, Western Australia  Ph +61 8 9444 8961  Fx +618 9264 8229 (fax -> email)
Products/Personal/Client web area at http://www.wantree.com.au/~erazmus
 (Current pics - trip to Malaysia to install equipment in jungle power site)

Some say there is no magic but, all things begin with thought then it becomes
academic, then some poor slob works out a practical way to implement all that
theory, this is called Engineering - for most people another form of magic.

----------------------------------------------------------------------------
To unsubscribe from diy_efi, send "unsubscribe diy_efi" (without the quotes)
in the body of a message (not the subject) to majordomo at lists.diy-efi.org