Old 486 Board for ECU??

[Mike from West Australia]

At 05:08 PM 1/5/2000 +0200, "Axel Rietschin"
<Axel_Rietschin at compuserve.com> wrote:

>How many concurrent PWMs (that's the WHOLE point of this entire discussion)?

How many do you want ?

As the PWM frequency is low then you can do it quite well in software,
you don't need to do it in hardware unless you really want to.

>One is not enough. Consider that you have to fire the spark at the
>appropriate time (assuming a DIS) and actuate a few valves in PWM all at the
>same time.

Firing the plug - at the same relative time for each cylinder can be done
real easily in software as can sequential injection. The number of cylinders
will fit in a single byte - dedicate a CPU register to the current plug
and another to the active injector etc etc yada yada

Remember you can execute a sizable amount of instructions in an engine
revolution :)

>> So all this hogwash about 486 boards not having timing is a software
>> issue around MSDOS not any limitation of the motherboard - the hardware
>> is much more capable then it needs to be with the exception of ADC's.
>
>For your typical low-function-one-PWM ECU, it may. Motorsports ECUs uses 332
>or even MPC555 microcontrollers with one or two built in Time Processing
>Units, microcoded for time-related engine mgmt functions, plus a couple of
>DSP chips for multi-channels detonation signal processing and, sometime,
>dedicated chips like the 67F687 to help with engine timing issues. That's a
>lot of timing-related hardware added to an otherwise standard CPU. The CPU
>runs the high level control functions, PID algorithms for
>closed-loop-everything and auxillary functions such as data logging. Again,
>it all depends of where you want to go.

Of course its obviously sensible to go for the latest hardware to make
the software development far easier and much more so for higher rpm
and more compensation/operational factors etc etc Part of this is the
push from the semi co to get devices into sale - sexy apps, marketing
etc etc.

Crikey ! The human race sent man to the moon on less computing power
then is in your average 99 model sedan !

None the less for a typical 6 (or even 8 cylinder) engine a 486 trundling
along at even 25Mhz can handle sequential injection *and* ignition and
run interrupts off a crank angle sensor set at 1 deg intervals. Software
can do wonders if you give it some thought (and it don't take much either;).

Consider this (street 6cyl RB26DET for example) :-

Engine at 9000 rpm = 150 per second = dist 75 per second
For a CAS with 360 deg pulses/rev thats 75*360 = 27KHz
A 25MHz 486 with L2 cache can handle 920 or so instructions between
1 deg pulses at max engine RPM. Allow 50 instructions for interrupt
overhead and you have 800 or so instructions per 'tick' to
decide which plug to fire (if at all) and which injector to turn
on or off. I'd say (having 22 years experience in assembler) that
you could make such decisions in 300 instructions or less and allow
some bit shuffling overhead to boot Now this is the time critical aspect
on interrupt, read on,

In the background you no such time critical requirement, you can handle
10mS ADC acquisition of AFM or MAP to put a number in a register and
do a table lookup - even a few multiplies if you want and even if this
is interrupted at the maximum rate you will still be able to handle
the rigours in software without the need for 'timing chips' which I
think most will come to the realisation have a 'marketing' angle (pun;)

Naturally the software will have to be well thought out, its not
actually a hard procedure to follow. Conventional flowchart design
will handle it well enough without having to resort to OOP or
any sort of complexity. CPU's core instructions are all you need to
handle simple logical decisions a multiply or divide as required,
which mask to apply to which IO byte - yada yada.

Now take a DX4/100, throw away the DRAM, use SRAM in its place and
load the kernal into cache on powerup. Only use the slow 70nS Flash
for table lookups. Crikey if you then keep the 16meg or so of DRAM
then use that for datalogging :)

>I'm not saying a 486 chip cannot do it. I say a PC won't be very good at
>this ECU job, unless you complement it with a lot of stuff.

You only need to complement it with an ADC and some thought, we won't
mention which is thinnest on the ground here ;-)

Don't get me wrong, I would still rather use peripheral features where
appropriate, I'm just pointing out you don't need to and that the core
hardware will do adequately with the addition of ADC and signal
conditioning.


Rgds


Mike Massen

Pictures of site installation at Mendulong near Sipitang, Sabah (Malaysia)
Remote Area Power System   http://www.wantree.com.au/~erazmus

Vehicle modifications on GMH Turbo, twin tyres, possible 175Kw at wheels
Preliminary pictures at http://www.wantree.com.au/~erazmus/Twin_tyre_vehicle/

Editorial on twin-tyre opinion and good reference about tyres:-
http://www.geocities.com/MotorCity/2195/ttyreopinion.html

Ancient Sufi saying:
        "Should your God save you from adversity, choose another God"
----------------------------------------------------------------------------
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