Please  look this over carefully and  if you
see any mistakes please let me  know.

Paul Lamar



Paul,
I'm new.
What layout  software are you using, ExpressPCB?

You are running traces at odd  angles.  I prefer to keep the angles on  my
boards evenly  divisible by 45.  It's easier to keep track of the   signals.

You have pads for parts in the middle of traces; for  example  your larger
caps (I assume they are caps; you have no  reference designators  or part
numbers showing).  Try to avoid  this wherever possible; the  trace wicks
the
heat away while you are  trying to solder the leads, and you will  be much
more
likely to  "lift" the pad and destroy the board.  By "lift", I  mean
breaking
the bond between the FR4 fiberglass and the copper.  Once  the  bond has
been broken, the board is scrap.

I do not know  why you are using such thin traces.  It costs no more to
have boards  made that have nice fat traces.  Narrow traces will be more
easily
damaged, and much harder to repair.  Just make certain you have  8  mil
separation from things that are not supposed to be  connected.  I run  the
widest
traces that I can while maintaining  8 mil separation.

I do not know what ICs you are using, but I do not  see sufficient bypass
capacitors.  The "rule of thumb" is a minimum of  one 0.1uF (100nF) metal
poly or ceramic bypass capacitor per IC, as close  to the power pins as you
can
get them.  It is also advised to use  bypass capacitors sprinkled liberally
between the power and ground buses  aka "rails" around the board.  If  you
don't do this, you may  regret the omission later.

You are far better off to include pads that  you don't need, than try to
cobble up the board  later.

Steve

Yes express PC.

Paul Lamar



OK, I came up with a slightly modified approach for the board.

Replace all the IC's with a PIC16F1933 that has the extended temperature
rating.  That makes most of the parts simply go away.  The PIC  itself costs
about $2 if you're buying one at a time.

It will also be more accurate than using a 555 timer, because the internal
clock is calibrated to better than 1% accuracy.

Among many other features, this PIC has an internal comparator that can set
 it's own reference level, and has software programmable hysteresis.

I've attached a preliminary schematic and board layout.  Note that  there
is no regulator as of yet, but I have a few in mind that are rated for  >60v
in.  At this point, the board is roughly 1.7" x 1.2".  The  connector is for
a 24-pin ribbon cable to go to the main board; every other  signal is
ground to help keep things quiet.

Your thoughts?

Steve

That will work very well indeed no doubt. In fact I just had a conversation
with a hangar neighbor that has built over 50 consumer products mostly
based on the PIC. That is exactly what he wanted me to do.
In fact it could be made to be interchangeable with my system.
You need a couple of bits so it can communicate with the processor
and tell the processor when it is ready to send the RPM data.
You will also need a way of telling it the number of teeth
on the flywheel. That number varies from engine series to
engine series. That function is done by the pot on the 555.

However it violates the premise that nothing about
this system needs to be programmed in assembly language.
Admittedly the PIC counter asm program is short and simple.

Anybody with a little instruction and with some recent real cheap
($30) USB high speed data acq  systems now on the market one can trouble
shoot it.

The whole idea here is to eliminate the biggest drawback
to do it your self EFI is the steep learning curve for
asm and C. That is why I am going with BASIC. If one wanted
to learn asm the Megasquirt system could be reworked to better
work with the A/C rotary.

As soon as you make an open source system easy and quick to program
the expansion capabilities and user interface ideas are limitless.

Paul Lamar