I notice that Tracy uses a separate ground for the injectors. He is
   not fooling. If you connect the battery ground to the injector ground
   it wont work. He should have mentioned that. It is an attempt to keep
   the injector spikes out of the computer. IMHO a few chokes here and
   there would also have worked without the uncertainties :)


   Paul Lamar

   Not necessarily -- putting the currents where you want them, where
   they don't share common impedance paths, is a better approach.
   "Chokes here and there" doesn't work very well unless you have a lot
   of energy storage in the power supply so it never sees the spikes
   anyway. Otherwise you will always be fighting the voltage spikes on
   the computer power that follow the current spikes from the injectors.
   Strongly recommend Henry Ott's book "Electromagnetic Compatibility
   Engineering." It's mostly about meeting EMC requirements but the
   issues are the same. His earlier version "Noise Reduction Techniques
   in Electronic Systems" is just as good for this problem.

   David Josephson


   True! Capacitors are needed. A 1000 Uf and a small choke on each
   injector would work well I suspect. Worth a try.  Of course the
   computer is well grounded to the engine.

   Even better is mount the switching transistor on the injector like a
   smart coil.


   Thanks for the book tip. I'll get it.

   Paul Lamar


   Maybe this discussion should be offline, but having wrestled this
   particular demon for decades in the audio and instrumentation biz I
   can assure you that "I suspect" approaches (even mine, or especially
   mine) are often a waste of time. It is tedious engineering but it is
   doable without too much exertion. Your number one question, for every
   instant of time for a given injector pulse, is "where does the
   current flow?" From battery + back to battery -, figure out the
   resistance of each path and see how the current through an injector
   gets there and back.

   I am not sure that putting the switching transistor at the injector
   would help. I would rather have the switching transistor where the
   source impedance for power feeding it was lowest. Yes, the loop area
   for radiated EMI would be reduced, but then you have another handful
   of connections that are more subject to engine vibration. What is a
   "small choke" and what function does it serve? Why 1000 uF and where
   would you put it?

   David Josephson



   Smart coils have no problems with ground loops nor does tracy use a
   separate ground for the smart coils. The switching transistor is in
   the coil.


   Paul Lamar

   ... I suspect the NSA is using hidden characters to identify certain
   class's of email users. It would be great if everybody used plain
   text. We get a wide variety of html messages. Especially from cell
   phones, tablets and Apple products. There is no effective standards
   in html. Every email editor program and cell phone manufacture wants
   to be different.

   David, Tracy uses separate grounds but the same 12 volt supply, He
   has had a lot of trouble over the years with glitches from the
   injectors despite the separate ground.

   A built in transistor switch with a large capacitor and small choke
   would suppress the spike  on the 12 volt supply. The choke need not
   carry the full 2 amps as the capacitor will handle the high peak
   currents. The choke will block the out going  spike. If I ever do an
   EFI that is the approach I will take. The injectors used to be shock
   and hold and now they are mostly just plain high current. It is when
   they turn off that causes the spike.  They are not designed for a
   100% duty cycle.

   Paul Lamar


   If the problem is a voltage spike when the transistor turns off
   rather than a current spike when it turns on (as I surmised from what I
   read earlier) then of course it is a different problem. You need to
   reduce the source impedance of the DC line that's shared between the
   injectors and everything else. A big capacitor will help as you suggest,
   as it will provide some of the energy for the inrush current, and it
   might absorb some of the turn-off transient when the transistor opens.
   BUT! In an airplane it's all about reliability and I would be very
   cautious about putting chokes and electrolytic capacitors in an engine
   compartment.

   Be sure to actually look at those waveforms with a fast scope, they
   could easily exceed the voltage rating of the capacitor, which then
   provides another failure path. And, I would be wary of adding yet
   another inductor in series, unless it was a known high reliability part
   like some from Renco. A spool of fine wire terminated to who-knows-what
   is not something you want in an engine compartment if you can avoid it.

   The whole point is to reduce the source impedance of the power buss, so
   that the energy in the coil cannot cause a big rise in buss voltage.
   With this approach, the current in the line at turn-off spikes more,
   which creates a large instantaneous magnetic field, so keeping that wire
   bundle away from the others, and preferably twisted with its return
   current line, is important. You can reduce the this with capacitors and
   chokes if you want to assume the risk. But you can also do it with
   bigger wire, transistors with lower on resistance, and a separate +12
   line back to the battery for the injector circuit. A fast zener across
   the injector to shunt the voltage spike could also be useful, but be
   sure the instantaneous peak current rating of the zener is not exceeded.

   David Josephson

The injector driver I have for my system operates much the same way as you are saying Tracys does with the separate ground. I tie the + terminal of the injector to a power source, and then run the ground side to the driver board. The ECU triggers an NPN transistor to fire the injector by connecting the ground. This way none of the power and spikes of opening and closing the injectors is actually run through the ECU system. There is a series of flyback diodes and other protections, but it sounds like Tracy has done it the same way that I am.

My LED strobe system that I built for my wing tip lights is the same way- 6W of LED flashes at 2 per second with no feedback.

Kevin Alderman




   The Rotary Engine News Letter. Powered by Linux.
    ACRE NL web site. http://www.rotaryeng.net
    You Tube http://tinyurl.com/beqqxas
    Copyright 1998-2014 All world wide rights reserved.