I got the system up and running with the L6388.
  Works like a charm.

  I never quite understood
  why the Aussies were claiming a minimum of two
  sparks at high RPM until I looked at the signal
  across the 2.5 ohm resister in series with the 1 uF
  cap. This resister is simulating the coil,
  for now, in the low voltage (12V) test set up.

  With  12 volts I am getting a peak current
  of .4 AMPS!! through that resistor when the lower
  FET turns on and ANOTHER .4 AMPS!! when it turns
  off. This is due to the totem pole configuration
  of the FET switches. Most CD system use only one
  switch.

  Imagine what the current would be with 300
  volts. Of course the inductance of the coil
  will make  a difference.

  In this set up these two current pulses are 20 usec
  apart. The pulse width and timing are controlled
  by the ZBasic computer with a simple BASIC
  program. RPM is about 9500. The final version
  of course will be stand alone with no
  computer needed.

  Here is a scope picture. The upper
  trace shows what the upper FET switch's
  input trigger is doing and the lower trace
  is the +/- 10 volts across the 2.5 ohm
  resister. E = I x R and I = E/R.

  In other words when the lower FET
  turns on the current pulse in the coil flows
  one way and when the lower FET turns off
  the current in the coil flows the other way.
  Duh!

  This means we will have a total of 8 ignition
  sparks for each leading spark plug. Groups
  of two 10 degrees apart from 22 deg. BTDC
  to 10 deg. ATDC. Up to 11,000 RPM or more.

  Next step is get the high voltage power
  supply and the stock Mazda 2nd gen coil
  working. I'll be very surprised and disappointed
  if this system does not significantly improve
  the BSFC of the rotary.


  Hello Paul;

  This looks like a nice design.  Relatively simple and I
  like the pulse pair with each trigger.  Even closely
  spaced pulses should increase your probability of getting
  a good burn started vs a single spark.

  But be careful when you choose the 300V supply.  Some
  quick calculations based on the energy stored/dissipated
  in the 1uf capacitor shows that 90mJoules is consumed for
  each pulse pair produced. ( energy, in Joules, stored in
  a capacitor is CxVxV/2 )   Each of the two pulse pairs
  uses the same energy.


   At 10k Rev/minute that translates to:
  10k/60 pulses/second x 9E-2 Joules/pulse= 15 Joules/second or 15 Watts.
  That's for one pulse pair per shaft revolution for one spark plug.

   Multiple by the total number of pulse pairs (you always get a pair)
  you will use per shaft rotation to get the total DC power
  you will need at 300V.

  Could be a significant amount of power.

  Craig Taylor


  I have a 400 watt 12 volt to 220 volt AC converter waiting
  in the wings :)

  Kind of old tech because of the 50 Hertz. Heavy and large
  but cheap.

  My new 300 volt 30 watt power supplies are
  converting at 120 Khz (120,000 cps).

  ---------------------------------------------

  I did away with the computer and used the actual
  hardware to control the L6388. The pulse width
  is now controlled by the LMC555. One of the neat
  things about this chip is it will not allow
  both switches to be on at the same time (disaster) so
  that eases the timing requirements. The doc is
  not great but fooling around with it will teach
  you how it works. I already blew the top
  off one but they are cheap :)

  I connected a 2nd gen RX7 coil just to see what
  would happen. It looks like it wants a 1 ms wide
  pulse but that may not be necessary or desirable.

  That allowed the coil oscillations to damp
  down after several oscillations. See the
  attached scope picture. Of course it is still
  running on 12 Volts but the time constants
  should be the same. This will vary of course
  and some coils will oscillate faster than others.

  I'll dig up some other coils and try them out.
  If you have any old coils lying around you
  don't need please send them to me.

  About the max we can do with this coil is 8 msec
  (7500 RPM) /36 or 222 usec if my math is correct.
  2.2 to 10 to the -4 or 0.000222222... 222 usec
  I think. We will probably get one or two oscillations
  going down and the same going back up.

  Next step is the 300 volt power. I'll use
  the 400 watt inverter as I don't want to risk the
  expensive power supplies just yet.

  BTW for dyno test I don't need the inverter.
  It will be possible to merely plug this
  ignition system into the 220 volt wall outlet
  found in the shop :)

  Paul Lamar


  Paul, I recall there being a "low inductance" ignition
  coil designed for CD ignition systems.  With the standard
  Kettering system, the energy to produce the spark is stored
  in the inductance of the coil, which makes a substantial
  amount of coil inductance desirable.  With the CD system,
  the energy is stored elsewhere and too much inductance can
  be a problem, as you are discovering.

  Craig Taylor

  Yes I think you are right. Probably the best bet
  is a MSD coil designed for CD systems.

  BTW I just noticed that Klaus (or Paul Lipps the designer) is
  using three Hall Effect sensors spaced 1/4 inch apart. Could he be
  using the same idea? :) In case somebody accuses me of copying
  Klaus's system I did not know about this until yesterday when
  I noticed it on his web site.

  http://www.lightspeedengineering.com/Products/MiniSensor.htm



  "Single systems require only one Mini Sensor.  On dual 4 cylinder
  engine installations, there are two Mini Sensors at 180 degree
  intervals. On dual 6 cylinder installations, a second Mini Sensor
  is mounted 120 degrees from the first."

  "The Mini Sensor consists of two billet aluminum machined halves
  that capture a small circuit board holding 3 Hall effect sensors
  at 0.250? spacing."

  End quote. I don't know what KLaus is using for coils but
  they appear to be off the shelf. Looks can be deceiving:)

  I have seen it many times but I never paid any attention.

  Klaus's excellent results indicates this idea will
  also improve the BSFC of the rotary.



  BTW Craig the output of my 220 volt bridge rectifier looks
  like this.

  220V____    ________    _______
          |  |        |  |
  0V      |__|        |__|

  Do you think I have a bad diode in  that bridge?

  Paul Lamar

  Paul,

Yes, I see what they are doing with the three hall effect sensors and the
groups of 4 magnets buried in the flywheel at 120* intervals (or 180* for
4-cyl apps).  It is important to see that each of the groups of magnets are
located on a different radius.  Each group of magnets aligns with one of the
three hall effect sensors.  That looks like a good solution for a 3-rotor.
  By separating the hall effect sensors by .0250", it solves the problem of
distributing the spark to the corresponding rotors at the proper time.  It
seems that the LSE would almost have to utilize a separate ignition circuit
for each hall effect sensor.  This would allow the circuits additional time
to recharge before the next firing event.  As I understand it, the LSE
treats a 3-rotor like three one-rotor engines tied together with a common
eccentric shaft.

Mark

A three rotor is equivalent to a six cylinder when it comes
to sparks per rev. The LSE might work on a rotary but I don't
thing so. The design for the rotary is more difficult due to
the much higher RPM.  He has a much easier time of it as the AC
piston engines run one third as fast as the rotary.

Do you know somebody running a LSE on a rotary? I'll ask Klaus
when I see  him again if he has sold any to Rotarians. I think I
would have heard about it from Klaus.

Paul Lamar


Hi Paul

Great work, Klaus uses motorcycle coils made by PVL I think it is commonly
designated 3, they are used on Triumph motorcycles and I think many others
about £50 each in the UK.  Picture attached.

Clive ........


Thanks for the tip. I'll check them out. They look real small
and light weight.

Paul Lamar