I am very pleased to have discovered this group, and have the utmost
 respect for the participants and their obvious expertise. One of the
 first things I saw was the 555 injector control.

 I have been planning for several weeks to use the 555 injector
 system on
 my project, so I was entering the schematic in preparation for laying
 out the circuit board. I was considering how to make provision for the
 initial setup when I realized the MAF sensor output is related to air
 mass per unit of time, whereas the 555 controls fuel mass per firing
 cycle. There is not a defined correspondence between the two except
 for
 a fixed engine speed..

 Assume an engine running wide-open-throttle (WOT) at 4000 rpm, and the
 555 delivering the proper pulse width for proper fuel delivery. Lower
 the nose and allow the engine to advance to 6000 rpm. The airflow
 increases by 50%, and the MAF sensor output increases 50%, lengthening
 the pulse by 50%. Because of the increase in rpm, there are 50% more
 pulses, each of which injects 50% more fuel, for an increase to 225%
 of
 the original flow instead of the desired 150%. In fact, absent a
 change
 in volumetric efficiency between the two rpm settings, the pulse width
 should not change, but deliver the same amount of fuel per
 combustion cycle.

 Fortunately there is a simple solution. The integral of the 555 output
 is proportional to fuel flow per unit of time. The mass of fuel flow
 per
 unit time (instead of per revolution) is easily obtained by
 multiplying
 the flow capacity of the injector by the duty cycle of the 555. Duty
 cycle is determined by the duration of the 555 pulse and the frequency
 of the trigger (a direct function of rpm), and is represented by the
 average voltage on the 555 output divided by the 555 output "HIGH"
 voltage.

 By connecting a resistor from the output of the 555 to the inverting
 input of the TLC279 op amp and a capacitor from that input to ground,
 the average voltage is presented to that input. The op amp now
 compares
 the (mass airflow)/time to the (mass fuel flow)/time, and generates an
 output to control the pulse width. It then becomes a matter of
 determining the correct component values to set the 555 timer
 characteristics and the op amp gain for the injectors and MAF sensor
 used.

 A variable resistor would still be used to allow  the pilot to alter
 the
 mixture would still be desired, but the system would automatically
 compensate for volumetric efficiency variations with rpm, because it
 would always maintain a fixed relationship between the air and fuel
 volumes (aside from injector latency, fuel pressure variations,
 etc.) We
 are obviously not measuring fuel flow directly, but are measuring
 directly the command to the injector. Installation-specific
 adjustments
 would be required for idle and full power, but everything in between
 would be automatic.

 Victor Roberts

 With all due respect I don't think you are right.

 This is just a rough approx. to keep the engine running over the
 rev range.

 The VE also  changes with RPM. It can go from 100% to 120% over a
 2000 RPM
 change in a organ pipe tuned system.

 I would not use a 555 system in a car
 as it will not tune the engine properly over a wide range of HP, RPM
 and load combinations under transient conditions. You also might not
 want to
 use the 555 system in an aerobatic airplane.

 I Think you have to go from a high angle climb to a vertical dive to
 get
 the engine to rev from 4000 full load to 6000 full load. I think
 that is an
 unrealistic scenario for most aircraft. The prop load  changes a lot
 over such a wide range during normal operation. The fixed pitch prop
 load
 is the square or cube of RPM. The engine RPM might change
 several hundread RPM in a cruise situation but then you must adjust
 the leaning just like any other aircraft engine.

 BTW The rotary unlike an aircraft engine  will run over an air fuel
 mixture ratio of 10:1 to 20:1.

 Paul Lamar

 Paul, Victor -

 Victor is right that with an Inject-Per-Revolution scheme (IPR - I
 made it up), a simple MAF sensor/555 timer would need some RPM
 feedback and compensation, making it ... not simple.

 Paul is wrong that it should be designed/built as "just a rough approx
 just keep the engine running".  It should be designed to match the
 fuel injected to the incoming air.

 Paul is right that VE can change.  However it can change for the worse
 over a given RPM range as easily as it can change for the better.  IPR
 schemes need crank angle and RPM feedback to work effectively,
 otherwise the air/fuel ratio will be all over the map (no pun intended).

 I had always assumed that the 555 timer would inject independent of
 RPM or crank angle.  The injectors would sit upstream enough in the
 manifold, happily injecting away in fine grained pulses, so the fuel
 ingested per intake stroke would average out and only slightly vary.
 This would be an Inject-Over-Time scheme (IOT - I made that up, too).

 NOTE: That is the beauty of MAF sensors - they tell you the amount of
 air going in - you simply inject a matching ratio of fuel, and it'll
 burn at 350 RPM, 4000 RPM, or 6500 RPM - hence the simplicity of the
 555 timer for controlling fuel injector pulse width.

 BUT... If you're going to start varying the injection RATE (tied to
 crank angle/RPM), then you need to COMPENSATE the fuel pulse width
 with the same variable (RPM), in a feedback to the FORMERLY simple
 system.

 Paul - just have the 555 Timer system inject in a relatively fast
 pulse stream independent of the RPM/crank angle, using the IOT scheme.

 In other words, keep it simple.  (Just my 2 cents worth.)

 -Vince Orton

 I you think it will work that scheme  solves a lot of problems with
 high speed
 injection. There is not a lot of time  to inject fuel once per rev at
 7500 RPM
 anyway. All we need is an oscillator to trigger the 555's.

 It does make it  a lot  simpler as we don't need injection triggers
 and the associated
 circuits. I am getting close  to trying the 555 system as the RX8
 engine slide throttle
 and intake tubes are near done. The motor mount is done and the prop
 dyno is near done.

 Paul Lamar

This works in theory but not in practice because of the latency (delay
between the leading edge of the pulse and the beginning of fuel flow) of
the injectors. In order to get even distribution from one firing cycle
to the next, there would need to be at least 3 pulses per firing cycle
at the highest rpm. The injectors can't work that fast.  I am planning
to use the spark trigger as the start of the injection cycle, as
injector timing is not critical. At low rpm this is not optimum, but is
works fine for higher rpm. Using one sensor for each injector/spark plug
gives four independent ignition/injection systems for a two-rotor. If a
sensor fails, it takes out only the related injector/plug. Turn the
mixture full rich and the rotor with the failed sensor runs lean but
still produces power, and the other rotor runs rich. Good for limp-home.

To Craig Taylor - I'm not an engineer, just a guy who spent 45 years
working on airplanes and longer than that on electronics. I used to tell
my technicians that a technician only need to know what direction a
change would go, but an engineer could tell you how far. In that regard,
could you tell us the minimum time constant needed for a simple RC
filter to provide feedback to the inverting input of the op amp as
depicted in the schematic for the leaning system for operation from 1500
to 7500 rpm?

Victor Roberts

The latency delay is only about 1 ms. The large variations are down in
the idle range 1 to 3 MS. I don't see that as a show stopper. We idle
the engine at 2000 RPM due to gear box rattle.

The wcf file needs to be converted to a jpg or gif to publish.

Paul Lamar



Hi all.
 This is a quick pass at the design using an average of the injector voltage as an analog of the fuel volume.  It compares the air mass flow to the fuel volume flow and adjusts the fuel flow (via Control on the 555) to force the fuel to match the air flow. The signals are summed in the integrator made up of the opamp IOP1 and C5, R8.
    I ran a simulation to make sure it was stable.  The injector pulse smoothing filter values ( R7 and C7) are approximate, but they shouldn't be critical.  Since the constant pulse width, one pulse per rev approach gives a first approximation of the fuel requirements of the engine, allowing 1 second or more  for the feedback loop to adjust to the new MAF sensor voltage shouldn't be an issue.

  For an actual system, somel component values need to be optimized.  I used 15k and 200nF for the one shot RC time constant.  For a 555, the RC product of the components is very close to the actual pulse width; here it's 1.5E4 Ohms x 2E-7 Farads= 3E-3 seconds= 3ms.
  That means that when the Control voltage is 2/3 of the supply voltage (6.66V in my case) the output pulse is 3ms. This base value can be easily adjusted by varying R1.
  You should be aware that increasing Control voltage above about 0.95xVsupply  will cause strange things to happen. So  shouldn't make that voltage part of the normal operating range.  Therefore I would pick RC to be the middle of the expected injector pulse width range.  That allows Control to adjust the pulse width over a reasonable range.

  Also note that I haven't make any attempt to "calibrate" the system.  I have no idea, for instance, what full scale ( 5V? ) on the MAF sensor output is equal to expected full speed-full throttle air flow.  I don't even know if the MAF sensor output is LINEAR with air mass flow.  I sure hope so!

Craig Taylor