Subject: Manifold pressure and the ECx controllers
From: Rotary Engine
Date: 1/8/2011, 9:20 AM
To: AAA Put this in the To box
CC: Tracy Crook <tcrook@rotaryaviation.com>

   As you are setting up the test engine, I would like to suggest a
 couple of ideas which grew out of my study of your excellent work in
 measuring MAP using the oiling holes, along with the improbable
 injector
 maps  others have arrived at for Tracy's ECx computers.  Your
 measurements showed the enormous variations occurring for each
 combustion cycle.

 I believe it is likely that these variations,
 transmitted through small tubing, account for drastic variations in
 measured pressure as standing waves are set up in the tubing which
 move
 with rpm changes. The standing waves either add to or subtract from
 the
 average pressure, and as rpm changes the measured pressure varies.

 I first recognized problems in electronic measuring of MAP several
 years
 ago when I bought an electronic engine monitor from one of the first
 small vendors fo such items, and found MAP readings not in agreement
 with my mechanical gauge.  I tested both on the bench with our
 altimeter
 calibration equipment, and they agreed.  In the airplane, at low
 altitude one read 20.0 when the other read 21.5.  At higher altitude,
 the error was reduced.

 I added a restrictor to  the MAP line, and the
 error was further diminished. I was surprised at the degree of
 restriction required to eliminate the error.  My assumption was
 that the
 system used an algorithm that produced a stable but inaccurate reading
 for a varying input. A couple of years later I installed a Dynon
 engine
 monitor in another airplane, and found the MAP reading very
 jittery.  It
 appeared they rapidly read and display a couple times a second, and
 each
 sample varied from the previous.

 Again, a restrictor solved the problem
 Teeing the two pressure taps of a rotary does not eliminate the
 problem,
 although it might reduce it.  The very sharp drops in pressure during
 intake mean either a very tiny hole in the tap or a substantial volume
 in the lines will be needed to provide a stable pressure free from the
 effects of standing waves. A separate line and accumulator to each MAP
 sensor of the EC2 might be the best solution.

 If my deductions are in error, perhaps some of the engineers here
 familiar with fluid dynamics can set me straight.

 Victor



 Your basic reasoning of as to why there would be different readings
 seems to hold water.

 Any combustion engine that is not continous, ( jet , turbine, engine )
 will produce it's vaccum in pulses. Even as the theoretical number is
 known, the true intake pressure will be a wave of rising and falling
 pressure.

 For the engine this is actually a good feature, and we want it,
 intakes
 are made with intake runners, tuned to the best performance,
 meaning, we
 really want the air column to stack up against the next intake
 opening,
 and give a superficial boost to fill the combustion chamber at the
 next
 intake stroke.

 And yes, this will be a conflict, with pressure readings, if we want
 them as even and straight as possible, because the pressure sensor
 will
 not read the theoretical average intake pressure, it will read
 violently
 swinging ups and downs.

 There are as you say ways to dampen the effect, by either restricting
 the flow, to the sensor, but you will pay in slower readouts.

  You put it close to the intake port and you will get a very quick
 readout, but instead a heavier pulse. You put them close to the
 beginning of the intake runner, and the pulses will be somewhat
 smoother, but now you pay in slower reaction time.

 Depending on how much you want to put into it, there are ways to
 get to
 an equilibrium, a four cylinder engine, would have four intake
 runners,
 a small tube from each one of them at the same location of each, that
 will get to a spot in the middle will have equilibrium ( not fully of
 course, but close enough) and I dont know rotaty engines that much,
 but
 if it has an intake for combustion, that opens and closes, the basic
 principles would be the same.

 Lets say a two rotary intake pipes are running paralell, a small pipe
 inbetween will give you that( lesser so the lesser combustion chambers
 you have, that are loading up) ....but here is the clincher...only at
 a certain spot at a certain RPM.

 I'm sure yo don't want to drill forty seven holes in your intake
 runners, just to find that spot, it's just not worth it....it's only
 good for a certain RPM on top of it all.

 Some kind of dampening or restricting arrangement is probably best,
 all
 of them will make you pay with a slightly slower readout, but that is
 the price.

 One way to go here, ...if I may....just a sugggestion, as you have
 experienced that it takes an extremely small hole to restrict the
 pressure sensor, maybe , if the pressure sensor had a little chamber
 ahead of it, that was internally designed for good flow, but, as a
 muffler was designed to stop shockvaves.

 I'm thinking...and this is highly experimental now....a muffler, has
 actually very good flow though it, and that is what we want, that will
 fill up and make the pressure sensor, feel the pressure quick. If we
 just make the ebb and flow slow down, with baffles. so if we go the
 same
 way, have a fairly good flow to the pressure sensor, that will
 enable it
 to quick reaction time, as we have bigger holes, we have now gone away
 from the very small restricting holes, but are aiming each shockwave
 towards maybe a series of small baffles that can prolong the effect of
 the shockwave long enough for the opposing wave force to take effect.

 I'm not thinking about something that has the size of a regular
 muffler,
 more the size of an ordinary D cell battery.

 Lets say you have seven or eight walls in that cylinder, with seven or
 eight 1 mm  holes, (Here you need to experiment)  in each wall that
 does
 not correspond to each other, that's a lot of caivities and holes for
 the shockwave to find. It will probably work better than a muffler,
 as a
 muffler is a flow through system, while this is a dead end system,
 once
 the "MAP muffler" is filled, it will now be a standing wave flowing
 back
 towards the next oncoming shockwave.

 The closer to the intake manifold the more of a standing wave, the
 closer to the sensor, the less of a wave, if any. The object would
 be to
 get no wave there.

  There might be some experiments involved, but I am pretty darn sure
 that will stabilize a pressure sensor.

 It might also be an idea to look at how gun silencers are made, they
 have come up with some pretty nifty technologies in that field, and
 still they have to have a big huge hole in the middle of it, to
 make the
 bullet go though, but they for sure have managed to treat that
 shockwave
 in a favourable manner. Maybe some of that technology can be applied
 between the intake runner, and the pressure sensor.

 You want as much airflow as possible for a quick reaction time, but as
 much as possible of the shockwave to dissipate, those two desires
 have a
 tendency to not agree, so inbetween there somewhere is a compromize
 you
 just have to experiment with to find.

 Roger


   It would seem to me that the /time constant/ of the orifice and volume
   of the system after the orifice must be long enough to cover several
   cycles at the lowest rpm to be used in order to provide a stable
   measurement of pressure.  A pressure sensor at the end of a conduit can
   only sense the pressure at the opposite end when there is no net inflow
   to or outflow from the conduit. Therefore, the measurement must average
   over at least one cycle for a constantly changing pressure, and as a
   practical matter over several cycles. My intuition suggests to me that
   this pressure would actually be the root-mean-square (RMS) value and
 not
   the arithmetic mean, but either would only be an approximate and not a
   precise indication of the mass of the charge.

   Ideally, the pressure sensor would be located directly in the wall of
   the combustion chamber, and the sensor would be connected to a
   peak-sensing amplifier.  The pressure in the chamber would be directly
   proportional to the mass of the charge (neglecting the contribution of
   temperature rise from compression), and would correct for volumetric
   efficiency.

   For a tuned system, the port would ideally close at the instant of zero
   flow, when the pumping action had forced every possible molecule into
   the combustion chamber and if the port were left open the air would
   begin rushing out. It is, after all, the pressure in the chamber when
   the gate is closed (apex of rotor passes intake port) to trap the
 charge
   inside that matters, not the pressure when the gate opens. For the case
   of an engine operating within the range of max VE, the peak pressure
   just outside the port would be very close to the pressure in the
 chamber.

   Again Intuitively, it would seem that the smaller the conduit, the
   longer the distance the sensor can be located from the port and still
   get a good approximation of the peak value, as the magnitude of the
   standing waves is related to the cross-sectional area. Working against
   this is the resistance to flow in the smaller tube. It is here that I
   have no idea how far removed from the engine the sensor can be located
   and still use a peak-sensing method.  I can see that the higher the
 rpm,
   the shorter the distance.

   Victor

   That is the reason I propose measuring the pressure here.
   This is the apex seal oiling hole. Most people don't use it.
   They mix the oil with the fuel 1 oz per gallon.
   So far this seems to be working best.

   Tracy does not tell me when (relative to e-shaft position) or how
   often he takes a 2 micro second snap shot of the pressure.

   The 555 system will automatically average things out
   as it is looking at this hole all the time. It takes
   a millisecond for the pressure sensor to respond to changes.

   The data was taken 1000 times a second with the same pressure
   sensor Tracy's uses.

   Paul Lamar

 I wonder if the data would look different  if the sensors were located
 directly adjacent to the taps instead of connected through tubes several
 feet long.  It appears the engine had little or no load as the MAP was
 around 13 inches.  Larger variations in pressure would be expected at
 higher power settings.   I notice the two rotors has slightly different
 waveforms. even allowing for only 9 samples per cycle. Were the tubes
 different lengths?  When you get the engine running, it would be nice to
 see if the length of the tubing alters the data.
 I would be surprised if Tracy did not average the output of the sensors
 with low pass filters before using the information internally.

 Victor

 That is easy to try. I can get the sensors within 6 inches  of the
 RX8 pressure taps. The hoses were about 18" to 2 feet on the
 first go around. The RX8 test engine rotor housing taps can
 be teed together.

 I have got to get these RX8 coils working first.

 Did Tracy averaged the output of the sensors? I would not be surprised.
 the EC1 had a LM339 quad op amp close by each sensor. Good call Victor.

 I have not looked at an EC2 as it is harder to second guess as it
 is surface mount.

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

 Good call Victor. It never occurred to me to look before.
 Sure looks like one. I wonder what the response is.
 If too slow the engine will not responds to rapid
 throttle openings. Could also be amps for the temp sensors.

 I could always mock one up on a proto board and test it:)

 What do you think it looks like?

 Paul Lamar


Not enough info to tell - most of the connections are on the other side
of the board.  It is highly improbable, however, that a designer would
do anything other than filter out the variations in the signal. If you
can't control the placement of the tap and the propagation delays in the
conduit to the sensor there is no way to know when to sample.

The easy way to handle rapid throttle application is a throttle position
sensor.  A potentiometer mechanically connected to the throttle linkage
gives instantaneous notification of throttle change. A 1k pot between
supply and ground with a capacitor connecting the slider to the output
of the MAP sensor. The MAP sensor has a high output impedance so the
signal is easily influenced.  Of course, Rule number 1 always applies -

Nothing is Easy. In order to have reasonably-sized capacitors, we need
an electrolytic and since either end could be positive at any given
time, we need two in series with a ground reference between. Since we
need the ground reference anyway, let's make it a gain control so we can
adjust the influence. Now when the throttle is moved, a temporary charge
is added or subtracted to the MAP sensor output. All components except
the TPS pot itself go in the controller box.
There are many other ways it could be implemented but the simple
addition of a pot to the throttle provides an immediate signal to be
processed for enrichment.

Victor.

We have considered that. Since we have a slide throttle here are
some ideas in implementing it in hardware. The photo cell trick
is solid state and no additional mechanical parts are necessary.

Another neat thing about it is one could compare the MAP with
the throttle position for academic reasons :)

BTW Come to think of it I don't need to mock up the low pass filter.
Just fire the EC1 up and use a scope while feeding in a source
of turbulent air to the MAP sensor.

Paul Lamar
-- 
The Rotary Engine NewsLetter. Powered by Linux.
ACRE NL web site. http://www.rotaryeng.net
Youtube key word PaulLamar2
Copyright 1998-2010 All world wide rights reserved.