>Paul,
   >
   >This week I will sand blast a housing, and fill the outside depression
   >around the oil injection location with DEVCON Aluminum epoxy.
   >
   >Thursday afternoon has become my access day for  Michiwa's school
   >machine workshop and I will be boring the housings and making the
   >aluminum inserts then.
   >
   >Pictured below are two types of injector cylinder housings, both with
   >upper and lower 0-rings.  The design on the left has a M-20 thread on
   >the top part though it isn't shown, and although it takes up less real
   >estate I have decided to go with Lynn Hannover's design suggestion.
   >Why?  O-rings tend to deform or fold when they are twisted several
   >revolutions while being pushed axially as well.
   >
   >I'll pressure test the housings with some glycol/water mix in the
   cavity
   >to confirm it works.
   >
   >FYI the injector interface to the inner liner hole is drilled and V
   >shaped from the backside to allow the injector tip to insert with about
   >0.5mm clearance of the apex seal path.
   >
   >Since I am supercharging the engine and wish to stick with MAP and
   >MegagSquirt ECU,  the question remains where to put the sensor or
   >fitting?
   >
   >Two choices.  The present oil injection hole, or in the back plate on the
   >slide throttle?
   >I am concerned that the oil injection hole may accumulate fuel in it
   >during certain engine events, so I am leaning towards the slide throttle
   >location now.
   >
   >Cheers
   >
   >
   >Doug in Japan
   >
   >Don't forget to prepare the aluminum with an etching solution.
   >
   >Question: when does megasquirt sample the manifold pressure
   >and is the short sample representative of the air going in the engine
   throughout
   >the RPM/tuning range? Have you studied the C source code to find out?
   >
   >What are you going to use for a throttle position sensor?
   >
   >The O2 sensor will fail if you use 100LL.
   >How important is the O2 sensor to proper operation?
   >
   >
   >Paul Lamar
   >

   Paul,

   I have not gotten deep into the megasquirt code as yet but I can tell you
   it is not C, but assembly lanuage which quite frankly is a relief.
   Megasquirt recently came out with the latest hardware version,
   Microsquirt in a small 2.6" by 3.5  seal AMP enclousure.  MAP sensor is
   remote.  Unit sells for under $500 completely assembled with surface
   mounted components.

   Without going into detail, Megasquirt allows you to roll your own and
   change any parameter in the  MegaTune.set-up software.  The code is open
   and freely available and many people have tweaked it to suit their
   applications.

   http://www.megamanual.com/MS%20FAQ.htm

   Megatune with any PC becomes a free dyno software, a feature that pays
   for the board in my opinion.

   I am still machining parts on my slide throttles  (one per port) but it
   will have a TPS built-in and the abilitity to be motorized if desired.
   Both the TPS and oxygen sensor are not really needed.  They are there to
   help in idle, cold start-up, and acelleration events and can be disabled
   manually or via software settings depending on the users wishes.  I will
   probably disable the O2 sensor at the 50% power level and go to manual
   leaning.  The O2 will then be used to monitor the A/F levels.
   Megasquirt system can be configured to be a simple as Tracy's or as
   sophisticated as the higher priced ECU units.  I am not 100% sure at
   this moment but it appears the user can control WHEN the sensors are
   polled.  Obviously at 7500 rpm there is not enough time to take more
   than one reading per intake event even though the computer can perform
   24000 calculations during that time.   The tach trigger signal is the
   reference from whence all sensors are sampled.  When that signal is
   generated can be altered of  via hardware or software set-ups.

   I am using E-85 and as a back-up auto gas. NO 100LL.   Meg. Sq. allows
   for use of a flex fuel sensor that tweaks the fuel map table depending
   on what percentage of ethanol it sees in the fuel.
   Sensor cost is about $300.

   For aluminum prep.  the directions on the DEVCON box said the metal had
   to be clean, free of oil and  surface roughened to a 3-5 mil profile.
   They advise cleaning, (I've used a caustic soda base solution before)
   sand blasting or sanding with 8-40 mesh grit, and then another cleaning
   to remove sanding grit.  I would then clean with alcohol or something to
   neutralize the caustic soda.
   Naturally no touching the metal with un-gloved hands and leaving oily
   finger prints. While the metal is still bright and warm apply the epoxy.
     DEVCON,  like any epoxy needs accurate proportions, mixing and surface
   prep.   It says on the carton for maximum strength to oven cure at 200F
   for 4 hours after room curing for 2 1/2 hours.

   Doug in Japan

   Early version of the MegaSquirt hard ware were assembly language and I commend
   you for using it because you have close control over exact timing. That is one
   thing you cannot say about C compilers. I did not know you knew how to program
   in assembly language.  I am very impressed. All the latest hardware versions
   however uses C.

   The problem, as far as epoxy adhesion is concerned, is aluminum oxidation which

   happens immediately. Back in 1965 when we were building epoxy bonded race car
   chassis we anodized the aluminum first as epoxy sticks better to anodized
   aluminum. Nothing to do with cleaning. I have been there and done that several
   times the hard way. The prep kit is called a conversion coating. It is a pain
   in
   the butt but absolutely necessary  and it is why I invented the welded in SS
   p-port. Makes two 1/2 inch wide test strips of aluminum and bond them together
   with a 1/2 inch over lap. Try to bend the bond area and you will see what I
   mean.

   Paul Lamar


Your aluminum prep kit looks to be the same as Dupont more than likely is
the same stuff different package.

Dan