12/26/97
- riding through Louisiana to the Low-Fangler's Ball with Robert Harris, 
  talking about fuel injection systems
- decided continuous flow, high pressure would probably give much better
  atomization than electronic injectors, viz poor spray patterns with the
  electronic stuff
- continuous injection nozzle could be small diameter and bent around corners
  to spray wherever we wanted
- could also have different ends for different spray patterns
- possibly steal Bosch CIS nozzles
- observed Hilborn/Crower/Enderle mechanical injection was RPM/fuel and 
  TPS/fuel, mechanical equivalent of basic Holley Pro-Jection.
- observed problem with Hilborn was VE changes with RPM, injector normally
  calibrated for WOT, so usually goes fat at midrange.  Idle controlled by
  throttle stop
- since default is rich, a simple control loop with an O2 sensor and a control
  valve in the bypass rail parallel to the pill would allow you to run at
  stoich at cruise and normal driving.  Talked about industrial control
  valves, then I realized an ordinary TBI type injector would work, 
  functioning as a "de-jector" in the bypass rail
- use a microswitch in the throttle linkage to open the injector power wire
  at 80% or greater throttle; let the control circuit keep doing its thing,
  just remove the dejector from the circuit to go full rich
- I suggested a Basic Stamp or similar microcontroller would do
- Kinsler et. al. have pill tables to set WOT fuel flow
- de-jector would be fail-safe; it could run even with complete electrical
  failure
- we talked about a "phase 2" setup for lean cruise monitoring EGT instead of
  O2.  Use vacuum for lean cruise actuation; >10% throttle, hot coolant,
  >15" vacuum, switch to EGT
- talked about throttle plates.  A barrel valve would have no restriction when
  open, a small hole in each through-tube and a single needle valve at the
  end could set idle air bleed.
- talked about a simple knock-only spark control system.  Set static advance
  to 30-35 degrees using drag strip or dyno to find max power.  System watches
  knock sensor and retards spark as needed.  A start switch would retard the
  spark to TDC for easy starting; a simple countdown timer would let it run
  a few seconds before switching to knock mode.
- Robert suggested the TI knock control chip, which is available with a
  developer's kit, might have enough smarts to run the de-jector algorithm
  in some unused space

01/09/98
- noted Hilborn style pumps run $350-$500, plus require bracketry and drive
  arrangement.  A Bosch CIS electric pump runs near 100 psi, available in
  junk yards.  A control circuit - use the signal from the coil '-' wire,
  a signal-shaping chip, a programmable timer, and a transistor to drive
  a "de-jector".  Make the electric pump a "replace as assembly" for the
  mechanical pump

01/11/98
- struck me the mechanical pumps, though supposedly positive displacement,
  probably increase in pressure with RPM due to back pressure from the 
  nozzles.  On the other hand, the increase in pressure would force more
  fuel through the pill, so overall line pressure might not change much...?



Date: Fri, 09 Jan 1998 12:13:45 -0800
From: Robert Harris 
Subject: RE: dejection
To: 'Dave Williams' 

Dejection Goals:

A. Self Tuning
B. Self Optimizing
C. "Good Enough" - i.e. nicely tuned Weber Carb performance level.

Constraints

A. No Complicating Shitting
B. No "timed" Injection
C. No Calibration of sensors etc
D. No outside services to make changes, etc.
E. No air flow measurements - maybe

Control Theory.

It is far easier to start from a known error and move to correctness than it
is to be "about right" and tweak to perfection.  Example - sailing.  You always
navigate to a point offset from your destination - say 50 miles north.  When
you get to that point, you simply turn south and home in.  If you steered to
destination and didn't hit it exactly, are you north or south???  For
Dejection, it is far easier and SAFER to start very rich and trim to lean.
This is why I liked the mechanical fuel injection system.

Philosophy:

Keep it very simple.  Use mechanical advantage whenever possible.  Example -
Rob White from Cosworth was talking about injectors in an article in one of the
yuppie mags.  The current trend is to multiple staged injectors, with the first
set as close to the port as possible, and the last set into the INLET of the
turbo.  More smaller injectors allow better control over mixture etc. This is
why I favor in bowl continuous injection.  Let the valve heat and turbulence
mix it for me rather than complicated spray patterns etc.

On to practical matters - going to find a book about CIS - but, if we are going
to eliminate mechanical - lets go all the way and get rid of most of the analog
- more after reading about CIS.

-----Original Message-----
From:   Dave Williams [SMTP:dave.williams@chaos.lrk.ar.us]
Sent:   Friday, January 09, 1998 5:08 AM
To:     bob@bobthecomputerguy.com
Subject:        dejection


Standing in the shower this morning, thinking about the control loop
for the dejector.  Basic Stamp might be the simplest way for computer
geeks like us to do it, but the setup is basically just a negative
feedback loop - voltage low, increase pulsewidth.  Voltage high,
decrease pulsewidth.  Have to average the O2 reading and pick a proper
sampling rate, but no big deal.  I bet this could be done with ordinary
logic chips, no microprocessor or software required.  Might not even
take much more in the way of chips, since there'll have to be some
interface stuff to the Stamp anyway.

Another thought was, we could use an electric pump and rig up a
separate, freestanding controller.  Instead of O2 to pulsewidth, do RPM
to pulsewidth, to control fuel volume by RPM.  This would be
functionally identical with the Hilborn pump without requiring bracket
fabrication, pulleys, etc.  Steal the pump, regulator, hoses, and
nozzles from a Bosch CIS setup out of the junkyard.

By going freestanding, we could make the electric pump interchangeable
as an assembly with the mechanical pump.  Probably cheaper too.





Date: Mon, 09 Feb 1998 06:48:00 -0800
From: Robert Harris 
Subject: Back to basics
To: Dave Williams 

Will leave you with brain salad stew while you conjugate Ammonium Hydroxide,
UAN-32, and how to shrink the V-12.  Am in an animated flame exchange with
FENATIC so expect that fuckhead to erupt like a bad pimple on the list.

Back to simple ass dejection.

Electric Motor Theory.  Conventional motor control consists of varying
current or voltage.  This changes the speed BUT I2R loss's are wicked,
heating both the resistive element and the motor.   Better way is to pulse
the motor with a full voltage pulse. Back EMF limits current flow to normal
levels.  Vary speed by varying pulse width.

Plan to use one or two Bosch (depending on flow) for each engine half.
Looking for the ones that will make up to about 90+PSI.   Will pulse the
motor each time the plug fires - thus the amount of fuel will go up with
rpm.  Will vary the pulse width under dejection control.  Output will then
go to Bosch Accumulator which smooths the output and on to the dejection
system.  This closely resembles the Hilborn pump.

Control output is pulse width to motor.  Sensor input is fuel pressure from
fuel distribution block.  Control input is desired fuel pressure.  Set point
is minimum of 50 psi ( needed to make Bosch injectors to work).  This means
no expensium pump needed.  450 * 2 buck savings.

Can use a PIC or a STAMP to sense and control pump.

More thoughts later.


Date: Wed, 11 Feb 1998 08:28:13 -0800
From: Robert Harris 
Subject: RE: they're heeere!
To: Dave Williams 

How much does she "covet" the book???  Extract a high price - like maybe a
Monica - cause it yours to give to her.  Consider it a down payment on the
V-12.  I WANT that engine in my Mustang - more low fangalange - more power -
more funkiness - and a unique signature line.

-----Original Message-----

Date: Mon, 09 Feb 1998 15:58:19 -0800
From: acrch 
Subject: Fw: Dejection - decomplicating shitting
To: dave.williams@chaos.lrk.ar.us

Simplicity is beauty.  Cheap is a virtue.

Consider this.  Use a Bosch K-Jet CIS electric fuel pump.  Runs
continuously - fuel flow cooled - makes about 90 to 100 psi. If its good
enough for Mercedes, I can live with it.  Turn it on with key, let it run.
Plumb next in line an accumulator (also from Bosch) to smooth pump
vibrations out and act as a backflow preventer.  Next a fuel filter.

This then splits into two lines going to two fuel distribution block's
(left bank, right bank), which splits the fuel out to each cylinder's late
KE air assisted injector nozzle.

The return line's run back to two more distribution block's at the tank.
Here each line splits into three valves - one on/off solenoid and two
dejectors.  The solenoid controls 1/2 the fuel return to the tank and the
dejectors the other half.  The two dejector valves spray directly back into
the tank.  They are alternatively driven off a optimum frequency binary
multivibrator.  Their pulse time is controlled by the computer.

What we have now is a closed loop, where fuel is recirculated to the tank,
with a controlled leak at the injectors.  No mechanical fitting whatsoever.
No FP regulator etc.,  Only one set of learning curves, complete with bank
tuning

Still thinking about the valving but this is pretty close to final for a
single injector system.  Will be working it out for a second bank next.
Thinking toggling between bleed valve and fixed jet.  This way pump and
line see no difference - just doubling the load range.  Think CIS injectors
at port, with secondary injectors kicking in upstream to take advantage of
some charge cooling.

Notice there is nothing mechanically moving in the fuel metering system. No
pressure regulator, no distributor, no throttle valve - no nothing.  How
much simpler can I make it?

Will describe the software later.  Got some working ideas sketched out -
all self tuning, self optimizing and self converging and SMALL and SIMPLE.

Date: Wed, 11 Feb 1998 22:32:46 -0800
From: Robert Harris 
Subject: RE: Back to basics
To: Dave Williams 

Deleted Stuff

You mentioned adding kerosene to your motor fuel.  I've thought of
that, considering the MTBE-swill "gasoline" could use something that'd
actually burn mixed in, but local kero prices are $1.75+/gal.  Diesel is
$1.06, about 10c higher than gas, but I worried about the cetane
additives and their effect on the octane rating.  Was always going to
experiment and see, but never found any round tuits.

I have found on my Dodge Truck that about 1 to 10 Diesel to gas works just
fine.  I just put two bucks of Diesel in the tank for every twenty bucks
gasoline.  Works fine for me.  Think that perhaps the lower ignition temp is
made up for by the slower overall burn and much more hydrocarbons.   Coming
to the conclusion that Octane may be hype and myth.  But I am not known for
conventional thought.

Besides, if I convert to full time EGR, I could give a shit about octane
anyway.

What do you think about the oil pump plan?

Date: Thu, 12 Feb 1998 08:14:20 -0800
From: Robert Harris 
Subject: RE: Klingons
To: Dave Williams 

On to fuel injection pumps.

All fuel injection systems can be reduced to pumps, restrictors and leaks.
It's the matching of the 3 that gives fits.  All injectors are simply leaks.
Fuel pressure regulators are generally a leak.  Mechanical systems generally
control by restricters, EFI by leaks, and Bosch of curse, along with the
Hilborn, uses both of them.

The Hilborn and some Bosch use a motor driven pump who's output is roughly
proportional to RPM. All of these pumps are positive displacement pumps.
Electronics and later CIS use a fixed output electrical pump to simplify the
engine at the cost of complicating the injection system.

Trying to run an electrical pump as a mechanical pump - i.e. varying its
output might work, but its Mickey mouse.  Too long hydraulic delay from pump
to nozzle, too much dependence on out of production parts remaining
available etc.  The mechanical pump is now found only on the expensium and
unobtainium dragster stuff.

Which leaves us back to a constant output high pressure pump - OOPS I JUST
remembered one other positive displacement pump whose output varies with rpm
(Klingon Letter).  Its the oil pump.  Since all dejection engines can use a
distributor less ignition system - its much simpler to tune and steal for,
we half a hole upon.

So, what I suggested was take an appropriate small engine oil pump, moly and
Teflon coat the puck out of it, tap it for the appropriate fuel line's in
and out, turn it upside down and stuff it into the distributor hole - and
use a modified distributor shaft drive to connect it to the cam. Use an
input restricter to limit flow.

What we haff now is a mechanically driven fuel pump whose output volume is
roughly proximate to the full throttle load of the engine at that rpm - just
like a Hilborn pump, only not taking accessory space, and available,
unexpensium - if not down right cheap. Remember part is parts. Steal freely
from other's and you vill be rewarded, mien furher.

With a pump setup like this, the electronics get much simpler - back to
stealing some other cheap mechanicals.  The problem with a constant output
electric is that the bleeding function is inverse to the power curve - with
the most leaking at idle. Actually its a constant leak system with the
unused output going straight into the tank.  Variable pump will be
non-linear and probably too much delay to be tightly coupled to engine
demand.

Your mission, should you choose to accept it, and this message having
survived the klingon translation is to unzip this pump and if not possible,
is this closer to the true agenda - not only simple, cheap - but emotionally
offensive to as many people as possible???

On alternate fuels, ammonium hydroxide (stop and rob ammonia cleaner) is
cool.  Tried it yet?

UAN-32 is over the top and on the edge.  If what I suspect is correct, this
only needs a good carbon fire to let it rip.  The blend appears to be Urea,
AN and water.  Remember Urea needs water to complete its reduction, AN needs
NOTHING and has oxygen to give away.  From Oklahoma City, we know there is
lots of surplus energy in AN and from physics in CO and ammonia (NH3).  So
if this fluid needs no "air" and just some surplus hydrocarbon to suck up
its surplus oxygen, is not this close to an ideal power adder?  Just double
dose 1x carbon, 1x UAN-32, instant say doubling the power.  With
Fertilizer???  We can tell the truth and thoroughly confuse the world by
calling it "Kat Piss" injection because it puts a "tiger" in your tank -
since ammonia is what makes kat piss reek.  How many Scott Schidel turbo
types can we send into apoplexy when told that they could have made more
power using "Kat Piss" than their multi-turbo stuff.

PS - MS Spell checker wants to change Klingon into Clinton - somebody in
Redmond had better go into deep cover if Bill ever finds out.

Date: Thu, 12 Feb 1998 14:06:08 -0800
From: Robert Harris 
Subject: RE: 4spd Flexplate?
To: Kemper Porter 

I generally prefer my sinus's kept clean, however your mileage may vary.

Since you showed me yours, I spouse I should show you mine - but I'll let
Dave explain it.  You won't have the only Dave William's built V-12 for
long.

Damn - now I reeealy reeeeealy need to become a wage slave again.  Now where
do I find a cheap 5-speed with the spread of an AOD, the bolt pattern of a
Windsor, and the ability to soak up the power of a 462 (four six two - no
relation to a 460 or 390)???


Date: Mon, 23 Feb 1998 08:17:17 -0800
From: Robert Harris 
Subject: RE: where great minds go...
To: Dave Williams 

Of curse - what he just described is the K-Lamda system (mid generation)
which uses Dejection to vary the control pressure.  Last generation used a
pulse driven assembly to totally control the setup.

The K-lamda, complete with everything runs about 350 400 in a junque yard.
Cheaper if you know where to look.  Biggest hang up is the air sensor.

My solution is to blow off the air sensor whether MAF or MAP.  Using a
rising rate fuel pressure regulator tied to manifold vacuum.  Physical
dampening of "cam" effects ad nauseum.  What I originally wanted was to
mechanically approximate the fuel mixture without "measuring" air flow -
thus having total immunity to all air flow related crap and the requirement
to duct everything thru one sensor.

Then instead of driving the fuel distributor off the intake air, connect its
plunger to a cam off the throttle.

If the constant speed pump works for Bosch, it works for me.  Can use 2 4cyl
regulators or 1 six for each V-6. Still thinking about how to make a
volumetric pump roughly approximating engine demand.  Problem is Bosch
Injectors require 50 psi breakover pressure.

Ideal System.

A. Volumetric pump to cover RPM related fuel demand
B. Throttle related distributor to feed forward load ( A and B = Hilborne)
C. Rising Rate fuel pressure regulator to grossly compensate to load (MAP
sorta) demand
(Replaces Pill - acts to self tune and limit total fuel flow to engine)
D. Continuos O2 feed back after 30 second warm up to continually trim
including full load.
F. Lie to O2 sensor to get mid full power by inducing oxygen into exhaust
ahead of sensor
G. Massive demand related EGR for cruise - mid range economy. ( Maybe Mini
Dragon Charger )
( Replaces leaning out for fuel economy. )
H. Secondary set's of power injectors - upstream for bettor charge cooling.
( Demand activated. Takes fuel from after regulator and before
distributor. )
I. Air and Fuel Preheaters for low mid range
J. GM type knock controlled ignition.
K. Misc. stamp computer controlled crap - like fan off/on etc.

Date: Tue, 24 Feb 1998 21:03:15 -0800
From: Robert Harris 
Subject: RE: Complicating Dejection Crapping
To: Dave Williams 

A. Sounds like a balance tube used on most quasi IR systems.  Good Idea.
B. Using it for Idle -low range air intake rather than thru the throttle
plates.  But direct heating it with exhaust and keeping the air flow
separate aint a bad idea.  Want to really control reversion so that when we
introduce EGR, we control it.
C. Pure household ammonia should be at janitorial or restaurant supply.
Stainless and Mirrors/Glass really hate the piney shit.  Worse comes to
worse, poor side of town stop and robs generally stock it for cleaning
glass - much cheaper than the blue stuff.