By John De Armond
Introduction
‘This article is the first of a series on building and using a fuel
injector testing and cleaning machine. The machines easily built,
isbased onan IBM-PC and can costaslitle as $200 to build. This
first article will show you how to build the machine and will
present some basics of testing, Subsequent articles will go into
considerable detail about using the machine in a high perfor-
mance environment.
“The purpose of this machine fs fairly simple - To allow you to
measure and record allimportant parameters of an electronic fuel
Injector. The design and construction of this machine is also fairly
simple, Some of the underlying concepts may not be. Of course,
thal is why this magazine exists, to de-mnustify this stuff,
Some Basics
What is the purpose of any fuel metering system? Keep it
simple now. That's right, to create the optimum fuel/air mixture
Photo 1 - The fuel
injection flowbench,
front & back views, |
Performance Engineering Magazineunder all conditions, See, that was easy. This holds true
whether the device metering the fuel is fuel injection,
carburetion or someone pouring gas down the intake, As
with many things in life, while the theory is simple,
implementation is quite complex.
Let's presume for amomentthat the best mixture under
all conditions is the chemically comect one, 14.7:1 for
gasoline. Thatis, for every one part by weight of gasoline,
there must 14.7 parts by weight of air to completely bum
the fuel. This is known as the stoichiometric mbxture. 1
tend to abbreviate this as “stoch”, pronounced just like it
looks.
‘Achieving this, particularly with fuel injection, should be
easy, Just measure the ait low, calculate a proportional
amount of fuel and shoot it in. Problem is, physics rears,
its ugly head. Both air and fuel have the inconvenient
tendency to change properties with temperature and
pressure. Air is even worse because it changes density
\with pressure. If we didn't have all these considerations,
a simple, all mechanical system such asthe Hilbon would
‘work perfectly underall conditions. Reality dictates all the
fancy electronics we have today.
In order to configure any kind of fuel injection control
| system, one must know the characteristics of the final
actuator, the fuel injector. That is, the tuner must know
how much fuel per unit time the injector will deliver, how.
much delay after he commands the injector open before
the actual delivery starts and how long after he commands
it shut it actually stops flow. See the sidebar for more
details on injector parameters.
futhare
coi Figure 1 - Piping
Performance Engineering MagazineOverview
‘What must an injector tester do? Not that much realy.
Itmust supply the injector with a stable supply of fuel (or
other working fluid) at a known pressure. It must supply
the proper voltage for the injector. It must provide the
means to precisely and repeatedly actuate the injector.
Finally it must provide the facilities to measure the
injector's operating parameters. Thanks to the ubiqui-
tous IBM-tyle PC and a touch of hardware we will
fabricate or inexpensively buy, these requirements are
simply achieved.
In addition to a PC and the hardware we are going to
fabricate, you will need a couple of other pieces of
equipment. The most expensive san oscilloscope. Since
‘we are dealing with relatively slow events, most any old
scope will work. The cacillac of scopes for automotive
‘workis the digital storage scope. This instrument isreally
a special purpose computer that digitizes the signal in
question and displays it on a screen. Because the signal
is stored in the scope’s memory, it is visible until either
replaced with a new trace or erased. That means very
«low events such ignition or fue! injection cycles can be
viewed at will It also means that with the right equip-
ment, hardcopy of the trace can be generated
Inthe automotive context, the cadillacof digital storage
‘scopes (DSO) is the Fluke 97 Scope Meter. This hand-
held unit combines the functionality of a dual trace, 50
MHz oscilloscope and a digital voltmeter. See Photo 2.
All the scope traces and data presented in this series of
articles are generated from PE's Fluke 97. At $1795
reiail, this scope is within the reach of the professional
Performance Engineer and even for many hobbyists.
Note that this scope is not necessary in order to build and
use the fuel injection tester. Nice but not necessary.
One of the tests you will perform on a fuel injector is
to measure the opening and closing time. Thisis done by
‘mounting an accelerometer on the injector and record
ing the vibrations generated by the injector opening. See
Photo 3. An accelerometer is a device that converts
acceleration to an electrical signal. Typically the output
is specified in terms of signal magnitude per G of
acceleration. The accelerometer we use here at PF is a
fairly expensive commercial piezoelectric accelerometer
made by Endevco. These units typically cost $200 or
more. However, avery satisfactory accelerometer can be
constructed from an ordinary knock sensor.
Tester Description
The tester is pretty simple. A miniature fuel supply
Joop, almost identical to that found on the car, is set up
under controlled conditions. The loop consists of a
standard Fl fuel pump, filter, bypass regulator and pres-
sure gauge. See Fig, 1. The regulator pictured is a fixed
pressure unit from a Datsun 2802. For a more general
‘purpose tester, use an adjustable regulator available from
Photo 2 - The Fluke 97 Scopemneter
6 Performance Engineering MagazinePhoto 3 - Accelerometer and in-
jector clamping detail
a variety of sources. See the Re-
source List.
The pump is supplied with 12
volts through a switch and a theo-
stat, The theostat is provided to
adjust the speed of the fuel pump
ntl the bypass régulatorjust starts
to bypass. The purpose is to intro-
duce as little heat to the fuel as
possible. Fuel is supplied to the
injector through a flex hose and is
secuted via a homemade quick
action clamp. The injector is held
in the fixture by a comrhon wood-
working toggle-clamp. The accel-
erometerisattachedto this clamp.
Photo 4 illustrates this assembly,
You will observe that there is a
small flowmeter in the line to the
injector, This meter is optional
and is not necessary to conduct
any test in this series. The meter
is convenient but very expensive
so the design wasdevelopedwith-
out it.
Below the injector mount is a
spray catch-cup used to catch the
spray from the injector when
‘observing the spray patter. It is
fabricated from a 4" CPVC pipe
cendeap. Itis designed so it can be
quickly moved out of the way for
testsso requiring. See photo 1. In
this photo the cup is
white so it will show up
better. Itshould be paint-
ced flat black with appl-
ance epoxy paint so the
spray pattern can be
mote easily observed
The cup hanger is fabri-
cated from a piece of
gas line flattened in a
vice and drilled for at
tachment to the panel
The bends are accom-
plished with an ordinary
tubing bender available
at the car parts store.
‘The loops that hold the
cup to the hanger are
ordinary nylon wire clamps avail-
able from Radio Shack or other
electronic supply houses.
A quick peek at the schematic in
Fig. 2 reveals the simplicity of the
electrical circuit. The reason this
circuit is so simple is that the PC
does all the work. The printer port
adapter isnicely buffered. The soft-
ware does all the actual testing
Ain't PCs grandl?! The circuit pro-
Vides the ability to test both saturat-
ed mode and peakhold mode
injectors. Which mode isin effectis
strictly the function of the software.
The potentiometers are aciusted to
provide the proper peak and hold
current in peakhold mode or VR-
Photo 4 - Regulator,
flowmeter, pressure gauge
and piping detail. Note the
homemade T-handle clamp
at bottom-center.
Performance Engineering Magazine1 alone is used to set the saturated ofthe liquid in the reservoir in order
current for saturated-mode opera- 0 minimize foaming and evapora~
tion, Welllcoverthisaspect ingreat tion. Be sure to plugthe opening to
detail in the next issue. the reservoir with some fiberglass
cloth as shown, This a vapor barri-
Construction er anda fire stop.
The electrical constructions sim
‘The chassis of the tester is con _plicity personified. See photo 6 for
structed from plywood. All joints details. Basically, everything s sim-
areglued, braced andscrewed with ply screwed to the wooden chassis.
sheetrock screws. This makes 2 The heatsinks on the transistors
very sturdy but easily “machined” are not necessary and were includ-
base. A heavy coat of aerosol ep- ed in an early stage of develop-
coxyapplianceppaint providesanon- ment. Theyremain completelycool
stain finish. Control legends were tothe touch. None of the wiring or
created by laying outthe pattem in layout is critical. I recommend us-
CorelDraw andlaser printingthem ing shielded wire for the lead to the
onoverheadtransparencyfilm. The PC.
legend is ghied to the panel using Power for my tester is supplied
ultraviolet curable epoxy. If you by a regulated power supply. This
don'thave such esoteric glue(Mine is convenient and may be neces-
‘was stolen from Doreen's stained sary for some of the SAE specified
lass studio), superglue willdoffine, testing but is not required. A car
The piping is three-eighths inch battery will work just fine. If youdo
brake line: Iused metal piping be- use a regulated power supply, be
‘cause it is sturdy and fireproof, a sure to include an outboard filter
consideration that should not be — capacitorasillustrated on the sche-
underestimated. The relur piping _maticand in Photo 6D, This capac-
should extend to below the surface itor filters the switching transients,
Performance Engineering Magazine
(Baan in es#
Fig. 2 - Schematic Diagram
|
|
|
created by firing the fuel injector. |
Jeamed from experience thal these
transients will burn out the power
supply unless trapped. Most any
value capacitor larger than about
| 30,000 uF will work fine
The Software
‘Thesoftwareisamazingly simple
considering what it allows us to do.
{Its basic function is to tum the fuel
injector on and off for precisely
timed intervals and to total the
amount oftimethe injectorisopen.
The software uses the hardware
timer bull into each PC. The timer
is used in a high resolution mode
that gives us the ability to measure
time in microseconds. The actual
timer routines were taken from
some public domain code written
bby David Kirschbaum. This soft-
ware is writien to be as portable as
possible and has run on every type
of PC tested to date.
The operation is simple, Four
|values must be specified. These
+ The time the peak current flows.
Tipeak)
+ The total on time. Tlon)
Tote >
PiTlocald
Voltage
CURRENT——> VOLTAGE ——>
Coil Current
(OPEN >
Pintle Motion
ve
Fig. 3 - Idealized voltage
and current waveforms and
pintle motion for Peak/
Hold (solid lines) and
saturated driver (shaded
lines.)
Performance Engineering MagazineB
+ The total eycle time. Tleycle)
‘+ The total number of cycles.
See Fig. 3 for details, Once
these parameters are specifi,
the go command can be issued
to fire the injector. During each
cycle, both transistors (bits 0
‘and 1) are fired for Tipeakl
Then the peak transisior is
fumed off (bit 0} and the injector
is held on for the time interval
equal to Tlon) - Tipeak). The
hold transistor (bit 1) is turned
off for Ticycle) - Tlon) and then
the cucle repeats until the total
number of cycles are rin or a
key is pressed. A keupress ter
minates the test. The total on
time is accumulated during test
10
ing and ts displayed at the end of
each cucle. This time is vital for
computing the fuel delivery of the
injector. Thistimeaccumnlatorcan,
be reset at any time,
A second mode is the current/
purge mode. This mode simply
fires the injector continuously so
that the peak and hold current can
be set, Some caution must be used
Iecause excessive curentcan quick
ly bum out injectors. Because the
peak current is often many times
the rating of the injector, the peaks
‘on time is limited to one second at
a time. A second use of this mocle
is to turn the injector on continu
‘ously so air may be pinged after an
injector change and iorelease pres
sure prior to changing the injector.
Performance Engineering Magazine
D
Photo 6 - Wiring Details
A third mode is the software
calibration mode. Because every
computer requires a finite amount
of time to execute a sequence of
instructions and hecause the time
intervals we are dealing with are
very short, a. compensation factor
must be developed for each ma
chine, The software ships with
compensation valueappropriale for
my 25 MHz 386sx laptop. The
calibration mode allows you to al:
temately generate a 0.5 ms and a
5 ms pulse from bit 0 while chang
ing the compensation factor. You
would place the program in this
mode and then observe the pulse
output on an oscilloscope while
dithering the compensation factor
Until the pulse widths are exact.This compensation factor only
matiers when creating intervals
shorter than about 2ms. The value
vyou develop may be polaced in an
environment variable called
FIFCOMP. See the documenta
tion that accompanies the code for
details Please refertothe Resomnce
List for software availabiity
A_5VDC. B SVOFF
500ms/DIV SINGLE Trig: A
Cyele Starts
a Cycle Cvcle| Testing
End Peak
Begin Hold Tsuggest you test the mechanical
components first. Install an injec
tor, fill the tank with gasoline and
fire off the pump at full speed
Quickly look for leaks and verify
the pressure requlator is function
ing, Vary thespeed knob and verify
ihe pump changes speed, Turn off
the pump and bleed the pressure
byloosening the pine clamp on the
injectorand allowing the gasto leak
| Fig. 4 - Lamp test waveform into the catch bow!
Flectricaltestsareconducted with
Performance Engineering Magazine 1a small 12 volt, approx icles
mately 2 amp (tailight) Seeeeee) | ara),
lightbulb installed in place oes eel ee
of the injector. Connect
the PC and start the pro-
gram. This is important
because thefirstthing the
program does isreset the
parallel por. In my expe=
rience, abouthalfthe time
parallel port will power
‘up with the output bits
set; Le., the transistors
fired. This could, of
cose, burn out the in-
jector. Apply power and
verify thereis 12-14 volts
available at the injector
power switch. Ifthat isin
‘order, turn on the injec-
tor power switch. The
lamp shouldbe off. Iinat,
check your wiring. Enter the eur see the Resource List fora known lamp come on dimly for a second,
rent/purge mode. Fire the peak good parallel port card. Turn the go full intensity for a second and
transistor. The lamp should light, hold transistor off. Retum to oper- tum off for a second. If you have a
Tum the peakpot. Thelightinten- ate mode. Set the following pa- scope connected to the monitoring
Fig. 5 |
Standard
connection
diagram
sity should vary. Turn the peak rameters point, you should see a waveform
transistor off and fire the hold tran- similar to Fig. 4. Testing is com-
sistor, The lamp should again light. * Total cycle interval 3000 ms plete. Press any key to terminate
Turn the hold pot. The lamp in- + Peale interval 1000 ms the test
tensity should vary. In each case, + On interval 2000 ms.
with the lamp at full briliance(min-» Repetitions 100 Operation
imum resisiance), verily that the + Peak potentiometer fully
collector of the applicable transis _counter-clockwise, maximumre- This is going to be brief because
tor is switched within about a half sistance the majority of the next issue's
voltof ground. Ifitisnot,eitherthe +» Hold potentiometer fully clock- article wll cover operation. This is
transistor has low gain or your wise, minimum resistance. just enough to get you going,
parallel port has insufficient drive. Install an injector. Connect the
If insufficient drive is the problem, Run the test. You should see the _scopeasillustratedin Fig, 5. Setthe
A svoc 8 soonvc Fig. 6 200mW OC
A voc 8
Tmg/OW SINGLE _Tig:AL meV SINGLE Teg A
aay
Fig. 7
eI Performance Engineering Magazinescope as follows:
* Chan A - monitoring point, 2
volts per division
+ Chan B - accelerometer, full
sensitivity, usually 5 mv/division
* Trigger - Chan A, DC cou-
pled, negative slope
+ Sweep - 1 ms/division, Normal
trigger.
Set the peakand hold pots fully
counter-clockwise (maximum resis-
tance},
‘Sart the softwareand apply pow-
ce to the injector. Leave the fuel
pump off at this stage. Set the
following parametersforpeak/hold
injectors:
* Peaks time 5 ms
* Total cycle time 15 ms
* On time 10 ms
+ Repetitions 10,000
Start the test. You should see
some activity on the scope. Adjust
the trigger level fora stable display
similar to Fig. 6. The injector may
bemakingsomenoise but probably
is not opening. Start tuning the
peak pot clockwise. You should
hear the sound of the injector
change. At some point the change
will become marked.-This will oc-
cur when the injector gets enough
peak current to actually start open-
ing. You should see some activity
con channel B, the accelerometer.
You should see a vibration spike
shortly after the current is applied
and perhaps a second spike as the
Injector closes when the peak cur-
rent tums off. Ifthe injector closes
when the peak current tums off,
gradually turn the hold current pot
Until the injector stays on for the
duration of the on time.
You have now determined the
approximate peak and hold cur-
rent requirements of that injector.
You can back the peak time off
Lunt the injector fails to open. This
uill determine the approximate
peak time required. Thisisnormal-
Iyabout 1-2 ms Fig. 6 illustrates a
typical peak/hold waveform. You
‘may now go to current mode and
evaluate the currents you have set
For Saturated Driver Injectors,
Use the following parameters:
+ Peak time 0 ms
* Total cycle time 15 ms
+ On time 10 ms
+ Repetitions 10,000
Start the test and observe the
scope. You should see a display
similar to Fig. 7. Start tuming the
Holdpotclockwise, Atsome point
the injector should start opening,
The accelerometer should show
someactivity and the sound should
change. Continue increasing the
currentby further tuming the hold
pot and observe the opening de-
lay. The delay should decrease
rapidly at first and then remain
fairly constant. Just a bit more
than what is necessary to turn the
delay “knee” isusually about right
If the injector is a high impedance
{no ballast resistor) unit, the hold
pot should end up near zero ohms.
If a ballast resistor is used in an
existing installation with low im-
pedance injectors, the pot should
end up near that value.
Since we don’t have any fuel
flowing at this point, don’t run
these tests for too long lest you
‘overheat the injector. Once you
get the approximate electrical pa-
rametersset, voumay start the fuel
ump and actually inject some full
Working Fluids
For most testing, you will want
to use the actual fuel you will be
‘burning in the engine. If you plan
‘on testing a lot of injectors, you
might want to consider using N-
Heptane as outlined in SAE
41832, It has a lower vapor pres-
sure than gasoline and is much
more stable over time, If you use
«gasoline, you should use new gas-
cline each day. Pump the old gas
cout and use it in your hotrod
lawnmower or something,
Performance Engineering Magazine
This tester Is very effective in
cleaning dirty injectors. The worl
ing fluid I've found very effective is.
a half-and-half mix of gasoline and
Chevron Techron® fuel
cleaner. Simply clean the
con the outside and then connect it
to the tester. Run the injector at
about a 50% duty cycle for 15
minutes. The vibration of firing the
Injector appears to assist in the
cleaning operation. More on this
next issu.
In Closing
This just about wraps up the first
installment in this series. Use the
time to build your tester, acquire
the software, acquire a copy of
SAE J1832 and get everything
working,
13Resource Guide
Software
The software may be obtained from the following sources
* Courts of Chaos BBS 501 985-0059
+ Bye Mall Send request to lstserveidixie com
* By ial from PE Magazine Send a LargeSASE and $5.00
Fuel Pressure Regulator
‘A stock regulator from your car of choice may be used or an adjustable regulator may be
ordered from: Kenne Bell, 10743 Bell Court, Rancho Cucamonga, CA 91730, 714 941
6646, 714 941 0985 (tech support)
Electronic Parts
Q1,Q2TIP-120 Transistors Radio Shack P/N 276-2068,
DI Diode, 64, 50PIV Radio Shack P/N 276-1661
V1.2 Rheostat, 1002,12.5 watt Newark P/N 87F3611-50 [1]
VR-3.— Rheostat, 200,25 watt Newark P/N 8716390-20 [2]
S1S2 Toggle Switch, SPST. 3 amp Radio Shack P/N 275-322
BNC 1 BNC panelinount connector Radio Shack
TP-l Dual Bananna Jack Radio Shack:
- ‘Compatable Printer Port Card Magibonies P/N 8109 [3]
Toggle Clamp
This is a standard 1" woodworking toggle clamp available from most hardware
Bere ele ate sapped Clea Meta Peive Tool Song Mem ta, GA
404 446-8390,
Accelerometer
The knock sensor mentioned is a standard automotive knock sensor. Mine is from
a Dodge Turbo Lazer. ‘Iiue accelerometers can be had from
Endevco PCB Piezotronies Ine
30700 Rancho Viejo Rd. 716 684-0001
San Juan Capistrano, CA 92675
714 661-7231
SAE
11832 may be ordered from the SAE, 400 Commonwealth Dr . Warrenclale, PA 15096,
412 776-4970, 412 776-0790 (FAX)
Hard-to-Find Parts
The above marked items are available from Rapid Deployment Systems Inc, PO Box
670386, Maricita, GA 30066, 404 578 9547. Add $4 to each order for S/H, Allow 6 weeks
{or delivery.
[1] VRA,2- $10.00 each.
IZ] VR3- $15.00 each
{3} Card- $19.95 each