Fuel System Modifications
Thanks to Carlos Iglesias for
providing generous assistance with the information
below!
Fuel
Pumps
If you're
going to upgrade your injectors in the process of seeking more horsepower,
most likely you'll need an aftermarket fuel pump in conjunction with
an aftermarket fuel pressure regulator. Once you do that, you'll need
other aftermarket parts to plumb it too.
First, you
need to calculate your engine's fuel requirements. Max Cooper, a fellow
RX-7 enthusiast, has an excellent webpage to do exactly that.
Using Max's
calculator in conjunction with the maximum "upper limit" of my desired
parameters, I've come up with a figure of 432 RWHP or 508 HP. This
requires 205 liters per hour (LPH) at 75 PSI fuel pressure, and 20 LBS
boost (at a conservative pressure drop).

Click
for bigger.
There are a
number of pumps out there that will flow that much at a lower pressure.
The key is finding a pump that will flow more than the desired amount at
higher pressures. Aeromotive, SX and others have pumps that fit the bill,
so to speak, but there is little data regarding the actual flow rates and
most or all of these pumps are external- something I do not want. I
decided to go with an industry and motorsports proven pump manufactured by
Bosch. Lots of folks use the big fuel pumps Bosch made for big HP Porsches
and Mercedes. There is quite a bit of flow data out there as well for some
of these pumps. Beware, however, that there is relatively little clarity
regarding which Bosch pumps are making these flow rates. This is because
Bosch made dozens of pumps that flow at various rates and at various
pressures, so one has to be careful when selecting the right pump. For
example, Porsche used many different pumps for their cars. Below is a
table depicting some of these pumps (commonly used high flow pumps in
bold).
Vehicle
Application |
Year |
System |
OE Ref.
No. |
VM Ref.
No. |
911/911S 2.7
litre |
73-77 |
Bosch K Jetronic |
0 580 254
984 |
911.608.102.00 |
911/911S 2.7
litre |
73-77 |
Bosch K
Jetronic |
0 580 254 985 |
911.608.110.02 |
911 SC/Carrera 3 litre (6
cyl) |
75-80 |
Bosch K Jetronic |
0 580 254
984 |
911.608.102.00 |
911 Carrera
3 litre (6 cyl) |
75-77 |
Bosch K
Jetronic |
0 580 254 985 |
911.608.110.02 |
911 Carrera 3.2
litre |
83-89 |
Bosch Motronic |
0 580 464
021 |
944.608.102.00 |
|
|
|
|
944.608.102.04 |
911 Carrera/Targa 3.6
litre |
88 on |
Bosch |
0 580 464
058 |
964.620.104.00 |
911 Carrera
RS 3.8 litre |
94 on |
Bosch |
0 580 464 058 |
964.620.104.00 |
911 Turbo 3 litre (6
cyl) |
75-77 |
Bosch K Jetronic |
0 580 254
990 |
930.608.111.00 |
911 Turbo
3/3.3 litre (6 cyl) |
78-82 |
Bosch K
Jetronic |
0 580 254 984 |
911.608.102.00 |
911 Turbo3/3.3/3.6
litre |
83-94 |
Bosch
Motronic |
0 580 254
979 |
930.608.111.00 |
|
|
|
|
930.608.113.00 |
911 Turbo 3.3/3.6
litre |
77-94 |
Bosch K Jetronic |
0 580 254
967 |
911.698.102.02 |
924 Coupe 2
litre (4 cyl) |
76-78 |
Bosch K
Jetronic |
0 580 254 976 |
911.608.102.00 |
|
|
|
|
928.608.102.00 |
924 Coupe 2
litre (4 cyl) |
76-85 |
Bosch K
Jetronic |
0 580 254 984 |
928.608.102.00 |
924 Coupe 2 litre (4
cyl) |
76-85 |
Bosch |
0 580 254
992 |
810 906 091
B |
|
|
|
|
893 906 091 D |
924/924 S 2.5
litre |
81-88 |
Bosch Motronic |
0 580 464
028 |
944.608.102.01 |
924 S 2.5
litre |
85-88 |
Bosch
Motronic |
0 580 464 021 |
944.608.102.00 |
|
|
|
|
944.608.102.04 |
924 Turbo 2
litre |
79-82 |
Bosch
Motronic |
0 580 254 979 |
930.608.111.00 |
|
|
|
|
930.608.113.00 |
924 Turbo 2
litre |
80-83 |
Bosch K
Jetronic |
0 580 254 967 |
911.698.102.02 |
928 4.5 litre |
77-82 |
Bosch Motronic |
0 580 254
979 |
930.608.111.00 |
|
|
|
|
930.608.113.00 |
928S 4.5/4.7 litre |
77-82 |
Bosch K Jetronic |
0 580 254
967 |
911.698.102.02 |
928S 4.5/4.7
litre |
77-83 |
Bosch K
Jetronic |
0 580 254 984 |
928.608.102.00 |
928 CS 5 litre |
87-91 |
Bosch LH Jetronic |
0 580 464
045 |
928.608.104.01 |
|
|
|
|
928.608.104.02 |
944 2.5 litre |
81-88 |
Bosch Motronic |
0 580 464
028 |
944.608.102.01 |
|
|
|
|
944.608.102.05 |
944/944 Turbo 2.5
litre |
81-91 |
Bosch Motronic |
0 580 464
021 |
944.608.102.00 |
|
|
|
|
944.608.102.04 |
High-Output
Competition |
|
Bosch KE
Jetronic |
0 580 254
044 |
993.620.104.80 |
Additionally, the published flow rates for some of these pumps
(From Bosch):
Fuel
Pump Technical Data |
Part Number |
Flow Qty liters/hour (pressure) |
Inlet Connection |
Outlet Connection |
Length (mm) |
Diameter (mm) |
Max Power Cons (Amps) |
Electrical Connection (-/+) |
0 580 254 023 |
168 (5 Bar) |
open base |
M10 x 1.0 |
169 |
60 |
10 |
M5/M6 |
0 580 254 040 |
102
(6.5 Bar) |
open base |
M10 x 1.0 |
169 |
60 |
11 |
M5/M6 |
0 580 254
044 |
200 (5
Bar) |
M18 x
1.5 |
M12 x
1.5 |
196 |
60 |
15.5 |
M5/M6 |
0 580 254 046 |
207
(3 Bar) |
M14 x 1.5 |
M12 x 1.5 (F) |
169 |
60 |
10.5 |
M4/M5 |
0 580 254
053 |
175 (5
Bar) |
12mm
(3/8") |
M12 x
1.5 |
180 |
60 |
11.5 |
M4/M5 |
0 580 254 909 |
148
(5 Bar) |
12mm (3/8") |
M12 x 1.5 |
180 |
60 |
10 |
M4/M5 |
0 580 254
910 |
130 (5
Bar) |
15mm
(1/2") |
M12 x
1.5 |
203 |
60 |
10 |
M4/M5 |
0 580 254 911 |
95 (4
Bar) |
15mm (1/2") |
M12 x 1.5 |
199 |
52 |
5.2 |
M4/M5 |
0 580 254
975 |
165 (5
Bar) |
15mm
(1/2") |
M12 x
1.5 (F) |
180 |
60 |
10 |
M4/M5 |
0 580 254 979 |
165
(5 Bar) |
M14 x 1.5 |
M12 x 1.5 (F) |
168 |
60 |
10 |
M4/M5 |
0 580 254
984 |
165 (5
Bar) |
12mm
(3/8") |
M12 x
1.5 (F) |
180 |
60 |
10 |
M4/M5 |
0 580 464 069 |
98 (4
Bar) |
12mm (3/8") |
M12 x 1.5 |
186 |
60 |
7 |
M4/M5 |
0 580 464
070 |
130 (3
Bar) |
12mm
(3/8") |
8 mm
(5/16") |
175 |
52 |
6.5 |
M4/M5 |
0 580 205 413 |
200
(8 Bar) |
M18 x 1.5 |
M12 x 1.5 |
196 |
60 |
15.5 |
M5/M6 |
We see that
a number of pumps flow high rates at high pressures, namely the pumps
ending in part numbers 979, 984, and 044. Most of the flow numbers
published out there are for the former two models (click below for
bigger).
Now we need
to go back to some of the flow data to determine how accurate the flow
estimates from Bosch compare to the tested rates. In general, Bosch flow
rates are conservative. This is primarily because they use N-Heptane to
derive these rates, a chemical with different viscosity and density than
gasoline. For example, the Porsche "979" pump (Bosch number 0 580 254 979)
is rated at 165 liters per hour (LPH), but has been easily flow tested to
190 LPH. Based upon the published and tested rates, I decided to go with
the Bosch "044" Motorsports pump (0 580 254 044) used in the 993 Supercup
cars and others. This pump is rated by Bosch at > 200 LPH at 5 Bar
pressure (73 PSI). The 044 pump has been flow tested to 228 LPH, which is
about 480 RWHP or 565 HP using my specifications (click below for
bigger).

Realistically, my target is less horsepower than the
upper limit to provide less strain on the fuel system and engine.
Using 400 RWHP as my target, the fuel system
requirements are 190 LPH or 3.17 liters per minute (LPM) at 15 PSI boost
(70 PSI fuel pressure), which is 20% lower than the conservative upper
limit.
The 044 pump should prove more than adequate. This pump
can be purchased from Truechoice.
Fuel Pressure Regulators
There are two type of fuel
pressure regulators (FPR) to be considered for the 3rd gen RX-7. One is a
linear or standard FPR and the other is a rising rate FPR. Both can be
adjustable. The standard regulator will provide x amount fo fuel pressure
at idle and will increase the fuel pressure one pound for every PSI
increase of boost. The second type will add fuel at a multiple of boost,
say 2 LBS of fuel pressure a 1 PSI, 4 at two, etc. There are a number of
considerations dictating what you should select, but I'm not going to go
into detail here. For my RX-7, a standard fuel pressure regulator with
little pressure drop and consistent fuel pressure regulation (with little
variance) will work fine.
You can find FPRs from SX, Aeromotive, Bosch, etc.
These are mostly high quality units with options for different fittings,
adjustability, and options for other items such as aftermarket fuel
pressure gauges. I decided on the adjustable SX regulator without the
pressure gauge port for simplicity. This regulator is purportedly good for
1000 HP. I don't know of anyone cranking out that much HP with one of
these, but I do know of many folks with 500+ HP RX-7 using them with zero
problems.
Fuel Filters
Same story for fuel filters. Lots of vendors with lots
of good items. I decided that the SX 41001 inline filter was the best
choice given there wide use on high HP cars (other good choices are the
Canton Mecca inline filters).
There are two models of SX filters: the 41002 slim line
series and the 41001 series. The 41002 is a smaller diameter filter that
is less expensive, while the 41001 is a larger filter with an indicator
for filter element replacement. Both of these provide almost zero pressure
drop (.15 PSI at 200 GPH or 757 LPH!) and flow the same rates, and can
easily support 1000HP. SX offers paper and metal elements which are easy
to replace. In addition, the filter can be mounted relatively
easily.
Installation The most difficult part of
the fuel system is mounting or mating the aftermarket parts to the stock
fuel system. I have to install a different fuel pump, filter, pressure
regulator, fuel rail, and secondary injectors. I will highlight the
details for the install as well as the parts required to plumb the new
system.
My initial plan is to use braided -8 lines from the pump to
the fuel rails. Prior to going to the rails, I'm going to split or T the
-8 to -6s. On the other side of the rails, I'll be using -6 lines T'd into
the FPR.
I'll be doing something similar to this using Aeroquip
fittings and hoses:

Final fuel configuration. Merely a suggestion,
BTW.
Parts list for fuel system:
Big parts
Bosch 044 fuel pump
SX fuel pressure regulator with -6 ports
SX fuel filter with -10 ports
K2RD fuel rail with 1/4 NPT ports
Bosch 1600 cc secondary injectors (2)
VDO 100 PSI fuel pressure gauge
Plumbing (all parts are Aeroquip unless
noted)
1/4 NP to -6 flare adaptors (4) for fuel rails
-6 90 degree hose ends (3): 1 from fuel dist block; 1 from
secondary fuel rail; and 1 from FPR
-6 T (1)
-6 straight hose ends (8)
-6 1/8 straight gauge adaptor from Earls (1)
-6 female coupling (1) for pressure gauge adaptor to -6 on
FPR
-6 to -6 flare adaptor (3) for fuel pressure regulator
-6 end cap (1)
-6 Startlite kevlar hose (estimated 10 feet)
3/8 NP to -6 flare (2) for fuel distribution block
3/8 x 2 by 1/2 x 1 fuel distribution block (1)
1/6 allen plug for fuel dist block (1)
1/2 -8 flare adaptor
-8 straight hose end (3)
-8 Startlite kevlar fuel hose (10 feet)
-8 90 degreee hose end (1)
-10 to 08 flare adaptor (2) for fuel filter
-8 bulkhead fitting with nylon washer and -8 nut (these must be
purchased seperately)
-8 female coupling
12mm 1.5 to -8 adaptor from Earls (1) for fuel pump
outlet
18mm x 1.5 to -8 adaptor for fuel pump inlet
Hoerr racing fuel pickup kit (includes 1/2 inch fuel hose -8 brass
barbed adaptor, and fuel pickup sock
-6 to 1/4 NP 90 degree fitting (to mate fuel pickup sock to 1/2"
fuel hose
-6 O rings for FPR and fuel rails
-8 O-rings for fuel filter
The stock primary fuel rail needs to be modified o accept the 1/4
NP -6 fittings. You must tap the rail (or have someone do it) or have -6
aluminum ends welded to it. I opted for the former.
The Installation
Fuel System Installation And How to Make a Surge
Tank

What a mess!
This installation took much more time than I had envisioned. I
planned on doing this over two weekends. The car was actually in the
garage for five. The reasons for this were that I was doing more than just
re-plumbing the fuel system, I was breaking things when removing or
installing the fuel system, and I discovered that I needed more AN
fittings as I got to the injector rails.
Instead of just running AN
lines from the fuel pump I attempted six “projects”:
1. Installing
new fuel lines to the injectors and fuel filter 2. Installing a new
fuel pump and pickup 3. Installing a new secondary injector rail with
injectors and fuel pressure regulator 4. Fabricating and improving upon
the stock fuel tank baffle- effectively making a surge tank. 5.
Simplifying the vacuum hoses and solenoids under the intake
manifold. 6. Configuring the ECU to accept the larger
injectors
What was I thinking? In fact, number “4” required two
weekends alone.

Basic setup is finalized except for the outlet from the
secondary rail. This is explained below.
I’ll review each project individually.
New fuel lines
to the injectors
This was the easiest part of the installation.
I simply ran a -08 AN line from the fuel tank cover around the rear
subframe and along the side of the stock metal fuel lines. The advantage
of using the Aeroquip kevlar StartLite hose is assembly. The hose is
amazingly simple to cut. Simply place the hose into a scissor type hose
cutter and whack, you’re done. No fraying stainless to deal with. Yes the
hose is more expensive than stainless braid, but it’s 40% lighter and well
worth the price given the simplicity of installation.
I had a
difficult time finding a place for the SX fuel filter, but decided to
utilize one of the existing fuel line brackets. The picture below depicts
the installation.

Fuel filter location. Inlet is on the right.
Looking toward the front of the car (passenger side is on the
right).
From the fuel filter, I ran a 90 degree -8 fitting and
StartLite hose toward the front of the car. I secured the hose with rubber
coated stainless steel straps and mounted it very close to the stock
lines. Thus, it’s safely tucked under the car and snugs up to the bottom
enough to prohibit scraping on large speed bumps, big obstacles in the
road, etc. Where the lines bend upward in the engine bay, I notched the
plastic fuel line cover and brought the hose inside and upward to the
engine bay.
I placed the fuel distribution block (FDB) right
next to the brake vacuum reservoir. From here, the fuel line splits into
two -06s, one going to each fuel injector rail. I’ll discuss the rear of
the car before coming back to the front.
Installing a new fuel
pump and pickup
Based upon my fuel calculations and desired HP
goal, the upgraded NipponDenso pump wasn’t going to cut it. I did an
extensive amount of searching and found that the Bosch Motorsports pump
would work just fine (see link). This pump is also an internal and
external unit, and I wanted to keep it in the tank given the space
constraints under the car. Since I planned on installing this pump in the
stock location, I needed to figure out the plumbing and the fuel pickup.
Aeroquip makes the fitting for the pump inlet, and 18mm x 1.5 to -8, but
they do not make a -8 for the pump outlet which is 12mm x 1.5. I found the
right fitting from Earls.

Stock fuel pump assembly. With this
configuration, the bottom of the pump rests at the bottom of the fuel tank
and baffle.
For the fuel pickup, I looked nearly everywhere to find a fuel
sock/strainer that would mate to a 90 degree connector and attach to the
bottom of the pump. I couldn’t find anything that would work, so I
modified a fuel pickup kit ($30) sold by Hoerr Racing (www.HRP world.com).
I believe this is the same ½” pickup up assembly sold by Fuel Safe. The
fuel sock was mated to a nylon barbed connector. Upon receiving this piece
I cut off the nylon barb and drilled out/tapped the outlet for a ¼ NPT
connector.
For the pump outlet, I mated the -8 12mm 1.5 to a
-8 female swivel and then to a -8 90 degree bulkhead fitting going through
the tank cover. To do this, you need to remove the stock outlet pipe. I
couldn’t get a cut-off wheel close enough to cut the top (outlet) of the
fitting so I could drill it out, so I used a hacksaw. I was amazed at how
thin this line was. After cutting the top of the line off, I drilled the
remaining hole to 3/8” which removed the lower part of the line (save this
line!). I used a ¾” unit to make the hole large enough for the -8 bulkhead
connector. The 3/4” hole is perfect.
Once this was done, I
test fitted the pump. I left the fuel level float and bracket attached to
the pump cover. This serves as support for the new pump as well as a mount
for the fuel level sensor (of course). It also is a handy measuring stick
as the 90 degree splayed bracket at the bottom indicates the bottom (or
nearly so) of the fuel tank/baffle.
The pump fit nicely to
the bulkhead connector, but the hose barb connector and ½” fuel line that
cam with the pickup kit was way too long. I decided to cut the hose about
1” long and mate the barbed connector to a 90 degree -6 to ¼ NPT fitting.
The NPT portion of the fitting was screwed into the tapped nylon fuel
inlet sock. Holding the stock pump with the old metal fuel line (remember
I said to save this) along side the new setup I could clearly see that the
pickup was too long and would not allow the tank cover to fit due to the
length. The stock length is about 11 3/8” to the bottom of the pickup. I
ended up cutting the hose barb and the 90 degree fitting a bit to adjust
the length. Then I used worm type clams to hold these two pieces together
with the shortened (again) ½” fuel line. I also used 3” worm clams to
secure the fuel pump on the metal bracket as I heard that the thank cover
can break where the bulkhead connector inserts.

Big Bosch pump on the stock fuel pump assembly.
Note how high the pump sits compared to the stock unit. This is
important as the bottom of the pump is approximately 3.7" from the bottom
of the tank. The baffle is 3.5" high. Therefore the hole for
the baffle cover (top) can be made smaller than using the stock setup
(since the hole does not have to be cut for the fuel pump) . UPDATE-
pump needs to be electrically isolated from the metal bracket. I
used 3-ply silicone and nylon tie wraps.
Installing a new secondary injector rail with injectors
and fuel pressure regulator
I purchased the secondary rail from
Xcessive Motorsports (now GroundZero Motorports). This is an aftermarket
3rd Gen secondary rail with ¼ NPT fittings and setup for top feed
injectors. The Bosch 1600 cc units I purchased plug right into the rail
with a little lube. I test fit the rail and injectors to the lower intake
manifold (LIM) to figure out the plumbing. Since I’m using parallel feed,
the inlet was simple, a ¼ NPT to -6 port fitting and a 90 degree -6 which
will connect to the top of the FDB.
The outlet wasn’t so
simple. A 90 degree -6 would hit water outlet (to the throttle body) on
the water pump. I thought about using a 90 degree ¼ NP to -6 connector,
but didn’t like its sealing properties. I wanted to use a port fitting for
both sides of the rail. I also couldn’t figure out how the outlet of the
primary fuel rail was going to connect to the outlet of the secondary.
After much (and I mean a lot) fiddling, I found that a tight radius 180
degree 06 fitting connected to the port adaptor would work nicely when
connected to a -6 tee fitting with another tight radius 180 and a 4”
(roughly) run of StartLite hose. The other side of the tee would mate to a
female swivel and 90 on the primary and the third to a 90 that will go to
the fuel pressure regulator. See pic below.

Final plumbing setup for the fuel return. The
line and connector to the FPR attaches to the left side of the T
fitting.
The other part of the equation was figuring out how to properly
seat the secondary injectors in the LIM. I’ve read how many people remove
the fuel diffuser and cut a piece of 5/8” hose to fill the gap between the
bottom of the injector (above the o-ring) and the LIM. Then you simply
drop the injector to seat on the ledge in the LIM. This didn’t look like a
good fix since the o-ring and injectors still have room to move around.
Additionally, the pintle cap on the injectors has a lip on the top of the
cap that interferes with the fuel diffuser and/or the smaller holes in the
LIM. Many folks simple remove the caps, which affects the fuel
distribution pattern.
After pinging a few folks I came up
with what I feel is the best solution. I ground the lip off of the pintle
cap so it would fit into the LIM/fuel diffuser and ground the ridge off of
the top of the fuel diffuser so it could be maintained and so the injector
o-ring would seat better (Remark- I have removed the fuel diffusers
altogether to good effect). To seat the injector in the LIM, a fellow
RX-7 buddy named Dave manufactured 2 aluminum spacers or sleeves that plug
snugly into the LIM and are radiused so that the injector o-ring will
slide into and seal perfectly. This should form a tight seal and will hold
the injectors firmly in place.

LIM/Injector sleeve

Sleeve resting in the LIM secondary injector
port.
The last part of this modification is what to do with the stock
fuel temperature thermosensor that was inserted into the stock primary
rail. The ECU, even the PFC, will adjust the fuel mixture as fuel temps
increase. There is no provision for the thermosensor in the aftermarket
rail. Some people have simply attached the sensor to the secondary rail
with silicone and left the ECU programming alone. This is an ignorant
solution as the outside of an aluminum fuel rail next to a very hot intake
manifold and engine block is going to be significantly hotter than a
sensor bathed in fuel inside the rail. I decided that I would tap the fuel
distribution block for the sensor at a later date and install the sensor
there. For now, however, I simply attached the sensor to the connector and
ziptied it to the top side of the engine harness. I adjusted the PFC
settings using the Datalogit to refrain from adjusting the fuel mixture
when temps increase (simply change “INJ vs. Fuel Temp” on Settings 2 to
1.000 in all three cells).
For the primary rail, I tapped the inlet
and outlets for 9/16-18 fittings. The inlet was a simple 9/16 18 Earl’s
port fitting (Aeroquip does not make a 9/16-18 fitting) and a -6 straight
connecting to the FDB. I chose the 9-16-18 as they provide a better seal
than NPT fittings. I didn’t want any leaks, particularly where I could not
see them. For the outlet I ran into another dilemma: every fitting I tried
interfered with the oil metering nut/barb. This is because the outlet of
the primary rail sits lower than the inlet. My solution was a 9/16 -18 90
degree port adapter that connects to the -6 tee described above. This
still hit the oil metering nut, but I was able to grind off enough excess
material on the nut to allow the injector to seat. Note that I always use
Teflon tape for the non AN fittings.

New primary injectors, just in case.

Final primary configuration. Outlet is on the left,
of course.
When I was tightening the oil metering nut after grinding, I
got a bit overzealous and snapped the aged and brittle plastic oil
metering line. Another call to Mazda Competition and more money/time
wasted.
The fuel pressure regulator was simple to plumb, but
difficult to find a place to mount. I occasionally use my cruise control
(believe it or not), so I didn’t want to mount the FPR there. There wasn’t
much room any where else, particularly with the fuel pressure gauge
requirement I had. I decided to mount it on the bracket that hold the
relays on the driver’s side- right under the strut brace. The line from
the tee under the intake manifold will go toward the firewall , run by the
brake booster, and connect to a -6 straight, then a -6 gauge port
connector with a 1/8” gauge port and VDO fuel pressure gauge to a -6 port
connector. The other side of the FPR is a -6 port connector and a -6 cap.
The bottom has a -6 port connector with a -6 45 degree fitting going to
the fuel return. This line is cut on the return side (i.e., no fitting),
and is double clamped to the metal return line on the firewall via fuel
hose clamps.

Final layout of the fuel system.
Fabricating and improving upon the stock fuel tank baffle-
effectively making a surge tank
Fuel starvation or fuel surge
(more like lack thereof) is a common problem for 3rd generation RX-7s.
Most, if not all, of the time this occurs when making sweeping left hand
turns under ½ a tank. I have experienced this with as much a 2/3 of a tank
on certain tracks.
A quick glance into the RX-7 fuel tank
reveals why this is the case. Mazda spend a good amount of time making a
fuel baffle that keeps gas around the pickup under moderate cornering, but
anyone with common sense can see that the passenger side of the baffle,
since it is not covered, will allow fuel to run ways from the pickup at
moderate to high cornering loads. Even thought the baffle is fairly deep,
it is not deep enough to prevent fuel from spilling over the top of the
baffle and toward the passenger side of the tank. During late 1995, Mazda
was apparently aware of the problem and revised the baffle. What they did
was to simply add a cover for this baffle so the fuel wouldn’t spill out
so easily. People that autocross with this improvement *(you can order the
96 and newer tank with the cover for about $350 from Mazda Competition)
note that fuel starvation is less frequent, but still occurs at about 1/3
tank. The reason that fuel surge still exists is that the hole in the
cover is too large and too near the passenger side of the baffle. Fuel
still spills out, but not as quickly. Additionally, fuel can only enter
the baffle (now affectively a “tank”) through the hole in top or via a
small inlet on the passenger front side.

Stock baffle pre 1996.
I decided to do what Mazda did, but to do it better. One
issue that needed to be addressed was getting the fuel into the tank when
it has a cover on it. A very knowledgeable road racer friend of mine told
me about this fix. What he did was closed off the stock inlet (see
pic) and punched three holes in the sides of the baffle adding three one
way ball type check valve. That way fuel can enter when decelerating,
decelerating a turning and steady state, but stay in there while turning
and/or accelerating. The other issue is to make the hole (in this case the
entire top of the baffle) in the top smaller so fuel cannot spill out so
quickly.
I found the check valves from Fuel Safe and
purchased some 18 gauge aluminum sheet to fab a
cover.

Fuel Safe one way ball/check valve. These are 1" in
diameter.
I drained the tank and crammed in a piece of cardboard to trace
the pattern for the cover. Once the pattern was drawn, I made a prototype
cover with extra material on the sides that would allow me to create tabs
for bending/affixing the cover. With much pain (the metal gas tank braces
inside and the tank access hole are very sharp) and blood, I had a working
cover. Unfortunately, I break things quite often. In securing the baffle
for test fitment, I broke two of the mounting tabs that hold the baffle to
the tank. More on that later.
To figure out where to cut a
hole in the prototype cover I had to mount the fuel pump and pickup on a
fabricated Plexiglas cover that mimics the stock fuel tank cover so I
could see where to cut. Once done I had to remove the cover and make a
real one. The cover does not simply fit through the access hole on the way
in or out. You must carefully bend it to get it out and straighten it once
inside the tank.

Plexi cover holding pump. You can see exactly
where the pump will drop into the baffle.
Now the check valves. I needed to find a way to cut holes in
the plastic baffle for the check valves. I used an air drill with a 1"
hole saw. Sparks are not a good idea, even with an empty tank. Make sure
you mount these valves so that they will not interfere with the tabs that
hold down the baffle. A picture of the check valves installed is below,
but incorrect. The valve in the upper right hand needs to be more toward
the rear of the car or the fuel pickup will hit. I had to use some Al
sheet and gasket to repair this hole in the baffle. I then moved this
valve rearward of the tab mounting spot- very close to the back of the
baffle The pic of the "cocked" check valve on the left is okay, it has to
be there so that it will not interfere with the valve in the back of the
baffle. It's cocked because the baffle is slightly curved at that
location. It does not leak, however.

Check valves installed into the baffle pan. The
valve on the upper right had to be moved. I placed it below the tab
(barely visible) so the fuel pickup sock would fit and rest at the bottom
of the lowest part of the tank (pretty much where the erroneously placed
valve is at the moment).
Once you install the check valves you have to plug that hole in the
upper right hand corner as in the pic (toward the passenger/front of the
car). I used a 3/8" piece of fuel hose and a zip tie.

Plugged stock inlet. That is, of course, other than the
huge hole on top!
Fixing the broken stuff. The mounting tabs have to be bend
downward to hold the baffle pan to the bottom of the tank. The bottoms of
these are welded to the tank. Since I broke two of them I decided the best
way to secure the baffle to the tank would be to employ the remaining part
of the welded tabs. I drilled a hole in the middle of these tabs and used
a marker to indicate where to drill the baffle. Once done, I used a 1/4”
bolts going through the baffle and into the holes in the tabs. I didn’t
secure the baffle tank as I still had to mount my finished cover. Mounting
the cover was easy- I stared by using sheet metal screws in a few choice
spots. Once the cover was reasonably secure, I used pliers to bend the
excess around the lip of the plastic baffle. This forms a good seal
against leaking fuel.

Busted and repaired baffle tabs.

Bolt that will mate to the drilled baffle tab. Holes
below (actually above as the baffle is upside down in the pic) need to be
plugged. An experiment gone awry. Note blood on baffle
:)
I secured the tank and dropped the pump in an attempt to finish the
job. Since the mounting hole for the pickup is intentionally at an angle
(to inhibit fuel from escaping while cornering), I had to rotate the pump
assembly roughly 30 degrees and use a long probe to stick the sock/pickup
in the hole. Once again Murphy reared his ugly head. The pump cover would
not fasten to the tank at the rear mounting point- something was
interfering. The back of the pump bracket was hitting the tank cover’s
access hole toward the front of the car. I tried pushing the lip of the
hole downward, but could not make the hole large enough. I really didn’t
want to remove everything and have to reinstall it. I ended up cutting off
the bottom of the bracket just below the bottom of the pump. This worked
very well, but with the bracket removed, I could have made the pickup
insertion hole even smaller. I guess another day after I get some testing
in.

Preliminary cover. Note how the upper right hand
side offers limited protection against surge due to the small amount of
material covering the baffle. The fix is below.

The fix was a sectioned and cut drain flange from a
hardware store. This cover offers better protection against left
handers as now the fuel is retained more effectively due to the lip/flange
that protrudes 1" below the cover.
How large is the "new" surge tank? The surge tank
measures roughly 9.5" x 6.5" x 3-4". The depth of the tank varies as it
slopes upward and has an additional amount of volume on the driver's side
(see check valve pic above). This is roughly equal to 210 cubic
inches or 3.5 liters. Assuming I'm running the pump at max capacity
for my desired goal this equals about 197 LPH of fuel consumption.
Thus, the tank would be pumped completely dry (assuming no fuel flow back
in to the tank) in 65 seconds. Of course, there will be some fuel loss due
to sloshing out of the top of the cover. Let's assume that 1/2 of
the tank was emptied due to slosh. This would still require 32.5
seconds to pump dry at full boost. I feel that this is more than adequate
as I cannot imagine turning left under steady state conditions and at full
boost for that amount of time.
Simplifying the vacuum hoses and solenoids under
the intake manifold
The last
part of the job actually occurred throughout the installation. As I was
removing things to get to the fuel rails and oil metering nuts, I decided
that now would be a good time to chuck most of the vacuum hoses since I
was going to be removing them when switching to single turbo. I had
already removed the ACV solenoid and double throttle and the EGR was
next. (Remark - all of this is gone now that I've removed the
sequential turbos)
Wade Lanham had done the very same thing and
provided generous explanation for this mod. He created a very nice
schematic depicting the hose layout. I stuck to this layout, but mounted
the turbo control solenoids and pressure tank a bit differently.
Essentially, I turned the solenoid rack upside down and rotated it 180
degrees to raise it higher off of the engine block. Everything is plumed
exactly as Wade indicates, however. I also removed a bunch of
miscellaneous crap from the solenoid rack assembly which required direct
plumbing of the fuel catch can (that is, I removed the metal line going
from the catch line to the throttle body) and some other
re-routing.
I used 1/8” ID by 3/32” Viton hose from McMaster Carr
to plumb the solenoids. This stuff is expensive, but is good to 400
degrees (silicone is about 250) and resists fuel better than silicone. I
mated all 3 way connections using nylon 5/32” hose barbs and zip tied
everything together.

Simplified vacuum rack. This rack hold four solenoids,
including the turbo control solenoid that used to mount to the lower
intake manifold (near the ACV). The rack used to house 8
solenoids.
ECU changes for the enlarged secondary
injectors.
Once that was done I put everything back
together. The only remaining item is to re-program the PFC to accept the
different injectors. This was very simple as Ralph Friend from Ground Zero
Motorsports in Portland, OR walked me through it on the phone.
The initial settings I used for the PFC are as follows:

Basically, the only thing I changed was the lag for the
secondaries. After tuning, I’ll no doubt have to adjust the overlap
settings and transition, but this should be good enough to run around for
a bit.
Remark- these setting work great. No revisions are
necessary.
Checking for leaks
After installing everything, I
started the car to check for leaks. It took a long time to get the car
started, primarily to prime the empty fuel lines. After running for a few
minutes, I noticed a healthy fuel leak at the back of the fuel rail where
the ¼ NPT port adaptor attaches. I had to really tighten this down to the
point to stripping the threads. Once that leak was fixed, I noticed
another at the front of the same rail/same port adaptor. Same process.
Unfortunately, you cannot see the primary rail to determine if it is
leaking in the same area. You can “sort of” see the front of the primary
near the flange and I didn’t see any fuel dripping at all. The rear is a
different story, even with a small angled mirror and a very bright light,
I could not see nor illuminate the area in question. This is a major
concern as an unnoticed fuel leak could spell disaster. I hope that the
9/16-18 threads seal better than the lousy NPT fittings. There were no
leaks on the AN connections whatsoever!
After the car got
good and hot, I noticed another leak in the coolant line where I replaced
the stock hose with an AN line. I had forgotten to tighten this line. Once
done, it didn’t leak at all, but by that time enough fuel had dripped on
the top of the engine block that I decided to give it a shot of water for
dilution.
I turned off the car and let it cool for a while. After
an hour or so, I started the car again. Same story with regard to taking a
long time to start. I think this is due to fuel pressure loss in the fuel
lines since there is no longer a check valve to hold
pressure.
I let the car rest for a day so that the puddles of
fuel and water would evaporate. Tried starting the car again, and nothing.
I ended up draining the battery!
I’m not sure if it was loss of
fuel pressure again along with getting connectors wet. To remedy the fuel
pressure problem, I installed an Earls once way check valve I had
purchased earlier between the fuel tank outlet and the fuel filter inlet.
This should hold pressure.
After doing this, I also disassembled
the intake manifold to double check the tightness of all of the fuel line
connections. I noticed that the coolant temperature sensor on the back of
the water pump assembly was completely broken. I’m not sure if this is why
the car wouldn’t start, but I had to order another and reinstall to see.
Doing some homework, I heard that a broken or malfunctioning coolant
thermo sensor will cause the ECU to richen the fuel mixture. I assume that
this condition and the higher base fuel pressure might be contributing to
difficult starting.
The next day I received my replacement
thermosensor from Mazda Competition (now Mazdaspeed). These guys are
great. You get parts at cost just for competing in two events per year. I
reinstalled the sensor, put everything back together and viola, the car
started right up! I let it run for about 20 minutes and did not see any
fuel leaks. Next test is a short drive.
After a test drive, I found no leaks, but the sequential turbo
operation was not working properly. After removing the intake
manifold twice, I discovered that I did not have the lines to the LIM in
the correct order. This is very important as the LIM is not simply a
source of vacuum or pressure. Some of these lines simply go right through
the manifold and must be in the proper order. After making the swap,
the car runs perfectly. Secondary injector transition is beautiful using
the settings listed above.
A week later I noticed a slight fuel
smell coming from the car, but it wasn't from under the hood. It turns out
that fuel was actually leaking from the tank cover where the 90 degree -8
bulkhead fitting was and one could smell it inside the car since that's
where the tank access point is located. I used two o-rings: one on
the outside and one on the inside to fix the leak.
Another week and another leak. I started noticing a strong
fuel smell when accelerating past 4,500 RPM. I concluded that it was
related to the secondary injectors as they are brought online at 4,500
RPMs I had to disassemble the intake manifold, remove the solenoid
rack, coils, and secondary rail where I found a cracked injector o-ring
where it mounts into the lower intake manifold. It appeared that the
injectors were misaligned when the mounting bolts were torqued causing
undue angular stress on the injector o-rings. Additionally, I was
not comfortable with the way the aluminum injector/LIM sleeves were
sticking above the LIM and cutting into the injector seals (above the
o-rings). I decided to remove the fuel diffuser/air separators so
the sleeves would fit entirely into the LIM and to utilize studs for
mounting the secondary rail. I used Honda manifold studs which were
the correct length and pitch for the LIM. The and injectors seat
perfectly now and I've no leaks after nearly two weeks.
Update 6-10-04
Starting to smell fuel again. After much searching, I found
another leak at the fuel tank cover -8 AN bulkhead fitting. I fixed
this leak with liquid teflon.
Update 6-19-04
More leaks, more problems. I'm starting to re-think this
mod. Car began stalling at stops. I found a bad dashpot and
improperly adjusted throttle cable. New dashpot and throttle cable
helped, but car still stalls occasionally. About the same time the
car started to hesitate a bit upon throttle tip-in. Since I haven't
driven the car very much after the initial fuel system install it's
difficult to determine if this was due to the actual install of the fuel
system, a leak, or something else. And...I smell fuel again. I
disassembled the fuel system and checked once again for loose
fittings. I couldn't find anything and have narrowed the problem to
two possibilities:
1. electrical or CPU related. That is, electrical such as
fuel pump grounding issues and CPU as tuning related
2. Leaky injectors. More likely the primaries since I have
slight hesitation upon throttle tip in.
I've a solution underway for number 1 which includes further
isolation of the Bosch fuel pump via a rubber fuel line (between pump and
AN bulkhead fitting as well as a rubber cover for the fuel catch
tank. While I figure that out, I ordered and installed new o-rings
for the newish primary injectors as well as a different AN fitting that
does not interfere with the rear OMP bolt (found interference at
last install) on the primary fuel rail. Also installed new primary
fuel rail isolater bushings.
We need to do some more driving to see if this helps at
all.
Update 11/04
Car runs great. Had it tuned at GroundZero where Ralph
Friend dialed out more fuel across the RPM range and at idle. No
more stalling! No fuel leaks either.
Fuel Pump Re-wire
I discovered a possible issue with the fuel pump as the fuel
pressure line deviated somewhat with boost around 5000 RPMs during my dyno
runs. However, I'm not sure if this was due to pump limitations, a
slipping clutch, or voltage drop to the fuel pump due to insufficient
wiring. Several colleagues have noted a substantial current drop to
the fuel pump due to the narrow gauge wiring and long electrical runs to
the pump. I'm going to re-wire the fuel pump for the next series of
dyno runs.
Below is a description of how to do this.

Typical Bosch SPDT Relay
The terminals of a relay are
defined as follows:
-
30 is the common or input voltage
(or ground) to be switched.
-
87a is not used
-
87 is the normally open
connection (switched voltage output when the relay is energized).
-
85 is connected to the ground of
the triggering voltage.
-
86 is connected to the positive
12V of the triggering voltage.
The above schematic will work using the stock fuel pump power lead
as the trigger for the relay. This way, you do not have to bypass
the fuel pump resistor as, even under low current (approx 9V), the relay
will be triggered.
I recommend a 20 amp blade type fuse for the positive from the
battery. If well hidden, this can serve as an added measure of
security as you can remove the fuse. Because the FD is a duel current
pump, this modification will likely raise base fuel pressure a couple of
pounds. Therefore, this is not recommended with a stock fuel
pressure regulator. With an aftermarket regulator, you can dial the
pressure down to prior levels. I'm using 40 PSI base pressure, which is
about 36 PSI under vacuum.
Some runs with my new wide band lambda sensor revels a slightly
richer air/fuel ratio at high RPM and boost- exactly what I was going
for.
|