RX-7
twin tower ignition coils installed in MR2 Turbo.
The goal of the project was to design and build an affordable ignition
upgrade for the MR2 Turbo (3S-GTE). The original plan was to build
my own ignition circuits using the Motorola MC3334P chip, and four ignition
coils (one for each spark plug). The MC3334P chip has current limiting
capability, and the ability to automatically adjust dwell time. I
did get the car to run with four coils, but the igniter circuit kept
overheating, and I, in general, had problems with it. The biggest
challenge to the project was designing the digital logic to take the
place of the stock distributor (ie a spark plug selection circuit).
The stock Toyota ECU (as it is on many Toyota's) uses a position sensor in
the distributor (G1 and G2) to determine the engine position. It then
adjusts the timing based on various engine parameters and sends a signal (5
volt pulse waveform) called IGT to the stock igniter. The signal goes
high to charge the coil, and then low to fire the coil. The stock igniter
returns a feedback signal to the ECU (called IGF) to tell the ECU that it actually
did fire. It is the IGT signal that needs to be modified and then sent
to different igniters/coils.
The first iteration of the circuit used a Flip Flop, Inverter, a comparator
( for signal conditioning of IGT, and G1), and a 3 to 8 line decoder (inverting). This
circuit did work, and the car ran with four coils, but there was some problems
(as stated earlier) with the four igniters. Since the charge time could
be so long for four coils, it was decided to try to build a system using a
'wastespark' design with twin tower coils. GM uses this type of coil,
as well as electromotive, and numerous other manufactures including the 1986-88
RX-7. These coils (RX-7 coils) in particular are very powerful coils
(ie have a low primary resistance). The downside to these coils is that,
new, they cost around $250.00. With igniter they cost around $350.00. Fortunately
I found a couple at a junk yard with igniters for $70.00 (for two!). It
seems very likely that the GM coils/igniters could also have been used. I've
been told that the GM coils can be had for around $30.00 - new. The digital
logic was then modified to fire just two coils instead of four. Because
of the simplified operation, the G1 signal was thought not to be needed to
make the circuit work. It would, however, probably be best to design
a circuit that would use G1; however no problems have been found with the current
setup. (new) - There is a slight problem - every once in a while during
start up the circuit does not 'sync up' with the firing order and the ignition
has to be turned off and restarted. This happens infrequently, and is
easily reminded by just trying again.
Here is a diagram (bad drawing sorry) of the setup:
The two coils (+12) are hooked to the stock wiring for the ignition. There
is a 40 amp (if memory served) fuse for this circuit, which is plenty. The
ground for each coil was wired to two separate ground locations using
12 gauge wire. The wiring from the selection circuit to the igniter
should be shielded, but does not have to be very think - 18
gauge wire is plenty. Up until now, I was taking the output of the stock
coil (which would normally go to the distributor) and feeding it into a 4.7k
resistor. This kept the IGF signal happy (and hence the ECU) as well
as the tachometer. However charging the additional coil didn't really
make sense. So the MSD tach adapter (PN8910 HEI) was purchased from Summit
Racing for about $26.00. This (basically coil) attaches to where the stock
coil originally connected. The nice thing about this setup, is that switching
back to the stock ignition is relatively easy (if nice connectors are used). I
used bullet connectors from Radio Shack.
What about charge time?
The charge time for each coil is basically doubled from what it would have
normally been with the stock ignitor/coil. The nice thing about this
is that even at 7200RPM the coils should give a nice strong spark (each only
'sees' 3600RPM). The bad thing is that it is somewhat wasteful as the
coils do not need to be charged that long. I have
monitored coil temperature, and even after a long drive (45 minutes) they are
cool to the touch.
Results
I do not have any concrete results yet. However, my 1991 MR2 turbo did
have a hesitation problem on cold starts on cold mornings. This is all
but gone with the new ignition! Gas mileage improved from 18.00 to 18.7
(first try) - OK so that's not that big a deal. I have new plugs which
I plan on running at a gap of 0.40 to see what happens. There does seem
to be more high end punch - but that's seat of the pants speculation. Since
there are no spark plug wires attached to the distributor, there will never
be a need to change the cap&rotor. The losses (from the spark jumping
from the rotor to cap) are eliminated since it is now done electronically.
How do I build one
I've spent a long time designing this ignition - starting the engine, looking
at signals - over and over etc.. The selection circuit is not built with
a micrcontroller - it is all 'in hardware', and is wire wrapped for durability. The
original plan was to make an ignition kit that could be purchased - but some
more development needs to be done for that can happen.
If you are interested in building a setup like this for your Toyota - I'll
be happy to help.
Just send me an email!
:-)
Old picture of ignition coils in car
Update 12/09/2002
Much work has been done on the ignition project since last reported. For
one, G1 and G2 both need to be used to keep the circuit in sync. The
big problem is that the G signals can occur before the signal to fire
(IGT) or after the signal to fire. This causes havoc! Signal
conditioning with comparators proved hazards, as changes in the outside
temperature caused the resistances to change, which caused the signal
conditioning to no longer work. Thankfully national semiconductor
has a nice chip, the LM1815, which conditions these signals (G1 and G2)
nicely. I still had the problem, with the G1 and G2 signals occurring
before or after the signal to fire the coil. This signal (G1 OR
G2) was used to reset the primary flip flop (the one that sends the signal
the the coils (Coil 1 and Coil 2) to fire (ie the modified IGT signals).
What Chris Conlon came up with (thanks Chris!) was to delay the G signals using
a series of flip flop circuits and the N1 signal to the ECU. N1 is yet
another signal to the ECU from the distributor. It is a many toothed
gear so the ECU gets numerous pulses for each revolution of the distributor. This
signal was also conditioned with the LM1815 chip from national semiconductor.
Using this signal and Chris's flip flop delay circuit, I delayed the (G1 OR
G2) reset signal two N1 signals. So, (G1 OR G2) would occur and the circuit
would wait to see two N1 signals before sending the reset signal to the primary
flip flop. This worked and the car ran, but occasionally would miss,
particularly under rapid throttle changes. I'm not sure why this current
circuit is not working out as planned. I'll try to update the information
here more with a complete circuit diagram soon. Please e-mail me with
any questions - I'd love to talk about it.
