Flyin' Miata : Dissecting the Mazdaspeed MX-5

Mazda has finally caught up with us. 14 years after we turbocharged our first Miata, they decided to try it. Big thanks to Bruce Cliffe and Skip Cannon for loaning us their cars for initial investigations.

We hooked up datalogging equipment to Skip's car. We had to get a little creative, using a variety of pieces right down to a modified Maglite flashlight and a 1996 cam angle sensor! The datalogging combined with our dyno has given us some good ideas.

Mazdaspeed MX-5 (red) with different exhausts
The first test. Our turbo exhaust bolts right up to the Mazdaspeed, but it doesn't really make much difference. The real restriction must be further up in the exhaust system and we'll investigate this later. Run 4 (blue) is with the FM exhaust, run 10 is stock. The car heatsoaked fairly quickly on the dyno.

Mazdaspeed MX-5 (titanium) with the boost solenoid disconnected
There's a boost control solenoid on the front of the valve cover of the Mazdaspeed. Just for fun, we disconnected it to see how much difference it was making. Even at high altitude (4700'), it's not doing much. Peak boost was 122 kPa with it plugged in, 117 with it unplugged. That's good for just under 4 hp. At sea level, that solenoid won't be doing much at all. You will notice that this car spools up a little better than the first one for some reason.

Mazdaspeed MX-5 (titanium) with the knock sensor having a problem
The knock sensor in the car has a lot of authority! Our datalogger was feeding in a bit of stray voltage causing a check engine light and putting the car into a limp-home mode - or was it reacting to the signals from the sensor? Either way, it was pulling out up to 15 degrees of timing at times. This had a big effect on power output.

Mazdaspeed MX-5 (red) vs FM I
Well, you know we had to compare these two. The FM I made 35 ft/lbs more torque at 3500 rpm. Note the hard fuel cut in the MX-5 at 6500 rpm. The titanium car that we had in later spooled up a little better, but not THIS much.

Mazdaspeed MX-5 (titanium) vs FM II
Yes, it's kind of mean. But here's the difference between what we consider a fast car and what Mazda considers a fast car. That's a 12 psi FM II on a stock engine.

Here's what we discovered.

Tuning:

The car is incredibly conservative in tuning. At 3500 rpm and 106 kPa - that's almost no boost at all - the car runs 13 degrees of timing.
Maximum boost seems to be 122 kPa and is electronically controlled. That's about 35 kPa above ambient here, or around 5 psi.
The intercooler heatsoaked so badly on some runs that it heated the air by 16 degrees C at the start of the run. We could see a big difference in IC effectiveness as our fans kicked in. Post-IC temperatures seemed to be at around 50 C when this is taken into account.
The car has a hard fuel cut at 6500 rpm. You can easily see it on the dyno runs.

Parts:

We have developed the parts needed to fit one of our intercoolers to the car. It made no difference in power but that's not surprising at the low stock boost level. We have intercooler upgrades coming soon.
The Randall cowl intake will fit. This might be a very useful modification as it will drop the intake temperatures fairly dramatically.
The muffler is the same as a "normal" Miata with a larger tip. Our turbo exhaust bolts up easily. Not that it made any more power due to the restrictive midpipe.
Our midpipe will NOT fit. Yes, we are looking into making something.

Specs:

Vehicle weight was 2506 with 3/4 of a tank of gas. With a full tank, it will be approximately 2525 lbs. The 2003 Club Sport was 2365 by comparison.
Wheel and tire weight was 37.25 lb.
Front sway bar diameter is 23 mm, rear is 14 mm.

Part of a dyno run. Two computers, three screens. On top, the dyno readout and data acquisition. On the laptop, datalogging from the car's ECU.
The intercooler in place.
The intake setup. The "snorkel" for the airbox is aimed right at a warm intercooler pipe.
Yes, we put a cam angle sensor on a 2004 Mazdaspeed. It's for our datalogging hardware.
The boost control solenoid on the valve cover and the bypass valve underneath.
"Why did they put the heater core in front of the radiator?" Bill takes a look at the intercooler.
The engine room. It's a nice looking valve cover, but it won't fit any other Miatas. There's no VVT (sorry, 2001-04 owners) and the coil-on-plug means a different plug hole (sorry, 1994-00 owners).
An oil catch can. Very interesting. Mazda must expect some blowby. Not good for emissions, of course. The car has a special oil pan with fittings for both the oil return and the catch can.
A view of the intercooler and piping from below.
The intercooler out of the car.
The underside of the car. Thanks to Jay Evans for stitching our photos together!
The intercooler in place. It's now for sale through FM
A nice touch - Thermal R&D (who produced this exhaust for us) even included brackets for the O2 sensor wire on a prototype exhaust! Now that's why we work with them. The production pieces will be fully polished.
The prototype exhaust in place. This exhaust was slightly louder than stock but not by much at full throttle. At idle it was very quiet.
The FM prototype exhaust (right) vs. stock. The FM part is designed to work with the stock midpipe if desired or as a complete unit. As a bonus, it's 13.7 lbs lighter than the stock piece!
Our intercooler versus the factory part.
The MSM turbo vs an FM II Garrett turbo.
Our modified MSM turbo. The compressor housing has been machined and a larger wheel installed. Everything else is the same so it bolts in to place.
We modified the wastegate actuator to find out why the stock turbo wasn't making any boost. The answer: it's not the wastegate.