MAZDA MIATA GT3 TOURING

“There was a moment when I was driving that I suddenly no longer felt the existence of the car. I thought, this is it.”

Tetsu Kasahara, Chassis Dynamics Department, Mazda

This is why Caterhams and Lotus Elises tend to be second and third cars for some. A great litmus test is if you have have a conversation in your car at a normal volume, and if you can drive the car for a few hours without feeling like the noise, vibration and harsheness has beat you up. Here is a comprehensive list of weight reduction (which I didn't follow), and you will find many of these will make the car more compromised as a road car: 

The original Mazda MX-5 ran on Bridgestone SF325 tires at a 185/60/14 size. Playful handling and progressive breakaway characteristics were the objectives of this build. Chris Harris exemplified this idea with a V8 AMG on space saver tires. 

For all out grip, the widest stickiest tires drop lap times on autocrosses. For longer circuts, a stock miata works best with 205 width, around 140whp works well with 225 width and around 200whp works till about 275 width. The main reason is that larger wheels have more of an impact on straight line speeds on those circuits.

However, chasing laptimes often comes at the cost of subjective feel on the road, as the limits of the car become raised. Moreover, the NA chassis is now over 30 years old, and adding grip exposes the lack of rigidity in the chassis even more. 

Instead, I'm more interested with the lightest possible wheel and tire combo that matches engine torque, and on limit behaviour, while looking good. The wheel maketh the car. A useful way of explore wheel and tire combinations is to look at what Mazda did with the NA, the NB and finally what time attackers do on the other side of the spectrum.

The chart below details the various wheels that I was looking at. The ideal size of wheel for my application is a 15 by 7 with a +25 to +35 offset. This size was chosen because the tire size of choice is 185/55/15, perhaps a sticky RE71 compound for a track day, and a less aggressive compound otherwise. A grippier tire generates a greater roll moment (more body roll), so I'm planning to optimise the spring rate around the road tire compounds, and then stiffen the rollbars and suspension as a compromise for track work. Anyway, here are wheel choices I considered.

I ended up with two sets of wheels. The street setup is a set of cast magnesium RS Wantanabe wheels mounted on  Bridgestone Potenza RE004 tires, and a set of forged Rays CE28 wheels with Bridgestone Potenza RE71RS tires. Both sets are 15x7 and 185/55/15 wheels. This allows a lot of flexibility to mix and match, particularly for drift events and studying car dynamics. Both of these wheels have about 10mm less offset than the original wheels, or about 5mm less than wheels that came on the NB.

Positive offsets below the stock rates increase scrub radius. Scrub radius affects how your front tire moves as you steer. If scrub radius is 0, then the tire pivots around the center of the contact patch as you steer. Nice and simple. If scrub radius is not zero, the tire gets dragged in an arc as you steer. That means more friction, dulls steering, adds kickback, tram-lining and soaks up a little power from the engine. The +40mmm offset is a starting point and Spec Miata race cars regularly run closer to +20mm.

I usually match the width of the tyre to the width of the rim for better response. It also gives the sidewall a very slight ‘stretch’ or preload, which will improve the tire’s response and break-away characteristics. It also reduces camber loss from sidewall deflection and allows for more linear steering response at max slip angle.

However, when the sidewall is too stiff, it acts like a tire with too much pressure and actually loses grip. Tires that are slightly stretched and narrower on the rim will have less phase lag (time of steering input and force reaction), which lead to snappier break-away. Generally, a rounder tire will have better transient response and a gentler break-away characteristic compared to a squared-off tire, so in this case, I have a little breathing room to move up a tire size if required.

We can't talk about wheels, without talking about hubs. OEM front and rearhubs typically last 50-100 hours depending on use. Cracked flanges and bearing play are not uncommon, and thankfully there are some aftermarket solutions out there discovered by Gene Splicer on Miataturbo.net. The Toyota MR-S hubs will fit the rear with some tweaking. They are 4x100 and have the same stud thread pitch and length as a factory hub while being thicker. A company called BRO-Fab modifies BMW E30 front hubs that are are stronger than the originals

Herb Adams (author of Chassis Engineering) defined torsional rigidity as “how much a frame will flex as it’s loaded when one front wheel is up and the other front wheel is down while the rear of the car is held level.”

The chassis is the backbone of the Miata and the more rigid it is, the more likely the wheels will point and move in the same direction.  For any car to be responsive and predictable, it does not help when the chassis twists slightly and alters the suspension geometry under load. 

One of the key features employed in the car is the power plant frame or ppf. It is an aluminum truss frame that connects the gearbox to the differential running along the right side of the drive shaft. It is pressformed from 6mm gauge aluminum sheet, and has an intericate shape and weights around 5kg. It greatly reduces drive train deflection during acceleration and deceleration resulting in a direct transmission of engine power to the driving wheels, improving the overall feel of the car. 

We can better understand the limitations of the NA miata by exploring the development work they did on the NB. According to a Japanese press release, a total of 114 people in 26 development departments weighed a total of 3,600 parts and searched for room for further weight reduction. While 196 parts were considered for weight reduction such as the air conditioning and suspensions components, they also employed functional integration by having a single part perform more that one function. The biggest challenge was to strengthen the chassis to meet the "safety standards", especially offset collision. Based on inital calculations, this would have added 37kg, but by employing 'gram strategy' they reduced that amount by 10kg.

Using the Mazda Advanced Impact Distribution and Absorption System engineering approach employed on the Mazda Lantis, side impact bars (one more than the NA) being placed inside the front doors and the center of the body was strengthened so the frame around it became a crushable zone. 

Due to this, the cabin of the NB Roadster is 70mm narrower than that of the NA Roadster (length NA 935mm NB 865mm). Due to the increase in rigidity, the brace bar at the rear of the cabin was removed, and there is a high hardess reinforcement built into the a-pillar to increase rollover protection. The side sill itself is also thickly reinforced, and the semicircular opening of the door also contributes to the strengthening of rigidity. Even the arm rest in the NB2 doubles up as a crash pad area.

In the image below, we can see some of the bolt on parts employed to improve the chassis stiffness further. 

There is very little information that is validated on torsional rigidity for all cars including the Miata. Original cars are now almost 30 years old. Unlike a coupe, where forces are dissipated throughout every corner of the car’s body, convertibles channel those forces through the door sills, placing larger amounts of stress around the floor of the car. 

There is a great article done by Christopher Shieh who analysed the Flyin Miata frame rails to ascertain it's impact on the torsional and bending rigidity on the car. His simulated frame rails showed to increase torsional rigidity in the chassis by 13% and bending rigidity by 17%.

In his analysis, he found that adding a hardtop to a Miata will increase the bending load and effectiveness of floor-mounted chassis bracing. This is because the hardtop will move the neutral bending axis of the Miata up, causing the frame rails to deal with more tensile and compressive stress. The Mazdaspeed "NR-A" dealer option package for club racing in Japan and included a 6-point roll cage for both safety and for the above reason. 

Emilio from Supermiata has complied a comprehensive list that serves to improve torsional rigidity and NVH. This list goes from most effective to least effective, but not all recommendations are suitable for road cars.

1. 8 Point Weld in Roll Cage
The ultimate for track, unsafe for daily driving.

2. Full Seam Welding
Requires stripping the car to a bare shell and welding it.

3. Door Bars
Reduces chassis jiggle, requires one piece racing seat. Creates oversteer.

4. Roll Bar
Any bar works which bolts to the rear bulkhead.

5. Frame Rails

6. Door Bushings
Allows the door to box the central tub.

7. Fender Braces
Reduces cowl shake and resonant frequencies.

8. Rear Chassis Braces
Very small gains

9. Front Chassis Braces
Reduces cowl shake

10. Subframe Chassis Braces
Negligible gain

Increasing the torsional rigidity of the Miata without significantly increasing the weight is the engineering challenge, and my personal approach is to use only bolt in parts. All the added bracing from the NB should be used as a starting point for this, as well as uprated door bushings.

The Flyin Miata frame rails and butterfly bars have a claimed 29% increase in torisonal rigidity, so that's pretty much a must as the company does great testing and validation on their products. Here is how they tested the new version of the butterfly brace according to Keith Tanner: 

"To test the twisting of the chassis, we supported the car on three jackstands. Two were at the front of the factory frame rails and one was on one of the rear control arm mounting points. The doors on the car were opened to give a worst-case scenario. The distance from the floor to the stock rear jacking point was measured. A jack was then put beside the measured point and the car was lifted until the front jack stand started to unload.

We performed this several times to ensure accurate measurements.The car was then lifted off the jackstands and all measurement and support points were marked. Pre-modification, measurements showed 5.5-6 mm of twist. Post-modification, we found 3.5-4 mm. A 2 mm change in twist over a 6 mm range is a 33% improvement, we rounded that down to 30%.

The original design has been improved with more mounting points and a lighter design. As a result of their effectiveness, we continued development and created our Butterfly Brace."

Interestingly there seems to be some evidence that the Skidnation frame rail may be of a superior design. However the only way to be sure would be to do a back to back comparison and simulation. Here is the analysis Skidnation provided to back this up:

PACO Motorsports has not produced any data on the efficetivness of their fender arms, and when asked by Maneki_Neko, Keith Tanner had this to say:

"We did attempt a test like you describe, by bolting a beam to the bumper supports and adding calibrated weights. Unfortunately, we found the diagonally opposite corner of the car was lifting even when we loaded up the trunk with a few spare engine blocks. Lacking a good, low-effort way to anchor the corner down we stopped the testing. As you noted, spending several thousands of dollars worth of labour to come up with some numbers on a $219 product simply didn't make good financial sense, and we weren't willing to drill into our concrete floor for it.I'm not sure our testing would have necessarily answered your question. Will they make a difference on a car with a mild suspension as opposed to a more aggressive setup? It's not a yes/no answer. They'll have the same effect, but less of one. Whether that effect would be worth the cost is a determination you'd have to make for yourself, and it's not easily distilled into numbers."

That bring said, since it is an area which is traditionally seam welded, the fender braces go on. The final piece of the puzzle is how to reinforce the door cavity, and for that, a door bar coupled with a reclinable bucket seat would be the way to go. A better option would be to seam weld the door sills and A-pillar for added strength. 

Here are some resources and images for chassis stiffening and seam welding:

I want to emphasise that this list below is unverified, but I am adding it here as a starting point to hopefully solving the mystery of torsional stiffness and the miata. If you have any resources that will help find sources for each of these, please let me know! 

Mazda MX-5 (NA, 1990 – 1993)
4,881 nm/deg (Unverified)
Source: http://www.roadster.blog/2015/06/mx-5-nd-skyactiv-body.html
Chassis weight: 211kg 

Mazda MX-5 (NA, 1994 – 1998)
5,152 nm/deg (Unverified)
Source: http://www.roadster.blog/2015/06/mx-5-nd-skyactiv-body.html
Chassis weight: 211kg 

Mazda MX-5 (NB, 1999 – 2000)
5,219 nm/deg (Unverified)
(+1.3% torsional; +7.6% bending; +35% dynamic torsional)
Source: 949 Racing Bare tub weight and picture
Chassis weight: 224 kg (verified)

Mazda MX-5 (NB, 2001 – 2005)
6,367 nm/deg (Unverified)
(+22% torsional; +16% bending);
Source: 949 Racing Bare tub weight and picture
Chassis weight: 224 kg (verified) 

Mazda MX-5 (NC, 2006 – 2008)
8,132 nm/deg (Verified Below)
(+47% torsional; +22% bending)
https://newsroom.mazda.com/en/publicity/release/2005/200502/050228.html
Chassis weight: 211kg  (verified) 

Mazda MX-5 (NC, 2009 – 2015)
nm/deg: ???
Source: ???
Chassis weight: 211kg  (verified)

Mazda MX-5 (ND, 2016 –current)
nm/deg: ???
Source: http://www.roadster.blog/2015/06/mx-5-nd-skyactiv-body.html
Chassis weight: 200kg  (verified)

Mazda MX-5 (ND RF, 2017 –current)
nm/deg: ???
Source: ???
Chassis weight: ???

It's helpful to think of suspension as a compromise based on the use of a car, and typically OEMs have to design something that works for many countries, driver abilities and cost.

Interestingly, race cars optimise ride compliance on a given surface to achieve maximum grip. High quality coilovers also increase the range of reaction contributing to a better ride and better grip. Coilovers that are lower quality tend to have higher internal friction, poor weather sealing, oil that deteriorates quickly and much more.

Good body control feels smooth, comfortable and predictable not harsh. It gets more confidence inspiring as you push the car to the limits. Some describe the opposite 'bouncy' feel as sketchy or unpredictable.The more grip there is, the stiffer the springs need to be in order to move transfer the weight. However if the springs are too soft for that same grip, the car will have greater roll and more likely to bottom out. Now most roads in the worlds are not as smooth as good race tracks, so there is tremendous variability in the way a single suspension setup will handle around the world. This guide is for roads more similar to Japan, which are really well paved and maintained. Here were the options I looked at. 

Chasing lap times costs money, and I'm more interested in fast road and street oriented builds. The OEM 185/60/14 NA6 tires would allow around .84g. Modern 200 tw tires in a 225/45/15 will easily generate 1.35g on the same NA6, and by that point the lack of stiffness in the chassis will mess up any alignment anyway. Tires that I tend to gravitate to are only slightly wider than the OEM tires that came with the Miata. 

There have been huge increases in compound technology, grip and rolling resistance since the car came out, so by going with a modest tire size, we can gain the benefits of lighter reciprocating mass as well as more predictable and lower limits for the driving dynamics of the car. Available grip is critical to choosing spring rates, which is the key to allowing weight transfer to happen in a more natural way. One the tire informs the spring rates (with ride height taken into account), shock damping needs to be valved and well-matched to spring stiffness, which works together as a system. A general principle is to run the most compliant springs you can get away with for your level of grip and your ride height until you start to bottom out against the bump stops.

Since this is an item that is working every time the car moves, it's an area which I am happy to spend money for a quality product. Adjustable coilovers that have adjustments for ride height, compression and rebound allow the car to be setup for the highway during the week, and for the track on the weekend.

It's important to note that the grip levels should not change drastically for both applications, because
we are trying to keep the roll moment (side to side) and pitch moment (front to back) matched through the correct relationship of tire grip and spring rate. Giving up the lap-time chase has allowed me to have more fun by using the same tires on the track and off it, helping me develop a better relationship with the car.

Low ride heights are a big compromise, as it changes the roll centers and reduces available suspension travel, especially in the rear, many companies offering extended top hats to compensate for this. Lowered cars become unstable over bumps due to bottoming out. The talented Andrew from Keisler Automation has developed a set of drop spindles specifcially for lowered cars. Measuring ride hight from the pinch weld is key to tracking this. 

Improvements in head flow can be achieved with back cutting the valves. The object of cutting more angles into a valve is to form a more curved radius from the valve face to valve seat angle while also creating a thinner valve head. Backcut valves will usually increase low lift flow, and often increase mid and high lift flow too. As this build will not be concentrating on high rpm power, the valve sizes will remain stock, however the valves themselves will be upgraded to Inconel material to increase its melting point from 1650°f to 1900°f.

Individual throttle bodies sound amazing, provide greater response and potentially peak power if headflow can be matched. Being a right hand drive one, the brake master leads to a packaging compromise. That said, Mazda produced a concept car called the MX-5 MPS with a 2.0L engine and ITBs. This provides some clues to how they would have designed an airbox if ITBs came OEM. 

I want to give credit to Emilio from 949Racing and Supermiata for his grounded views on engine building and really balancing performance with value. 

Here are dimensional differences between the 5 and 6 speed gearboxes.

Another dream that I think might be fascinating to explore is an entire transmission and drivetrain swap from the ND. Since it was engineered to the original car's dimensions, I think that it might all fit with some minor modifications. You can see how the weight has changed from the NA to the NC to the ND below. 

The steering wheel is a Momo Mod 78 330mm, which is a little smaller than stock to create a slightly quicker steering response. Shifter is a Raceseng Circuit Sphere 100, which feels like a metal version of the Porsche GT3 shifter. These are highly personal choices and I'd recommend you find what works for you. 

Part of a nice interior is having some consideration given to noise, vibration and harshnesss. Driving a round in a stripped out race car gets old quickly, and some consideration for comfort makes for a place that you might want to be more often. Here is a list of general and Miata specific areas that you can use as a basis of investigation for any strange sounds. 

1. Tires Out of Round / Bent Wheels

May be balanced on the surface, but not perfectly round.

2. PPF Height Incorrect

PPF has specific height required by the service manual.

3. Tunnel Blocking 

Sealing the trunk cavity on the right with insulation or a pool noodle. 

4. Worn Shocks

May transmit unwanted frequencies to the driver.

5. Light Flywheels

Magnifies rattling in old gearboxes.

6. Flywheel Moment of Inertia

Larger diameter, generally dampens noise.

7. Trunk Soundproofing 

Adding sound absorbtion material to the trunk and rear bulkhead. 

8. Clutch Hubs

Sprung hubs quieter than rigid hub discs.

9. Fuel Line and Gasket 

Rubber gasket from the fuel line to the gas tank can create noise. 

10. Muffler Heat Shield 

Highly resonate part prone to rattling at the rear of the car. 

11. Engine and Differential Bushings 

Polyurathane bushings increase NVH. Rubber, less so. 

12. Worn Transmissions 

High mileage ones naturally rattle. Magnified by light flywheels. 

13. Worn U-Joints 

Known to rattle as they age. 

14. Bolts, Catches & Tabs

Look for any loose bolts, screws, catches and tabs. Tighten. 

I've since sold the car, and it has been reverted back to stock. However, I'm already thinking of the next build, which will be based on the NB. In addition to many of the ideas reflected here, these are a few additional areas I will cover.

1. Cooling and Heat Management
Coolant reroute with hard pipe
Coolant block of kit
37mm Aluminum Radiator w/ shrouding and fans
Distilled water with antifreeze
Hood vents for better cooling (high pressure areas) 
Heat shielding for heater hose
Heat shielding for master cylinder

2. Electrical & ECU
Wideband O2 sensor (correct location)
Digital gauges in single screen with logging
Air/Fuel ratio gauge
Exhaust gas temperature gauge
Oil temperature gauge
ECU safety cutoff
MR-S Power steering retrofit
ND airconditioning retrofit
Coil pack conversion 

3. Reliability and Random  
High quality screws and bolts
Inconel manifold studs
High quality hose clamps
Expoxy throttle plate screw
MR-S rear hub swap
E30 front hub swap
Mazdaspeed engine and diff mounts
(prevent exhaust cracking)
Dei shifter insulation kit
Engine lowering mounts
Quick release steering wheel 

The next car will be my third Miata, and it is safe to say that I really love these little things. If there is any information here that is wrong, or not correctly cited, please to get in touch and I'll be happy to update it!