Mazda MX-3

Mazda MX-3 0.6

Login or Register an account to download this content
I got a 3D scanner!
1729543026438.png

And I'm really impressed with its precision. If you look at the above image closely, you can see the outlines of the race numbers on the door, as well as a few other decals.

While I thought my exterior model was decent before getting the scan, it's far better now. Seeing the new and old models side by side makes the comparison pretty clear.

ba0.jpg
ba1.jpg


Oh, and the lights work now, too!

lights.jpg


On the interior, I refined the floor slightly to align with its 3D scan (and remove some spurious details), added the roll cage, and moved the steering wheel and seat (though now the driver's back goes through the seatback). I also replaced the instrument cluster with this digital one I made for my real car, though so far I've only added the tachometer and speed (in mph).

Screenshot_mazda_mx3_lilski_road_america_21-9-124-15-35-37.jpg


While the model is diverging more and more from any MX-3 other than my own, I only have the one on which to base my model, and getting that car into the sim was my goal with this whole project. I will still add a V6 option (and if you look through the car's files, you'll find a few you can swap in to get the V6 now), though I don't plan to add a stock interior. I'm having enough trouble making this one look good, I don't need another whole model to make.

Attachments

  • 1729541401767.png
    1729541401767.png
    766.5 KB · Views: 14
  • ba0.jpg
    ba0.jpg
    241.2 KB · Views: 13
  • ba1.jpg
    ba1.jpg
    267.5 KB · Views: 11
Immediately after uploading 0.4, I had an epiphany regarding the rear suspension: I can approximate the trailing arm design with a STRUT geometry, using the trailing arm and forward lateral arm as the "wishbone" and the rear lateral arm as the "tie rod." The trailing arm and lateral arms do not meet the upright at the same point, so this is an approximation, but I doubt it changes much and it feels about the same as it did with the ML suspension. STRUT just works better than ML, so I decided to stick with it.

While I was looking at the suspension with the visualizer, I noticed the top of the front struts was too low, so I raised it a bit without changing anything else. This reduces front caster, and the steering feels noticeably more vague, but that might be appropriate. I'll measure everything with a 3D scanner to make sure all the points are truly accurate in a future update.
Last month, I took my MX-3 to a 24 Hours of Lemons race at Road America. Since the last time I raced the car, I've added a datalogger that was able to record telemetry throughout the race. Fortunately, lilski made an excellent model of Road America for Assetto Corsa, which I used to set the torque curve and drag coefficient for this car to match the real world data as closely as possible. As it was an endurance race, I have little data above 5500 RPM, however the data I do have matches quite closely the torque curve provided by another user in the discussion for this mod (multiplied by 0.82), so I continued to use the values from that torque curve at higher RPM.

After the race I put the car on a lift to check the brakes, suspension, and a few modifications I'd made, and took several suspension measurements (detailed below) while I was at it. Incorporating these into the Assetto Corsa model made it feel much more realistic.

I made a few other adjustments subjectively, based on the way the car feels to drive. Yes, the rear brakes like to lock up now, but they do in real life (though that could be because rear brakes are one of the few upgrades I've made).

Additionally, I added mirrors so I could drive it in a league race I set up a few weeks ago. Few other changes have been made to the model.

The following is the full list of changes, with the reason for each given in parentheses.
- Adjusted drag coefficient (Road America telemetry)
- Adjusted torque curve (Road America telemetry)
- Adjusted brake bias (subjective feel, from Road America race)
- Adjusted ARB stiffness (measured by hanging a known weight on the ARBs and measuring deflection)
- Adjusted spring rates (measured by jacking up front and rear suspension, with ARBs disconnected, with the car on a lift and a bathroom scale on the jack)
- Adjusted damping rates (used values found in the factory service manual)
- Adjusted suspension travel limits (measured with a ruler)
- Added Gearset options for F25M-R and G25M-R transmissions
- Added mirrors!
- Adjusted graphical offsets (suspension changes)
- Adjusted steering ratio
- Adjusted max fuel (Road America fuel consumption)
- Adjusted fuel consumption (Road America telemetry)
- Adjusted engine inertia (real world telemetry)
- Adjusted idle RPM
- Adjusted LOD settings to prevent the car disappearing beyond 15m
- Adjusted CG height; it's still mostly a guess, but seems slightly more reasonable now
- Adjusted wheel size
I suppose it should have been obvious from looking at the car what the problem with the rear suspension was - the ride height was way too high and the suspension was near what is in reality full droop all the time.

Screenshot_mazda_mx3_lilski_road_america_28-1-124-22-20-13.jpg

Clearly this is not right
Screenshot_mazda_mx3_lilski_road_america_28-1-124-22-14-18.jpg


Lowering that improved handling tremendously. I've also made a few small changes to the model (mostly using my own wheels since the ones I bought aren't centered around their rotation point properly) since the last update but the suspension is the big one with this update.

Screenshot_mazda_mx3_lilski_road_america_28-1-124-22-20-48.jpg

The rear is probably too low now, but I can just adjust the visual offset without changing anything in the suspension geometry to fix that.
Screenshot_mazda_mx3_lilski_road_america_28-1-124-22-11-13.jpg

Much better.
It's been a long time since I've updated, but I've finally made enough progress it feels like it's worth doing.

I bought a model of an MX-3 on CGTrader, and while it has a lot more detail than the one I put together, lots of things just don't seem quite right about it. Some of that is certainly because it's a Euro spec car and has different bumpers and lights, but some things just don't seem to line up. I was, however, able to use the wheels, so it wasn't completely worthless (and I may yet be able to use the body with a bit of stretching and shrinking to make the proportions right).

In addition to the wheels, I now have the beginnings of an interior. There's a crude model of a seat, the black background of a dash, a steering wheel, the steel dash bar, and the two clamps that held my switch panel on. One of them is white for reasons unclear to me; Blender has both assigned to the same material.

On the exterior, I cleaned up many (but not all) of the gaps and added sharper edges in a lot of places. I also updated the lights, so they now have clear lenses over reflectors like actual car lights do.

I slightly modified the suspension setup as well, as the car seems to need a lot of rear camber to feel happy. With -2° rear and -2.5° front camber, it feels reasonably close to my actual car. I believe the engine is either making too little power or the body too much drag, as the high end speeds are just slightly lower than I experienced in the real car (101mph vs 106mph into T5 at Road America) but overall laptimes are faster. Corner speeds don't seem to be any faster so it may just have better low end acceleration. At some point I'm going to check how the acceleration compares to my race videos, and perhaps when the weather gets better I'll do some more road tests to really fine tune it. Getting on a dyno would be nice, but that takes more money than I want to spend on this at the one local tuning shop.
Made the textures a lot better. They're still not great, but they're reasonable.

Attachments

  • Screenshot_mazda_mx3_road_atlanta2018_12-10-122-13-6-29.jpg
    Screenshot_mazda_mx3_road_atlanta2018_12-10-122-13-6-29.jpg
    319.8 KB · Views: 184
Back
Top