Teasing Setup Secrets out of the Formula Mazda

I spent all day on Tuesday pounding around Summit Point in iRacing's Formula Mazda. Squandered a perfectly good day doing 200 laps, tinkering with every parameter in that setup menu.

I ended up taking nearly a second off my personal best (lowered it to a 1:07.4). Also I think my latest setup is a good race setup as well as having some potential for further development.

Richard Towler's Setup, Low Tire Pressures, and Ground Effects

I started with Richard Towler's setup from an iRacing.com member forum thread called Re: Formula Mazda @ Summit Point.

It's a challenging setup - if I get a little sideways and try to recover, I wind up in a violent tank-slapper from which there is no return. But it has what I thought was fantastic balance through the final turn and also good balance through Turn 3, two of Summit's most critical corners.

Richard's setup also is very unusual in some ways, with a quarter inch of toe-out at the rear and extremely low tire pressures - 17.0 front and 17.5 rear! Tire temps show the middle temp far below the inner and outer.

In the forum, people theorize that this setup oversteers because it has a quarter of an inch of toe-out at the rear. Although this makes sense, I wasn't sure and wanted to verify this.

The first thing I did was raise the tire pressures. No other change. And bang! The Mazda's characteristic understeer came back. I've established to my own satisfaction that the oversteer - or at least lack of understeer - is related to the low tire pressures, not the rear toe-out.

I suspect what's happening is that the ultra-soft tires are letting the car squat and roll so much that the outside edge of the sidepod is touching the ground, taking away load from the outside rear tire and thereby reducing rear grip. Or, possibly, the extreme roll and squat are causing the diffuser to stall, taking away rear grip. Either way, I believe the reasons Richard's setup oversteers are related to his very low tire pressures, not toe.

Actually, I think the phenomenon of the chassis squatting under aero load explains why Richard, Wolf Woeger and others have been arriving at very soft tires on the Mazda: because when they lower the tire pressures, they feel more grip, but I believe this is because the chassis squats more at speed on the softer tires.

So as they experiment they feel that lower pressures give more grip, when actually it's the fact that the lower pressures cause lower ride height at speed, which gives more downforce, which is where the extra grip is really coming from. Make sense?

Tire Temperature Spread

But tires in the real world actually make the most grip when the temperatures across the surface are even. This is logical; if the edges are hotter than the middle, than the middle isn't being squeezed against the pavement as hard as the other areas of the contact patch, so it's not generating the grip it could.

So really, the car should generate the best grip with the tires at optimum pressure: with temperatures even across the tread, or slightly higher on the inner edge and the middle temp halfway between those of the inner and outer edge. This is the way it works in the real world.

(And yes, Grant Reeve has said that the Mazda's tires need some fine tuning, but the temperature spread relationship to grip is so elemental that I can't believe Dave K. would have made a mistake on this aspect of the tire model.)

So, I figured, I should be able to set the pressures to get even temps and drop the static ride height to get back the downforce I'd lost because of lack of squat due to the higher tire pressures.

The Formula Mazda's Underbody Design

I also believed that the aerodynamic center of pressure (see below) was too far back, and I thought that was what was causing the understeer I dislike so much. So, as I mentioned before, I thought that increasing the chassis rake should allow me to dial in better balance by moving the CP forward.

Wrong. Or, only partially right.

In fact, at one point I was so baffled by what was happening that I looked up the Formula Mazda on the web.

I realized I'd been thinking of the Mazda as a tunnel car (like the Lola/Reynard generation of Champ Cars) but really it might be flat-bottomed with a diffuser like an F1 car.

I found a lot of photos and even the aero parts list at Star Mazda. Unfortunately there are no photos of the bottom part of the sidepods in the catalog, but there are photos of the diffuser (at right). That and other photos of the entire car elsewhere (look closely at the photo of the white car below) that convinced me that this is a flat-bottomed car with diffuser, no tunnels.

So that made me think about it a little differently. Flat-bottomed cars are notoriously sensitive to ride height - and the downforce they generate isn't linear as the ride height goes up and down. Lowering the chassis does increase downforce - to a point - but then you may reach a level so low the airflow under the chassis stalls and because of this downforce begins to diminish rather than increase as you lower it further.

Real-world race engineers have information generated in a wind tunnel called ride height maps. These show drag, downforce, and aerodynamic center of pressure at various arbitrarily chosen ride heights.

I asked Dave Kaemmer in his AutoSimSport interview if he was planning to supply us iRacing members with ride height maps and he said no, these are regarded as trade secrets by the teams that allow him to scan their cars and supply him with data.

So we are going to have to rely on guesswork and experimentation to arrive at info that the real-world race engineers for these cars have at their fingertips. Dammit!

But I am assuming that Dave himself does have those ride height maps and that iRacing's aerodynamic modeling of the Mazda consists of a set of tables which mathematically represent the content of those ride height maps. I figure the iRacing physics engine, having calculated velocity, ride height, pitch, roll, and yaw at any given instant, simply looks up drag, downforce, and center of pressure in those tables, resulting in behavior that mirrors the real-world wind tunnel results.

Anyway. I found that lowering the chassis and adding some rake did get back some of the downforce, but you can't go too low or the front end just washes out completely. In Turn 3, this might be due to the front of the chassis bottoming, but it also happens in the final turn, which is nearly level. So I think possibly getting the front of the chassis too low causes it to stall, aerodynamically.

So I think maybe you can only go so far with rake and with general lowering to get downforce. I believe you have to start using the wings, too.

Fortunately the front wing should be fairly low drag, as its primary function is as a trim tab, so it shouldn't hurt too much to add a few clicks of front wing to get the aero balance into the range where you want it.

I tried going up one degree on the rear wing and several degrees on the front, and lost only about 1 MPH at the end of the straight. I think gains in balance would produce improvements in corner exit speeds that would more than offset that loss, so for now I'm sticking with 14 degrees at the rear and whatever it takes at the front to balance it.

The Impact of Caster and Camber

I continued experimenting and came to the realization, much to my surprise, that camber (and caster, because of the camber gain it causes) is critical. And none of the setups I'd tried had optimized this. This turns out to be a really big contributor to the understeer.

(This is right in the area where my friend Ricardo Nunnini began to explore a few days ago, so as usual he is way ahead of me!)

I had been operating under the assumption that at high speeds, aerodynamic downforce was so great that it pretty much overwhelmed any mechanical factors like camber. But now I think this assumption was wrong. Aero is important, but so are camber and caster.

I continued tinkering. I could dial out the steady state understeer by cranking up the front wing but this made the car unstable under braking and turn-in. For a little while I went down a blind alley by trying to tune Richard Towler's damper settings for better turn in with what I consider optimal tire pressures. Didn't work; I got slower.

After that I focused on front camber and caster. The rear temps looked good - a degree or two higher on the inner edge of the left-side rear, and the middle in between - so I figured I could assume they were reasonably close to optimum.

I started tweaking the fronts, and to my astonisment I was able to completely dial out the dreaded understeer by getting closer to optimum with the front camber and caster! I'm not sure I have optimized these yet; it's tricky because of the camber gain which occurs when you turn the wheel. But the setups I ended up with are, I think, getting close.

I posted these setups in the same thread I mentioned above. (Pesonally I was a little faster with the x1g setup, but it understeers more than I'd like. I think the x1i setup is potentially faster; I just haven't been consistent enough as a driver to access it yet.)

One of the things I like about these setups is that, unlike my earlier setups with a lot of rake and stiff rear springs, I'm not giving up rear grip, so the braking stability remains fairly good and the traction remains excellent.

The stability of the aero platform is still important, but it's not as critical as when I was trying to balance the car with rake and running the front very low. I've stuck with Richard Towler's spring settings of 600 front, 700 rear, which I think are about the only things left from his original setup. But I should also say that it still owes something to Richard's setup, particularly in the area of damper settings.

I've been experimenting with going softer on the dampers, and I think that helps calm the car down with no downside that I can see. I started with Richard's extremely stiff settings and I've backed off four clicks on all the settings except front rebound, which I think I've backed off more like eight clicks.

Richard's front dampers were extremely stiff in rebound, probably to calm down the chassis response to steering inputs. Since the car with my setup isn't as volatile in its responses to steering, it doesn't need such stiff rebound damping at the front.

Aerodynamics and Setup in a Ground Effects Car

I feel that we are still just scratching the surface of what the Formula Mazda has to teach us. It's the first downforce car in a simulation whose behavior I actually trust to reasonably closely approximate the behavior of its real-world counterpart. Therefore it's extremely interesting to me as a learning tool.

Downforce - especially downforce from the underbody - radically increases the complexity of setting up a car. You have all the usual mechanical parameters that us old GPLers are familiar with (toe, caster, camber, spring and damper rates, tire pressures, roll bars, etc.) and you also you have this enormous aerodynamic effect as well - and it varies drastically with speed. And the mechanical factors and the aerodynamic loads interact, making things even more complex.

On top of that, not only does the total downforce vary with speed, but where - relative to the wheelbase - all that downforce is being delivered also varies with speed.

This is known as aerodynamic center of pressure. If it's forward of the car's center of gravity, then at speed the car is going to tend to oversteeer, because the front tires are getting more downforce and therefore more grip relative to the rear. If the aero CP is aft of the CG at speed, you're going to get high speed understeer.

An interview with Sebastien Bourdais on autosport.com woke me up to this factor. In it, Sebastien pointed out that the Toro Rosso/Red Bull chassis has extreme migration of its center of pressure throughout the speed range. At low speed, the CP is very far forward, so the car oversteers in slow corners. But as the speed rises, the CP migrates toward the rear. At very high speeds, it's very far rearward, so the car understeers in very fast corners.

Bourdais says he doesn't do well with either of these characteristics. Mark Weber and Sebastian Vettel have been able to adapt their driving styles to drive around these problems, but Bourdais hasn't (and neither has Coulthard, apparently).

The reason "Sea Bass" was so quick at Spa is that almost all the corners are medium speed, so the car's CP was in the middle, giving it good balance in almost every corner. And so he flew.

Other Factors Affecting the CP

Speed, of course, is not the only thing which impacts the center of pressure. Ride height, chassis rake, pitch, roll, and yaw all have their impact on the airflow over and under the car. The underbody of a flat-bottomed diffuser car like the Mazda is particularly sensitive to these things. The CP might be migrating all over the place, for all we know, and it might not always be doing this in a way that is intuitive. It makes sense, for example, that increasing the chassis rake would move the CP forward, but does it always?

And, of course, spring rates, damper settings, and tire pressures all have an impact on dynamic ride height and rake, as does the track surface. A steeply banked corner is going to squat the chassis down more than a flat corner of the same radius; stiffer springs are going to reduce the squat compared to softer springs, and stiffer damper settings will impact both overall ride height changes and rake changes during transients.

This is why a downforce car is so much more complicated to set up than a non-downforce car like the Skippy. Almost anything you do to anything affects other things. And it's often very difficult to pin down exactly what is causing a particular reaction to a change. Is it the thing you changed, or is there a ripple effect, with the greater impact coming from a secondary or even tertiary factor?

The Mazda's Center of Pressure

My issue with the Formula Mazda is what I perceive as relentless understeer, particularly at high speeds. I figured this pointed to an aerodynamic center of pressure that was fairly far aft and/or migrated further aft at high speeds.

Therefore, I've been trying to balance the car by moving its CP forward, both by increasing the chassis rake (raising the rear/lowering the front, which should move the CP forward) and by running more front wing.

My experimenting the other day suggested that - lo and behold - tire alignment, both camber and caster (via its dynamic camber change) are also very big factors in the balance of the Mazda, even at high speeds.

It turns out that those big wide bias ply slicks need to have their surface pretty much flat on the pavement. With the front camber and caster in the ball park, the balance of the car is transformed - and radical tricks like very soft tires or extreme chassis rake aren't necessary to get its handling into the neutral zone.

But, as I say, I feel we are still just scratching the surface.