Three and Four Wheel Drift
- Thread starter GhostZ
- 130 comments
- 9,750 views
- 2,151
- Germany
- x_XNyordiaX_x
Never rely on a secondary source. Always find the primary documentation.
My primary source: my own experience and brains. My secondary source: Wikipedia, because I'm too damn lazy to type it out
Downhill Dino
Premium
- 11,753
- Maryland
- DownhillDinosaur
Haha. 
Good one.
Good one.
- 291
- Oxfordshire
- Bread_45
@ AC: Was the Z4 you used stock or had you tuned it at all? I took a stock one out for a go in the High Speed Ring tunnel - got loads of 4 red wheel drifts but was unable to get the outside front tyre to go white. Also what sort of angle were you pulling? Could be I wasn't letting her go far enough before countersteering.
Cheers,
Bread
Cheers,
Bread
- 133
- MI
@ Bread82: I tune my cars myself, so that might be why. I was going to the angle where my car was holding at max counter steer and just barely pulling back, so I twitch the stick back and forth very slightly to keep it at that angle. When I twitch the stick back and forth is when the outside wheel turns white.
- 291
- Oxfordshire
- Bread_45
I've been messing around with drifting, this thread is a bit old but I think stuff is still relevant to here.
@AC, I've observed some drift lobbies, and one lobby had a team (not sure which) of US players, they were getting white outside front tyres all the time. They were discussing tuning over the mic so it looks to be a consequence of a deliberate setup, they were also using several different cars. I've been able to get it on my S13, usually with a lot of front toe out; do you use a lot of front toe out? I guess this makes sense, with toe out as you countersteer, the front outside will be "in line" with the path of your drift, while the front inside will be at a greater slip angle as the tyre is facing more in the direction of your bonnet = front inside gets roasted.
@ Ghost/anyone else that could offer some advice : I've been mucking around with 4 wheel drifts, generally with minimal countersteer, I think it was happening a lot more than I realised before - I was obviously not looking at the tyre indicators much. What I generally do is turn the car in to the drift, usually with moderate trail braking and then a pronounced hard stab on the brakes, this loosens the rear and starts rotation but also often locks up the front, so only the front tyres only go red but the rear is also slipping, although not with locked up burning red wheels (it is now lighter due to the braking). I then add throttle and this encourages the rear to lose traction more as well. Get it balanced = 4 wheel drift. I usually return the steering to centre immediately after the turn in. Providing you've pointed the car in enough to start with, this "smudging sideways" can be on a decent line. This seems to be what Ghost described in the OP
I can sustain this for quite a while, but I can't often get the "understeer on exit" effect described in OP, particularly on longer corners. Either too much gas = oversteer and I have to countersteer loads to keep things stable, or I let off too much and the drift grinds to a halt. In short corners (eiger) it's not really a problem as I have lots of momentum from the recent entry and I can seem to get the car pointed correctly to exit smoothly.
What I have been trying to do to help in longer corners is using more rear toe in, to prevent exit oversteer, and also to get more "forwards" traction to enable the exit to be faster.
With suspension tuning in general (for drift), I tend to use front toe out to help turn in, and rear toe in to help get the power down and exit the drift. Basically I am attempting to "oversteer in, understeer out".
I sometimes set brake balance a little front biased e.g. 7/5, generally if the car breaks loose too easily at the start. This could seem counter to the point of using front toe out, but it seems to feel right for me if I keep the toe. I want some rotation but not an uncontrollable amount!
I read a lot of tuning articles, but don't profess to any real ability at tuning - any comments on setup or technique would be welcome.
BTW, using DS3, all aids off, ABS off. S13 Silvia '91 premium. Will post setup if people want (I'm not on PS3 at the moment and don't have a record of it to hand)
Cheers,
Bread
@AC, I've observed some drift lobbies, and one lobby had a team (not sure which) of US players, they were getting white outside front tyres all the time. They were discussing tuning over the mic so it looks to be a consequence of a deliberate setup, they were also using several different cars. I've been able to get it on my S13, usually with a lot of front toe out; do you use a lot of front toe out? I guess this makes sense, with toe out as you countersteer, the front outside will be "in line" with the path of your drift, while the front inside will be at a greater slip angle as the tyre is facing more in the direction of your bonnet = front inside gets roasted.
@ Ghost/anyone else that could offer some advice : I've been mucking around with 4 wheel drifts, generally with minimal countersteer, I think it was happening a lot more than I realised before - I was obviously not looking at the tyre indicators much. What I generally do is turn the car in to the drift, usually with moderate trail braking and then a pronounced hard stab on the brakes, this loosens the rear and starts rotation but also often locks up the front, so only the front tyres only go red but the rear is also slipping, although not with locked up burning red wheels (it is now lighter due to the braking). I then add throttle and this encourages the rear to lose traction more as well. Get it balanced = 4 wheel drift. I usually return the steering to centre immediately after the turn in. Providing you've pointed the car in enough to start with, this "smudging sideways" can be on a decent line. This seems to be what Ghost described in the OP
I can sustain this for quite a while, but I can't often get the "understeer on exit" effect described in OP, particularly on longer corners. Either too much gas = oversteer and I have to countersteer loads to keep things stable, or I let off too much and the drift grinds to a halt. In short corners (eiger) it's not really a problem as I have lots of momentum from the recent entry and I can seem to get the car pointed correctly to exit smoothly.
What I have been trying to do to help in longer corners is using more rear toe in, to prevent exit oversteer, and also to get more "forwards" traction to enable the exit to be faster.
With suspension tuning in general (for drift), I tend to use front toe out to help turn in, and rear toe in to help get the power down and exit the drift. Basically I am attempting to "oversteer in, understeer out".
I sometimes set brake balance a little front biased e.g. 7/5, generally if the car breaks loose too easily at the start. This could seem counter to the point of using front toe out, but it seems to feel right for me if I keep the toe. I want some rotation but not an uncontrollable amount!
I read a lot of tuning articles, but don't profess to any real ability at tuning - any comments on setup or technique would be welcome.
BTW, using DS3, all aids off, ABS off. S13 Silvia '91 premium. Will post setup if people want (I'm not on PS3 at the moment and don't have a record of it to hand)
Cheers,
Bread
- 271
- United States
- The__Ghost__Z
It all has to do with weight management over the tires.
You cannot let off gas entirely, because you do not want weight to transfer to the front, and if you have too much gas, weight will transfer to the back, so you need to slowly let off while the car is rotating into the corner and slowing down. This is particularly difficult with a DS3 because you only get a few millimeters of error allowed, or less, to do it effectively. If you do it right, you may even have a moment where both brake and gas are on during entry. Simply put, any correciton is over correction.
If you do it correctly, you can just hold the wheel steady and the throttle steady in a neutral position and continue drifting indefinitely. That is the sign you have done it properly. The car continues on a circular path without any steering input, the rear is giving smoke, and the front is slipping (although the front will not smoke because there isn't enough heat and power to make noticeable wear on the tires.
Try to tune your car for even, predictable weight transfer. You can use spring rate formulas for this. This means it should ride, compress, and roll flat. That will optimize the grip during a four wheel drift, and give you the most buffer zone between failure and success at managing the drift.
You cannot let off gas entirely, because you do not want weight to transfer to the front, and if you have too much gas, weight will transfer to the back, so you need to slowly let off while the car is rotating into the corner and slowing down. This is particularly difficult with a DS3 because you only get a few millimeters of error allowed, or less, to do it effectively. If you do it right, you may even have a moment where both brake and gas are on during entry. Simply put, any correciton is over correction.
If you do it correctly, you can just hold the wheel steady and the throttle steady in a neutral position and continue drifting indefinitely. That is the sign you have done it properly. The car continues on a circular path without any steering input, the rear is giving smoke, and the front is slipping (although the front will not smoke because there isn't enough heat and power to make noticeable wear on the tires.
Try to tune your car for even, predictable weight transfer. You can use spring rate formulas for this. This means it should ride, compress, and roll flat. That will optimize the grip during a four wheel drift, and give you the most buffer zone between failure and success at managing the drift.
- 291
- Oxfordshire
- Bread_45
I've been running my spring setup based on static weight distribution, so in steady state cornering the front and rear would want to roll the same amount. One thing that has always eluded me is how to quantify anti roll bars properly given the 1-7 scale in GT. As they effectively act as springs that only work in roll, not pitch, I'd like to be able to add their effective spring rates to the "proper spring" spring rates when calculating roll.
Do you suspect/know if they are a proportional to something like the middle spring rate on the slider (or chosen spring rate?) for that axle?
Also, when comparing one ARB setting to another, I'm guessing that a "7" is not simply 7 times as stiff as a "1" but as to what the actual relationship is, I couldn't say!
Interestingly, a log scale is suggested here.
https://www.gtplanet.net/forum/showthread.php?p=5164799#post5164799
Cheers,
Bread
- 271
- United States
- The__Ghost__Z
I believe in GT5 it says that it affects anti roll bar diameter.
Anti-roll bars are still calculated like springs. However, instead of being given the final product of Hooke's Law (as we are in spring rates) we are given diameter of the roll bar. So to actually calculate the resistance force of the bar, you need precise measurements of the resistance to twisting that various metals have, and exactly how the in-game one is made. This is called the torsion constant I believe, and this is obviously impossible to calculate with what we are given. But we can still guess that if the material's constant stays the same, and we know that the resistance has to do with the total mass within the bar, and that torsion constant, the scale should be the same scale as the volume of a cylinder, or at least close. It's not quite a log scale, but that's not too far off.
http://en.wikipedia.org/wiki/Torsion_spring
However, the relationships that determine when to increase or decrease roll bars is easy enough to understand. It dampens lateral movement through a spring, so the greater that movement (either in the diameter that the torsion spring is "twisted" or the weight, or the cornering forces, or the harmonic motion incurred beyond the cornering forces) then the greater the value you want for roll bars to achieve the same 'feel' or grip.
Interestingly enough, and I believe GT5 does model this to some extent, a car's chassis functions as a roll bar front-to-rear. This means a loose chassis with stiffer anti-roll bars will resist roll similar to how a stiff chassis will resist roll with looser anti-roll bars. This is really important in comparative tuning. This means that the rear's roll at initial turn-in is dampened by the cornering forces on the front, until it 'catches up', since the rear wheels do not turn. That explains why front anti-roll bars tend to be stiffer, because the length of the body initially resists the rear's roll difference to the front at corner entry. So as you said, statically, the car should roll evenly as a unit. However, it approaches that max roll slower on the rear than the front because of the twisting of the chassis.
Long explanation short: Choose anti-roll bar measurements based on the difference of weight transfer over that particular set of wheels (front or back).
So to answer your question, since I believe that the diameter of the roll bar increases as the value goes from 1-7, It should increase like volume does as diameter goes up, assuming length is held constant. Of course, since we only have 7 options to choose from, finding precise values would result in statistically insignificant changes to our tuning, so its better to just remember that as weight, body roll, and cornering force over an axle goes up, the roll bars should too.
Edit: Looking at the post you linked, again the idea is there, but the log scale doesn't give numbers that would correspond to higher limits of grip, but what would be predictable. I'm not entirely sure how accurate it is, and while my guesses are based on experience in-game and physics, but that may not translate accurately if PD cut enough corners.
Sometime in a few months I'll release my tuning method, and it will explain all of this a little more detailed, but I'm still going to get some mileage out of it yet.
Anti-roll bars are still calculated like springs. However, instead of being given the final product of Hooke's Law (as we are in spring rates) we are given diameter of the roll bar. So to actually calculate the resistance force of the bar, you need precise measurements of the resistance to twisting that various metals have, and exactly how the in-game one is made. This is called the torsion constant I believe, and this is obviously impossible to calculate with what we are given. But we can still guess that if the material's constant stays the same, and we know that the resistance has to do with the total mass within the bar, and that torsion constant, the scale should be the same scale as the volume of a cylinder, or at least close. It's not quite a log scale, but that's not too far off.
http://en.wikipedia.org/wiki/Torsion_spring
However, the relationships that determine when to increase or decrease roll bars is easy enough to understand. It dampens lateral movement through a spring, so the greater that movement (either in the diameter that the torsion spring is "twisted" or the weight, or the cornering forces, or the harmonic motion incurred beyond the cornering forces) then the greater the value you want for roll bars to achieve the same 'feel' or grip.
Interestingly enough, and I believe GT5 does model this to some extent, a car's chassis functions as a roll bar front-to-rear. This means a loose chassis with stiffer anti-roll bars will resist roll similar to how a stiff chassis will resist roll with looser anti-roll bars. This is really important in comparative tuning. This means that the rear's roll at initial turn-in is dampened by the cornering forces on the front, until it 'catches up', since the rear wheels do not turn. That explains why front anti-roll bars tend to be stiffer, because the length of the body initially resists the rear's roll difference to the front at corner entry. So as you said, statically, the car should roll evenly as a unit. However, it approaches that max roll slower on the rear than the front because of the twisting of the chassis.
Long explanation short: Choose anti-roll bar measurements based on the difference of weight transfer over that particular set of wheels (front or back).
So to answer your question, since I believe that the diameter of the roll bar increases as the value goes from 1-7, It should increase like volume does as diameter goes up, assuming length is held constant. Of course, since we only have 7 options to choose from, finding precise values would result in statistically insignificant changes to our tuning, so its better to just remember that as weight, body roll, and cornering force over an axle goes up, the roll bars should too.
Edit: Looking at the post you linked, again the idea is there, but the log scale doesn't give numbers that would correspond to higher limits of grip, but what would be predictable. I'm not entirely sure how accurate it is, and while my guesses are based on experience in-game and physics, but that may not translate accurately if PD cut enough corners.
Sometime in a few months I'll release my tuning method, and it will explain all of this a little more detailed, but I'm still going to get some mileage out of it yet.
Last edited:
- 291
- Oxfordshire
- Bread_45
@AC: It'd be cool to try your tune, wonder if I'll get the same effect or if its mostly down to personal technique.
@Ghost: Diameter makes sense. The other thing is IRL a lot of adjustable ARBs have arm linkages with different holes to use, effectively altering arm lengths.
I've just thought of a rough test for ARB stiffness in GT5... On the cape ring loop, with low ride height, it is easy to bottom out. Checking if the car bottoms at a fixed speed with different spring and ARB settings should give insight into how much an ARB click is worth, eg if 7 ARB and 2 spring just bottoms out, but 1 ARB requires 10 spring etc etc.
Cheers,
Bread
@Ghost: Diameter makes sense. The other thing is IRL a lot of adjustable ARBs have arm linkages with different holes to use, effectively altering arm lengths.
I've just thought of a rough test for ARB stiffness in GT5... On the cape ring loop, with low ride height, it is easy to bottom out. Checking if the car bottoms at a fixed speed with different spring and ARB settings should give insight into how much an ARB click is worth, eg if 7 ARB and 2 spring just bottoms out, but 1 ARB requires 10 spring etc etc.
Cheers,
Bread
