Jump to content


Photo

pull to damp


  • Please log in to reply
78 replies to this topic

#51 cheapracer

cheapracer
  • Member

  • 10,388 posts
  • Joined: May 07

Posted 30 March 2012 - 03:26

Zero warp stiffness!


Indeed it has.

Roll centres are quite low. Did you consider making them adjustable? .


I think he could go lower easy enough, maybe not ideal though.

Very keen to hear how it drives.


It uses an engine.


Advertisement

#52 Johan Lekas

Johan Lekas
  • Member

  • 52 posts
  • Joined: May 09

Posted 30 March 2012 - 06:22

You are quite right, in principle. Controlling the roll (moment) distribution provides a method of controlling lateral balance in a turn. A roll moment distribution can be expressed as a combination of roll & warp, and it is the warp component that provides balance control. So far as I'm aware, there are three main methods of controlling roll moment bias in a turn; one is to have different (adjustable, ideally) roll centre heights, another is to control warp load directly & the last to controll the torque distribution (last added for completness).

In a conventional suspension a change in roll moment distribution is achieved by roll stiffness distribution, which yields the required warp load because the suspension has a warp stiffness. In a suspension that truely has zero warp stiffness, changing roll stiffness distribution will not (I think) change warp load, because the chassis will simply move to null the effect. What is required in a vehicle that doesn't roll, or has zero warp stiffness, is a mechanism to provide the stabilzing turn dependent warp offset.

Having different roll cente heights would be one way of achieving this. Active suspension provided a way of achieving a low warp stiffness with a directly controllable warp offset. The Lotus system, for example, embodied an algorithm that used the steering input and vehicle responses as a "trajectory" command - effectively making the vehicle neutral, at least whilst all four wheels remained in contact.

I personally think that Johan's design probably requires easily adjustable roll centre heights, or something equivalent.

I actually think the roll distribution is changed when the rocker geometry is changed.
And from your description of the three possible ways of controlling roll moment bias, it's controlling torque distribution
I'll try to describe my thinking:

Starting from the symmetrical 50/50 case lets assume the MR ( = spring displacement/wheel displacement) is 1 from the axle to the roll-spring for both front and rear.
ie when the axle rolls (left wheel down, right wheel up) 1" at the wheel, one end of the roll-spring is moved 1".
In a turn a force F is acting at each outside wheel, the roll spring is compressed by F at each end, and the total "roll-opposing force" is 2*F.

If MR is changed to 1.5 for the front and to 0.5 for the rear (MRf = 1.5 and MRr = 0.5), the force on the outside front has to be 1.5*F to give a force F on the roll spring. Similarly the force on the outside rear is 0.5*F to give a balancing force of F on the roll spring, for a total "roll opposing force" of 2*F (as in the 50/50 case).

The displacements of the wheels are then inversely proportional to the forces, ie the front axle rolls 2/3" at the wheel, and the rear 2".
Looking at the wheel displacements (after the above movements) the outside front is +2/3", inside front -2/3", outside rear +2", inside rear -2".

But since the axles can move freely in "warp", with the roll-spring still compressed, the outside rear can move -1" (which is MRr*(-1)" = -0,5" at the roll spring) and the outside front can move +1/3" (which is MRf*1/3" = +0.5" at the roll spring) so all the wheels are in the same plane with outside front at +1", inside front at -1", outside rear at +1" and inside rear at -1"



#53 Johan Lekas

Johan Lekas
  • Member

  • 52 posts
  • Joined: May 09

Posted 30 March 2012 - 06:37

Roll centres are quite low. Did you consider making them adjustable? Of course your roll resistance and damping is fully adjustable. Are you confident that roll damper has identical push and pull characteristics?

Very keen to hear how it drives.

The wheels that are on now are too small, specially at the rear. It's designed for a ground clearance of around 75mm so roll centers will be 90-100mm.

The only drawback I could see with this principle is that he'd possibly need 'non-conventional' dampers to avoid different damping characteristics in left and right turns (I think 'normal' car dampers should have difference between bump and rebound damping by some 3x)...

It's true I had to do some searching to find 1:1 dampers. Most people where saying that it could only be achived with trough-rod dampers (and they are probably right if you want perfect symmetry and adjustability) but Protech shocks made one (non-adjustable) from a standard body that is symmetrical enough for my purposes at a very reasonable price. Leda offered to make a non-adjustable trough-rod damper for £400.
I'll post the damper-curves when I find them



#54 Johan Lekas

Johan Lekas
  • Member

  • 52 posts
  • Joined: May 09

Posted 30 March 2012 - 06:57

These guys have to be mentioned also http://www.creuat.com/
They have been doing systems with separate springing/damping for roll/warp/pitch/heave for some time now.
SAE paper
http://www.creuat.co..... (screen).pdf


#55 cheapracer

cheapracer
  • Member

  • 10,388 posts
  • Joined: May 07

Posted 30 March 2012 - 08:39

It's true I had to do some searching to find 1:1 dampers.


You could go as simple as 100's of off the shelf 50/50 dampers, fill it completely with oil and merely add an isolated nitrogen gas chamber you could pick up from any motorcycle dismantlers (to allow for oil displacement by the shaft).

A number of bike shocks I have rebuilt could easily have the shim stack either side of the piston built up the same for 50/50 as well - they may not offer the travel you want though, hmm, you could get 2 and make one Frankenshocker out of them. In fact 2 in hand would give you the 2 equal shim stacks.

Most people where saying that it could only be achieved with trough-rod dampers


Rubbish.


Actually I just realised I was handling a large machine stop damper just yesterday, hmm I will have a closer look at it tomorrow, didn't relate it to this application at all.

Edited by cheapracer, 30 March 2012 - 08:42.


#56 Johan Lekas

Johan Lekas
  • Member

  • 52 posts
  • Joined: May 09

Posted 30 March 2012 - 08:59

A number of bike shocks I have rebuilt could easily have the shim stack either side of the piston built up the same for 50/50 as well

As been described to me the problem is the volume of the rod on one side of the piston. Moving the piston 1 cm in one direction does not push the same amount of oil trough the shim stack as moving it 1cm in the other direction. So the shim-stacks have to be different to compensate. The pressure areas are different because of the rod, so the shim-stcks have to be different to compensate
Have also thought of MC steering dampers, but they are to weak and don't have linear characteristic


you could get 2 and make one Frankenshocker out of them. In fact 2 in hand would give you the 2 equal shim stacks.

I think using 2 in parallell is the easiest way of getting adjustability. But it adds weight and some complexity

Edited by Johan Lekas, 30 March 2012 - 09:06.


#57 cheapracer

cheapracer
  • Member

  • 10,388 posts
  • Joined: May 07

Posted 30 March 2012 - 09:39

As been described to me the problem is the volume of the rod on one side of the piston.


... and your seat isn't in the middle of the car either.

Offset the shock's compression and rebound adjustment to compensate.


Have also thought of MC steering dampers,


Well there's another option, MC's aren't the only machines to have steering dampers, there's a huge range of them from VW Bug's through to the biggest trucks. Try Hot Rod and especially 4WD sites.

http://www.snakeraci...-tough-dog.html

Edited by cheapracer, 30 March 2012 - 09:47.


#58 gruntguru

gruntguru
  • Member

  • 7,637 posts
  • Joined: January 09

Posted 30 March 2012 - 10:04

I think for this application an adjustable damper would be very useful. You could have lots of fun optimising the spring and damper rates in roll.

What are your initial selections for the front and rear spring/damper units? There is a world of opportunity there to play with dive and squat without affecting wheel rates a great deal and almost no effect on under/oversteer.

EDIT. Have you analysed your extended damper/linkage for buckling? The L/R ratio looks large but I am guessing the peak compression loads will be low enough to avoid buckling.

Edited by gruntguru, 30 March 2012 - 10:08.


#59 Johan Lekas

Johan Lekas
  • Member

  • 52 posts
  • Joined: May 09

Posted 30 March 2012 - 13:29

I think for this application an adjustable damper would be very useful. You could have lots of fun optimising the spring and damper rates in roll.

What are your initial selections for the front and rear spring/damper units? There is a world of opportunity there to play with dive and squat without affecting wheel rates a great deal and almost no effect on under/oversteer.

EDIT. Have you analysed your extended damper/linkage for buckling? The L/R ratio looks large but I am guessing the peak compression loads will be low enough to avoid buckling.

You're maybe right about the adjustability of the roll-damper, but this was the simplest way forward. And I can still adjust the roll stiffnes with the rockers (and have a lot of fun  ;) )
The MR for roll from axle to roll spring is 0.51 both front and rear, but is adjustable. The roll spring is 80 N/mm. In roll it's compressed from both ends, so it acts as two 160N/mm springs i series. So what one axle "sees" is a 160N/mm spring. with MR = 0.51 it gives a an "axle roll ratio" of 160*(0.51)^2 = 42N/mm, and a total for the car of 84N/mm
Front spring is 30N/mm, rear is 33N/mm.
Target weight with driver 450 kg
43/57 % distribution front/rear.
46 kg/146 kg unsprung/sprung weight at the front and 51kg/207Kg at the rear

Regarding buckling the long rod to the front is maybe risky. I'll put a "safety support" halfway to ensure that if it starts to bend out it won't be catastrophic




Advertisement

#60 GSpeedR

GSpeedR
  • Member

  • 35 posts
  • Joined: March 12

Posted 30 March 2012 - 18:52

Johan, fascinating design!

I probably need to spend some more time thinking about your design, but I don't see how warp is decoupled from roll for anything but equal (50/50%) roll stiffness distribution. It seems to me that adjusting the rocker ratios front or rear will create stiffness in the warp mode. Therefore changing roll stiffness distribution would indeed change your warp load (or cross-weight) and thus your balance, but you have have a nonzero warp stiffness.

I could be mistaken, and it certainly has 'low' warp stiffness if not.

#61 gruntguru

gruntguru
  • Member

  • 7,637 posts
  • Joined: January 09

Posted 31 March 2012 - 08:17

I probably need to spend some more time thinking about your design, but I don't see how warp is decoupled from roll for anything but equal (50/50%) roll stiffness distribution. It seems to me that adjusting the rocker ratios front or rear will create stiffness in the warp mode. Therefore changing roll stiffness distribution would indeed change your warp load (or cross-weight) and thus your balance, but you have have a nonzero warp stiffness

I think it wll still have zero warp stiffness (statically) but the sprung mass will have to roll a little, ie it will follow the axle with the greater roll stiffness allocation. Of course in a dynamic sense this will add some resistance to warp due to roll inertia of the sprung mass.

#62 cheapracer

cheapracer
  • Member

  • 10,388 posts
  • Joined: May 07

Posted 31 March 2012 - 08:28

I think it wll still have zero warp stiffness


Only in perfect 50/50 harmony.

No, we were to quick to judge - everything that GSpeedR speaks is correct.

I had a migraine yesterday (serious), what's your excuse?

Edited by cheapracer, 31 March 2012 - 08:28.


#63 gruntguru

gruntguru
  • Member

  • 7,637 posts
  • Joined: January 09

Posted 31 March 2012 - 09:01

The MR for roll from axle to roll spring is 0.51 both front and rear, but is adjustable. The roll spring is 80 N/mm. In roll it's compressed from both ends, so it acts as two 160N/mm springs i series. So what one axle "sees" is a 160N/mm spring. with MR = 0.51 it gives a an "axle roll ratio" of 160*(0.51)^2 = 42N/mm, and a total for the car of 84N/mm

I am not familiar with standards for roll stiffness but I would imagine it should be expressed in Nm/degree.
Trying to get a feel for your roll stiffness, I assumed track = 1500mm, CGH = 400mm, RCH = 100mm, Sprung mass = 350 kg, LatAcc = 1G

Roll moment = 350 x 9.8 x (0.4-0.1) = 1029 Nm
Sprung weight transfer = 1029/1.5 = 686 N
Displacement at wheels = 686/84 = 8.2mm
Roll (deg) = tan-1(8.2/1500) = 0.3 degrees.
Sounds very small. I may have missed a factor of two somewhere by using full track width instead of half but still a very small amount of roll. Can you point out my mistake?

#64 Johan Lekas

Johan Lekas
  • Member

  • 52 posts
  • Joined: May 09

Posted 31 March 2012 - 12:08

I am not familiar with standards for roll stiffness but I would imagine it should be expressed in Nm/degree.
Trying to get a feel for your roll stiffness, I assumed track = 1500mm, CGH = 400mm, RCH = 100mm, Sprung mass = 350 kg, LatAcc = 1G

Roll moment = 350 x 9.8 x (0.4-0.1) = 1029 Nm
Sprung weight transfer = 1029/1.5 = 686 N
Displacement at wheels = 686/84 = 8.2mm
Roll (deg) = tan-1(8.2/1500) = 0.3 degrees.
Sounds very small. I may have missed a factor of two somewhere by using full track width instead of half but still a very small amount of roll. Can you point out my mistake?

There is a factor of two because I was unclear in my MR definition. The MR = 0.51 is for axle rotation, ie 1mm up at the left wheel and 1mm down at the right wheel gives 0.51mm at one end of the spring. This means that the 84N/mm is when measuring displacement at one wheel, so over the track it's 42N/mm. Track is 1400mm
This gives roll = tan-1 (16.4/1400) = 0.7 degrees




#65 gruntguru

gruntguru
  • Member

  • 7,637 posts
  • Joined: January 09

Posted 31 March 2012 - 22:56

Thanks Johan, that was exactly what I was unsure of.

Do you know how that roll figure (0.7 degrees at 1G cornering) compares to cars with conventional suspension? Anyone else out there have figures for typical performance or race cars?

EDIT. Roll stiffness in Nm/Deg comes to 1029/0.7 = 1470 Nm/Deg


Edited by gruntguru, 31 March 2012 - 22:59.


#66 Greg Locock

Greg Locock
  • Member

  • 6,353 posts
  • Joined: March 03

Posted 01 April 2012 - 05:35

Thanks Johan, that was exactly what I was unsure of.

Do you know how that roll figure (0.7 degrees at 1G cornering) compares to cars with conventional suspension? Anyone else out there have figures for typical performance or race cars?

EDIT. Roll stiffness in Nm/Deg comes to 1029/0.7 = 1470 Nm/Deg

0.7 is very low for a road car, and is around what you'll get from the tires. So it sounds about right.

#67 mariner

mariner
  • Member

  • 2,329 posts
  • Joined: January 07

Posted 01 April 2012 - 13:14

Johan, can I first say I think your project is brilliant, a radical concept that are putting your real money and time into, looking forward to its first test.

As I am not an engineer aolot of the comments are beyond me but I do have one question ( not criticism).

You quote an all up weight of 450kg and give the matching unsprung weights. These are quite high versus some norms as your sprung weight is very low.
The front ratio is about 1:3 and the rear 1:4 . In addition you will have to face the problem of gyro precession on a beam front axle which may be minimised by your small wheels but can't be eliminated.

My question is do you yet know to what extent the damper/spring rates will need to be set to control unsprung weight movement versus a very light chassis rather than controlling the heave/roll /. pitch dynamics of the sprung mass.

The fact that the unsprung control is focussed on a single spring/ damper in bounce and just two in roll or heave will put extra unsprung control loads on the dampers compared to the usual set up I would think.

Just a thought

#68 gruntguru

gruntguru
  • Member

  • 7,637 posts
  • Joined: January 09

Posted 02 April 2012 - 03:16

Good point Mariner. As I see it the single-wheel-bump mode has a very low effective spring/damper rate and the warp mode virtually zero. Imagine a cyclic warp input exciting the relatively large unsprung warp-mode inertia. What will happen? Large contact patch load oscillation?

More realistically, what about the same scenario with a cyclic input to a single wheel.

#69 Johan Lekas

Johan Lekas
  • Member

  • 52 posts
  • Joined: May 09

Posted 02 April 2012 - 07:43

Johan, can I first say I think your project is brilliant, a radical concept that are putting your real money and time into, looking forward to its first test.

As I am not an engineer aolot of the comments are beyond me but I do have one question ( not criticism).

You quote an all up weight of 450kg and give the matching unsprung weights. These are quite high versus some norms as your sprung weight is very low.
The front ratio is about 1:3 and the rear 1:4 . In addition you will have to face the problem of gyro precession on a beam front axle which may be minimised by your small wheels but can't be eliminated.

My question is do you yet know to what extent the damper/spring rates will need to be set to control unsprung weight movement versus a very light chassis rather than controlling the heave/roll /. pitch dynamics of the sprung mass.

The fact that the unsprung control is focussed on a single spring/ damper in bounce and just two in roll or heave will put extra unsprung control loads on the dampers compared to the usual set up I would think.

Just a thought

Thanks for the kind words. I appreciate anyone putting time and thought on this and writing comments. It's valuable design review

I'm aware the unsprung/sprung ratios is not so favorable, but haven't done any deeper analysis because I find it too complex.
I think the springing is quite soft with natural frequency of 2.4 Hz at the front and 2 Hz at the rear.

Also, in a few aspects the situation is not as bad as the ratios indicate.
- In a one wheel bump the whole unsprung mass does not move, more like 50% of it
- The free warp mode lowers the wheel rate for a one-wheel bump (see Creuat paper http://www.creuat.co..... (screen).pdf)
For the damping I've used general advice (that I've found on OptimimG Technical Papers/ Springs and Dampers and by consulting Mark Ortiz) and hope the adjustment range will cover
Speed (mm/s)	Damping ratio (part of critical damping)
	  0					  0,3
	  20					0,6
	  100				  0,3
The total damping is divided on 1/3 compression and 2/3 on rebound

As gruntguru notes the warp mode is undamped which maybe could cause some "interesting" behavior, but I decided to not do anything about it and see what happens. Damper(s) could be attached somewhere between the linkage and the chassis if it proves to be a problem

Edited by Johan Lekas, 02 April 2012 - 10:38.


#70 DaveW

DaveW
  • Member

  • 431 posts
  • Joined: January 09

Posted 02 April 2012 - 09:14

What will happen? Large contact patch load oscillation?

I rig tested a "quad" with a similar layout a couple of years ago. It was not fitted with dampers of any description (not required, apparently), but it was not devoid of (friction) damping. Unsurprisingly, perhaps, the hub modes were some way from being in control (damping ratio's < 10 percent). However, when the vehicle was subjected to a "kerb" input (a phased input down one side of the vehicle), roll was the only sprung mass mode requiring additional damping.

Interestingly, that vehicle did not have zero warp stiffness (so far I could see), which leads to a question, with zero warp stiffness, what controls mean suspension position, especially when the c.g. is not on the vehicle centreline?

Edited by DaveW, 02 April 2012 - 09:18.


#71 gruntguru

gruntguru
  • Member

  • 7,637 posts
  • Joined: January 09

Posted 02 April 2012 - 10:26

Interestingly, that vehicle did not have zero warp stiffness (so far I could see), which leads to a question, with zero warp stiffness, what controls mean suspension position, especially when the c.g. is not on the vehicle centreline?

It has roll stiffness to centre it in rotation about the x axis and pitch stiffness for the Y. However, if the road has a "twist" in it the "wheel weights" will remain the same as for a flat road.

#72 DaveW

DaveW
  • Member

  • 431 posts
  • Joined: January 09

Posted 02 April 2012 - 12:37

It has roll stiffness to centre it in rotation about the x axis and pitch stiffness for the Y. However, if the road has a "twist" in it the "wheel weights" will remain the same as for a flat road.

Your quite right, gg. My muddled thinking....

#73 gruntguru

gruntguru
  • Member

  • 7,637 posts
  • Joined: January 09

Posted 15 September 2012 - 05:26

Bump. Very keen to hear of any progress Johan???

#74 carlt

carlt
  • Member

  • 4,169 posts
  • Joined: June 09

Posted 15 September 2012 - 16:04

Bump. Very keen to hear of any progress Johan???

me too

#75 carlt

carlt
  • Member

  • 4,169 posts
  • Joined: June 09

Posted 15 September 2012 - 16:07

it appears that the horizontal 'roll' coilover has to work the spring both in compression and tension ?

going back to this - are std compression coil springs designed to work in tension [ experience of light springs suggest they stretch-]


#76 Greg Locock

Greg Locock
  • Member

  • 6,353 posts
  • Joined: March 03

Posted 15 September 2012 - 21:49

going back to this - are std compression coil springs designed to work in tension [ experience of light springs suggest they stretch-]

A standard coil spring is happy to work in tension, except you need to think about the ends.

#77 Johan Lekas

Johan Lekas
  • Member

  • 52 posts
  • Joined: May 09

Posted 16 September 2012 - 07:01

Bump. Very keen to hear of any progress Johan???

Missed this bump somehow
Not to much progress to report I'm afraid. Since this is the first car I build every design problem is new so to speak, which means it takes a lot of time. Currently working on fuel system and cooling. Left do do before it's a drivable rolling chassis is basically exhaust, driveshafts, electrics, brakelines, clutch, floor. Here's a build thread i update more regularly
http://rejsa.nu/foru...7...c&start=483



#78 Bloggsworth

Bloggsworth
  • Member

  • 9,394 posts
  • Joined: April 07

Posted 16 September 2012 - 17:16

That makes more sense. To be honest I think that that is one of those drawing board arguments, yes you have a rationale, but who says equal is best? Most production cars are underdamped in roll when correctly damped for pitch and bounce.


I refer you to the Austin 1100, cornered as flat as a pancake...

#79 Fat Boy

Fat Boy
  • Member

  • 2,594 posts
  • Joined: January 04

Posted 21 September 2012 - 16:01

Paging Dr. Goldberg, Paging Dr. Goldberg...