Mazda MX-5 Miata banner

1 - 20 of 52 Posts

·
Registered
Joined
·
289 Posts
Discussion Starter · #1 ·
As prompted by Tony over in t'other thread, a look at how I go about developing an upgrade kit from scratch. I'll endeavour to keep the maths out of it, and I'll not be recommending any one product over another.

Firstly, a few thoughts on what we're actually trying to do. Ride and handling is all about controlling weight transfer as the driver fiddles with the controls and the car goes across uneven ground. Weight transfer is an immutable physical law, thanks to Isaac Newton, so you're going to get it anyway. Even if you stop the body from moving around, weight transfer still happens. The difference with a non-moving body is that the reaction to that weight transfer is

a ) taken out in twist of the body, which isn't much but can mess around with your carefully crafted suspension geometry and calibration
b ) is effectively undamped, so can carry on for a long while which really messes around with your cunning engineering and really messes with the drivers brain because you're getting big variations in the contact patch forces and a less consistent feel.

If you watch the ultra slow-motion footage of an F1 car you'll see what I mean - most of the suspension movement is in the sidewalls of the tyres (something that may well disappear next year if Michelin get their way and we get 18" rims for F1 cars) and the bodywork twisting rather than the wheels actually moving up and down relative to the body.

So, I work on the premise that you're going to get weight transfer anyway and its better to have suspension travel which we can control with the dampers rather than over-limiting movement and having the chassis take away some of your control.

That said, some of the weight is transferred directly up the suspension arms into the body, so you can't control that. The amount is a function of the alignment of the various suspension arms and the amount varies as the suspension arms move and the car rolls.

Anyway, back in the room...

Step 1 is always measurement. If you're developing an upgrade kit, you've generally got to work with what the original equipment manufacturer has designed in rather than expecting your end user to fork out for new suspension arms, driveshafts and a whole heap of welding and stiffening to put new hardpoints in. One of my earliest jobs out of university involved a vehicle that used applique plates for bolting the suspension arms to the body (not to mention that it was built out of 75mm thick plate and torsional rigidity wasn't a big issue) but generally you've got to work with what the manufacturer has designed in.

So, it's up on a ramp and out with the tape measure to locate in three dimensions where the suspension arms are relative to the ground, the location of the wheels relative to the outer links, the position of the inner track rod ends and a visit to an alignment rig to find out what the basic alignment values of camber, caster and toe are.

I also measure suspension motion ratios (how much the spring and damper move in relation to the wheel moving), the spring sizes and the anti-roll bar diameters and lever arms.

Once I've got those numbers, then I feed the data into a suspension kinematics program. I use SusProg, mainly because it's cheap, accurate and the developer is the sort of bloke who will customise things just for you. The output of SusProg lets me know how the suspension moves and also what effect changing ride height (and if it's a vehicle like the 5 where the alignment is adjustable by moving inner pickup points around) has on the suspension alignment.

I'm effectively looking at two things:

a ) The roll centre height. It's effectively the point where the load transfer is applied to the body of the car via the suspension and controls how much of the total weight transfer is fed into the suspension and how much is fed directly into the body by the suspension arms
b ) Kinematic changes - things like bump steer (where the toe of the wheel changes as the suspension moves up and down relative to the body), roll steer (where the toe changes as the body rolls) and camber and caster changes.

Sometimes, you get something that is a show stopper for doing sensible things like lowering the car (a lower c of g means less load transfer which is a good thing, but can result in more load transfer into the body at the expense of the suspension which is generally a bad thing). I know of one Teutonic manufacturer where a canny race engineer found large chunks of time by returning the car to stock ride height because the suspension geometry was so duff in its lower position. Similarly sometimes you find that the stock ride height is the best compromise (believe it or not, most R&H engineers working for the OE actually know what they're doing) so you're looking then at optimising one feature of the car's behaviour at the expense of others - for example if you go very extreme with the kit, you may make the car significantly less stable if it carries your weekly shop in the boot.

Step 2: Testing a representative vehicle. It really helps if you know what you're trying to improve, so I endeavour to take the test vehicle to somewhere taxing and measure its handling performance, both objectively and subjectively. In other words, fit a load of data logging equipment to the car and take it to Millbrook Proving Ground. I use Millbrook for several reasons, not least of which is that it's very close to my house. The other main reason is the fact that it has the Outer Handling Circuit. It's a short (about 0.8 mile), narrow (the width of a British B-road) handling circuit designed to expose any frailty in the capabilities of a car (and driver, but that's another matter).

The purpose of testing is to measure the capabilities of the car and its response to the wheel. It's not about setting an outright laptime but actually seeing what happens if you put some very quick changes of direction in of combine a long sweeping bend with a load of camber changes. We're looking at setting a baseline against which we can measure the improvements that the finished kit will bring. One of the key things for most kits is driveability - while we all might like to think we're somewhat more talented than Lewis 'n' Jenson, the fact of the matter is that we're not and while you can engineer a kit with a significantly increased performance envelope, if you need their skills to exploit it without a metal-hedge interface, then it's not an appropriate avenue to explore.

Here's where the subjective stuff comes in. How easy is it to get a 'hooked up' lap? How easy is it to repeat that lap? How easy is it to fix problems created by the wrong speed and line? Are there some nasty behaviours (lift-off oversteer, power understeer and the like) that you might want to try and tune out? There's a whole range of international standard manoeuvres for this sort of thing but you need a subjective assessment as to how well the stock car deals with these things.

Once we've got these things measured, we're ready to start the real work...
 

·
Registered
Joined
·
1,483 Posts
Very nice right up, i might have to rethink a few ideas. I have a set of tein ha`s and to achieve a ride height of 12 and a half inches at the front and 13 inches at the rear i have to wind up the springs untill the helper springs are almost fully compressed, so i thought i might extend the bottom shock mounts on the wishbones by an inch, that i thought might allow me to unwind the springs and release the coil-bound helper springs, what do you think, cheers tony.
 

·
Registered
Joined
·
289 Posts
Discussion Starter · #6 ·
Very nice right up, i might have to rethink a few ideas. I have a set of tein ha`s and to achieve a ride height of 12 and a half inches at the front and 13 inches at the rear i have to wind up the springs untill the helper springs are almost fully compressed, so i thought i might extend the bottom shock mounts on the wishbones by an inch, that i thought might allow me to unwind the springs and release the coil-bound helper springs, what do you think, cheers tony.
In finest 'expert' fashion, it depends
There are a few questions that need to be answered:

a ) What are the respective rates of the main and helper springs? The kit may have been designed so that the helper springs are fully compressed under normal operation. A pair of springs acting in series has an overall rate less than either of the individual spring rates so this has an effect on the damper rates you need to control movement

b ) How much travel the damper has? Would moving the damper mount give too much reduction in bump travel.

c ) The stiffness of the new mounts. Will they buckle with the lower lambda value?

d ) Will the modified motion ratio turn the kit into a pile of poo as the rates are then wrong?

Lots of questions, no definitive answer...
 

·
Registered
Joined
·
289 Posts
Discussion Starter · #8 ·
Step 3 - Picking a spring rate

The first thing to recognise is that each corner of the car has two springs attached - the coil spring and the sidewall of the tyre. Because you have two springs you have to deal with two resonant frequencies: The first is where the wheel bounces up and down against the spring, referred to the as the 'ride frequency' and the second is where the whole corner bounces up and down against the sidewall, called the 'wheel hop frequency'. Wheel hop is something we can't really help - you're limited by the fact that in wheel hop the only thing moving is the sidewall, so you're governed by what damping the tyre maker has built into the rubber.

Ride frequency is a function of three things: the spring rate, the suspension motion ratio and the weight on that particular corner of the car. If we set the frequency too high, then we'll have issues with human response (there are some fundamental frequencies that you don't want to vibrate at, lest you create amusing vomiting effects) and dynamic contact patch load - the body effectively pulls the tyre up and down and you get poor vertical loading and thus less grip. Softer is better, but with a softer suspension it takes longer for everything to reach a final position, which leads to a 'sluggish' feel to the driver. So effectively you want to go as stiff as you can while maintaining decent contact patch loads. For a car without downforce, you're looking at a practical limit for ride frequency of around 1.5 Hz.

There's a second effect at work here - the human body is less sensitive to straight up and down movement and much more sensitive to pitch motion. If you go over a single bump the front moves up before the rear, causing pitch. If you have a higher ride frequency at the rear then it takes less time to reach its bump position, so you lessen pitch. There's a 'magic' ratio of ride frequencies which limits pitch, so if you can arrange things thus it's better for the end users appreciation of ride - even though it might be worse (in terms of pure vertical g under the driver) than OE, reducing the pitch makes it feel better.

So once you've picked a suitable ride frequency at front and rear, and converted that back into a spring rate. you then have two more checks to make:

1) Suspension workspace. It's no good picking a spring rate that will give lovely ride and consistent contact patch load if you haven't got the suspension travel to work with it. Plenty of times I've had to compromise a kit by going for a stiffer spring rate just to keep the suspension out of the bump stops. Similarly careful attention to the bump stops in a kit can make all the difference.

2) Roll moment distribution. Without delving too far into the theory, an axle is less efficient at generating grip if more of the weight transfer is carried on that axle. At a constant speed and radius, i.e. mid-bend on a long sweeper, this varying efficiency means you need either more or less steering. Nobody wants to fit a kit that means they need more steering on every bend, so you have to carefully balance to overall spring and ARB rates at each end to cover this. Normally you can get in the ball park by setting the ratio of front:rear stiffness to approximately 5% greater than the front:rear weight distribution, but I've known a +/-5% ratio either side of that to be optimal.

If you're creating suspension from scratch you tend to work the other way round: set the ride frequencies then set the ARB rates to suit.

So, once you've run around the calculation loop a few times, you've got some suitable spring rates to play with... now you're ready to start considering the dampers.
 

·
Registered
Joined
·
1,483 Posts
b ) How much travel the damper has? Would moving the damper mount give too much reduction in bump travel.

c ) The stiffness of the new mounts. Will they buckle with the lower lambda value?

d ) Will the modified motion ratio turn the kit into a pile of poo as the rates are then wrong?

Lots of questions, no definitive answer...
[/quote]

I think you have given a definitive answer there, although we use riser plates on motor cycles which replace the existing shock to swinging arm plates, they dont appear to alter the overall shock travel in either compression or extension, the way they are made does`nt appear to alter the speed of travel either. That does`nt mean that my idea would work either, i did think i might make a spacer to fit on the top mount, but as you say the helper spring may in fact need to be heavily compressed as part of the overall design, in conclusion you have saved me a lot of time fixing a problem that probable did`nt exist, thanks for your time kind regards tony
 

·
member
Joined
·
9,153 Posts
hi mate, can you post sometime on how front / rear ride height changes the front / rear roll centre - by 'how' i mean in terms of a few numbers, rather than principle. eg 13" front ride height = x front roll centre height, 12" front ride height = y front roll centre height etc. also, is there any way you can measure the position of the c of g?

lastly, i'm assuming that at stock ride height, we have a rising rate suspension set-up just in terms of geometry. are you able to post on the suspension ratio as you lower ride height (or just how the suspension ratio changes over suspension arm movement).
 

·
Registered
Joined
·
407 Posts
I love the academic approach, very entertaining !!!!
Picked up a new Fiat Abarth last week at Brands, now you would think that Abarth would have spent quite a few quid developing this car, wouldn't you?.
Well.... what they did was put some uprated springs on the Brazilian made rear shocks and Eastern block front struts and sent it out for use by the general public. It is underdamped and oversprung on the rear and undersprung and overdamped on the front with the understandable effect of severe understeer on track and pogo effect on B roads.

Fitted new SAS front struts and springs and kept the rear springs with new adjustable rear dampers.

Job done, car transformed on B roads. Back to Brands for track testing next Wednesday, I bet it doesn't understeer !!!!!!.
 

·
Registered
Joined
·
289 Posts
Discussion Starter · #13 ·
hi mate, can you post sometime on how front / rear ride height changes the front / rear roll centre - by 'how' i mean in terms of a few numbers, rather than principle. eg 13" front ride height = x front roll centre height, 12" front ride height = y front roll centre height etc. also, is there any way you can measure the position of the c of g?

lastly, i'm assuming that at stock ride height, we have a rising rate suspension set-up just in terms of geometry. are you able to post on the suspension ratio as you lower ride height (or just how the suspension ratio changes over suspension arm movement).
I can do, but you realise it's a real can of worms:

a. Rake affects roll centre height - the inner pickups move as the body pitches
b. Alignment affects roll centre height as the position of the inner pickups is moved by the adjusters
c. Static roll centre height is less important than what it does as the suspension moves in roll and bump
d. Similarly bump steer is more of an issue than roll centre height.
 

·
Registered
Joined
·
289 Posts
Discussion Starter · #14 ·
I love the academic approach, very entertaining !!!!
Picked up a new Fiat Abarth last week at Brands, now you would think that Abarth would have spent quite a few quid developing this car, wouldn't you?.
Well.... what they did was put some uprated springs on the Brazilian made rear shocks and Eastern block front struts and sent it out for use by the general public. It is underdamped and oversprung on the rear and undersprung and overdamped on the front with the understandable effect of severe understeer on track and pogo effect on B roads.

Fitted new SAS front struts and springs and kept the rear springs with new adjustable rear dampers.

Job done, car transformed on B roads. Back to Brands for track testing next Wednesday, I bet it doesn't understeer !!!!!!.
Doesn't surprise me at all - the accountants run the show these days so you get disparate bits from all over the world flung together. Combine that with the difference in road surfaces worldwide (I don't think Fiat do much UK testing) and a realisation that track days aren't that important in many markets and you get cars that don't handle that well.

If it were any different, we'd all be out of a job...
 

·
Registered
Joined
·
289 Posts
Discussion Starter · #15 ·
Step 4 - Picking Damper Rates

If you thought there was a little too much maths in the explanation of how to pick a spring rate, then it gets worse when you get to the dampers. Much of the complexity comes from the fact that you've got two springs bobbing up and down (the tyre sidewall and your carefully selected spring) with two masses (the wheel and tyre assembly; and the body) but only one damper to try and control the whole lot. If you factor in the difference in frequencies, it all gets very complex.

In general you divide the rate of operation of a damper into four distinct zones:

Frictional (0 - 5 mm/s) - the really small movements that are resisted mainly by friction in damper shaft and seals and stiction in the suspension bushes. If you have too much of this you can get a phenomenon called 'boulevard jerk' where the car rocks up and down on tyre sidewalls on smooth surfaces because there's not enough energy going into the suspension to break through the friction. It's a very annoying thing to have and can only be fixed by finding ways to reduce friction - new bushes, low friction piston seals and the like.

Inertial (5 - 25 mm/s) - most of these movements are the body moving relative to the wheels, i.e. weight transfer effects. Generally called 'low speed' - it's this range of movement that controls the handling more than anything else.

Road Input (25 - 200 mm/s) - movement of the wheels relative to the body - here you want the suspension to move quickly to absorb the energy and then control it well on the way back.

Abuse (> 200 mm/s) - big inputs caused by doing something dumb, such as taking air over a kerb. If you're in a race series its important. On the road is more about ensuring you don't blow the damper up.

So, distilling all that down, you effectively have three things that you tune in a damper:

1. Gradient of the low speed damping curve
2. Gradient of the high speed damping curve
3. The point at which you transition from high to low (the 'knee point')

In general when I'm designing rates, I tend to design each segment of low speed damping to do something different. I use low speed bump for traction purposes - it's about keeping the tyre in contact with the road over small undulations. I specifically don't use low speed bump damping to control handling, because if you set it stiff enough to control the quarter-tonne of corner weight coming down onto it, it's way too stiff to control the contact patch loads so you lose out horribly in traction. I've driven cars over the years that are so stiff in bump damping they can break traction crossing motorway expansion joints (which in a car with <100 bhp is quite something). Instead I use low speed rebound damping to control handling by controlling the weight at which the suspension extends on the opposite side of the car. High speed bump is set to ensure that most of the time the suspension just stays out of the bump stops and high speed rebound is set to let the wheels come down smoothly after a bump without being so stiff as to jack the whole of the car downwards over a succession of bumps.

Sound a bit like voodoo and other dark arts? In reality, the initial stab at damping rates is based on percentages of critical damping. This critical damping rate is a function of weight and spring stiffness and is in theory the maximum damping that can be applied before it starts to hinder the natural motion. Anything less than critical damping and you get some oscillation of the spring after a bump and with more than critical damping you get a situation where you're artificially slowing the movement down, which can be useful if you've got limited wheel travel but wrecks both ride and handling because the movement is slower than you'd want it. I normally further distil the four sections into two main values:

%age of critical damping (and 60-70% is a good starting figure) and ratio bump:rebound (I've seen ratios from 1:1 all the way up to 1.5 before).
Many of the aftermarket dampers I try tend to go too agressive in the latter metric - by the time you've got bump set, you've got too much rebound and if you get bump right you've got too little rebound. Annoying, and no amount of playing with the adjuster is going to get things 'right'.

Anyway, once you've picked damping rates and presumably cajoled a manufacturer into building them to those specs, it's off for development driving. Just as we did earlier, it's a case of a lot of mileage on various roads and test tracks. I combine objective measurements - datalogging what the car is doing in relation to control inputs - with subjective tests - is my (allegedly) calibrated behind telling my nice things about the new setup or not. I'm not a racing driver by any stretch of the imagination (IME they don't make great development drivers because they simply drive around issues in the handling that lesser mortals will find either disconcerting or just plain dangerous) so I'll often get a few trusted individuals to drive it - in the case of PureDrive I chucked the keys to Don Palmer and let him loose on Millbrook's Hill Route. Based on those numbers it's then a case of tweaking the damper settings until I get the best compromise.
 

·
Registered
Joined
·
1,483 Posts
Bloody good read that, the puredrive shocks, if i did`nt want to use mk2 top mounts would i have any chance of using tein pillow ball on the front an fm on the rear or would i be wasting my time, i`m on track at rockingham with the tein ha`s but i`m thinking of bringing the puredrive with me.
 

·
Registered
Joined
·
407 Posts
Interesting but theoretical.
I'd be interested in the following.

1) Who is going to manufacture them ?
2) Who is going to service them ?
3) How many valving solutions are available ?

Dave, I'm presuming that if you are looking for that sort of hip curve and performance you must be thinking of a monotube variation and that straight away creates problems with a single adjuster because of the lack of bump adjustment on mono's, our new Lotus Elise and TVR offering will have to have at least 3 or 4 valving solutions and that means heaps of road and track testing and quality monotubes with decent hard anodising don't come cheap.

The only solution to cost is to go the Far East route like so many of the current offerings and that route leaves you with valving that doesn't suit all and buying a product and then stripping to revalve puts the price back out of reach of 95% of owners.

I'm not even going to consider making 1 or 2 way monotube shocks for the MX-5 market because I don't believe it exists in sufficiant numbers to warrant the time and effort, I'd rather spend more time playing with blow off valves on the more adjustable twin tube option to get better monotube type response.(and I'm testing that Wednesday)
 

·
Registered
Joined
·
289 Posts
Discussion Starter · #19 ·
The only solution to cost is to go the Far East route like so many of the current offerings and that route leaves you with valving that doesn't suit all and buying a product and then stripping to revalve puts the price back out of reach of 95% of owners.

I'm not even going to consider making 1 or 2 way monotube shocks for the MX-5 market because I don't believe it exists in sufficiant numbers to warrant the time and effort, I'd rather spend more time playing with blow off valves on the more adjustable twin tube option to get better monotube type response.(and I'm testing that Wednesday)
It isn't the only solution... DIY if you don't have the overheads of a large factory can make things affordable. Like I said, the UK is still a very good manufacturing base and price is getting more competitive thanks to the government!

As far as monotube and twin-tube are concerned I'm in no way a zealot for one over the other. I've gone down the monotube route for a range of reasons, mainly to do with friction, hysteresis and the ease of modularity - i.e. I have effectively a single design that can be scaled to pretty much any application and be expanded to 1,2 and 4-way operation.
 

·
Registered
Joined
·
1,047 Posts
measure an na and an nb - Mazda fixed the bump steer at the front on the nb. there are also two sorts of track rod end for the na, some limited editions used different ends to correct steering geometry, the ones that are banned in spec miata...
 
1 - 20 of 52 Posts
Top