Is a rigid chassis needed w/ actively adjustable suspension?

[Tommy Cookers]

Isn't the trump card of Active the capability to detect and treat a breakaway quicker & better than any human driver, like traction control but more so ?

Maybe you have referred to this ?

[DaveW]

Actually, the trump card(s) were the irreversible actuators (which moved only on command), I think. This delivered control over the suspension to the ingenuity (or otherwise) of the engineers, with relatively few compromises.

A number of "driver aids" were coded. One reduced front ride height proportional to steering rate & angle, to increase front downforce during entry to a corner. Another was to reduce the roll moment carrying by the rear axle during power down (actually this used longitudinal acceleration as a measure, because throttle position was unmeasured). A third turned steering wheel position into a trajectory demand (using ground speed, lateral acceleration & yaw rate). We felt the last was a brilliant addition (you just turned the steering wheel & let the system sort out the trajectory) but, as I said earlier, Senna hated it. I think it (know it be) true that Senna had no problems controlling anything we threw at him, he just didn't like the feeling that something else was in control.

Incidentally, we did try ride height control, but we felt that the devices were not without issues & (in any case) they tended to couple with the suspension algorithms. One of the Lotus engineers also suggested that we "memorized" a lap, but that was never implemented for a variety of reasons.

[hardingfv32]

One of the Lotus engineers also suggested that we "memorized" a lap, but that was never implemented for a variety of reasons.

Do you have any knowledge of lap 'memorizing' every being used in F1?

Can you expand on some of the implementation issues?

Brian

[DaveW]

Do you have any knowledge of lap 'memorizing' every being used in F1?
Can you expand on some of the implementation issues?

No, but it would not surprise me.

The objective (in our case) would have been to identify the lowest ride that could be run around the circuit without running out of suspension travel, or grounding. To be successful:

We had to have a reliable measure of front & rear ride height (we didn't),
We had to be able to store & update time histories of target suspension position (to make it easy, we needed the capacity to store two target positions every millisecond).
We needed a reliable "start of lap" signal (not completely reliable at the time).
We needed a reliable way of estimating distance travelled in real time (forward & backward optimization not permitted).
The routine had to converge on a solution in one lap (qualifying).
Ride heights had to be updated to account for tyre pressure/temperature changes.
Ride heights were only really important though corners, and the trajectory through each corner had to be repeatable (arguably).

I can't think of any more at present, but that should be enough, I think.

[Tommy Cookers]

A number of "driver aids" were coded.
A third turned steering wheel position into a trajectory demand (using ground speed, lateral acceleration & yaw rate). We felt the last was a brilliant addition (you just turned the steering wheel & let the system sort out the trajectory) but, as I said earlier, Senna hated it. .

This sounds most interesting, please could we have more on this feature ?

[DaveW]

This sounds most interesting, please could we have more on this feature ?

The algorithm, which was devised by Peter Wright & developed by Steve Green, was based on the idea that warp load (Woff) controlled the lateral balance of the vehicle. In the USA the phenomenon is known as "weight jacking", I believe.

The algorithm, which included some bells & whistles, used r/V (yaw rate per unit velocity) as measure of track and Tan(steer angle) as a measure of track demand. Then,

Woff was controlled as a function of ny*r*V[tan(beta) - K*r/V].

ny was lateral acceleration, and K was a geometery function. tan(beta) was measured steer angle. Positive Woff moves balance towards oversteer.

The algorithm can be observed in action in the video I posted a while ago - here it is again. Note the trajectory of the active car during the slalom clips. The initial oversteer gets the trajectory estabished early, and the later move towards understeer stabilises the turn. A similar effect can be achieved passively by the intelligent use of damping...

[Tommy Cookers]

Thanks again !


Jacking/corner weighting (fixed) is a 'cheap and dirty' way of getting a balance, eg for a road car on a trackday where you can balance for righthand bends and manage the (few) lefthand bends . Just fine for ovals though.
It will always work better with a (relatively) stiff chassis, if only for practical reasons.

Driver adjustable (anti-roll/sway bar) balance was brilliant, but is now banned in F1? (in-race balance now done by tyre pressures). Needs a stiff chassis.

Dynamic jacking via Active is the answer to (almost) everything handlingwise, the car optimally rebalanced for best roadholding in each phase of each corner (F1 does this partly and indirectly via real-time driver selection of diff settings).

Such a DJ system (if fast enough and intelligent enough) has the potential to contain an incipient breakaway ie drive better than a human (road driving and F1) ?
There's always a breakaway threshold, even if its criteria are now determined by tyre preservation ?
A stiff chassis would help a system working at this level of performance

[WilO]

Some interesting reading can be found here: http://papers.sae.org/930266
http://papers.sae.org/890081
http://papers.sae.org/880799

@Dave,
Thanks for expanding on the concept of using steering wheel position into trajectory demand. Great stuff. One of the papers I referenced above discusses the use of active suspension and active rear steer (I believe you told me about this particular car at one point). Makes for some great reading.

[DaveW]

Some interesting reading can be found here:

Thank's Wil.

I have to reveal one of my weaknesses by declaring that, as a matter of honour, I refuse to top up the coffers of the SAE by buying papers that might (or might not be - hard to tell from an abstract) of interest, but were in any case prepared and presented at little or no cost to the SAE. It is true that the SAE has costs but these are relatively minor, and without its publications the SAE would not be the organisation it is. So.. I will have to guess.

I suppose that the rear steer vehicle would be the 205 16T; incredibly stiff & quite the most rewarding car we modified (until it was spoiled by installing a racing clutch). A case for stiff structures.

On the other hand, the first customer car we modified was a relatively soft front drive 1983 Chevy Park Lane. That also had its moments - one I recall was when it was driven around a handling loop in Pheonix and a Corvette attempting to follow fell of the road (to be fair, the handling loop had ripples in the corners, which the Corvette wasn't happy with). So perhaps the case for a stiff structure is over-stated. The only compliance problem I was conscious of was to find good hard points to mount transducers.

Edit: My memory returns (kind of). Richard worked with a Lotus dynamics research vehicle, called SIDD. My comment on the Peug remains valid, I think.

[WilO]

This sounds most interesting, please could we have more on this feature ?

The algorithm, which was devised by Peter Wright & developed by Steve Green, was based on the idea that warp load (Woff) controlled the lateral balance of the vehicle. In the USA the phenomenon is known as "weight jacking", I believe.

The algorithm, which included some bells & whistles, used r/V (yaw rate per unit velocity) as measure of track and Tan(steer angle) as a measure of track demand. Then,

Woff was controlled as a function of ny*r*V[tan(beta) - K*r/V].

ny was lateral acceleration, and K was a geometery function. tan(beta) was measured steer angle. Positive Woff moves balance towards oversteer.

The algorithm can be observed in action in the video I posted a while ago - here it is again. Note the trajectory of the active car during the slalom clips. The initial oversteer gets the trajectory estabished early, and the later move towards understeer stabilises the turn. A similar effect can be achieved passively by the intelligent use of damping...

Dave,

When you have a moment, could you expand on the use of damping to achieve an effect similar to that of the demand/trajectory system? I have a theory, (linear rear damping, digressive in bump at the front?), but your post included the word 'intelligent' in it, which excludes my participation, I'm afraid.
Thanks in advance.

Wil

[Smokes]

Can you run active suspension from a smart phone and how much would cost to kit a track day car with such a system in mo
dern times?

[DaveW]

When you have a moment, could you expand on the use of damping to achieve an effect similar to that of the demand/trajectory system? I have a theory, (linear rear damping, digressive in bump at the front?), but your post included the word 'intelligent' in it, which excludes my participation, I'm afraid.

I don't think that is the case, but I don't want to discuss details as yet because I need to validate the procedure (& offer my customers an improved rig testing service).

I have been aware of the possibility for some time, but it has become apparent to me only recently that I do tend to push set-ups in that direction (there's intelligence for you). More importantly, perhaps, it would appear to be something I can now quantify. Inevitably, this happened whilst I was developing an analysis procedure for something else.

[DaveW]

Can you run active suspension from a smart phone ...?

There's a thought....

Actually, it's not that far fetched. The first F1 system was designed before DSP (Digital Signal Processors). The controller was a fixed structure analogue computer (remember those?) using multiplying DAC's in place of potentiometers. The DAC's were updated sequentially use a low power single chip processor - I'm sure that a smart phone could be modified to do that. Later systems used DSP's for the complete control activity. They represented a major increase in computer power.

I'm not sure about costs, but am tempted to say that if you have to ask then you probably can't afford it. It is worth noting that Moog invested heavily in developing production intent hardware with a target production cost of less than $5K. They were within a year, perhaps, of succeeding before the project was canceled for various reasons.

[Smokes]

I was think more allong the line of a home brew kit, buying off the shelf parts. Or finding citroen xantia with the active suspension. ts would be a cool project for someone to do and test in a unlimited hillclimb class.

back to complance how easy is it to control compliance on a road bike any material ( I mean one with pedals not a engine) and would there every be a shaker rig designed for this purpose?

[DaveW]

Oops, I misunderstood.

It is a long time since I looked at a Citroen system. The actuators should be able to take the pressure (designed for 160 bar, or thereabouts), but would have thought they were too long (Zantia fronts are McPherson struts, but I'm not sure about the rears). The expensive parts in a Lotus system were, I suppose, the servo-valves, certain transducers (load cells for example), & the controller (particularity I/O). I believe that hill climb cars are fairly light, so weight (hence cost) would be an issue, I guess. Would be fun to try, though (not that I'm offering).