Bose Suspension

[Tommy Cookers]

apart from the mechanics a linear motor really isn't different from a rotating motor, servo motors run much in the same
low speed high torque conditions as a linear motor would in a suspension
but I think for a suspension you would try to incorporate something in the mechanics to take up some of the static forces

rotating motors are typically 95-99% efficient overall, because they have efficient magnetic circuits
this is due to the relatively tiny air gap inherently realisable with the rotational form of motor construction
impossible in most linear motor applications ie those without precision and rigidity in their guidance (trains included)
loudspeaker or solenoidal technology is around 25-50% efficient, the air gap being so large

no motor that can run at a significant speed will be efficient at low speed/high torque, servo motors are no different
force/torque is related to current, high current is expensive and inefficient (relatively)
electric linear motion generally uses rotational motors with efficient mechanical reduction, ballscrew technology as was in car steering

[Tim.Wright]

A different system, but with a similar effect is already on Mercedes road cars. They call it active body control.

I call this the Mumford solution, largely because he was responsible for prototyping it on a Jaguar in the late 1970's, I recall, although a quick search failed to reveal a reference to the fact. In control system terms, it is similar in principle to the Citroen system, or indeed the system developed by Williams F1. It is quite difficult to stop the controlled part fighting the passive part if displacement is controlled actively. Limiting the bandwidth of the controlled part is key, I think ....

Yea, I agree about the bandwidth part. To me its pretty clear they are only working to control low frequency body movements because if the system was capable of working at high frequencies, then it wouldn't necessarily need the spring/damper in series.

I'd also guess that the increase in comfort/isoation is mainly due to them being able to run softer springs and then compensating the extra roll and pitch with the hydraulic system. Not so much from the hydraulic system itself.

[Tim.Wright]

apart from the mechanics a linear motor really isn't different from a rotating motor, servo motors run much in the same
low speed high torque conditions as a linear motor would in a suspension
but I think for a suspension you would try to incorporate something in the mechanics to take up some of the static forces

rotating motors are typically 95-99% efficient overall, because they have efficient magnetic circuits
this is due to the relatively tiny air gap inherently realisable with the rotational form of motor construction
impossible in most linear motor applications ie those without precision and rigidity in their guidance (trains included)
loudspeaker or solenoidal technology is around 25-50% efficient, the air gap being so large

no motor that can run at a significant speed will be efficient at low speed/high torque, servo motors are no different
force/torque is related to current, high current is expensive and inefficient (relatively)
electric linear motion generally uses rotational motors with efficient mechanical reduction, ballscrew technology as was in car steering

I see the voice coil style linear acutator and the ball screw style linear actuator as two opposites of a spectrum in terms of how the system forces are reacted. In the voice coil (electrical linear actuator) system, pretty much all of the force is reacted by the EMF occuring between the coil and magnets. In the ball screw system, most of the force is reacted by the ball screw threads, with little passed through to the electric motor.

In terms of them working as a spring damper I see:
The voice coil solution would be the best at absorbing random force inputs but worse at precisely controlling the displacement
The ball screw solution would be the worst at absorbing random force inputs, but would be the best at precisely controlling the displacement

An interesting compromise could be to use a high speed motor and a very long lead ball screw such that maybe 50% of the force is reacted by the motor and 50% through the balscrew thread. You would also have the effect of an inertia damper coming from the rotor which might also help the isolation.

In any case, I think its worth noting that while all of these systems seems to be aimed at ride comfort, they could potentially have performance applications to since they allow;
1. Reduction of the contact patch load variation
2. Lower static ride height, and therefore lower CG
3. Control of the body in a much more ideal way to help the aerodynamics

[DaveW]

I have also seen a company offering a dampener design that uses a linear electrical absorber, with the electrical power produced used to recharge the battery.

I need to be careful how I answer this... The idea was greeted with concerted opposition from the technical establishment. However, it is difficult to be too critical because it so often happens that good things pop out from pursuing crazy ideas. As Chapman used to say, if you are in a queue of traffic, then you are travelling in the wrong direction....

[Tim.Wright]

There was a discussion on regenerative damper somewhere on here recently. Dave, would you have any estimates as the the power disappated by a damper from a rig test? My simple calcs puts it down as less than 10W on average over a lap for a GT car. It seems a bit low, but if its true, it would be completely useless and not worth the extra weight.

[DaveW]

An interesting compromise could be to use a high speed motor and a very long lead ball screw such that maybe 50% of the force is reacted by the motor and 50% through the balscrew thread. You would also have the effect of an inertia damper coming from the rotor which might also help the isolation.

I like the suggestion, although I'm not sure it would help isolation. But (without doing the maths) I guess it could well function as an inerter (not quite the same thing as an inertia damper, I think).

[DaveW]

Dave, would you have any estimates as the the power disappated by a damper from a rig test?

The nature of the test was fairly unrepresentative of a real track data, but the F3K example I used elsewhere consumed a peak of 131 (front) & 266 (rear), with averages over a test run of 37/55 Watts, respectively. All numbers were per axle.

Overall power consumed varied with set up, of course. For the test I chose, peak power varied between 453 & 809 Watts. It takes more fuel to drive a poor set-up around a track, but the "best" set-up is not minimum power.

[Tim.Wright]

Well the degree of isolation would depend on the controller. I.e. whether you give a lot of reaction force when you see a displacement change or a little.

To me an inerter is just a type of inertial damper. Perhaps the language used in the industry around them treats them differently.

[DaveW]

To me an inerter is just a type of inertial damper. Perhaps the language used in the industry around them treats them differently.

I think the terminology is quite confusing. For me, a damper is a device designed to absorb energy. A properly set-up "mass damper" does just that, but an inerter is intended to store, and then release, energy (rather like a spring, I suppose).

[Tommy Cookers]

...... In the ball screw system, most of the force is reacted by the ball screw threads, with little passed through to the electric motor.
In terms of them working as a spring damper I see:
The voice coil solution would be the best at absorbing random force inputs but worse at precisely controlling the displacement
The ball screw solution would be the worst at absorbing random force inputs, but would be the best at precisely controlling the displacement
An interesting compromise could be to use a high speed motor and a very long lead ball screw such that maybe 50% of the force is reacted by the motor and 50% through the balscrew thread. You would also have the effect of an inertia damper coming from the rotor which might also help the isolation.
...... could potentially have performance applications to since they allow;
1. Reduction of the contact patch load variation
2. Lower static ride height, and therefore lower CG
3. Control of the body in a much more ideal way to help the aerodynamics

sorry if this sounds a bit negative but .......
the benefit of the ballscrew is its (efficient) mechanical reduction such that most of the reaction work is done by the motor
the motor torque being amplified by the reduction, the current will be small for the required force
(BTW a planetary roller screw like the Rollvis could well be the best for this job)

in bandwidth potential the right kind of electromechanical actuation will be (a close) second only to servo valved hydraulics
(bandwidth won't be better without reduction, inductance is much higher as the electrical machine is bigger)
anyway, the dominant issue for high bandwidth (race) active ride suspension is the control intelligence, or insufficiency thereof
which makes academic the bandwidth inherent in the systems configuration and components ?

if current purposes are served with less bandwidth then electromechanical route is easy, energy efficient and inexpensive
inertial damping seems naive compared with EMs possibities via suitable controlled variations in motor excitation
naturally flowing from the '4 quadrant' motor operation that is the heart of a modern electric servo system
(whether the nominal controlled variable is eg force-controlled, position-controlled)
4 quadrant drive has the 'motor' motoring and generating, energy is capacitavely stored and reused from within the drive unit

[DaveW]

... the dominant issue for high bandwidth (race) active ride suspension is the control intelligence, or insufficiency thereof which makes academic the bandwidth inherent in the systems configuration and components?

Fully digital control systems have been available for active suspension since 1983, to my knowledge. These were capable of controlling four hydraulic actuators with a transport delay of <150 microsecs, control the hydraulic power supply and generate their own data stream with an overall iteration rate of 1 millisecond. By 1987, they also ran monitoring algorithms and could survive a single transducer failure. The most complex vehicle system we built controlled active suspension, active rear steer, active front steer, active steering wheel "feel", active throttle & brakes: 10 EHSV's in all controlled by a single computer iterating at 1 millisecond.

Today, there should be no issues with controlling a fully active suspension, though incorporating it into the existing MES might be a stretch, I suppose.

p.s. I like the idea of an electrically driven suspension. Two issues to be addressed are actuator life & failure mode.

[Tim.Wright]

sorry if this sounds a bit negative but .......
the benefit of the ballscrew is its (efficient) mechanical reduction such that most of the reaction work is done by the motor
the motor torque being amplified by the reduction, the current will be small for the required force
(BTW a planetary roller screw like the Rollvis could well be the best for this job)

I'm not talking of reaction "work" but reaction "force". If you consider a 20mm ballscrew with a lead of 4mm per rev, you have a helix angle of 3.5deg. Therefore, with any axial force on the ballscrew, only sin(3.5deg) = 6% of the axial force actually is reacted by the motor torque, with the rest reacted by the threads. During high frequency inputs, unless the motor is moving at a ridiculous rate, the ball screw is effectively a solid link.

[Greg Locock]

I like the idea of an electrically driven suspension. Two issues to be addressed are actuator life & failure mode.

Right. It is curious to watch the mind of electronic engineers and the like as they consider the safety implications of Electric Power Assisted Steering Systems on passenger cars. eg

http://www.linkedin.com/groupAnswers?vi ... _140208214

The downside of failure of an electric suspension on a race car is probably less than in a production car,admittedly.

[Tommy Cookers]

sorry if this sounds a bit negative but .......
the benefit of the ballscrew is its (efficient) mechanical reduction such that most of the reaction work is done by the motor
the motor torque being amplified by the reduction, the current will be small for the required force
(BTW a planetary roller screw like the Rollvis could well be the best for this job)

I'm not talking of reaction "work" but reaction "force". If you consider a 20mm ballscrew with a lead of 4mm per rev, you have a helix angle of 3.5deg. Therefore, with any axial force on the ballscrew, only sin(3.5deg) = 6% of the axial force actually is reacted by the motor torque, with the rest reacted by the threads. During high frequency inputs, unless the motor is moving at a ridiculous rate, the ball screw is effectively a solid link.

what I had in mind was a family of systems that had relatively very much higher lead screws (that I now think were made to order)
that is much less reduction than above, potentially with a better bandwidth but greater cost etc for the electrics
more readily backdriveable (not as readily as I claimed) ie less mechanically damped
and similar to your suggestion on 25th of a high lead ballscrew eg giving a 50/50 split (which I had missed)

the low lead system will have more mechanical 'damping', both frictional and inertial
(BTW at 4mm lead a (planetary) roller screw would for the same load capacity have far smaller diameter and inertia)
this is potentially disadvantageous ie either may at times promote instability
inertial effects will only be damping if they are tuned ie act directionally against motion (as in 2Cv and F1 Renault)

if we have chosen a suitable lead .....
the actuator 'motors' will at every moment have an excitation voltage (eg proportionate to 'error' of the controlled variable)
so (within the nominal bandwidth) there always be an actuator force appropriate in size and direction ie the system is stable
above that bandwidth the system response will decay stably and the actuator would become less and less compliant
(the motor will resist being backdriven ie act as a generator)
but then the actuator will have its compliance varied by pulsed 'open circuiting' of the motor to modulate its resisting load
this is in effect continuously controllable damping and springing

the big question is ..... what is the necessary or desirable bandwidth ??

[Tim.Wright]

The 4mm lead was only an example to prove that most of the work is done by the ballscrew not the motor in a typical motor + ballscrew applications.

What you then described is more along the lines of what I was thinking. With a long lead, the motor and the ball screw have more of an equal share of the load.

My feeling is that bandwidth will be limited by the mechanical inertia of the system, not the electrics. As Dave pointed out, they are already very fast.