The future of car suspension is here: DSC Sport’s active shocks in action

It constantly samples data from the car's network, adapting the damping at 100Hz

Read the whole story

 

[FreeFire]

Penforhire[/url]":3hx04p0i]I love this stuff.

Curious about the choice of CAN bus though. I'm no electrical engineer but I've read CAN uses data packets, not a continuous stream. That 1 MBPS (CAN max data rate) has to be considered much slower than you'd expect from 1 MBPS. If I understand what I've read correctly, we're talking around 1.2 milliseconds minimum time between packets of 8 bytes (11 bits, or even more, consumed with header info), actual rate dependent on data compression - how many consecutive zeros or ones in the data).

1.2 ms sounds plenty fast since this suspension system is responding in 6 ms intervals. But it isn't enough for everything in a car. I happen to know manufacturers often want data from steering sensors in 1 ms intervals. That want that frequency of data to detect errors reliably, to shut a faulting system down for safety.

I am an electrical engineer, and while I'm not that familiar with the CAN bus, I can tell you that at 1MHz, 11 bits takes 11 microseconds, not 1.2 milliseconds. That's around a hundred times faster, and would leave plenty of headroom for other activities, depending on CAN's other overhead.

 

[metalsheep]

Errum[/url]":208elfom]

metalsheep[/url]":208elfom]

Janeazy[/url]":208elfom]I'm still curious to see/feel/drive VW's adaptive chassis control (http://en.volkswagen.com/en/innovation- ... y/dcc.html). Probably not as advanced as this one, but still sounds like it really changes the car's behavior when turned on

I tried their DCC system in the process of buying a new GTI, and was actually pretty underwhelmed. I preferred the traditional (and incidentally $1000 cheaper) suspension. I felt the car had too much nervous body lean when you first start steering (the feeling of soft anti roll bars). It firmed up quickly in the corner but it didn't inspire as much confidence as regular suspension.
It's probably faster on a track, but to me it didn't leave me feeling as comfortable in the car because I couldn't predict how it would act as well.

If so — I haven't driven a GTI with the DCC — then it's a function of how VW chose to tune that car's suspension. I own a 2015 Golf R with DCC, and it's notable for the minimal roll, dive or squat. Another feature of DCC is the ability for the driver to switch between comfort, normal and performance suspension settings. The differences between these are quite noticeable.

We were also on a very poorly paved road, so I think it might've been something of an edge case where there really wasn't a correct answer. And once you actually got in the turn everything was very tight and well sorted, it was just right after the initial control input that it felt wonky. Perhaps the R samples at a higher rate.
Also comfort mode was very nice but I'm not the sort of person who would use it.

 

[Cheesewhiz]

Jimmy McNulty[/url]":2g7bqirs]How is this better than MagneRide which use a ferro-magnetic fluid and electro-magnets instead of hydraulic pumps? MagneRide is 4th gen and is available from the factory in a variety of cars from chevy, audi, ferrari and land rover.

Agreed. I have not actually owned or driven a car with GM's magnaride but from all accounts it is quite excellent. So good that other car manufacturers have licensed it from GM (like Audi).

I'm no expert, but from what I have gathered, magnaride is an elegant solution. As in, far less complex as compared to others. No reservoirs or pumps. It sends electric current to alter the viscosity of the shock fluid. It is simpler and response is fast.

And while this DSC system might be an advance and/or quite good, I'm failing to see the radical advance. Active suspensions, as other have pointed out, have been around for decades. Citroen was using a hydraulic self-leveling suspension in the 1950's.

The issue has always been cost and complexity. Today, many cars (mostly luxury/sport) offer an active suspension of some sort as an option. The issue is that it is often an expensive option on an already expensive vehicle, often with only a marginal benefit (magnaride excepted, I'd consider it a must have option). Typically, you have to get into a top of the line sport model, like an AMG, M or RS to get active standard, and maybe not even then.

 

[jdecastro]

This has been going on for a while in most European cars. I don't understand's American car manufacturer's love for leaf springs (but reasonable in crummy pickup trucks).

 

[foxyshadis]

That car is a Porsche Boxster Spyder, a lightweight (2,900lb/1,315kg) special edition that saw out the last-generation Boxster).

Not quite sure what's going on with this sentence, Jonathan. Looks like a bad cut/paste.

Anyway, for everyone pissing all over this tech because it's not new, it says right in the article that this is about a refinement in the control systems to make it better at a much lower price than ever before, not a magical new technology. Most of what Ars writes about are refinements, because that's how the state of the art advances, get over it.

 

[Dr Gitlin]

foxyshadis[/url]":2i2kqswu]

That car is a Porsche Boxster Spyder, a lightweight (2,900lb/1,315kg) special edition that saw out the last-generation Boxster).

Not quite sure what's going on with this sentence, Jonathan. Looks like a bad cut/paste.

Anyway, for everyone pissing all over this tech because it's not new, it says right in the article that this is about a refinement in the control systems to make it better at a much lower price than ever before, not a magical new technology. Most of what Ars writes about are refinements, because that's how the state of the art advances, get over it.

Thanks, it's fixed now.

 

[abend s0c0]

Any hard data on the performance advantage? Accelerometer data for example, with and without the active control, same vehicle, same conditions?

Edit: Would also be curious about relative performance on different road surfaces: smooth blacktop, striated, potholed, etc.

 

[BigMattyC]

FreeFire[/url]":1kvuhpnf]

Penforhire[/url]":1kvuhpnf]I love this stuff.

Curious about the choice of CAN bus though. I'm no electrical engineer but I've read CAN uses data packets, not a continuous stream. That 1 MBPS (CAN max data rate) has to be considered much slower than you'd expect from 1 MBPS. If I understand what I've read correctly, we're talking around 1.2 milliseconds minimum time between packets of 8 bytes (11 bits, or even more, consumed with header info), actual rate dependent on data compression - how many consecutive zeros or ones in the data).

1.2 ms sounds plenty fast since this suspension system is responding in 6 ms intervals. But it isn't enough for everything in a car. I happen to know manufacturers often want data from steering sensors in 1 ms intervals. That want that frequency of data to detect errors reliably, to shut a faulting system down for safety.

I am an electrical engineer, and while I'm not that familiar with the CAN bus, I can tell you that at 1MHz, 11 bits takes 11 microseconds, not 1.2 milliseconds. That's around a hundred times faster, and would leave plenty of headroom for other activities, depending on CAN's other overhead.

Several points to make:

1. CAN has an upper limit of 1 Mbps, but no automotive company that I know of runs it faster than 500 kbps due to noise issues. Further, the overhead on an 8 byte packet (CAN can send data packets with lengths of between 0 and 8 bytes) is a little under 100% (off the top of my head 112 bits +/- fuzzy memory for a 64 bit payload). The Bosch spec on CAN is the canonical definition in case anyone wants more information.

2. A CAN bus isn't stable beyond 60-70% load due to the inefficiencies of having a multi-master bus like that. You'll drown in bus errors if you attempt to fully load the bus without some sort of external synchronization. This further reduces the effective data rate in your comment.

3. Not all data need arrive over the CAN bus.

 

[azazel1024]

Solidstate89[/url]":3gdanqrh]McLaren's hydraulic suspension trickery is such that their super cars (P1, MP4-12C, 650S) don't even have anti-sway bars, instead relying entirely on the hydraulic system to not only control the suspension, but the stiffness as well.

That would be sweet. Reduces curb weight as well as unsprung weight, which has a lot of benefits.

I wonder when active suspensions will be much more a de facto suspension?

 

[jbrizz]

superchkn[/url]":6zjixcdk]

araemo[/url]":6zjixcdk]
<snip>... and I believe most modern 'passive' shocks actually use 'variable viscosity' fluids - specifically, shear-thickening fluids, as that is how they react differently to fast bumps than slow bumps.

They actually have different mechanical circuits for high and low speed dampening. Certainly shock oil can effectively behave that way when used in something like an old dampening rod fork in a motorcycle due to its relatively tiny orifice. One of the ways is described on page two in this Penske shock manual:

Spoiler

In the state of low shaft velocities (i.e. corner entry, exit, and power down), oil is
displaced within the damper in direct proportion to the volume of the shaft
entering the body. The displaced fluid passes through the compression adjuster
where it is metered through a fixed, low speed bleed orifice. Due to the small
diameter of this orifice and the viscosity of the damper fluid, a pressure loss
occurs across the orifice. This loss of pressure is a loss of energy in the fluid
due to friction and the subsequent opposing damping force is generated.
As the shaft velocities increase, the same amount of fluid must pass through the
low speed bleed orifice, but at a much higher rate. The viscosity of the fluid
causes a greater resistance to flow at the orifice entrance which in turn
produces a large internal force on the CD housing. The other major internal
components, namely the piston and shim cage, are designed to handle this
extra force by allowing the shims to “blow off” proportionally to the extra force
generated, much like a coil spring compresses proportionally to the axial load
applied. With this arrangement, the low speed bleed orifice still meters fluid
during high speed shaft movements, but the extra forces generated are handled
with the shims which have less resistance to flow at higher velocities. They are
designed to virtually bypass the low speed orifice and form a new fluid circuit.
The force at which this occurs can be varied by turning the compression
adjuster in or out, which preloads the shims. Therefore, as the preload on the
shims increases, the static force required for them to activate is increased as
well. The name designation for the parts also clue one in to their purpose, with
the low speed bleed orifice handling low velocity bleed flows and the piston/
shim arrangement handling high velocity flows. This principle originated in the
main shaft piston/shim arrangement and follows similar behavior.

Bilstein, Koni and many other upper end car shocks work this way, as do many bikes forks and rear shocks. Without a variable oil flow they just hydraulic lock out on fast impacts, not much good for anything remotely high performance.

 

[Abhi Beckert]

Pyros[/url]":950buuvv]Tapping into the existing CAN bus to do this seems like an unnecessarily complication. Just put an accelerometer on it to detect braking. Should probably have that anyway to figure out what is going on with roll, etc.

If you use an accelerometer the suspension won't be adjusted until after the car has started to dive, and by then it's too late to stop the weight transfer.

By using CAN bus you can harden up the suspension the instant the driver steps on the brake/accelerator and keep the car perfectly level.

I imagine he's doing tricky stuff too, like for straight line acceleration it probably softens up the rear suspension to get as much weight onto the rear wheels as possible, but corner-exit acceleration you want lots of weight on the front wheels and stiff rear suspension.

I wonder if this would make the car hard to drive. The brake/accelerator are usually used mid corner to deliberately induce or recover from over/understeer - but this system would harm the driver's ability to do so.

I suppose with experience you'd get used to it.

 

[longhornchris04]

Penforhire[/url]":e5yxsbuo]
Curious about the choice of CAN bus though.

The reason for using CAN is to get access to all the data that is going to/from the ECU. On modern cars everything is on the CAN bus. This means the suspension controller (potentially) has access to all the data coming from the stability & traction control sensors as well as the driver inputs. This allows for some level of prediction (like someone else mentioned) stiffen the suspension when the driver hits the brakes rather than waiting for the nose to dive.

Is CAN the best system... maybe not. It is, however, the current industry standard and it was designed to drive an network of semi-independent controllers. I used it on the experimental rig for my master's thesis to drive a series of motor controllers and it works well when properly implemented.

Also, 100Hz is probably overkill for driving a valve given the viscosity of your typical hydraulic fluid. The same is true of magnetohydrorehologic (can't spell) fluids - the viscosity isn't going to change that quickly. At that rate and higher, you really need to start looking at voice-coil based system that would be 'fully active.' Said systems have been demonstrated, but are both expensive and energy hungry.

 

[NovoRei]

Dr Gitlin[/url]":2fdfvhpt]
Magnetorheological fluid can suffer from hysteresis (lag, basically). Levitas also told me that those systems aren't as fast to reach as tractive's conventional valved dampers.

If Levitas system is truly superior it's not by the assertions he made.

Hysteresis is energy dissipation within the fluid.
Lag would be the case if the fluid was compressible. It's not.

1.
Any damper suffers from hysteresis as long as it uses oil. A MagnaRide system and a regular system both use oil, both will have hysteresis. What Levitas said is a moot crafted statement.

2.
Not only the MagnaRide controls the damper at 1000Hz (as opposed to 100Hz Levitas' system), the fluid hysteric model is added to the control module, the fluid itself reacts faster than 5ms (it's actually sub-ms) according to the fluid manufacturer, and the system response time of newest MRF dampers are under 5ms (as opposed to 6ms for only the damper of Levitas' system)
http://www.lord.com/emea/products-and-s ... ical-fluid

3.
The MagnaRide may have an individual controller for each damper and it can be custom tuned by the OEM. The DSC only has one controller.

The only advantage his system may have is custom tuned semi-active suspension at a reasonable price for vehicles that don't have such a thing.

 

[OmegaXII]

tetrapyloctomy[/url]":3foibcq6]But can it jump over obstacles in its path?

The only thing that went through my mind before i watched the video, was the mach 5's auto-jacks from speed racer.

 

[Kommet]

Minor nit, but 100Hz is not equal to "every 6 milliseconds"; 100Hz means 100 cycles per second and thus one cycle per 10 milliseconds.

The article should either read "controls the shocks at 100Hz—every 10 milliseconds" or "controls the shocks at 167Hz—every 6 milliseconds".

 

[citpeks]

So, strictly speaking, DSC's system (as well as the others) is still semi-active, since it has passive springs.

The last time active suspension was promised as the future, those systems had full hydraulic actuators at each corner, requiring a power unit to drive them (which was part of their downfall).

Who remembers the pictures of those cars leaning into corners?

Smart dampers are still only part of the equation.

 

[mjb86]

Common semi-active suspension control algorithms (Skyhook and Groundhook) respond to suspension jounce and rebound. Groundhook might be the more interesting strategy (especially for road holding) as another input is the profile of the road. I've never come across a manufacturer stating they use Groundhook (Skyhook is more common, here is a good article on Ducati and Aprilia bikes that use it). I've wondered if production implementations of Groundhook, assuming they exist, estimate the road profile with a linear observer.

Surprisingly, the goddess of active suspensions might just be French

 

[blacknoise]

Abhi Beckert[/url]":10b6qj4n]

Pyros[/url]":10b6qj4n]Tapping into the existing CAN bus to do this seems like an unnecessarily complication. Just put an accelerometer on it to detect braking. Should probably have that anyway to figure out what is going on with roll, etc.

If you use an accelerometer the suspension won't be adjusted until after the car has started to dive, and by then it's too late to stop the weight transfer.

By using CAN bus you can harden up the suspension the instant the driver steps on the brake/accelerator and keep the car perfectly level.

I imagine he's doing tricky stuff too, like for straight line acceleration it probably softens up the rear suspension to get as much weight onto the rear wheels as possible, but corner-exit acceleration you want lots of weight on the front wheels and stiff rear suspension.

I wonder if this would make the car hard to drive. The brake/accelerator are usually used mid corner to deliberately induce or recover from over/understeer - but this system would harm the driver's ability to do so.

I suppose with experience you'd get used to it.

My concern with the computerization of cars is that the driver involvement is more and more removed from the car. This sort of technology is awesome for a comfortable family car, or a cruising vehicle, but I think it can be detrimental for the "fun" of driving.

Automated manual transmissions, active suspension, electronic diffs, AWD with torque vectoring etc. all make a car faster, but rob the driver of part of the experience. You can get used to it, but you're still not really controlling the weight transfer of the car with your feet and hands, you're not matching the revs with a heel-and-toe as you brake into the corner, not smoothly dialing in the throttle as you leave the apex... it's the computer doing all of that partially or completely for you.

Of course, for 99% of drivers this isn't really an issue; my experience is most people with performance cars never track them or ever push anywhere near the edges of the performance envelope. And of course the vast majority of car sales are for family/utility vehicles - for which comfort and safety are paramount. There is also the issue that high performance cars now are so ludicriously powerful that it would be extremely difficult for most drivers to control them without the aids (or impossible in the case of some of the crazy high power front wheel drives out these days).

That aside, the technology is pretty cool, and I'm enjoying computers finally getting a bigger role in cars.

 

[eggie]

Normally, a car's weight transfers from front to back as it brakes and accelerates, and from side to side as it corners. As the force on each corner changes, so does the grip available to that wheel; less weight equals less grip.

I hate these sentences. It’s a complicated subject & non-metric units are partly to blame, but we can do better here at Ars. It’s a technical area! Are we not technicians?

First, the word "weight" never means weight in this quote. "Force" in the 2nd sentence is correct: we’re interested in the normal force pushing the tire onto the pavement. So, let's talk about tire "load" & "load transfer." (Thought experiment: if weight truly transferred, would trailbraking encourage oversteer? There would be less rear mass for the rear tires to accelerate laterally…)

Second, any time a car accelerates in any direction, load transfers among the tires. Heck, it doesn’t even need a suspension for this to occur! So let’s delete “normally” from the 1st sentence, at least until we get the center of gravity down to road height.

Finally, yes, less load on a tire reduces its grip. However, we are discussing racing cars, not unicycles; less load on 1 tire implies more load on another. Under typical loading, a tire’s grip/load ratio is fairly linear. But as a tire approaches its limits at higher loading, the grip increases less quickly than the load. This digressive response explains exactly why this is a limit situation: the outside tires’ grip increases less than the inside tires’ grip drops. IOW, load transfer reduces the vehicle’s total grip in every real-world scenario I can think of.

High-tech suspensions offer advantages, especially in transient conditions on roads with varied corners. (IOW, they’re better suited to Laguna Seca than Indianapolis.) But they don’t effectively change the fundamentals of load transfer & grip.

Apologies to Dr. Gitlin. I’m sure you know this stuff, and it’s damned hard to explain concisely. I certainly can’t do it in 2 sentences. Now, if you’ll excuse me, I'm needed to explain that temperature is not heat…

 

[eggie]

Solidstate89[/url]":1agtmscw]Magnetorheological dampers were invented by GM engineers who worked for Delphi, they're - at the time - wholly in-house owned development house and parts supplier.

And Zora Arkus-Duntov taught Porsche about the anti-roll bar in the '50s!

 

[eggie]

controls the shocks at 100Hz—every 6 milliseconds, which Levitas describes as "near-enough real-time."

Close enough for driver inputs - especially from practiced, smooth drivers who anticipate. Probably not for bumps & the unexpected.

 

[blacknoise]

eggie[/url]":2dni2mgu]

Normally, a car's weight transfers from front to back as it brakes and accelerates, and from side to side as it corners. As the force on each corner changes, so does the grip available to that wheel; less weight equals less grip.

I hate these sentences. It’s a complicated subject & non-metric units are partly to blame, but we can do better here at Ars. It’s a technical area! Are we not technicians?

First, the word "weight" never means weight in this quote. "Force" in the 2nd sentence is correct: we’re interested in the normal force pushing the tire onto the pavement. So, let's talk about tire "load" & "load transfer." (Thought experiment: if weight truly transferred, would trailbraking encourage oversteer? There would be less rear mass for the rear tires to accelerate laterally…)

"Weight" is a concept that laypeople intuitively understand. Weight is already a force, even though laypeople use it as a term when they really mean mass. So, "weight transfer" is almost correct, and the effect of increased or decreased downward force on each tyre/contact patch is immediately grasped without losing any useful information - everyone gets that "weight" acts "downwards".

Calling it force or load transfer may be more accurate, but it can be confusing for people not accustomed to these terms. It could be the difference between your audience learning something, and just letting their eyes glaze over. Sometimes it's more important to get your idea across than make sure all of your terms would be correct if used in a physics textbook. Even automotive engineers use the term "weight transfer".

 

[Dr Gitlin]

eggie[/url]":3p9k57h4]

Normally, a car's weight transfers from front to back as it brakes and accelerates, and from side to side as it corners. As the force on each corner changes, so does the grip available to that wheel; less weight equals less grip.

I hate these sentences. It’s a complicated subject & non-metric units are partly to blame, but we can do better here at Ars. It’s a technical area! Are we not technicians?

First, the word "weight" never means weight in this quote. "Force" in the 2nd sentence is correct: we’re interested in the normal force pushing the tire onto the pavement. So, let's talk about tire "load" & "load transfer." (Thought experiment: if weight truly transferred, would trailbraking encourage oversteer? There would be less rear mass for the rear tires to accelerate laterally…)

Second, any time a car accelerates in any direction, load transfers among the tires. Heck, it doesn’t even need a suspension for this to occur! So let’s delete “normally” from the 1st sentence, at least until we get the center of gravity down to road height.

Finally, yes, less load on a tire reduces its grip. However, we are discussing racing cars, not unicycles; less load on 1 tire implies more load on another. Under typical loading, a tire’s grip/load ratio is fairly linear. But as a tire approaches its limits at higher loading, the grip increases less quickly than the load. This digressive response explains exactly why this is a limit situation: the outside tires’ grip increases less than the inside tires’ grip drops. IOW, load transfer reduces the vehicle’s total grip in every real-world scenario I can think of.

High-tech suspensions offer advantages, especially in transient conditions on roads with varied corners. (IOW, they’re better suited to Laguna Seca than Indianapolis.) But they don’t effectively change the fundamentals of load transfer & grip.

Apologies to Dr. Gitlin. I’m sure you know this stuff, and it’s damned hard to explain concisely. I certainly can’t do it in 2 sentences. Now, if you’ll excuse me, I'm needed to explain that temperature is not heat…

Yes, as you have found, it's hard to do it consisely in a way that readers with a casual interest will grok.

 

[eggie]

blacknoise[/url]":za1v4m7c]...everyone gets that "weight" acts "downwards".

Perhaps, but doesn't this make more difficult to understand why a car's handling balance changes? Any thoughts on my trailbraking comment?

blacknoise[/url]":za1v4m7c]Calling it force or load transfer may be more accurate, but it can be confusing for people not accustomed to these terms. It could be the difference between your audience learning something, and just letting their eyes glaze over.

Do you read any of the more pure-science articles posted here? Ars strikes me as a forum that can handle a bit more rigor.

Dr Gitlin[/url]":za1v4m7c]Yes, as you have found, it's hard to do it consisely in a way that readers with a casual interest will grok.

Hilarious! I had "grok" in an earlier - longer! - version of my rant.

 

[pcfidelity]

How does 100Hz equate to every 6 milliseconds? Surely that would be every 10 milliseconds.

 

[HellDiver]

joetf[/url]":3hhvg5dk]
Yuuup, and Mercedes Benz uses such system in production cars for nearly 2 years now.
It's amazingly funny how US based Ars can pick technology from decade ago and run with it as if it was cutting edge.

You guys are in bloody car stone age compared to Europe

This was an option on the Mk5 Opel/Vauxhall Astra. They came out in 2004. A GM product.

So, 11 years ago, this was a £800 option on a family hatchback.

Tell me why this is news again?

 

[blacknoise]

eggie[/url]":2gcex3o8]

blacknoise[/url]":2gcex3o8]...everyone gets that "weight" acts "downwards".

Perhaps, but doesn't this make more difficult to understand why a car's handling balance changes? Any thoughts on my trailbraking comment?

Yes, it is a sub-optimal term. I guess there's a point where sticking with "weight" for familiarity starts to hurt more than it helps. I'd guess this happens once you get more in depth than "when you brake, load transfers to the front tyres; when you accelerate, the rear". Explaining the dynamics of trail braking or throttle balancing would probably be past this point.

blacknoise[/url]":2gcex3o8]Calling it force or load transfer may be more accurate, but it can be confusing for people not accustomed to these terms. It could be the difference between your audience learning something, and just letting their eyes glaze over.

Do you read any of the more pure-science articles posted here? Ars strikes me as a forum that can handle a bit more rigor.

A good point. I would prefer that we all use the correct terms for these things, all the time, but I've just given up on that battle for the most part. The Ars articles are interesting - some of the science articles are super in depth (past my understanding), but a lot of the car articles are quite simplistic. Upon reflection, I probably took this stance because the car articles make me subconsciously feel that the audience is less informed.

 

[eggie]

blacknoise[/url]":sqrm69gf]The Ars articles are interesting - some of the science articles are super in depth (past my understanding)..

And mine! But I keep trying...