To learn and share about dampers / shock absorbers

[Belatti]

Some years ago I did a model that took into account shims and areas and was not close to reality.
In pages 200 and 201 of the shock absorber handbook (2nd ed) you will see why.

[DaveW]

The rear is locked initially. When it finally starts to move the apparent friction falls, suggested that once the friction is overcome, it takes time for it to reset. The characteristic is reminiscent of a pre-loaded shim valve (which requires time for the fluid boundary layer to be ejected after the valve closes, during which time it still leaks fluid).

Dave, I am Having a hard time wrapping my head around this wondering if you could explain a little more or an example. I haven't heard of this before.

I know its a little off topic.

Not here, however.

First a restatement of the the phenomenon. The plot shown in red was taken from 13.8 seconds into a run, when the input frequency was 1.4Hz. Here the suspension was static over a load range of 600 N. The load was high for a damper (much higher than the front suspension of the same vehicle, and also much greater than the same plots for a standard vehicle at both axles. Over-plotted is a trajectory, taken at 26 seconds into the same run, when the input frequency was 4.2 Hz. Here, the damper has a digressive characteristic, but the damper keeps moving throughout. It would seem that there is a "time" element to the friction force.

I first encountered the phenomenon when using a single corner rig to characterise a range of dampers. One damper (from a Lotus Esprit, I recall) was digressive and heavily rebound biased. Sweeping up through the frequency range, the spring mass was pulled down and then, at a frequency of around 15 Hz, the sprung mass suddenly returned to a ride height that was (close to) its static position. The effect was entirely repeatable, and even did the same thing (in reverse) when the frequency was swept down. It followed, I think, that the damper was able to support a pressure drop across the piston (pulling the sprung mass down), but only up to point. In this case, the point appeared to be a frequency, but more properly, it was time. Below that time, I argued, the valves did not close properly. The feature seemed to be analogous to the "Stribeck effect".

I struggled to find an obvious example, but here is one that is interesting in its own right. It shows trajectories extracted from the same run of a road car application with a Koni damper fitted with an "FSD" widget in rebound only. The term "Frequency Sensitive Damping" is something of a misnomer, because it is time, rather than frequency sensitive. In the present case the rebound trajectory follows a low rate path, gradually stiffening up with time. Thus the rebound trajectory follows a counter-clockwise path in the plot. At low frequency, shown in red, the average damping is high, gradually reducing as the frequency increases, see the green and blue plots. Arguably, the "FSD" widget attempts to emulate the phenomenon under discussion (not completely successfully, because it happens rather slowly). The black trajectory is rather different. The structure of the damping has all but disappeared, leaving a slightly hysteretic liner trajectory. The min, max & mean time histories are shown here. The dampers pulled the vehicle down by around 10 mm at 22 seconds (+ve is compression). At 36 seconds the the mean value suddenly increases (in the blue circle). The last two damper trajectories (blue & black) were taken either side of the hiatus. Arguably, the lack of structure in the black sample suggest that the valves are no longer in control of the trajectory.

The phenomenon is quite commonly found in road car dampers, but is less obvious in race car dampers. Here is another example - all plots are again taken from the same run. The dampers are quite digressive at low frequencies (the red trajectory was taken at a frequency of 2.3 Hz), the digression reduces as the frequency is increased (the green trajectory was taken at 7.7 Hz), and finally all but disappears at higher frequencies (the blue trajectory was taken at 12.7 Hz). I would argue that development drivers spend a lot of time devising stack builds that use the phenomenon to achieve control at low frequencies, whilst washing out the digression at high frequencies in the interests of ride.

[Belatti]

A nice read Dave, thanks.

The analogy you made with the Stribeck effect makes me think about how much of this is taken into account by damper designers when they set the shim-seat contact areas.

Would be nice to test in the rig dampers with the same (or as similar as you can go) force-velocity digressive curve, but with different approach in the shim-seat design.

[Luke]

Sorry to but in on this conversation, but I have been thinking about hysteresis and trying to understand why on TRDs quite often the Compression open forces exceed the compression closed forces unlike on non TRDs.

Often I have seen a reduction in the hysteresis in the low speed region when using a TRD but often in the higher speed range its worse than a non TRD damper.

any ideas?

[WilO]

Higher seal friction (on a TRD)?

[DaveW]

.. I have been thinking about hysteresis and trying to understand why on TRDs quite often the Compression open forces exceed the compression closed forces unlike on non TRDs.

Apologies, I'm not familiar with TRD's. If they are what I think they are, then there is much information already posted in this thread. See here, or here, for example. (Apologies, Wil).

[WilO]

Certainly no reason to apologize to me, Dave. I owe Luke an apology, if anything, for posting incorrect information.
My thought was that a through-rod damper might exhibit hysteresis due to the necessity of having greater sealing forces and associated friction, and that the friction force might be greater in one direction than the other.

Thanks for the correction Dave.

[Luke]

Certainly no reason to apologize to me, Dave. I owe Luke an apology, if anything, for posting incorrect information.
My thought was that a through-rod damper might exhibit hysteresis due to the necessity of having greater sealing forces and associated friction, and that the friction force might be greater in one direction than the other.

Thanks for the correction Dave.

Thanks for the replies people. Sorry I am a bit unclear as I can't post an FV plot. A prime example is on page 32 of this:
http://www.motorsportsspares.com/files/ ... design.pdf

you will see the lines "cross over" on most TRD graphs. I see it mainly on the compression side, but this also can occur on the rebound side as displayed here.

Where I work, we have been running twin tube non-trd dampers for years and have in the last couple of years converted to TRD. So for the similar average FV plot we see an improvement in the low speed range of hysteresis, but often worse in the mid speed range where the curves cross over. The difference between the dampers being the removal of the foot valve, and of course the fitment of the through rod parts.

[DaveW]

....you will see the lines "cross over" on most TRD graphs. I see it mainly on the compression side, but this also can occur on the rebound side as displayed here...

Apologies, now I understand. TRD = Through Rod Damper. There is still much information posted in this thread, though some of the diagrams are now missing, unfortunately (come back 747heavy). You might start looking here.

Ohlins TTX are very good dampers, probably only improved by Multimatic's DSSV (showing my prejudice). The only issue with the TTX (in my view) is that the low-speed controls are often too powerful, so low speed settings should be adjusted with care.

Damper velocity trajectories can usually be modelled quite accurately with a pure damper operating through a series spring. The series spring is often different (slightly) compression to rebound. A 40mm TTX typically has series spring value of around 4.5KN/mm. It is the result of oil compliance (with dissolved air). The "shape" of the hysteresis loops (as shown in Figure 11.0 of your reference) is (mainly) the product of the shape of the pure damper distorted by the series spring.

Interestingly, the series spring stiffness of a TTX sometimes appears higher than my estimates, possibly because of valve dynamics (a valve time delay appears as negative hysteresis). On one occasion I "corrected" a TTX trajectory by removing a time delay of 2.5 milliseconds (I recall). Perhaps this is the cause of your complaint.

[Luke]

....you will see the lines "cross over" on most TRD graphs. I see it mainly on the compression side, but this also can occur on the rebound side as displayed here...

Apologies, now I understand. TRD = Through Rod Damper. There is still much information posted in this thread, though some of the diagrams are now missing, unfortunately (come back 747heavy). You might start looking here.

Ohlins TTX are very good dampers, probably only improved by Multimatic's DSSV (showing my prejudice). The only issue with the TTX (in my view) is that the low-speed controls are often too powerful, so low speed settings should be adjusted with care.

Damper velocity trajectories can usually be modelled quite accurately with a pure damper operating through a series spring. The series spring is often different (slightly) compression to rebound. A 40mm TTX typically has series spring value of around 4.5KN/mm. It is the result of oil compliance (with dissolved air). The "shape" of the hysteresis loops (as shown in Figure 11.0 of your reference) is (mainly) the product of the shape of the pure damper distorted by the series spring.

Interestingly, the series spring stiffness of a TTX sometimes appears higher than my estimates, possibly because of valve dynamics (a valve time delay appears as negative hysteresis). On one occasion I "corrected" a TTX trajectory by removing a time delay of 2.5 milliseconds (I recall). Perhaps this is the cause of your complaint.

Interestingly attached is another brand dual tube through rod damper with a 40mm inner tube. As you can see the series spring required to correct the damper curve are 7kN/mm and basically infinite on the rebound.



whereas this damper below subscribes closer to your estimates of piston size determining effective series spring rates



The first damper has no valve block type adjustment as per an Ohlins/Sachs, so that would be my first thoughts of why it is better hysteresis wise.

[DaveW]

Your first example is an impressive damper.

Two comments: I suspect that your software has over-estimated the compression stiffness. Replacing the identified 7KN/mm with a (more likely) 4.5KN/mm would probably (i.e. might) get rid of the "negative" hysteresis at higher velocities. The rebound stiffness is proably not identifiable.

Here, for comparson, is a "dyno" test of a simple model of a damper complete with a series spring set at 4.5KN/mm (everywhere).

It looks as though your second example has several valve issues.

[JeffHNY]

Great thread, highly educational. On the topic of hysteresis, this is a dyno for a pair of Penske 7500 DA dampers with a VDP/VDP piston which were built new. Compression range demonstrated. Without any data to fall back on is this considered a normal amount of hysteresis or would it be considered excessive for a new damper? From reading through this thread it looks like there may be cavitation occurring.

Also for another vehicle I can see myself going with a ratio favoring more low speed compression but as DaveW mentioned the design may not be conducive to this. Any tips on how to improve damper performance?

[DaveW]

Without any data to fall back on is this considered a normal amount of hysteresis or would it be considered excessive for a new damper? From reading through this thread it looks like there may be cavitation occurring.

First, I don't think the dyno trajectories indicate cavitation (but that can always be checked by increasing charge pressure), but they do suggest to me that a better comprise might be obtained if you feel the need to increase compression damping.

At this point I would normally call on a damper expert to advise....(cue damper experts.)

Alternatively, this document might help, starting at Page 10.

[mlubbock]

Dave, I agree,
Jeff's trajectories don't appear to indicate any signs of cavitation. Assuming compression is in the upper window there is no 'plateau'-ing effect to be seen. As the adjustment is made via a remote reservoir (another assumption on my part) the hysteresis grows as a product/result. The hysteresis is unique to compression and commonly found from either hose compliance and/or compression valve ('CV'-reservoir) compliance. Both of which are sensitive to compression pressures which are are product of the CV being reacted by the shaft area only and not the somewhat larger main piston area in the main body (F=PA).

'Un-Mirrorred' plots would be more useful and as 'raw' as possible with no avaraging and/or filtering to assess 'performance' a little easier.

Apologies in the delayed response, I've just read this infomative article on FSD stuff from Dave and thought I ought to sign up.

Cheers.

[DaveW]

At last. A response from a damper expert. With that in mind I am happy to comment on JeffHNY's question, with the knowledge the mlubbock with correct my mistakes. The question was:

Also for another vehicle I can see myself going with a ratio favoring more low speed compression but as DaveW mentioned the design may not be conducive to this. Any tips on how to improve damper performance?

Compression biased damping is generally a good idea.

Compression damping for a Penske "shimmed" damper is normally split between two shim stacks.

The first, installed on the piston, provides damping, but no adjustment.

The second, installed in the reservoir, provides adjustable damping.

The first offers low hysteresis, but the second (acting at a distance, in series with the fluid column and pipe compliance) introduces hysteresis, which can be significant if the reservoir is remote from the damper.

The solution to your question, assuming you want to minimize hysteresis (also a good idea, I think), is do most of the damping on the piston, leaving just enough adjustable damping for your needs.