To learn and share about dampers / shock absorbers

[WilO]

Dave,

Thank you for the clarification...As an aside, having read the introductory pages of the Dixon book it's not difficult to come to the conclusion that the propensity for dampers to be rebound biased is steeped in tradition....

Wil

[tok-tokkie]

Dave, thats exactly what I suspect that is happening with those dampers and, of course, they are twin tube with nothing physical separating the fluid & the air reservoir.

As using pressure is forbiden by the rules, the solution Im gonna try here is to use some kind of rubber, maybe like this one:
http://en.wikipedia.org/wiki/Ethylene-vinyl_acetate
to fill the air volume that you need to have but letting the oil compress it at the same time...

Maybe next week I can plot the results...

Rubber is incompressible. What it does is it deforms but the volume stays constant. Foam rubber is compressible - it is the air in the bubbles that compresses.

[Belatti]

OK sorry tokkie, was the closed cell (doesnt absorb the oil) foam rubber I was talking about, the thing is its hard to get it right here...

[Belatti]

Well, sorry to tell that the test with the foam didnt work out... I used some packaging foam with very little air bubbles and I left a small place with air because I didnt know how much it was goning to get compressed, I didnt want to destroy the damper or the seals... anyway, I think the material absorbed some oil.

Will get some other material later...

Another test I did was using different pressures in a presurized 2 way penske damper and here is a plot:




There you can see 5 tests overlayed with 250, 200, 150, 100 and 50 psi. Not the difference I was expecting...

Another thing I did was to cycle the damper at 2Hz and 200psi with the manometer connected: there was a delta of 10psi between max compressed and max extended.

[Ciro Pabón]

Well, if cavitation is a problem, there you have a trick, used in water spillways, in hydroelectric dams: we (that is, civil engineers) use aeration as a way to minimize cavitation.

As aerated water is compressible, when the bubbles created by cavitation explode, the effects on the concrete surfaces are minimized. The shocks are not so powerful nor the temperature of water is risen so much (I believe that the temperature effect on the liquid can be a major problem in dampers, we have spoken about that ages ago).

I have no idea if it is possible to aerate the mixture of liquid and air in the damper cylinder, but a first idea that comes to my mind is to use a mesh before the valve (you know, the one that controls the flow of liquid) to try to mix the air and the liquid passing through that restriction. This COULD minimize cavitation effects. Can you check something like this, Belatti? You have the test rig, surely you can think of a way to mix gas and liquid before the restriction, you're the smart one...

The basic idea would be to use a "foam" instead of distinct "virtual" cameras with liquid and air (altough anyone who has dismantled a damper knows that the liquid and the air mix a lot inside the thingie).

[xpensive]

..., when the bubbles created by cavitation explode, the effects on the concrete surfaces are minimized.
...

Actually Ciro, my dearest friend in this mean world of unwarranted hostility, the problem occurs when those bubbles implode?

[Belatti]

I wish I had a test rig! But I only have a dyno shock...

I never said I have cavitation problems in the dampers, I allocate the compression lag to aeration, not cavitation wich I think its not the same.

Regarding the "mesh" idea, that could increase the turbulence before the restriction. Not sure what happens there... I´ll have to search for videos of transparent dampers

[Belatti]

Another test I have done in my spare time:



In blue you can see a damper cycled from 0,1Hz to 6Hz in steps of 0,1Hz.
In red you can see the same, with the difference that it is an average value of 10 cycles of every step in frequency (0,1 to 6).

One of the things I tried to guess with this is the natural frequency of the machine itself: I have noticed in prior tests that when frequency is rising there is a moment it shakes a little bit. That could lead me to change the way its mounted, but I dont see any weird stuff in the curves...

Another thing I wanted to know is the linearity of the asymmetry in the damper forces across the different freqs.

[thisisatest]

Belatti,
For a short while Cannondale used some closed-cell foam in one of their suspension forks for volume compensation.
http://www.cannondale.com/usa/usaeng/Ca ... ent_en.pdf
page 23 has pics, two of the three damper styles explained used the foam, the last one used a pressurized bladder.
Unfortunately, they don't use it anymore, and the end of the manual did not list a part number for that. But maybe some queries to Cannondale can help you find the most suitable type.
-GvT

EDIT- a better pic http://www.cannondale.com/usa/usaeng/Ca ... ote_en.pdf

[DaveW]

In blue you can see a damper cycled from 0,1Hz to 6Hz in steps of 0,1Hz.
In red you can see the same, with the difference that it is an average value of 10 cycles of every step in frequency (0,1 to 6).

The largest difference I can see between the two data sets is a load offset....

[Belatti]

Thanks tester!

The largest difference I can see between the two data sets is a load offset....

Indeed! The thing is, was that an error of load cell calibration or its the transients from the 1st cycle that in the blue case is computed and in the red one is averaged over the other 9 cycles?

I checked the calibration of the load cell according to the suppliers procedure and everything seems to work fine.

[747heavy]

it may be due to the temperature rise during the test --> leading to an increase in gas pressure --> leading to an offset in the force axis as the loadcell will see an increase in load/force due to the increase in gas force.

most damper dynos, will measure gas force (gas pressure*shaft area) at the start of an test, and take the value off the measured data, so that the graph is centered around the "Zero" force at "Zero" speed speed point/axis.
If the gas pressure varies during the test, it will/can lead to an "Zero" drift.
Normally it is not a big problem, but if you run a test over a long(er) period of time, it may lead to an drift which will/can affect your measured data, epecially if the damper warms up considerably during the run.

This problem will only be relevant with "normal" dampers (not with through rod/shaft or other constant volume dampers) and will be more pronouced with larger shaft diameters and with smaller gas volumes --> more progressive characteristic of the gas spring.

What is your test procedure?
Do you "re-zero" the load cell betwenn the runs, or only when you change dampers?

Just another question (unrelated to this), do you know if your damper dyno samples the data time or position dependent?
If time dependent, to you know what the sample rate is?
Thx

[Belatti]

it may be due to the temperature rise during the test --> leading to an increase in gas pressure --> leading to an offset in the force axis as the loadcell will see an increase in load/force due to the increase in gas force.

Check my post from november 5th... 200psi difference between tests and nothing!

Just another question (unrelated to this), do you know if your damper dyno samples the data time or position dependent?
If time dependent, to you know what the sample rate is?

Position (It has an encoder of 600 pulses per lap) and a freq controller.

[747heavy]

I did, and on the same tokken if you look here, you see a similar offset for a temperature increase of 20°, a graph you provided earlier, and which looks very similar to the damper shown in your last graph.




A increase of 200 psi gas pressure with an 5/8" shaft should result in a force of ~275N (28kgf).

I assumed (maybe wrongly) that your dyno software would perform a gasforce/friction test before each run, thereby taking the increased gas force into account, and offsetting it from the resulting graph.
This would explain that your graphs are centered around the 0 force line.(test from 5.11.)
If your damper heats up considerably during a run, the dyno can´t adjust for the change in extentsion force.

If your dyno has a "view live data" feature, you can see what I mean, when you cycle a damper at constant rpm for a extented period, you will/should see that your
football/potato graph drifts towards more bump force, as the damper heats up.
By I could be wrong, after all.

B.)
good stuff, there are not that many dynos which use a position based system.
Allowed, I would have thought, 720 samples per revolution (or a number based on 360°) to be more common/logical.
But, it does not matter much.
In this case it means you take a measurement every 0.6° crankshaft angle, instead of 0.5° (720 samples per revolution)
But in any case, this method should aid the repeatability of the tests, provide more consitent data, as the amount of data points per revolution is constant and does not change with rpm (which is the case with time based systems)
more/closer spaced data points @ lower rpm´s/frequencies.

on another note, you may want to check the opening characteristic of your rebound valve, or investigate other causes.
the rebound shape of both dampers, show an transition event when the damper is accelerating from 0 velocity to max rebound velocity, which looks a bit "odd".

[Belatti]

I did, and on the same tokken if you look here, you see a similar offset for a temperature increase of 20°, a graph you provided earlier, and which looks very similar to the damper shown in your last graph.

Oh, I tended to see it mor like in the temp graph there was a force decrease (both bump and rebound) and then in the "average" graph there was not the decrease but an offset.

The soft can also provide you of a table with soma values. One of the columns is "preload" and it correlated well in the "different pressures" test.
Thus your assumption of offset corerction maybe right

I use to watch out the live data when I perform a test to make sure that everything is going fine (I have a little paranoia of getting an oil bath ) and I will try to follow it in a test.