Interpreting and Comparing Shock Dyno Results

[Crazy Bored]

I'm currently trying to understand how to properly compare different settings on a shock that I have just had on a dyno. The dyno took the shock through roughly two full cycles, but the data points from the first cycle are noticeably different from the second. My issue is shown here:


How should I treat this data if I want to compare multiple different shock settings?
1. Leave it alone?
2. Average the forces and velocities from the different cycles and plot that?
3. Offset the y-axis to 0 force so that I can compare different settings that way?
4. Offset the x-axis to 0 velocity?
5. Both 3 and 4?
6. Use 2, 3, and 4?

Just to play around, I took the point with the red arrow in the first picture, and offset the plot so that point is at 0 on the y-axis (option 3 above). I chose the point with the red arrow because it was closer to the y-axis than the other point at near zero velocity. I then deleted the data from the other cycle, to get this plot:


To me, this is much easier to work with. But is it the best way to go about this? Am I missing out on important data by doing this? Again, my goal here is to accurately compare different shock settings.

[Jersey Tom]

I feel like that depends very much on specifically what you're trying to accomplish in the end... I think the way the question is right now is very vague.

[flynfrog]

option 7 figure out why your data is not repeatable.

[KeithYoung]

Were the shocks brought up to temperature before the runs? Maybe the first run they were cold.

[RideRate]

I'm currently trying to understand how to properly compare different settings on a shock that I have just had on a dyno. The dyno took the shock through roughly two full cycles, but the data points from the first cycle are noticeably different from the second. My issue is shown here:
http://i.imgur.com/rQqOL7d.jpg

You say your issue is shown in the first plot. There is no issue there. That looks exactly right and just fine.

I assume you are using a crank dyno and thus a sine input.

What you are seeing is not two cycles, but one cycle. That cycle has four distinct components: compression open, compression closed, rebound open, rebound closed.

Compression open is from bottom dead center (bdc) to the center point of the stroke. Basically 0 -> 10 ips and accelerating.
Compression closed is from center point of stroke to top dead center (tdc). Basically 10 -> 0 ips and decelerating.
Thusly the rebound open and rebound closed follow completing one cycle.

Your second plot only shows a half-cycle; compression closed and rebound open.

Think of how the input behaves as it goes bottom to top and then top to bottom. The difference in data points is your rate dependent hysteresis (shock lag) and some gas force variance since both velocities of a particular half-cycle occur at different displacements.

You do realize your plot is absolute velocity and not signed velocity, right?

Maybe run some different amplitudes or speeds and see how this all behaves. May help you grasp what's happening. The higher the accelerations (higher peak speed or smaller amp) the greater your hysteresis loop will become.

Again, nothing wrong here. Looks fine.

[DaveW]

Forgive me, but I think you are showing one cycle of a damper with "hysteresis". Quite likely you are using a sign convention where compression forces are negative. It is a bit of a long read, and some content is now missing, but I think this might be helpful.

[Crazy Bored]

I assume you are using a crank dyno and thus a sine input.

You do realize your plot is absolute velocity and not signed velocity, right?

Maybe run some different amplitudes or speeds and see how this all behaves. May help you grasp what's happening. The higher the accelerations (higher peak speed or smaller amp) the greater your hysteresis loop will become.

Yes, yes, and thank you for that suggestion.

Thank you all for putting up with my lack of knowledge. This post is a result of the first lab in a course I'm taking called Motorsports Instrumentation. Basically the lab involved calculating the critical damping value for one quarter of the car, and then putting the shock on a dyno, and determining which setting is best to start with for a baseline setup. The shock is only adjustable in rebound. All of this work was very straightforward and I did was was required for the report, but it doesn't seem terribly useful.

However, now I have unlimited access to the shock dyno, and I want to use this for my volunteer work on the FSAE car. The shocks on it are four way adjustable, which will obviously make things much more complicated. What lead me to making this post, (and my incorrect understanding of the shock dyno results) was this example image provided by the professor relating to the lab:


With the raw data from the different clicks, I was recreating this plot in excel, and ran into the issue I explained in my first post. I am not sure how the software came up with the above plot, since it only has one line for each setting. I get the feeling that this class isn't going to be very useful and I'm going to have to study on my own if I want to learn anything of value (this is the extent of damper related content). I really appreciate all of your quick replies, as well as the thread DaveW linked to.

[KeithYoung]

I am NOT an expert at this. However I got access in a race team and got to watch one of these being used.

Basically there is an actuator that pushes and pulls on the shock. This is done at varying speeds. Unfortunately I can't post the video I got of it, but I have some links to other videos:

http://www.youtube.com/user/ASTSUSPENSION?feature=watch

So the actuator pushes and pulls the shock at incremental speeds. For each speed increment, it takes force data and records it. The force in the shock (or damper) is proportional to the speed that of compression/rebound.

So for each "click" line you see there, someone took that damper, incrementally adjusted it with clicks, then put it in the machine which incrementally subjected the damper to different speeds and then the machine output the resulting force for each velocity and each click.

Does this help a bit?

[Crazy Bored]

Yes, all of that is true. However, I'm not inquiring about the actual process of operating a shock dyno. I was physically there, and adjusted it one "click" at a time myself, and started the machine. Zeroed the load cell, started the test, exported the results for use in Excel. I know how to use it, but that is meaningless if I am not able understand the data that I get from the tests.

[DaveW]

I know how to use it, but that is meaningless if I am not able understand the data that I get from the tests.

This might help. Here is a plot of a pure damper model, defined by the nodes listed. Here over-plotted in magenta is the output from a simulated dyno test. The difference is caused by a series spring placed in line with the damper (Ki).

Personally, I prefer to see the measurements as received, and then try to understand them. I am deeply suspicious of dyno software that generates processed plots that "somebody" thinks I would like to see.

[Jersey Tom]

I am deeply suspicious of dyno software that generates processed plots that "somebody" thinks I would like to see.

Could not agree with this more, and applies to more than just damper dyno's.

[RideRate]

However, now I have unlimited access to the shock dyno, and I want to use this for my volunteer work on the FSAE car.

Which FSAE shocks?

With the raw data from the different clicks, I was recreating this plot in excel, and ran into the issue I explained in my first post. I am not sure how the software came up with the above plot, since it only has one line for each setting.

Dyno software just tends to have many options for viewing the data it has created (good or bad). Again, all those plots are just a half of the cycle showing compression closed/rebound open. Another half-cyle view option would be rebound closed/compression open. Other common viewing options for the whole cycle include force vs velocity, force vs abs(velocity), mean(force) vs abs(velocity), and force versus displacement.

Actually I now see that you're using Roehrig's stuff in the pic u posted. You can download their software from their website and play with it to see what it does to make different plots.

[DaveW]

This might help. Here is a plot of a pure damper model, defined by the nodes listed. Here over-plotted in magenta is the output from a simulated dyno test. The difference is caused by a series spring placed in line with the damper (Ki).

You may think that my damper simulation is reasonably close to your first post. If so, then you can start to build a half reasonable model of the damper.

Since you have the results tabulated in Excel, you can postulate a series spring stiffness and compute a column of position errors (Zerror = load/Ki), add a column of corrected positions (Zc = Zmeas - Zerror), differentiate to obtain DZc, and finally add a column containing the areas of the triangles enclosed by two adjacent load/DZc ordinates and the origin. The sum of the areas can then be used in a "goal seek" with Ki as the variable.

You may find that a single estimate of Ki is not quite right (compression and rebound stiffness are often different). In that case split the trajectory into two & identify 2 values of Ki.

The resulting load/DZc co-ordinates are usually a reasonable estimate of the pure damper trajectory. They can be used as a look-up table directly, or used to indentify a functional approximation. A regression polynomial would be one option, but they tend to be unstable at extreme values. I use a cubic spline - I select a velocity array & then use regression to identify the associated load vector. A bezier spline might also work...

The result should be a computationally efficient description of the damper using small array defining a set of nodes and scalars defining one (or two) series springs.

That should, with luck, keep boredom at bay.

[Jersey Tom]

...I still feel like step 1 should be figuring out what they're trying to really accomplish.

[Crazy Bored]

Which FSAE shocks?

Actually I now see that you're using Roehrig's stuff in the pic u posted. You can download their software from their website and play with it to see what it does to make different plots.

They are Cane Creek shocks but were custom built for our cars, I think three years ago. And yes, it is a Roehig machine. I'll investigate the software more after I get through some of my higher priority schoolwork.

...I still feel like step 1 should be figuring out what they're trying to really accomplish.

To figure out baseline settings for the shocks to go into the initial test with, using simple calculations with critical damping and damping ratios. I am aware that these settings will be adjusted after gathering test data.