REstackor

[asnyder2]

Isn't there a difference between the dampening force and dampening coefficient? My restackor has seperate graph for force. Seems I remember reading that in user manual...

[LSWP]

Can somebody put another set of eyes on this?

 

2008 YZ 250, shock rebound.  Piston parameters were taken from this thread, shim stack info is from Kevin's website.  First picture is the setup, second is the results.  You can see that the ReStackor predictions are WAY off the measured data.  Can anyone find errors in my ReStackor setup that might be driving this error?

 

 

seem wrong values of force of the sheet, the extension of direct value to the shock is about 1200 lbf, maybe the maximum force should be set to 80 lbf.

Edited December 12, 2013 by lskyb

[Kyle Tarry]

Isn't there a difference between the dampening force and dampening coefficient? My restackor has seperate graph for force. Seems I remember reading that in user manual...

 

Yes, there is a difference!  The damping coefficient is to the damping force, sort of what velocity is to displacement.

 

http://en.wikipedia.org/wiki/Damping

 

seem wrong values of force of the sheet, the extension of direct value to the shock is about 1200 lbf, maybe the maximum force should be set to 80 lbf.

 

Are you talking about Fmax?  If so, Fmax has no influence on the damping force/coefficient results.  All that Fmax does is set the force level that is applied to the stack for the deflection image, and set the maximum value that is shown on the shim edge lift and flow area plots.

 

Furthermore, your arbitrary force choice of 80 lbf doesn't make any sense.  I chose 200 lbf because that is approximately equal to the shim stack force at the maximum velocity of my simulation.  That way, all of my plots have approximately the same range.

 

It looks like you were unclear about this back on page 7, over a year ago... Clicked and KevinStillwell both had a bunch of good explanations back there: https://www.thumpertalk.com/topic/792432-restackor/?p=10194745

[LSWP]

sorry, it does not matter.

?

 

However Yeah, my reasoning is different, if there is an error in the calculation of the bending of the stack, for example flexes too much, the only solution is to reduce the Fmax.

[Kyle Tarry]

I don't completely understand what you're saying, but I don't think that you understand how the software works.

 

Fmax determines what data is displayed on the charts that have stack force ("Force") shown on the X-axis, and the deflected shim picture.  FMAX IS NOT USED IN THE CALCULATION OF THE DAMPING FORCE OR COEFFICIENT.  Changing the Fmax value will have no effect on either of these forces!

 

I don't know what you mean about "an error in the calculation of the bending of the stack."  If there is an error in this calculation, the program is not working correctly, and changing the Fmax value won't fix that.  However, I have not seen anything that suggests that this aspect of the sheet isn't correct; in fact, ReStackor output matches fully meshed FEA models pretty well (but is way, way faster).

 

A really easy way to show this is to run a simulation, set it as the baseline, than change the Fmax value (but nothing else!) and run the simulation again.  None of the results will change (except for the picture of the deflected shims).

 

This is all described right in the ReStackor users manual (http://www.shimrestackor.com/Code/User_Manual/Sections/Input/input-ReStackor.htm):

 

"ReStackor pro calculations compute the damping force as a function of suspension velocity up to the value specified by u.wheel. If the maximum wheel velocity produces a shim stack force greater than F.max additional calculations are made internally to compute the stack deflection and flow area. Decoupling F.max and u.wheel in ReStackor allows the stack structure to be displayed at a low applied force, like examining where a crossover gap closes, while computing the damping force of the suspension at a much higher suspension velocity."

 

[LSWP]

Of course, the theory is very valid, but a given instrument remains the only reliable source in this case, and the solution proposed by Restackor is too discordant with the one proposed by Dyno.
However, if the speed of u.wheel is the parameter that controls the strength of the opening of the stack, while the value of F.max is only for setting down (at this point I do not understand the utility) is that only modify the parameters until to obtain the same curve Dyno and then figuring out the mistake.

[Kyle Tarry]

I don't understand what you're trying to say.

 

I know that the solutions are "too discordant."  That's why I posted this!

 

Can somebody else take a look at that sheet and see if they see anything obviously wrong that would explain the error?

[LSWP]

Moreover , if what the manual says Restackor must be observed :
•
F.max [= ] The maximum force to be applied to the shim stack face . Shim ReStackor calculations display the stack structure , compute the edge lift and shim stack face flow area at forces up to F.max . F.max is in pounds force .

•
u.wheel [= ] The maximum wheel velocity in inches per second . ReStackor pro calculations compute the damping force as a function of suspension velocity up to the value specified by u.wheel . If the maximum wheel velocity Produces a shim stack force greater than F.max additional calculations are made ​​internally to compute the stack deflection and flow area . Decoupling F.max and u.wheel in ReStackor Allows the stack structure to be displayed at a low applied force , like examining where crossover gap closes , while computing the damping force of the suspension at a much higher suspension velocity.

The relationship of bump height , and suspension bike speed velocity for different wheel sizes is Given here . With an estimate of the suspension velocity , say a three inch bump hit at 30 mph, you can design crossover gaps, backing shims and the stack taper to produce a specific damping forces at the suspension velocity produced on a Given bump height .

it follows that the relationship you have set on the sheet is totally wrong, or I have misunderstood how it works.

[Clicked]

Kyle,

The ’08 yz250 uses an 18mm shaft and a different needle than the older 16mm yz shock. You didn’t enter either of those in your calculation setup. Kevin also lists three different stack configurations for a “stock” ’08 yz250 setup.

 

The bottom line is it doesn't matter what Kevin lists for a stock shim stack – or the valve geometry. What matters is the actual stack configuration and valve geometry that was used in the dyno test. Posting photos of the valve along with the dyno data is really helpful so that everyone knows what shock configuration you are dealing with when looking at dyno data.

 

Getting configuration info from anywhere else is likely to be unrelated to the configuration that was actually tested on the dyno. Like you demonstrated in this example, nice work.

[Clicked]

Getting back to your original post: An important result of dyno tests in the 20 in/sec range is the data is taken right at the point where the shim stack cracks open. At that velocity the dyno is measuring the combined effect of a bunch of tunable parameters in the fork:

 

 

Since you have already worked out all of that geometry info for a drz400, dr650, te610, yz250 and a wr300 it would be useful to post that info along with the your dyno data so other riders can use that as sort of a touch stone to start the tuning process. To make that info useful you need to post the dyno data along with photos of the valves and needle geometries so everybody understands what fork configuration you are dealing with for each dyno test.

 

Valve face dish is a very important parameter to get the peak of the damping coefficient curve right. For a stock valve the dish is usually around 0.001”. That works out to be about ¼ of a shim thickness and delays the point where the shim stack cracks open until the applied force to the stack deflects the stack ¼ of a shim thickness. That preload forces all of the fluid to be crammed through the bleed circuit and jacks up the low speed damping. Since your calculations keep missing the peak of the damping coefficient curve it will be interesting to see what you have measured for dish on these five different valve configurations.

 

The other thing that is going to be interesting is the effect of the different needle geometries on the shape of the damping force curve. The yz250 and te610 use very different needle configurations so it will be interesting to see the measurements and photos of those needles and how the different geometries effect the shape of the damping force curve for these five different bikes.

 

Posting the dyno data along with the fork geometry configuration info should go a long way in terms of demonstrating how the various tunable parameters in a fork affect the shape of the damping force curve. Hopefully your posts here are aimed at that goal, as you say.

[Kyle Tarry]

it follows that the relationship you have set on the sheet is totally wrong, or I have misunderstood how it works.

 

Yes, you have misunderstood how it works.

 

 

Kyle,

The ’08 yz250 uses an 18mm shaft and a different needle than the older 16mm yz shock. You didn’t enter either of those in your calculation setup. Kevin also lists three different stack configurations for a “stock” ’08 yz250 setup.

 

The bottom line is it doesn't matter what Kevin lists for a stock shim stack – or the valve geometry. What matters is the actual stack configuration and valve geometry that was used in the dyno test. Posting photos of the valve along with the dyno data is really helpful so that everyone knows what shock configuration you are dealing with when looking at dyno data.

 

Getting configuration info from anywhere else is likely to be unrelated to the configuration that was actually tested on the dyno. Like you demonstrated in this example, nice work.

 

Good info, thanks for chiming in.

 

The test that we are looking at is supposedly a stock shock, so the configuration should match the stuff that Kevin lists.  However, I have asked him to look into this specific test number, so we should be able to confirm what the correct settings were.  This is test number 1493s, for the record.  I don't have a high enough access level to check these.

 

I fully agree with all of your assessments; that's why I posted my sheet here, so that hopefully the right people can give feedback on what aspects of the setup are incorrect.  As an alternative, if you have a different set of dyno data and matching ReStackor sim, I'd love to look at it. 

 

In the end, I am just trying to understand the tool, and more importantly, understand how to get results that accurately mimic the real world and can be used and compared across platforms.

 

As an aside, if you come at this from a very different point of view, there is some discrepancy between the dyno results and what I would expect from a "vehicle dynamics" point of view.  In other words, taking ReStackor out of the equation entirely, the amount of rebound damping that Kevin has measured seems low to me.  Does anyone know what the linkage ratio is for an '08 YZ250?  I'd like to refine my damping ratio calculations, if possible...  I am not suggesting that the dyno results are "wrong," because, well, they essentially can't be.  Rather, I am trying to discern the vehicle dynamics targets the OEM engineers were shooting for.

[Clicked]

Kyle,

 

Since you have the dyno data for all of those bike configurations it will be interesting to see how the wheel damping coefficients work out. You are going to have to work out a weight split between the front and rear wheel to do that and remember the low speed damping is for chassis control and the high speed damping is for wheel damping. So you are going to end up with eight wheel damping coefficients for each bike, twelve if you include the rider weight in chassis damping.

[Kyle Tarry]

Good post Clicked!  Let me try to break it up a little.

 

Getting back to your original post: An important result of dyno tests in the 20 in/sec range is the data is taken right at the point where the shim stack cracks open. At that velocity the dyno is measuring the combined effect of a bunch of tunable parameters in the fork:

 

 

Since you have already worked out all of that geometry info for a drz400, dr650, te610, yz250 and a wr300 it would be useful to post that info along with the your dyno data so other riders can use that as sort of a touch stone to start the tuning process. To make that info useful you need to post the dyno data along with photos of the valves and needle geometries so everybody understands what fork configuration you are dealing with for each dyno test.

I have worked out the "base" geometry for most (not all) of the forks for these bikes. The DR and YZ stuff I have been looking at is really on the shock side (DRs are damper rod forks, so they're totally out of the equation). However, I have a lot of data for the TE and WR, and can collect more as necessary.

I certainly have not looked at check plate (or midvalve, as the case may be) spring stiffness or preload. That's easy stuff to measure, but there isn't a simple way to put that into ReStackor, so it's difficult to work with it. On the mid, you could do some estimates by running simulations at different values for float, and look at the results for each in a different speed range, in order to estimate what the "actual" curve looks like.

As an aside, the ability to put a spring rate and preload behind the mid would be a really cool feature. For example, I've been doing some work on a sprung midvalve retrofit (where the midvalve spring controls the opening, instead of just the shim stack) and it's tough to simulate in ReStackor.

Valve face dish is a very important parameter to get the peak of the damping coefficient curve right. For a stock valve the dish is usually around 0.001”. That works out to be about ¼ of a shim thickness and delays the point where the shim stack cracks open until the applied force to the stack deflects the stack ¼ of a shim thickness. That preload forces all of the fluid to be crammed through the bleed circuit and jacks up the low speed damping. Since your calculations keep missing the peak of the damping coefficient curve it will be interesting to see what you have measured for dish on these five different valve configurations.

This is a really interesting comment! I have not measured valve dish, but I have the tools to do so, and I will look at that.

I don't have all of my sims available (at work), but looking at a simulation for a set of Marz 45's, it looks like the force/deflection curve is approximately:

Force Yport

[lbf] [in]

0.000 0.0000

0.055 0.0004

0.222 0.0015

0.499 0.0034

0.886 0.0051

1.385 0.0068

1.994 0.0090

And the BV speed/fstack curve is approximately:

U.clk Fstack

[in/sec] [lbf]

0.8 0.09

3.2 0.09

5.0 0.09

7.7 0.51

11.2 0.83

14.4 1.11

17.8 1.38

21.3 1.64

25.2 1.91

So, if the piston face has 0.001" of dish, it wouldn't be cracking open until somewhere around 0.15 lbf on the stack, which would occur somewhere around 5 inches per second. Of course, the U.clk vs F.stack curve is going to change if you add the preload caused by piston dish, so this is only an estimate, but the preload will probably increase the fluid pressure for a given fork velocity by forcing more flow through the bleed, so if anything I'd expect this "crack open speed" to go down.

In any case, I'll try to measure the valve dish ASAP.

 

The other thing that is going to be interesting is the effect of the different needle geometries on the shape of the damping force curve. The yz250 and te610 use very different needle configurations so it will be interesting to see the measurements and photos of those needles and how the different geometries effect the shape of the damping force curve for these five different bikes.

I have not looked at clicker needle geometry. My perspective is that, even if the geometry isn't right, the results with the valve fully open and fully closed should be fairly accurate, and thus should "bracket" the fork's effective range. Of course, if we want to have accurate results at all clicker positions, we need to look at the needle geometry.

 

Hopefully your posts here are aimed at that goal, as you say.

Just to be clear, I don't doubt any of ReStackor's backend calculations; fluid dynamics is fluid dynamics, there's nothing fuzzy about that. What it comes down to is I am trying to understand two things:

1. How well ReStackor predicts actual behavior

2. How to properly set up a ReStackor sim to maximize accuracy

Note that #1 is not a negative assertion; I've done a decent amount of work doing computational simulations for engine bahavior and suspension behavior (not this fluid damping side of things though), and I fully understand the difficulties involved. This isn't intended to portray the tool in a negative light, but rather to simply understand the tool and use it to the best extent possible.

What happened is I had some dyno results (DR shock, DRZ shock) that matched the software really well. Then I had some other results (forks, YZ shock) that didn't match at all. I thought "that's odd!" and wondered what was going wrong.

[Kyle Tarry]

Kyle,

 

Since you have the dyno data for all of those bike configurations it will be interesting to see how the wheel damping coefficients work out. You are going to have to work out a weight split between the front and rear wheel to do that and remember the low speed damping is for chassis control and the high speed damping is for wheel damping. So you are going to end up with eight wheel damping coefficients for each bike, twelve if you include the rider weight in chassis damping.

Yeah, super interesting. I will try to summarize the data that I have this evening.

I don't see 12 ratios, though? I see the following:

Front/Rear:

Chassis (sprung mass)

Chassis w/ rider

Wheel (unsprung mass)

So that would be 6 total. What other permutation are you counting?

Bikes are really intersting from this perspective because they are such a (relatively) complex system. Automotive applications are easy; you have sprung and unspung mass, and if you pop open Milliken or Dixon, they'll give you a range depending on application. For example, around 0.3 maximizes ride comfort (minimum transmissibility), whereas 0.5-1.0 is more common for high performance stuff. Bikes aren't this simple, because calculating the mass isn't as easy, and the masses aren't rigidly connected.

Bikes look simple but are actually quite complex... Cars are boring!

[Clicked]

I have not looked at clicker needle geometry. My perspective is that, even if the geometry isn't right, the results with the valve fully open and fully closed should be fairly accurate, and thus should "bracket" the fork's effective range. Of course, if we want to have accurate results at all clicker positions, we need to look at the needle geometry.

 

 

Kyle, I think that statement shows some good insight. But it also raises a concern – if ReStackor calculations are off by a factor of two, is that relative to the closed clicker calculations –or- calculations using a clicker setting of 10 with a needle geometry that has nothing at all to do with the needle geometry actually used in test? That is something that needs to get straightened out.

 

Frankly, when I see damping force errors on the order of 10 lbf for a fork producing 20 lbf of damping I immediately become suspicious of seal drag and how that was treated in the dyno data. For your tests, sense you are interested in wheel damping coefficients, seal drag is a part of the damping force so it would make some sense to leave the dyno data uncorrected for seal drag.

 

On the other hand, if you are looking a shim stack tuning you need to take the seal drag out of the dyno data so that you can look at test-to-test changes in the shim stack and how that effects the damping force. For an open chamber fork where you have to test with the stanchions tubes and all of the seals in place it would not be surprising to see 10 lbf of drag from the seals and sliders. That drag makes it difficult to sort out the difference between hydraulic damping and seal friction.

 

But there is no way to figure out if the drag force was in or out of the dyno data you have been talking about until you post your data.

[Kyle Tarry]

Frankly, when I see damping force errors on the order of 10 lbf for a fork producing 20 lbf of damping I immediately become suspicious of seal drag and how that was treated in the dyno data. For your tests, sense you are interested in wheel damping coefficients, seal drag is a part of the damping force so it would make some sense to leave the dyno data uncorrected for seal drag.

Totally agree. This is a big part of why I haven't been pushing the fork data as much, there are lots of variables. Shocks are a lot simpler, so I've been trying to focus there.

 

I also don't have much good fork data.  The other guy I have been collaborating with can only go to 20 ips on his Roehrig, which frankly is way too slow for forks, in my opinion.

 

But there is no way to figure out if the drag force was in or out of the dyno data you have been talking about until you post your data.

Yeah. The problem is that my dyno data (well, my source's dyno data) is incomplete. No insult to the guy I'm working with intended, but he didn't collect the data with the intention of doing these comparisons. So, for example, he doesn't always have good notes about what the setup configurations were. For that reason, I am hesitant to bank on his data, because it's hard to correlate.

I think that Kevin (Stillwell) has been very methodical about collecting good data and keeping track of everything. That was my primary motivation for looking at the YZ stuff; I don't have a YZ, but they are common, runored to have the best stock suspension, and Kevin has lots of data on them.  A big part of the battle here is getting a set of forks that we have good measurements from AND good dyno data.  I can get good measurements on the stuff in my garage, but nobody wants to dyno old Marz Shivers; conversely, there is some dyno data out there, but it's not necessarily for stuff that was thoroughly measured and recorded.  Frustrating!  This is why I started poking around in here, hoping to find someone who is active with both ReStackor AND a dyno.

I'd love to buy a dyno and put my own forks on them, but that's about an order of magnitude more expensive than what I would consider cost prohibitive.?

[Adammoto]

Good to have you back with us Clicked.  

 

Any comments on post #172?  Perhaps I should mention the "plush" curve has mid-valve float of .3mm paired with a two-port piston, and the "harsh" curve has float of .95mm paired with stock WP three-port.piston.

[dkbl5]

Too bad a good engineer does not necessarily a good tuner make.

[MarekB]

Sorry for interrupting (this thread is getting more and more interesting), but I have a question concerning following:

 

(...) in fact, ReStackor output matches fully meshed FEA models pretty well (but is way, way faster).

 

Have you actually done such comparison? What FEA modelling program did you use?

[Kyle Tarry]

Have you actually done such comparison? What FEA modelling program did you use?

 

Yes.  Autodesk Inventor Simulation (just because we have it at work).

 

It's a little bit difficult to make a direct comparison because ReStackor makes some assumptions about how/where the bottom face of the shim is loaded, but yes, the results I have gotten match up within reason.  I haven't done extensive work, just checked a few stacks.  It takes forever to run (lots of boundary conditions) and doesn't give you nearly as much useful data (curve shapes, damping, etc) as ReStackor, so I didn't see any reason to spend more time.