Dyno data –And what it tells us about how to tune a shim stack and control the shape of the damping force curve

[Kyle Tarry]

Clicked, there seem to be some discrepancies between your two statements?  In the latter, you said:
 

These two stacks are the same. They have the same clamp, the same taper and the same crossover. The face shims were set by the thickness cubed rule to be the same. With everything same, same, same you don't need shim factors to figure out the stiffness of those two stacks should have been the same.

However, in the post I was quoting from, you said (emphasis added at bottom):
 

From shim factory theory the difference in face shim stiffness is:

• SF14: Baseline stack of 14x40.2 face shims with a shim factor of sf=112
• SF4: Replacement stack of 4x40.3 face shims with sf=108
• The replacement face shims are about 4% softer

To figure out the change in damping force you need some way to estimate of the stiffness of the overall stack. Using tapered stack shim factors:

• SF14: The baseline stack with 14x40.2 face shims has a tapered stack shim factor of tsf= 262
• SF4: Shim factor scaled stack with 4x40.3 face shims and a tapered stack shim factor of tsf= 258
• The replacement stack is about 2% softer in overall stack stiffness

...
The replacement 4x40.3 stack was expected to be softer and the dyno shows that. To get the same damping force the 4x40.3 stack needs a shaft velocity that is about 8% higher. That 8% shaft velocity difference implies the stack was about 8% softer, not the 2% expected from shim factor theory. Shim factors missed damping force change by a factor of 4..

 

 

I agree with you that the stacks should be comparable without needing the shim factors, by using pure thickness cubed factors instead.  However, in your post, your comment of the 4X factor of error on damping force change was specifically referring back to the shim factor numbers (see above).  So, I think that if the shim factor equations are +/- 30%, that smaller error (8% difference vs 2% difference) is completely in the noise, and the results actually show good predictive power of the shim factor equations.

 

If we look at the purely thickness-based comparison, I think that does beg some questions, as you indicated.  In that case, the results are quite close though.  The thickness-cubed calculations suggest that the stack should be 4% softer, and the dyno indicates about 8% softer, meaning that the total system error is only 4%.  That is a really good match!  I think that a 4% error for this entire system (different shims, different build, re-assembly of the damper, possibly new fluid and nitrogen, dyno calibration error, dyno measurement error, repeatability, etc) is a really really good result!

 

In any case, I still really like your question.  We like to simplify shim stacks down to the "beam stiffness theory" behind them, but in reality, in the application, we know that things are a bit more complicated than that.  There are fluid flow effects, there are friction effects, there may be preload between the shims, and the deformed shapes of thick shims and thin shims may be slightly different too.  IF we can believe the small error between these two tests (I am still on the fence on that), I think it would show that these "other effects" can have an impact on the order of 4% on overall damping, outside of the stack stiffness effects.  However, I struggle to separate to what extent this 4% is due to other effects (friction, fluid dynamics, etc) and to what extent it's just misc. error.

 

I think it would be very valuable to do the same exercise with a large difference between stacks (say, 20% or more) and see the predictive power in that case.  Large differences between the stacks allow us to more easily separate the small errors inherent in any test, and the true differences between predicted and actual damping that we are trying to investigate.

[Clicked]

The above MXScandinavia dyno tests used an Ohlins four port valve and Ohlins 19 cSt oil with a SpGr of 0.85 because MXScandinavia hints he used that oil.  

The oil was assumed saturated with ambient air and compressed to 10 atmospheres. That works out to a dissolved gas volume fraction (xLsat) of +/- 0.015. The shock was assumed perfectly bled along with a nominal shim friction value of cf.stk= 1.0. All of the MXScandinavia analysis curves in this thread use those same values.

[Clicked]

MXScandinavia Shim Stack Stiffness Measurements

By thickness cubed theory the face shims of the 14x40.2 stack should be 4% stiffer than the 4x40.3 stack. Add in the tapered section and that difference drops to about 2%. The dyno shows 8%. There are a lot of numbers floating around here along with the associated uncertainty Kyle brings up. So, was the stack stiffer or not?

 

That kind of thing can lead to pages of debate on TT. In this case we don not have to guess. MXScandinavia went out and he got the data!

 

MXSCandinavia put the two stacks on a finger press and directly measured the stack stiffness. The finger press pushes a steal rod through the valve port and directly measures the force needed to deflect the stack using the setup shown in the picture below. With those direct measurements MXScandinavia wanted to find out if the actual stack stiffness was closer to the 2% difference expected by shim factors or the 8% difference measured in the dyno tests.  

You don't have to look at those results very hard to figure out the shim stack stiffness difference was larger than 8%. Debating 2%, 4% or 8% would be pretty silly.

Edited February 21, 2015 by Clicked

[Clicked]

Finger Press Data

To make use of the finger press data some corrections are needed. First, at zero force, the finger press data shows the shim stack was lifted off of the valve face. The data needs a zero offset correction to get the finger press to read zero force at zero lift.  

The valve face on a shock valve is usually dished so the shim stack is preloaded against the valve seat. In that case the press should read a positive force at zero lift. MXScandinavia says he lapped the valve faces flat for these tests so the finger press should read zero force at zero lift.  

[NastyR1]

 

 

 

 

That's an interesting picture and confirms a theory I always had about the reaction of a shim stack in a 3-port valve.

 

Correct me if I'm wrong but are'nt the shim stackor calculations base on Belleville washer formulas? That's fine but a Belleville washer acts as a variable spring load because the force is directed at the raised center of the washer and the force is distributed evenly around the OD of the washer as it collapses due to force. 

 

From the picture you can see that as oil passes through the valve ports the shim stack OD flexes unevenly and resembles a "wave". How could this uneven force be calculated by using Belleville washer formulas? This may explain why a calculated shim stack force can't be replicated on a shock dyno (excluding the effects of seal drag, nitrogen pressure etc).

 

To use the Belleville washer formula you would have to install a plate directly against the valve and have a Belleville washer underneath it - that way when oil passes through the valve ports the plate causes the washer to collapse and a calculatable force against the flow of oil can be obtained.

 

Does this make sense?

[Clicked]

ReStackor does not use Belleville washer formulas. ReStackor uses FEA calculations to determine shim stack deflections.

 

Either way, the goal of this thread is pull together whatever dyno data we can get our hands on and try to figure out how the stiffness of shim stacks actually behave and how that effects the shape of the damping force curve measured on a dyno.

For shim stiffness there seem to be two camps:

• Belleville spring washer theory: Shims become stiffer as lift increases making the curve non-linear
  
• Shim factor theory: Shim stiffness follows the thickness cubed "beam theory" of conventional springs creating a linear stiffness curve
  

Draw a straight line through that MXScancinavia finger press data and you can see that wave shaped shim stack deflection did not produce a linear stiffness curve. Instead the stack got a little bit stiffer as edge lift increased. A “little bit” stiffer is going to be an important distinction.

That “little bit” of stiffness increase I think is a part of the reason why the dyno data for that stack showed an 8% increase in damping force instead of the 2% expected by the linear relationships of shim factor theory.

 

That wave shaped stack deflection has an important influence on the stiffness of shim stacks. That wave shape also affects the damping performance of wide and narrow valve port geometries. The MXScandinavia link at the top of this thread shows some dyno data comparing different valve port geometries. That will be some interesting stuff to check out latter in this thread.

[Clicked]

So lets get on with checking out this finger press data.......

 

The second correction needed to make use of the MXScandinavia finger press data is a correction for the measurement location. The finger press measures stack lift at the location where the steal rod contacts the stack. In this case that is the port jet center line. For shock absorber calculations, stack lift is usually measured at the valve port edge since that is where the flow is metered. Some FEA calculations report lift at the shim stack edge. Either way, corrections are needed so that everything is measuring stack lift at the same location. In this case that is the valve port seat edge.

[Wonderspoon]

Just spittballing here, but since the thicker stack is, well, thicker....... can that  create a small fluid pressure increase (via restriction as the oil moves past the face shims) before the area widens into stack taper? 

[DHS_sponsored]

Clicked, or anyone else....

Slightly off topic, but still along recording differences/variations.

We look at shim stacks as 2 dimensional a lot.....

Even the tester above, finger test unit, is only testing through one port, of possibly 3 or 4 total.

It is hard to explain, but surely a shim, can only flex/bend, EVENLY, as far as the 'maximum size square' in a 8 port (4 each way....), or Max triangle in 6 port Pistons......

In an 8 port piston, at low shim lift, will you get an even lift on all 4 ports to a certain point, and then when pushed harder (beyond maximum square dimension) will that then act like a 4 port piston, bending/flexing the shims through a point/section/line through the centre of the piston......

So initial shim movement will be in four points, up to maximum square dimension (of the largest[thickest?] face shim) then revert to 2 points as flex/bending gets higher.....?

[DHS_sponsored]

.....would a 4 finger tester yield different results as it would be simulating flow through all ports, thus the 'whole' story of shim deflection, bend, flex, wave profiles and limitations etc etc. Pete Russell has xv6 Pistons that work as '4 port' so to speak. (Though multi port, but designed it seems to only flex shim stacks in 2 positions opposite each other).

[Clicked]

The finger press has four fingers. On the left and right hand sides of the photo you can see all the way under the stack to the other side. That is because the sides of the stack are lifted. The post fingers lifting the edge of the stack are out of frame in the picture.

You make an interesting point. If you tested shim stacks using one, two, three or four fingers the stiffness measurements of the stacks should be different. The stiffness should change after the edges of the wave intersect. After that the press has to lift the whole stack including the drooped wave section in between the valve ports/fingre press posts.

 

The other thing going on with these stacks is MXScandinavia tested a 14x40.2 and 4x40.3 stack. The thicker shims in the 40.3 stack transmit more force tangentially. That changes the wave shape and the stiffness progression of the stack compared to the thinner shims in the14x40.2 stack. You can see that difference in the finger press curve.

 

Those wave effects are not included in the stack stiffness estimates of shim factors. That might be something useful to consider adding to get more accuracy out of the shim factor equations. But you need to figure out some simple way to do that short of a full on FEA analysis. That is something you just can't scratch out on the back of an envelope.

[Clicked]

Finger Press Data

 

Third, the finger press applies force to the shim stack at a location that is different from the force applied by the fluid. The finger press rod sticks through the valve port and applies a point load to the stack at the port jet center line. The fluid pressure force is applied over the entire valve pocket port area. Converting the pressurized pocket area into an equivalent center of pressure gives a much smaller torque arm for the fluid pressure. In other words 100 psi of fluid pressure on a one in2 valve pocket is no where near a 100 lbs of force applied by the finger press at the outside edge of the valve pocket.

The three corrections described above end up with the finger press measuring a force that is about 85% of the fluid force applied to the stack when installed in a shock. That correction is specific to the valve port geometry used in this test. Change the port jet location, valve pocket area or valve seat location and the correction factors are going to change. Those corrections can make finger press data difficult to interpret in terms of the installed stiffness of a shim stack in a shock and the effect that is going to have on damping force.  

[Clicked]

Dyno Stack Stiffness Compared to Finger Press Data

We have dyno data for the two stacks MXScandinavia tested and we have finger press data for the stacks as well. That just makes you wonder: how close are the stack deflections derived from the dyno data; to the stack deflections measured on the finger press; and are those consistent with the deflection differences expected by shim factors?  

The Roehrig dyno MXScandinavia used only had enough power to get to a damping force of about 800 lbf. That was good enough to generate shaft speeds of 37 in/sec and a stack lift of just over one shim height. That’s pretty good, but only touches the very bottom of the finger press data.

 

At 35 in/sec the dyno shows a shim stack stiffness difference of about an 8%. For the finger press data you are down in the noise of the zero offset correction right at the point where these two stacks are just starting to differentiate themselves. To get any convincing data of the difference between the stacks they really need to be tested at higher shaft velocities. The data above is a good example of the usual frustration of dyno testing. You just get to the interesting part, where the shim stacks may – or may not - behave differently and the dyno runs out of balls so you just can't get the data you were after. DaveJ touched on the earlier.  

 

For the data MXScandinavia could get: the dyno data; finger press data; and FEA calculations all line up pretty well. But only at the very bottom end of the finger press data. So that just begs the question: “What happens if these stacks could be tested at higher speed?”

 

[Clicked]

High Speed Dyno Data

MXScandinavia did a clever thing for this first dyno test. The baseline 14x40.2 stack turns out to be a standard factory Ohlins stack. That stack has factory dyno data and that data goes way out beyond the shaft speeds MXScandinavia could get to speeds of 120 in/sec. That’s about three times higher then the velocities MXScandinavia could measure on his dyno.  

 

Building the dyno tests around one of the standard factory Ohlins stacks let MXScandinavia demonstrate a couple of important points:

• MXScandinavia could compared his data to the factory dyno data and make sure his tests were good.   

• MXScandinavia could compare his finger press data to the factory data at shaft speeds that are about three times higher to see if his finger press data was any good.

The MXScandinavia dyno data (blue squares) are spot on with the factory data (blue circles). The dyno data from both sources line up with the FEA calculations out to shaft velocities of 120 in/sec. Those calculated results are the blue line. The damping force from that shock has now been verified using three different data sources: MXScandinavia dyno; Ohlins dyno data and FEA analysis.

 

So the damping force measurements from MXScandinavia look pretty good against the factory Ohlins data. What about the finger press stack deflection data, does that line up with the high speed dyno data?

[Push Industries]

It's not really particular to the discussion necessarily, but if you check out the link below you can see our EMA running actual suspension data collected using our on-board data loggers. Heck, you'll even get a look at one of our data loggers.

http://www.vitalmtb.com/photos/features/First-Look-Push-Industries-Elevensix-Coil-Shock,8629/Slideshow,0/bturman,109

Darren

[Clicked]

What we are trying to do here is figure out how to use shim factors to compare the stiffness of shim stacks.

 

For this MXScandinavia data we have kind of a unique data set with both dyno data and a direct measurement of shim stack stiffness using a finger press. We need to figure out if the shape of the shim stack stiffness curve measured on the finger press makes any sense when compared against the shape of the damping force curve measured on the dyno. If that works out then we can compare the finger press stiffness curve to shim factors and see how it all lines up.

[Clicked]

What does the 120 in/sec factory dyno data mean in terms of real world suspension speeds?  

For a shock the rear suspension linkage ratio is around 3:1. So, a dyno shaft speed of 120 in/sec is equivalent to a wheel speed of about 360 in/sec. Trig that out in terms of bump height and a shock dyno test at 120 in/sec is equivalent to hitting a four inch bump at 50 mph for desert applications or an eight inch bump at 30 mph for MX. That's real world stuff.

The Ohlins data at 120 in/sec is pretty much the limit for what you can do with a crank dyno. Run a crank dyno beyond 120 in/sec and the continuous motion at 9.6 strokes/sec on a crank dyno will vaporize the oil in a couple of seconds. To get higher shaft speed data you have to use an EMA dyno and get all of the data in a single stroke before the oil foams out and vaporizes. Darren has an EMA dyno at Push Ind.

[sandymilo]

What does the 120 in/sec factory dyno data mean in terms of real world suspension speeds?  

For a shock the rear suspension linkage ratio is around 3:1. So, a dyno shaft speed of 120 in/sec is equivalent to a wheel speed of about 360 in/sec. Trig that out in terms of bump height and a shock dyno test at 120 in/sec is equivalent to hitting a four inch bump at 50 mph for desert applications or an eight inch bump at 30 mph for MX. That's real world stuff.

The Ohlins data at 120 in/sec is pretty much the limit for what you can do with a crank dyno. Run a crank dyno beyond 120 in/sec and the continuous motion at 9.6 strokes/sec on a crank dyno will vaporize the oil in a couple of seconds. To get higher shaft speed data you have to use an EMA dyno and get all of the data in a single stroke before the oil foams out and vaporizes. Darren has an EMA dyno at Push Ind.

 

 

Can you show the dyno graphs of runs that you have personally undertaken that prove oil vapourisation in a few seconds?

[Clicked]

High speed dyno data compared to finger press data

At shaft velocities of 120 in/sec the damping force computed by the FEA calculations are spot on with the factory Ohlins data. That implies the FEA stack deflections calculations were accurate. How close are the FEA stack deflections to the finger press data?

 

At the max shaft velocity of Ohlins factory dyno you only get about a third of the way up the finger press curve. The FEA stack deflections look pretty good compared to both the Ohlins factory dyno data and the finger press measurements.

 

But the finger press data goes way out beyond the 120 in/sec limit of the factory dyno. That makes you wonder:  

• At ultra high shaft speeds how well do the FEA calculations compare to the finger press data measured at high lift?

• What suspension velocity would you need to match the upper limit of the finger press data?

• What does the finger press data say about the shape of damping force curve at ultra high shaft speeds way beyond the limits of conventional dyno testing?

Edited March 3, 2015 by Clicked

[Clicked]

Damping force curve at ultra high shaft speeds

Replot the finger press stack deflection data in terms of shim thickness and the 4x40.3 stack was lifted to 4 shim heights and the 14x40.2 stack was lifted to just over 7 shim thicknesses. Those are large stack deflections and well into the range where shim distortions are expected to produce strong nonlinear effects in a shim stack.

 

To get a stack lift of 7 h/t the FEA calculations show you would need a shaft speed of 500 in/sec. That produces a damping force of about 5 tons. Putting that in terms of bump height and a shaft speed of 500 in/sec is equivalent to a suspension speed of 1500 in/sec (38 m/s). That is like hitting a 7 inch bump at 140 mph or a 4 inch bump at about 200 mph. That kind of stuff just doesn’t come up in suspension tuning. But it is good to know the FEA calculations are capable of matching finger press data at those ultra high 38 m/sec suspension speeds.

 

Dyno testing 5 tons of damping force means you would need a 700 hp dyno. Putting 700 hp of work into the shock would vaporize the oil in about 1/4 sec. That is the basic problem with a crank dyno. You can't test at high speed without foaming out the oil.

 

Can you show the dyno graphs of runs that you have personally undertaken that prove oil vapourisation in a few seconds?

None! But here's the deal: Work is force times distance. A shock dissipates shaft work as heat. No arguing that point, it's just basic physics. Google on work conversion and the ASME shows one hoarse power produces 0.7067871 btu/sec of work dissipated in the shock. At 700 HP that works out to 495 btu/sec. If you do not get that answer in test you did something wrong.

 

Using the finger press MXScandinavia was able to measured stack deflections that are pretty far beyond the limit of practical suspension tuning. But that's OK. The whole point of testing is to figure out how suspensions behave. Understanding that behavior when the shim stack is pushed beyond practical tuning limits is useful to know. In this case the measured data shows there is no sudden increase in stack stiffness at high lift and no unexpected non-linearity at high stack deflections as is commonly thought. The FEA calculations on that shim stack show the same thing.

 

So what does the finger press data say about shim factors? If shim factors are going to work for anything it is this simple case of replacing a straight stack of face shims. The 4x40.3 stack was supposed to be 2% softer. The finger press data, at a stack lift of 1.2 mm, shows the stack was actually 21% softer. Shim factors missed the stack stiffness by a factor of 10 at high lift.

Edited March 3, 2015 by Clicked