Page 1 of 2 12 LastLast
Results 1 to 50 of 58
  1. #1
    mtbr member
    Reputation:
    Join Date
    Sep 2010
    Posts
    220

    Does anybody have Suspension Velocity Data?

    Back when I was doing race car suspension, I had the ability to data log suspension position over time, which a quick bit of calculus (happily, done by a math channel in the logger software) could turn into suspension velocity.

    With that data, you could see what kind of shaft speeds the shocks were seeing, and tune accordingly. For example, it became clear that speeds under 3 in/sec were suspension moving in response to dynamics, and over 3 in/sec was bumps in the pavement.

    I'd love to see some MTB suspension data. Does anybody have any?

    DG

  2. #2
    Save Jesus
    Reputation: beanbag's Avatar
    Join Date
    Nov 2005
    Posts
    2,731
    Maybe I'll get around to it when I feel like making some mounting brackets and buying a netbook.

  3. #3
    mtbr member
    Reputation: two-one's Avatar
    Join Date
    Aug 2008
    Posts
    594
    I'd love to work with this kind of stuff too... creating some nice charts, and using shimrestackor.com to create the perfect damper for my riding

  4. #4
    mtbr member
    Reputation: PUSHIND's Avatar
    Join Date
    Jan 2004
    Posts
    1,864
    We use data loggers here at PUSH. Here's some data from a local trail.

    Darren
    Attached Thumbnails Attached Thumbnails Does anybody have Suspension Velocity Data?-travel-bottoming.jpg  

    Does anybody have Suspension Velocity Data?-velocity.jpg  


  5. #5
    mtbr member
    Reputation: tune's Avatar
    Join Date
    Aug 2011
    Posts
    157
    I'm also interested about this information for shock tuning purposes. Excellent info Darren, thank you for sharing that! Very interesting data and charts.

  6. #6
    moaaar shimz
    Reputation: tacubaya's Avatar
    Join Date
    Jan 2006
    Posts
    9,103
    Quote Originally Posted by two-one View Post
    I'd love to work with this kind of stuff too... creating some nice charts, and using shimrestackor.com to create the perfect damper for my riding
    Shim Restackor has limited usefulness in a market full of dampers with position sensitive features, secondary compression adjusters etc.

  7. #7
    mtbr member
    Reputation:
    Join Date
    Sep 2010
    Posts
    220
    80 in/sec peaks. Wow.

    That's a lot of range to cover.

    DG

  8. #8
    moaaar shimz
    Reputation: tacubaya's Avatar
    Join Date
    Jan 2006
    Posts
    9,103
    Quote Originally Posted by RecceDG View Post
    80 in/sec peaks. Wow.

    That's a lot of range to cover.

    DG
    That's what speed sensitive damping is for

    Btw, those kind of speeds are not rare.

  9. #9
    mtbr member
    Reputation:
    Join Date
    Sep 2010
    Posts
    220
    That's what speed sensitive damping is for
    All dampers are, by definition, speed-sensitive.

    It is position sensitive dampers that are rare (like the Bilstein off-road race truck dampers)

    DG

  10. #10
    moaaar shimz
    Reputation: tacubaya's Avatar
    Join Date
    Jan 2006
    Posts
    9,103
    Quote Originally Posted by RecceDG View Post
    All dampers are, by definition, speed-sensitive.

    It is position sensitive dampers that are rare (like the Bilstein off-road race truck dampers)

    DG
    Yeah, rare, only half the shocks out there have it (Boostvalve).

    I thought it was accepted among the suspension tech community to call variable port "size" (using shims, poppets etc.) speed sensitive in the sense that as speed, and thereby flow increases, the flow area increases but whatever, if you want to be all smarty pants then let me rephrase:

    That is what shimmed dampers are for.

  11. #11
    mtbr member
    Reputation:
    Join Date
    Sep 2010
    Posts
    220
    It is accepted in the suspension tech community to call dampers "speed sensitive" only so far as that is how they work. A damper produces no force unless it is moving; unlike a spring that produces force whenever compressed. A damper has no choice but to be "speed sensitive" so calling a damper "speed sensitive" is meaningless.

    Flow area must increase with increased damper speed, because if you don't, the force curve is sharply progressive and at some point the force produced is so high that the damper effectively locks solid.

    A basic, no-frills damper scales flow area in proportion to shaft speed such that the force/speed ratio is linear. More sophisticated dampers provide one or more regions of digressive forces (a "knee" in the force curve) to better react to faster speeds. We typically digressed from 0.65 DR to 0.3 DR at about 3 in/sec in the race cars, and anything over 30 in/sec was very rare.

    A shimstack damper is one way to get a linear force curve, yes. Any mechanism that produces larger flow areas in proportion to increased shaft speed can, in theory, do the same job.

    DG

  12. #12
    moaaar shimz
    Reputation: tacubaya's Avatar
    Join Date
    Jan 2006
    Posts
    9,103
    Quote Originally Posted by RecceDG View Post
    It is accepted in the suspension tech community to call dampers "speed sensitive" only so far as that is how they work. A damper produces no force unless it is moving; unlike a spring that produces force whenever compressed. A damper has no choice but to be "speed sensitive" so calling a damper "speed sensitive" is meaningless.

    Flow area must increase with increased damper speed, because if you don't, the force curve is sharply progressive and at some point the force produced is so high that the damper effectively locks solid.

    A basic, no-frills damper scales flow area in proportion to shaft speed such that the force/speed ratio is linear. More sophisticated dampers provide one or more regions of digressive forces (a "knee" in the force curve) to better react to faster speeds. We typically digressed from 0.65 DR to 0.3 DR at about 3 in/sec in the race cars, and anything over 30 in/sec was very rare.

    A shimstack damper is one way to get a linear force curve, yes. Any mechanism that produces larger flow areas in proportion to increased shaft speed can, in theory, do the same job.

    DG
    No sh1t...

  13. #13
    Save Jesus
    Reputation: beanbag's Avatar
    Join Date
    Nov 2005
    Posts
    2,731
    The maximal shock/fork velocity data is not that interesting because all you need to do is make sure you are not cavitating or hydrolocking on the compression stroke. You can determine this even if your dynometer has a much lower maximal speed for testing. The rebound stroke also has an upper limit because of the maximum extension force of the fork (maybe about 1.5-2x body weight).

    Beyond that, tuning seems to be mostly about knowing what LCS, LSR, HSC, HSR, etc feel like on the trail. In the little time I spent on it, I haven't been able to get much out of analyzing shock velocity data like histograms, transient response, fourier transform, etc. This is why I am not a professional shock tuner, LOL.

  14. #14
    mtbr member
    Reputation: Dougal's Avatar
    Join Date
    Jan 2004
    Posts
    4,397
    Quote Originally Posted by RecceDG View Post
    Flow area must increase with increased damper speed, because if you don't, the force curve is sharply progressive and at some point the force produced is so high that the damper effectively locks solid.
    The bike industry continues to be lumbered with orifice dampers that do the above.
    Owner of www.shockcraft.co.nz and NZ Manitou Agent.
    www.dougal.co.nz Suspension setup & tuning.
    SPV Devolve

  15. #15
    moaaar shimz
    Reputation: tacubaya's Avatar
    Join Date
    Jan 2006
    Posts
    9,103
    Quote Originally Posted by Dougal View Post
    The bike industry continues to be lumbered with orifice dampers that do the above.
    We can thank Marzocchi for that!!

  16. #16
    Save Jesus
    Reputation: beanbag's Avatar
    Join Date
    Nov 2005
    Posts
    2,731
    ok DG, now that you saw some suspension velocity data, what are you planning to do with it?
    Normally, I would ask this question of Push, but I figure they won't want to give out their secret sauce recipes.

  17. #17
    Save Jesus
    Reputation: beanbag's Avatar
    Join Date
    Nov 2005
    Posts
    2,731
    Quote Originally Posted by RecceDG View Post

    I'd love to see some MTB suspension data. Does anybody have any?
    yes

    Can you say 150 ips?


  18. #18
    mtbr member
    Reputation: dklopp's Avatar
    Join Date
    May 2009
    Posts
    52
    Awesome! Is that really from a handlebar smack like the image file name suggests?!?

  19. #19
    mtbr member
    Reputation: tyrebyter's Avatar
    Join Date
    Sep 2008
    Posts
    538
    All dampers are speed-sensitive? Your world has much better suspension than mine. Orifice-tube is more common here.

  20. #20
    "El Whatever"
    Reputation: Warp's Avatar
    Join Date
    May 2004
    Posts
    18,831
    Quote Originally Posted by tyrebyter View Post
    All dampers are speed-sensitive? Your world has much better suspension than mine. Orifice-tube is more common here.
    Awww... come on...

    As explained before, all dampers are speed sensitive in strict theory as they change damping and behaviour according to shaft speed.

    However, it's a given that the term "speed sensitive" is applied to dampers that change orifice size according to shaft speed (either by shims, poppet valves, etc.)

    Let's get back to discuss expected shaft speeds, please instead on fruitless discussion about terminology... which sometimes is unique to bike industry as the same thing may be called something else in other industries.
    Check my Site

  21. #21
    Save Jesus
    Reputation: beanbag's Avatar
    Join Date
    Nov 2005
    Posts
    2,731
    That "handlebarsmack" file was me forcing a hard landing to flat, which I think is one of the higher IPS events. The reason is that hitting a flat surface maximizes the spring constant of the tire, as opposed to hitting a sharp object that pinches the tire inwards. Here is some other token data and speeds:

    Ramming a 2" curb - 50 ips
    slowly dropping off a 6" curb: 50 ips
    flying off a 6" curb: 70 ips
    standing and mashing (pedal bob): 5-10 ips

    braking and swerving:


    Just Riding Along a smooth trail


    Concrete rock garden, maybe 1-2" roughness
    compression 3 clicks underdamped:


    my usual setting:


    3 clicks overdamped:


    velocity histograms:


    Minor comments for now:
    In the JRA data you can see the horrible horrible stiction in my Minute Pro fork, where the velocity likes to go to zero between little compression and rebound strokes.

    The velocity histograms are statistics for how much time the fork spends at each velocity. So for example, if you have too much compression damping, the fork never gets a chance to reach high speeds. It is some technique from 4 wheeled motorsports which is supposed to help set your damping. I don't know that much about them, but my guess is that you are aiming for a symmetrical histogram which somehow implies that the wheel is tracking well over bumps.

    This is my baller data logging setup.
    Last edited by beanbag; 12-02-2012 at 05:39 AM.

  22. #22
    mtbr member
    Reputation: tyrebyter's Avatar
    Join Date
    Sep 2008
    Posts
    538
    OK. I'm probably like the ape in Space Odyssey staring at the obelisk, but that is the coolest thing I've seen in a long time. What is it, what does it interface to, and where did you get it?

  23. #23
    mtbr member
    Reputation: dklopp's Avatar
    Join Date
    May 2009
    Posts
    52
    Yeah more info on your set up please. I have been considering buying an arduino and accelerometer (along with some other sensors) to play with on my bike - a goal of eventually setting up a cheap data logger for suspension data.

    Got any tips?

  24. #24
    PMK
    PMK is offline
    mtbr member
    Reputation: PMK's Avatar
    Join Date
    Oct 2004
    Posts
    2,505
    Beanbag, some of our previous topics / posts I thought you were of the opinion that there was not much need for more than 50 IPS information.

    Not specific to Beanbag but everyone.
    The capability to constantly log IPS and dyno runs to at least a max of 150 is not stuff I made up. I still laugh when a shop flaunts they use a dyno of 20 IPS to build and test shim stacks and the sheep follow drinking Kool Aid along the way.

    The other part of the dyno puzzle that you must have to validate any working data is force. You can have huge IPS capability but if the dyno stalls at real world force values you will never see the true IPS.

    Here in lies the problem, how do you also gather force data? Consider also, forks are easy at a 1:1 leverage rate, rears are more difficult on account of instantaneous leverage rates. Force calculations based solely on measured weight are moot also in that you encounter weight bias constantly changing.

    Dynos and all the engineering is really cool and makes things look impressive but in the end, It still comes down to the setup needs to be ridden... and unfortunately each rider and terrain is uniquely different. So really it is nothing more than a best compromise, provided the rider is capable of offering up good feedback to make changes for the better.

    PK
    Reps! We don't need no stickin' reps!

  25. #25
    Save Jesus
    Reputation: beanbag's Avatar
    Join Date
    Nov 2005
    Posts
    2,731
    Quote Originally Posted by PMK View Post
    Beanbag, some of our previous topics / posts I thought you were of the opinion that there was not much need for more than 50 IPS information.
    I am still of the opinion that the dyno does not need to go over 50 ips, for a couple of reasons:
    The rebound will never see more than 50 (or so) ips.
    The compression curves you can extrapolate out

    It's not worth 4x the money to get 2x more speed for 5% more information value

  26. #26
    Save Jesus
    Reputation: beanbag's Avatar
    Join Date
    Nov 2005
    Posts
    2,731
    Sensor is this one:

    Linear Velocity Transducer 3 4 inch Stroke DC Voltage | eBay

    It is only good for 3.4", but if you need longer stroke, I think you can just connect two of them together.

    Laser cut some brackets out of acrylic: (faster than using the cnc machine)


    datalogger is NI USB 6008 + netbook with SSD drive.

    Right now I am putting together an arduino + SD card datalogger coz I don't like jostling the netbook around.

  27. #27
    PMK
    PMK is offline
    mtbr member
    Reputation: PMK's Avatar
    Join Date
    Oct 2004
    Posts
    2,505
    Quote Originally Posted by beanbag View Post
    I am still of the opinion that the dyno does not need to go over 50 ips, for a couple of reasons:
    The rebound will never see more than 50 (or so) ips.
    The compression curves you can extrapolate out

    It's not worth 4x the money to get 2x more speed for 5% more information value
    Agree the IPS on rebound need not be real fast.

    As for extrapolating, I am not so certain the compression results will always plot in that fashion. For a bicycle with few shims it probably will. For good suspension designs utilizing crossovers, midvalves, HSC devices and low friction coatings, the results may not be that accurate.

    PK
    Reps! We don't need no stickin' reps!

  28. #28
    "El Whatever"
    Reputation: Warp's Avatar
    Join Date
    May 2004
    Posts
    18,831
    Quote Originally Posted by beanbag View Post
    It's not worth 4x the money to get 2x more speed for 5% more information value
    Agreed for our use.

    If you're in the business of suspension or you're a racing team trying to win, it may be worthy. 5% is easily the difference between win and lose.
    Check my Site

  29. #29
    mtbr member
    Reputation:
    Join Date
    Sep 2011
    Posts
    265
    A few graphs from the Shim Restackor site:




    This is all calculated mathematically as a guideline for their tuning program. It's not a field test. I don't think they'd be able to find someone to smack a 5" bump head on at 40mph.

  30. #30
    Save Jesus
    Reputation: beanbag's Avatar
    Join Date
    Nov 2005
    Posts
    2,731
    Self-contained Arduino system up and running


  31. #31
    PMK
    PMK is offline
    mtbr member
    Reputation: PMK's Avatar
    Join Date
    Oct 2004
    Posts
    2,505
    Beanbag, very impressive. My electronics skills were left behind growing up since my dad was the go-to guy for electronic stuff. My skills let me understand how it works overall, but build only simple circuits when needed.

    Does that small unit collect and store the data, or is it an interface for a laptop?

    If you build another and want to get some serious numbers, the Ventana tandem at speed may see the bike IPS values, assuming you would lend it out.

    I should just find a Shockclock and data log for grins.

    PK
    Reps! We don't need no stickin' reps!

  32. #32
    Save Jesus
    Reputation: beanbag's Avatar
    Join Date
    Nov 2005
    Posts
    2,731
    The Arduino is a microprocessor system, i.e. mini computer. It also happens to have analog-to-digital converters, so it can measure signals. I added an attachment to let it store data to a micro SD card. To transfer the data, you can yank out the card, or connect a USB cable. It can be battery powered, so no need for any external connections while riding.

    It is actually not all that much work to get it up and running. You mostly buy the pre-made parts and do some programming / modifying code others already wrote. The circuits part of it is pretty minor.

  33. #33
    mtbr member
    Reputation:
    Join Date
    Feb 2010
    Posts
    1,756
    Nice Beanbag. What are you using for transducers?

  34. #34
    Save Jesus
    Reputation: beanbag's Avatar
    Join Date
    Nov 2005
    Posts
    2,731
    see post 26

  35. #35
    mtbr member
    Reputation: PUSHIND's Avatar
    Join Date
    Jan 2004
    Posts
    1,864
    Pretty slick!

    Darren

  36. #36
    mtbr member
    Reputation: nybike1971's Avatar
    Join Date
    Oct 2005
    Posts
    1,875
    Last year I put together a suspension data acquisition system based on an Arduino sampling at 500Hz and logging to SD card. The position sensors are linear potentiometers by Active Sensors. I wrote the data analysis software in Matlab, drawing from the Shockclock user interface and adding other capabilities I was interested in.

    This Winter I am planning on bringing the sampling rate to 2kHz per channel using fast ad high resolution ADCs, add GPS logging capability (position and bike velocity), and improve the overall ergonomics of the setup for broader mounting opportunities on a range of bikes.

    Here are some pictures of my setup:



    The guts of the logging box:



    Mounted on the bike:







    An example of suspension position and velocity trace (Reba 29er 120mm RLT Ti fork):



    Position and velocity histograms and cumulative distribution functions (integrated histograms):



    This is a way I like to look at suspension data that is pretty common for analysis of oscillators. It's called phase space and you have position on the x-axis versus velocity on the y-axis. Time becomes a parameter along the plotted trajectory. Positive velocity is compression and negative is rebound. You can see how rebound speed and position are tightly correlated (spring drives the rebound circuit, unlike on the compression side). On the compression side one can see the evolution of "fast events" (high velocity) some of which become deep (lots of travel used), but the deepest events are not necessarily the fastest. The distribution of events is also interesting and can be used for tuning.

    Last edited by nybike1971; 12-07-2012 at 07:33 PM.

  37. #37
    mtbr member
    Reputation:
    Join Date
    Sep 2011
    Posts
    265
    The phase space graph is very interesting.

    I noticed many of the high shaft speed/low travel events seem to have a sharp edge at which deceleration begins, rather than a smooth curve like the mid/low shaft speeds. Do you figure this is because of hydro-lock in the damper at high shaft speeds?

  38. #38
    mtbr member
    Reputation: Dougal's Avatar
    Join Date
    Jan 2004
    Posts
    4,397
    Can you share any more details on your arduino? I really need to get into those to solve a few completely unrelated instrumentation problems.
    Owner of www.shockcraft.co.nz and NZ Manitou Agent.
    www.dougal.co.nz Suspension setup & tuning.
    SPV Devolve

  39. #39
    Save Jesus
    Reputation: beanbag's Avatar
    Join Date
    Nov 2005
    Posts
    2,731
    What details do you need? It most likely is not the best tool for the job.

  40. #40
    PMK
    PMK is offline
    mtbr member
    Reputation: PMK's Avatar
    Join Date
    Oct 2004
    Posts
    2,505
    So then,and don't get me wrong since I am in favor of what data you guys are gathering, let's go to the next step. You have real world runs illustrating, position, IPS, but no force. So in the big picture, the IPS and such need to be run into a dyno run with the suspension components, and the dyno run will provide force values.

    The force values along with IPS will allow a better guess at what to change.

    On the Yeti, for the rear suspension, I believe you also need a third transducer on the rear damper. This way you can overlay the wheel travel over damper travel and build the true plot of the rear end movements.

    The data collected is good, but without a worthy test rider providing accurate feedback on feel the information is incomplete.

    I truly enjoy it all and not being negative, rather would like to see your ideas in the overall picture.

    Beanbag is gathering data and tuning / test riding his own stuff. Is NYBIKE building his own suspension or having others work that for him?

    What would really make this all tie together is to see the force spikes like a dyno, except while riding. That takes the test rider somewhat out of the equation. If the data collection unit, and it should be able to, could read load cell or strain type measurements via the shock mount bolt on the frame, or cartridge rod inside the fork, this would provide the three dimensional plot to get the suspension dialed in. In other words, the actual bike / rider is the dyno. The need for a dyno would be moot except to replicate matching setups for suspension spares.

    PK
    Reps! We don't need no stickin' reps!

  41. #41
    "El Whatever"
    Reputation: Warp's Avatar
    Join Date
    May 2004
    Posts
    18,831
    Quote Originally Posted by PMK View Post

    What would really make this all tie together is to see the force spikes like a dyno, except while riding. That takes the test rider somewhat out of the equation. If the data collection unit, and it should be able to, could read load cell or strain type measurements via the shock mount bolt on the frame, or cartridge rod inside the fork, this would provide the three dimensional plot to get the suspension dialed in. In other words, the actual bike / rider is the dyno. The need for a dyno would be moot except to replicate matching setups for suspension spares.

    PK
    In an air shock/fork, I can easily see a pressure transmitter collecting pressure change (force) data.

    More complicated for a coil shock/fork, but as you say, it can be done.

    The GPS idea is pretty cool too..

    Another parameter that could be collected by available methods are wheel velocity (tells you if your wheel is tracking or skidding) and power. You could see the relationship between chain tension and power over the suspension.

    Wow... that escalated in sampling channels really quick!
    Check my Site

  42. #42
    PMK
    PMK is offline
    mtbr member
    Reputation: PMK's Avatar
    Join Date
    Oct 2004
    Posts
    2,505
    Quote Originally Posted by Warp View Post
    In an air shock/fork, I can easily see a pressure transmitter collecting pressure change (force) data.

    More complicated for a coil shock/fork, but as you say, it can be done.

    The GPS idea is pretty cool too..

    Another parameter that could be collected by available methods are wheel velocity (tells you if your wheel is tracking or skidding) and power. You could see the relationship between chain tension and power over the suspension.

    Wow... that escalated in sampling channels really quick!
    Problem with reading pressure change or spring change is that it dos not tell force. The measurement needs to be taken via strain gage or loadcell on the actual damper rod or equivalent.

    Pressure change can be calculated via position and rate or base pressure to compression ratio to position.

    If the bike feels harsh, then that is a force transmitted to the rider and can be a spike in damping, this assumes the basics of proper spring rates have been justified and applied.

    PK
    Reps! We don't need no stickin' reps!

  43. #43
    mtbr member
    Reputation: nybike1971's Avatar
    Join Date
    Oct 2005
    Posts
    1,875
    Quote Originally Posted by Matty F View Post
    The phase space graph is very interesting.

    I noticed many of the high shaft speed/low travel events seem to have a sharp edge at which deceleration begins, rather than a smooth curve like the mid/low shaft speeds. Do you figure this is because of hydro-lock in the damper at high shaft speeds?
    The sharp edge is because of the sampling rate. At higher shaft speed, this sampling rate fails to achieve a nice smooth curve in phase space (shaft position changes significantly between two sample). This is the reason why I want to use fast ADCs to achieve a 2kHz sampling rate.

  44. #44
    "El Whatever"
    Reputation: Warp's Avatar
    Join Date
    May 2004
    Posts
    18,831
    Quote Originally Posted by PMK View Post
    Problem with reading pressure change or spring change is that it dos not tell force. The measurement needs to be taken via strain gage or loadcell on the actual damper rod or equivalent.

    Pressure change can be calculated via position and rate or base pressure to compression ratio to position.

    If the bike feels harsh, then that is a force transmitted to the rider and can be a spike in damping, this assumes the basics of proper spring rates have been justified and applied.

    PK
    It sees the force at the spring end, but not on the damper shaft, which is where you'd want to see it. I get it now.

    Indeed, you'd want a strain gauge at the shaft and maybe an accelerometer on it but I suppose you can get away with only the strain gauge.

    The pressure gauge could tell you if there's air/gas expansion that is not related directly with shaft position and would be a more accurate indicator of the force excerted over/by the spring. In a coil over that would be moot and you can get away without it.
    Check my Site

  45. #45
    mtbr member
    Reputation: nybike1971's Avatar
    Join Date
    Oct 2005
    Posts
    1,875
    Quote Originally Posted by PMK View Post
    So then,and don't get me wrong since I am in favor of what data you guys are gathering, let's go to the next step. You have real world runs illustrating, position, IPS, but no force. So in the big picture, the IPS and such need to be run into a dyno run with the suspension components, and the dyno run will provide force values.

    The force values along with IPS will allow a better guess at what to change.
    That's what the fancy EMA dyno Darren just bought is for! I wish I had access to one of those toys.

    What I use for tuning so far are the dynamic ride height, the shaft position histogram and CDF (you may achieve full travel once and almost never use more than 60% of travel, as was the case on my Reba because of how the Floodgate was set). Velocity histogram provides some information about the threshold between high and low speed.

    I am still experimenting and figuring out how to data mine this logs and how to translate the data into tuning actions.

    Quote Originally Posted by PMK View Post
    On the Yeti, for the rear suspension, I believe you also need a third transducer on the rear damper. This way you can overlay the wheel travel over damper travel and build the true plot of the rear end movements.

    The data collected is good, but without a worthy test rider providing accurate feedback on feel the information is incomplete.
    That is correct. After mounting the rear position sensor, I remove the shock and put a short position sensor between the mounting bolts to map the motion ratio in the laid down geometry (using Racetech's language).

    I also use this setup to measure the leverage ratio of linkages of bikes. Here is an example for a Yeti SB-66:



    So far, I have only used my data collection system on my bikes. I would like to collect some data with local DH racers next season. I tore my ACL last April on the first day on the DH bike and that slowed down the whole project. I hope to have more interesting data and analysis methods next season.


    Quote Originally Posted by PMK View Post
    Is NYBIKE building his own suspension or having others work that for him?
    I assume you are asking if I am the pilot and as I said so far the answer is yes.

    Quote Originally Posted by PMK View Post
    What would really make this all tie together is to see the force spikes like a dyno, except while riding. That takes the test rider somewhat out of the equation. If the data collection unit, and it should be able to, could read load cell or strain type measurements via the shock mount bolt on the frame, or cartridge rod inside the fork, this would provide the three dimensional plot to get the suspension dialed in. In other words, the actual bike / rider is the dyno. The need for a dyno would be moot except to replicate matching setups for suspension spares.

    PK
    I have been thinking about statistical analysis methods that could approximate your ideas over large datasets. If you can characterize the harmonic content of a specific riding location, just like tire manufacturers do for roads, you could use this to deconvolve the data statistically and have an average "dyno" curve for the the rider and location. Still scratching my head around this one and trying to see if there is a way to back that information out.

    The other measurement I have been thinking about adding to the logging tool is an accelerometer/gyro/magnetometer connected to the sprung mass and unsprung mass (main frame of the bike) to measure force transfer through the suspension. This could be up to 9 channels per sensor, so the data logging capability (and especially data writing) would have to be brought up quite a bit. Kalman filtering techniques used in dead-reckoning applications could be used to put together the acceleration and rotation data to extract attitude and acceleration in the plane of the frame at all instants of time. This combined with GPS data could also allow you to look at behavior of the suspension under extreme yaw angles (drifting).

    So much to do! If only I didn't have a dayjob!

  46. #46
    PMK
    PMK is offline
    mtbr member
    Reputation: PMK's Avatar
    Join Date
    Oct 2004
    Posts
    2,505
    As cool as his EMA dyno is, my opinion is your data collection IPS will be greater than the max IPS the dyno can accomplish.

    As for having accelerometers, I don't believe you need more than what you already sample. IPS plotted over a timeline will provide that data.

    I'm pretty certain that with minimal effort a double element shock bolt could be fabricates with a loadcell capability.

    The fork would be more difficult, but may be possible by making a fork cap and damper rod extension to build the forks loadcell into. This would be modular and external. Provided the bars cleared it would be a longer fork leg with wires out of the top.

    With that data, the dyno is more of a toy to save time when testing stacks. Also, your method surpasses the dyno in it's ability to generate true IPS, in a random sampling. Let's say braking bumps of 20,35,100,60,150,20,85,then it smooths out exiting the turn 25,15,10,10,12,15,10... I know the EMA is good, but not so certain it will be able to run that random of an IPS plot over a few seconds time, or 20 feet of race track.

    PK
    Reps! We don't need no stickin' reps!

  47. #47
    Save Jesus
    Reputation: beanbag's Avatar
    Join Date
    Nov 2005
    Posts
    2,731
    Quote Originally Posted by nybike1971 View Post
    The sharp edge is because of the sampling rate. At higher shaft speed, this sampling rate fails to achieve a nice smooth curve in phase space (shaft position changes significantly between two sample). This is the reason why I want to use fast ADCs to achieve a 2kHz sampling rate.
    IMHO you should move to a LVT system. Pots are noisy, and then when you differentiate to get velocity, it gets worse.

  48. #48
    mtbr member
    Reputation: nybike1971's Avatar
    Join Date
    Oct 2005
    Posts
    1,875
    Quote Originally Posted by beanbag View Post
    IMHO you should move to a LVT system. Pots are noisy, and then when you differentiate to get velocity, it gets worse.
    I am using a high-order numerical differentiation formula (7 taps instead of 2 taps), which is equivalent to 2 taps convolved with a smoothing kernel. It's quite stable and doesn't compromise resolution too much.

    The resolution with the 10bit ADC of the arduino is in theory 0.2mm but in practice mechanical mounting is only good to a couple of mm's and that's the dominant source of measurement error.

    LVT are nice but significantly more expensive than linear pots. The Active Sensor pots are very well manufactured and packaged for motorsports applications. Mechanical stability is excellent and readings are very reproducible. If nothing moves, the reading is stable to the least significant bit of the ADC.

  49. #49
    mtbr member
    Reputation: nybike1971's Avatar
    Join Date
    Oct 2005
    Posts
    1,875
    Quote Originally Posted by PMK View Post
    As cool as his EMA dyno is, my opinion is your data collection IPS will be greater than the max IPS the dyno can accomplish.

    As for having accelerometers, I don't believe you need more than what you already sample. IPS plotted over a timeline will provide that data.

    I'm pretty certain that with minimal effort a double element shock bolt could be fabricates with a loadcell capability.

    The fork would be more difficult, but may be possible by making a fork cap and damper rod extension to build the forks loadcell into. This would be modular and external. Provided the bars cleared it would be a longer fork leg with wires out of the top.

    With that data, the dyno is more of a toy to save time when testing stacks. Also, your method surpasses the dyno in it's ability to generate true IPS, in a random sampling. Let's say braking bumps of 20,35,100,60,150,20,85,then it smooths out exiting the turn 25,15,10,10,12,15,10... I know the EMA is good, but not so certain it will be able to run that random of an IPS plot over a few seconds time, or 20 feet of race track.

    PK
    I am not advocating using accelerometers to measure force on the shock, but to measure transmissibility from the unsprung to the sprung mass.

    I will look into your suggestions of bolts with a load cell, that sounds really cool and would open up a whole new range of data collection.

  50. #50
    Save Jesus
    Reputation: beanbag's Avatar
    Join Date
    Nov 2005
    Posts
    2,731
    Here's some more pictures of what the setup looks like.



    This is the inside of the box. Arduino + SD datalogging shield. The circuits are just an instrumentation amp and two followers. Yay for hot glue gun for extemporaneous strain relief connectors.




    This is a little switch/indicator module I made. I figured that instead of one big dataset, I would toggle the switch for only capturing the interesting stuff. The display shows me which dataset I am logging to, so it will be easier to keep track. It is made from a LED display and a MAX7219 chip. The jumble of wires is just connecting pin A to pin A between the MAX and display pins. There is already a library written to drive this display, so it is not that much work to get it up and running.



    The whole thing barely fits in a waterbottle cage.



    A little 3-axis accelerometer mounted behind the headtube. This is the MMA7341 with analog outputs, but in retrospect I should have gotten the ADXL345 digital one instead.

Page 1 of 2 12 LastLast

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •