How to measure jerk

You really should have posted this on April the First..you could have reeled 'em in by the bucket load.

Obviously, I could take the signal from a an accelerometer and infer jerk

I think this would be the only way.

(Must admit I'd not heard of 'jerk' as rate of change of acceleration....)

Del

I've heard that roller coaster designers measure it, but how? There's got to be something more technical than my standard chocolate milk test. That's where you convince the Missus to put on her most expensive white dress and go for a drive with you while she holds a brimming full glass of chocolate milk. When she screams, "You jerk!", you know you just did.

In the Rail Industry they refer to Jerk as "Shunt". e.g. 5G Shunt (anything underslung under the sole bar has to withstand a 5G Shunt as manditory).

Go read the last few threads I have posted to. I am strarting to become one perhaps and if you measure my rate of change then maybe that will answer your question...

That's a good way to do it. So simple and easy to make. Thanks.

It is my pleasure, please take only care to place the coil on the accelerometer base and the PM on the inertial mass it must measure the relative displacement first derivative.

Rhabe thought that the measure will be absolute and of course in such a situation his remark is correct.

I wrote a comment with the equations in order to have his correction if I made an error but I do not know why it does not appear.

I am writing this in light of some recent threads about engineering. In my opinion, this shows engineering at its very best. I thought jerk measurement would require something new that might not be very accurate.

Nick name applied a proven technology (the suspended magnet in a coil was used in seismograph heads years ago) to a new problem, and, in doing so, provided a solution that has lots of research to back it up, has techniques for dealing with bandwidth and noise and nonlinearity, and is probably available on e-bay. And, when you add RHABE's suggestion of a derivative circuit, it is clear that you can also measure SNAP. I have no idea why I would ever want to measure SNAP, but the technique is there.

Thank you Nick Name and RHABE and Del. I asked a simple (I thought) question and got an education. And, just in case anybody in a bar ever bets me I can't measure SNAP, I got 'em.

The only thing I can come up with is using a device that measures G force across time. Given a long enough run the 'jerk' would show up as a spike in the G force meter (mind you, I'm assuming such a device exists). Either that or the 'jerk' is me.

Orpheuse

Hi,

post 4 is on error: induced voltage is proportional to velocity so to the integral of acceleration.

Last post (6) shows a usable road: take a high-pass filter and remove any (?) low frequency components from measured acceleration.

In the region where the gain of the low-pass filter is proportional to frequency this filter acts as a differentiator, in this region you can use the signals as indication of "jerk".

So first think about the frequencies you are interested in: the involved mass and maximum acceleration allows some estimate.

Then get the accelerometer(s) suited for this frequency, then add the low pass filter then add an extra high pass filter that shall remove unwanted very high frequency components and noise.

RHABE

The processing is easy peasy....

An analogue differentiator is easy enough with an op-amp and if it's all digital then it's easy enough with software to look for difference between successive samples (or difference between running averages of samples)

Del

Hi Del,

shure you are right, but don't forget to limit bandwith as the differentiator will amplify the input noise by open loop gain if not cutoff at upper useful frequency.

RHABE

Unless I'm missing something (other than hair), post 4 is correct and elegant in its simplicity.

In a spring type accelerometer, the extension of the spring (position) is proportional to the acceleration acting on the mass(Hooke's law). So the first derivative of that acceleration equals the first derivative of the spring extension equals also the first derivative of the mass's position (velocity).

So, if the mass is a magnet in a solenoidal coil, the velocity of the magnet equals dB/dt which is proportional to the induced Voltage. Then all you need is a nice little amp (Thanks, Del) and you've got a jerkmeter.

Hi TVP45,

"equals the first derivative of the spring extension"

and this is velocity!

This type of accelerometer is in reality a velocity meter and is widely used as "geophones" in oil exploration to detect the waves that are triggered by test explosions.

So better to take an other type that has a readout of the flexures deflection and add either differentiation or low-pass, high-pass - filtering with frequency limitation. These two possibilities are essentially the same.

The usefulness of this moving magnet type of instrument is the possibility to connect long lines of many of these in series - today no longer a big advantage - but in the 50ies to late 80ies a must.

RHABE

RHABE,

Thanks for the reply. I think you're missing how this works; the very simplicity of Nick Name's solution made it hard for me to grasp for a few minutes.

If you take a simple spring scale and hang a weight on it and measure how far the spring stretches, that's an accelerometer. If you then measure the rate of change of that spring stretch, that's a jerkmeter.

I'm sure your way works with greater precision, but I am naturally drawn toward something I can build in the basement.

Thanks again.

I hope you don't mind me batting in.

I am generally a mechanical guy and I think along the lines of Rhabe.

What is the definition of jerk? Would in not be m/s^2 / s or m/s^3 ?

As far as I can tell it still is m/s^2. It just may reach high values, depending and the "jerk" that is applied. The next step would be a derivative of jerk in m/s^4.

Sure a rate of change in acceleration can be expressed this way, but at the end we still have Acceleration in m/s^2 as a value. I agree m/s^3 can be expressed as the rate the acceleration changes.

Phew, did I make myself clear?

Regards to all.

I think I understood you, so I'll take a shot at an answer. And, if I misundertood you, you'll tell me and I'll try again. OK?

The units of jerk are m/s³. But it does affect acceleration, just as acceleration affects velocity, and so on. Jerk is a real physical thing that we can feel and we can observe the damage it can do to machinery. There was a previous challenge question on smooth braking (something about a high school girl arguing with her driving instructor) and that's a good example. If you brake a car at constant acceleration, there will be a perceptible jerk as you reach zero velocity. See my earlier post in this thread about chocolate milk - it happens.

The second derivative of acceleration, snap, is another story. There's not even universal agreement on what the proper name is, or what it physically means. But, I now know how to measure it.

Hi RHABE,

I have collected some data from accelerometer for jerk analysis. How should I approach for jerk analysis.

I have calculated the derivative of acceleration (but I have taken only X direction as sample sample) with the sampling rate of 100Hz but the jerk so calculated seems to be very high and is not perceivable during driving.

Is it because of the noise? and How to remove such noise from data? (is low pass/high pass filtering the solution?)

Rhabe has not been online for 3 years. I doubt he will get back to you. It was a good idea to open your own thread on this one.

yea. just saw that. I have made a new thread. Thanks.

ERROR CORRECTION

Hi to all and thanks to NickName who did send me a complete calculation for this type of accelerometer:

The statements of NickName are ok for an accelerometer (mass and spring - system with low damping). This is used well below its natural ferquency.

(my post No.12 is not correct for this type of accelerometer sensor,

but valid for a different type of sensor, see below.)

Thank you NickName!

There is another use of this accelerometer: used well above its natural frequency. It is often named also an accelerometer (but this is misleading). It is measuring displacement. For this instrument my statement is correct: moving magnet is measuring velocity. (Used as "geophone").

Once more in short:

a sensor comprised of a mass and a spring inside a housing and displacement of mass to housing is measured is:

a. an accelerometer if used well below the natural frequency,

b. a displacement-sensor if used well above natural frequency.

RHABE

TV45 I thought that a jerk was a jerk irrespective of his height so why measure the idiot.

The idea of measuring jerk is interesting, but now I pose this question. Even if you measure with the modified accelerometer, How do you take in consideration the direction of the acceleration. This acceleromer has limited capability to be in only one direction of motion which may not work for the equipment you are testing. If lets say you were in a zero g enviroment, I could see 3 of these inter-twined in perfectly perpendicular to eachother to give relative x,y,z acceleration signals which can then be interpreted to account for jerk.

Meanwhile back on earth, or any other place with gravitational forces, the measurements can change based upon the location, i.e. elevation, distance away from the CG of the planet.

I know it is a lot to think about, but I am sure that in all the years that people have known about jerk, and have not come up with a "simple" solution to measuring it directly, I do not believe someone to post a world changing solution on this website.

However, if you do get it to work in your garage, please notify the Nobel prize committee of your findings and I am sure you will get recognized for your work.

Please to not take this the wrong way, I am glad to see people out there doing research on their own, that could be groundbreaking ideas, but it does not happen to everyone.

I would say, leave it to the pros at Cambridge, MIT, etc. and other research professionals to come up with the device.

Good luck, and keep CR4 up to speed on your progress, I will keep an eye out.

1- do not underestimate creativity of engineers

2- do you know where we come from ? may be from one of the mentioned organisations ? who knows?

3- it is not a research project but only an application of very well known principles from physics.

Actually, I already built a simple version and tried it. I observed jerk, but my device was too crude to calibrate (I spent less than $1 US on this so what should I expect?). So maybe I can get a Noble prize; how about Argon? Or perhaps Krypton?

How do you take in consideration the direction of the acceleration?

Hi,

if you need precise 3dimensional measurement you take 3 accelerometers - this is established since the first inertial navigation systems.

In geologic measurements most often it is sufficient to have one direction and this only roughly oriented.

"perfectly perpendicular" is not achievable as any housing or mounting is much worse than necessary. Very good accelerometers are expected to measure DC components down to 1 µg and have a diameter of 1", so calculate the allowable tilt of mounting and housing to flexible element and you will see: not possible. (The 1µg is wanted in moving vehicles: aircraft and ships!)

You make a cluster of 3 to 6 instruments, measure the cross axis sensitivities and compensate for non orthogonality. If accuracy is mandatory then many more error terms are to be measured and compensated for.

See IEEE gyro and accelerometer panel for more information.

It is good praxis to use minimum 4 of these instruments tilted as the faces of a tetrahedron to have some fail-safe information.

Below there are some accelerometer flexures we tried, very different performance.

RHABE

I just ran across this on the Purdue site. You gotta lose the vegetable oil, but otherwise it looks good, and it's cheap.

http://web.ics.purdue.edu/~braile/edumod/handseis/handseis.htm