×
INTELLIGENT WORK FORUMS
FOR ENGINEERING PROFESSIONALS

Log In

Come Join Us!

Are you an
Engineering professional?
Join Eng-Tips Forums!
  • Talk With Other Members
  • Be Notified Of Responses
    To Your Posts
  • Keyword Search
  • One-Click Access To Your
    Favorite Forums
  • Automated Signatures
    On Your Posts
  • Best Of All, It's Free!
  • Students Click Here

*Eng-Tips's functionality depends on members receiving e-mail. By joining you are opting in to receive e-mail.

Posting Guidelines

Promoting, selling, recruiting, coursework and thesis posting is forbidden.

Students Click Here

Jobs

How to accurately measure rotation?

How to accurately measure rotation?

How to accurately measure rotation?

(OP)
I need to know the type of component I should use to determine the rotation of a shaft. Basically, I'll be building a pendulum which will hang down at 0' (due to gravity) with an axle at the top of it where I'll also mount the sensor (basically I'll look like the face of a clock with only one arm that always points to 6o'clock, due to gravity as I say, it'll only need to move in 2D). If I was to push that pendulum in an anti-clockwise manner, I want to calculate the positive rotation angle as it moves, likewise the rotation angle will be negative when it passes the 0' point in the other direction.

I'd like to use a mechanical method, such as a rotary encoder, rather than optical methods which don't seem to be as applicable to my application.

My problem here is that I don't know what attributes of a rotary encoder I need. Being able to measure 360' would be a plus, although I don't expect the pendulum to do a full rotation, being able to measure at least every 1' is necessary, anything more accurate would be a plus, and I need the shaft to rotate freely, not to 'click' round like I've seen in my switches which come in the same physical package.

I've been looking at something like this:

http://uk.farnell.com/bourns/pec11-4220f-s0024/incremental-encoder/dp/1653381?Ntt=PEC11-4220F-S0024

Would that be viable?

Cheers!

RE: How to accurately measure rotation?

Quote:

My problem here is that I don't know what attributes of a rotary encoder I need.

Yep, that's a problem.

No one can help you until you figure out what you need in a quantifiable way.

Encoders are pretty much commodity items.  Start reading data sheets and ask yourself "Is that feature useful or necessary for me?  How about that feature?"

RE: How to accurately measure rotation?

So, at the minimum, we can at least write down the various sensors that could do some sort of rotation measurement job:
> gyros
> optical encoders
> capacitive encoders
> synchros
> other inductive sensors
> resistors
> magnetometers
> accelerometers

Once you figure out what it is that you're trying to measure, resolution, accuracy, etc., you can narrow down the range, or not.
 

TTFN

FAQ731-376: Eng-Tips.com Forum Policies
Chinese prisoner wins Nobel Peace Prize

RE: How to accurately measure rotation?

(OP)
Cheers for the response, guys. Basically I'm actually using the pendulum as a 'golden standard' to determine how accurate some IMU's I have are. So using gyro/accelerometers wouldn't make for a very good comparison.

I've been looking at mechanical methods for determining the angle as my mind is making me believe that nothing could be more accurate than a physical representation giving the angle. I'm not concerned with velocity nor acceleration, just the angle.

In terms of numbers for resolution I'm still clueless at this time, doing lots of Googling, but I basically want to be able to calculate the angle several times a second as the pendulum swings.

A friend of mine has suggested using potentiometers as they are apparently the most accurate and are susceptible to poor resolution and bad calculations of velocity/acceleration (not that I'm concerned about them anyway), he's also suggested servo-potentiometers.

I think he's slightly biased towards this method as he's used it before, but the calculations to determine the angle seem pointless and unnecessary as using a rotation encoder works almost identically to potentiometers (in terms of mechanics and physics) but can determine the rotation angle much easier.

I've been looking at absolute rotation encoders as they seem to be just what I need, but what do you guys think?

RE: How to accurately measure rotation?

would it help to convert the rotation motion into a linear motion ?

how about sine/cosine ? ie measure one (or both) sides of the triangle ??

RE: How to accurately measure rotation?

You have a couple of not-too-expensive options:
  • A simple pointer mechanically attached to your rotary shaft and indicating onto a scale, or alternatively a scale on a disk attached to the shaft with a stationary pointer.  Both with appropriate graduations.
  • An electronic rotary encoder with sufficient pulses-per-revolution specification, directly connected to rotary shaft or attached with something like timing belts & pulleys to give you a higher resolution.  You'd have to output the encoder to an appropriate device of some sort to read the pulses.
  • An analog potentiometer attached to shaft.  Also would require sufficient hardware to read the signal and convert to something useful.
  • ...others, I'm sure, but....
First things first, though.  Before you go any further, you need to do your homework and scope your requirements:
  • range
  • resolution

TygerDawg
Blue Technik LLC
Virtuoso Robotics Engineering
www.bluetechnik.com

RE: How to accurately measure rotation?

Use a rotary table intended for milling machines. It's already calibrated and will do everything you require including supporting your IMU. You just turn a hand wheel to whatever angle you desire and it will hold that angle very accurately. It's as compact and portable as any alternative and you can easily resell it later.

RE: How to accurately measure rotation?

Rotation -> linear... Use a gear on the pendulum pivot point coupled with a linear "gear" (think rack and pinion.) Then you can decide on potentiometers or other means of measuring/recording the linear position.

The larger diamter gear on the pivot point, the higher linear displacement (analogous to resolution) you can achieve.

Not knowing the size of your pendulum, this might be a little too obtrusive due to the friction of that assembly.

RE: How to accurately measure rotation?

This is quite easy and inexpensive to do, measure shaft rotation.  Get a small magnet and embed it to the rotating shaft.  In the housing, a Hall Effect Sensor measures the magnet passing by as an electrical spike, i.e. oscillating magnetic field generates an electrical field. Done.

Componets are bought off-the-shelf or online a website hardware stores.  You can probably get to it by your website browser.  This is a dead simple problem.

Kenneth J Hueston, PEng
Principal
Sturni-Hueston Engineering Inc
Edmonton, Alberta Canada

RE: How to accurately measure rotation?

'golden standard' to determine how accurate some IMU's I have are"

But, you haven't said what accuracy level you want.  What parameter are you even trying to measure?  And, unless you're building a replacement for the Minuteman, it's highly unlikely that you've got an IMU that cannot be compared against other IMUs.

TTFN

FAQ731-376: Eng-Tips.com Forum Policies
Chinese prisoner wins Nobel Peace Prize

RE: How to accurately measure rotation?

You could possibly just meaure the horizontal displacement of the pendulum and apply a little trig.  Linear encodoers are not hard to come by.

RE: How to accurately measure rotation?

if you measure the linear displacement the resolution/accuracy of your results changes with the angle. This may be a problem.

NX 7.5
Teamcenter 8

RE: How to accurately measure rotation?

1gibson- using gears might be detrimental to accuracy because of the play between gears (esp. when pendulum changes direction). There are ways of fixing it, but might not be worth the trouble...

RE: How to accurately measure rotation?

"if you measure the linear displacement the resolution/accuracy of your results changes with the angle. This may be a problem"

Yes.  A problem solved by trigonometry, as Tick suggested.

Cockroach's idea is a good one.  If you need to know positively which direction the pendulum is swinging, use two hall effect sensors at some angle to each other.

RE: How to accurately measure rotation?

I agree with Cockroach. I have done this before with much success. The more magnets you have the more accurate the reading is. The components can be all together less than 5 bucks too...

peace
Fe

RE: How to accurately measure rotation?

>Oh, I forgot to note that the iPad2 has compass, gyro, and accelerometer onboard, along with an app that does recording. You could tape the iPad to the shaft and record away.

That seems quite inconvenient given the size of the iPad. How accurate is it?

RE: How to accurately measure rotation?

It was a joke smile

peace
Fe

RE: How to accurately measure rotation?

Very true IRstuff

peace
Fe

RE: How to accurately measure rotation?

(OP)
Hi all,

Sorry for the delay in response, wasn't working yesterday.

Lots of good suggestions, thanks all for that. I've looked into hall effect sensors, which look promising, but unfortunately due to magnetometers on-board the IMU (although I'm currently not using them at the moment, they will be incorporated in the future) the magnets will completely throw off the results.

I think the iPad is rather impracticable for my application, one because of its size, and two because it uses accelerometers and gyroscopes and would only be as accurate as my IMU anyway. Using accelerometers and gyroscopes to measure the accuracy of accelerometers and gyroscopes seems a little redundant to me.

The gear suggestion, linear displacement, and use of trig, are also good suggestions, but ideally I want a digital result which can be recorded and then compared with the IMU results using some sort of software.

Apologies for my lack of knowledge in terms of resolution and  range that I need at this time, truth of the matter is I just don't know. Will continue to look into it and will let you know when I know a bit more.  

From my searches as what you guys have said I still think an encoder of some sort is the way forward; absolute rotary or optical.

RE: How to accurately measure rotation?

For linear measurement, couldn't you use a gear
and a belt in tension.  No backlash that way.

RE: How to accurately measure rotation?

"but ideally I want a digital result which can be recorded" ... you could video yourself taking the measurements, doing the calcs !  

RE: How to accurately measure rotation?

Quote (btrueblood):

"if you measure the linear displacement the resolution/accuracy of your results changes with the angle. This may be a problem"

Yes.  A problem solved by trigonometry, as Tick suggested.

Cockroach's idea is a good one.  If you need to know positively which direction the pendulum is swinging, use two hall effect sensors at some angle to each other.
trigonometry is the exact reason you'll get a change in accuracy.

Assume you measure the horizontal displacement of a rotating beam. What you'd actually be measuring is the cosine of the angle, multiplied by it's length. Since the relation between cosine and length isn't linear, your accuracy will be bigger near the middle (=vertical positioning) of the swing, and smaller near the extremities.

Or am I missing something?

NX 7.5
Teamcenter 8

RE: How to accurately measure rotation?

So use two transducers at 90 deg.  Assumes the OP will be swinging his pendulum at fairly high angles. Within small angles, say +/- 15 degrees, the cosine error can darn near be neglected.

RE: How to accurately measure rotation?

You haaven't even described what parameter you're trying to measure.  

And again, unless you've got a strategic missile class IMU, there are plenty of IMUs that are capable of being used as transfer standards.  Even the Earth can be used as a rate standard at 72.7 urad/s rotation rate.  If your IMU cannot measure Earth rate or see Earth rate, then there are indeed tons of IMUs and gyros that can be used as transfer standards.

And if you're truly serious about using a pendulum for measurement, then your IMU is not as good as you infer.

TTFN

FAQ731-376: Eng-Tips.com Forum Policies
Chinese prisoner wins Nobel Peace Prize

RE: How to accurately measure rotation?

(OP)
What parameters? What more do you need to know? I've stated I'm interested in measuring rotation angles. It's as simple as that. At this time I don't know resolution but am merely asking for suggestions on technologies that could be up to the job.

An IMU will be strapped to a pendulum arm and swung, by hand, to any random degree, 1DOF to start with. As I have stated, I don't expect it to do a full rotation, but the device would ideally be able to do one full rotation. For example, I push the pendulum arm in one direction, it records all the angles, from 0 (pointing downwards with gravity), to a maximum angle, say 67 degrees (depending on how hard I push it, obviously), the results will be digitally read and recorded on a computer. The IMU will also be recording at the same time, then the results will be compared, to see if they both obtained the same maximum angle at the same time.

Again, I'm not interested in measuring the IMU with accelerometers and gyroscopes, they'll be subjective to exactly the same real-world problems the IMU will be applied to, it doesn't make sense, what would I then use to calibrate the accelerometers/gyroscopes which are then suppose to calibrate the IMU?

I see no reason why the pendulum idea wouldn't be a good idea, even if it's considered an initial test. What would you suggest to determine the accuracy of my IMU's, IRstuff? Chances are my IMU's are very accurate already, but I won't know that until I compare them with another reliable source.

Many thanks, sreid. From my findings and peers recommendations I agree that optical encoders are the way to go. I'll check that link out and get back to you.

RE: How to accurately measure rotation?

I was looking at this
http://www.asm-sensor.com/asm/product_detail.php?lang=en&det=win_abs
to measure cam angle on steam turbine.  they apear pretty ruggly made
I was thinking the the ones with housing and shaft bearing could be used to support the test pedulam
If I recall the quote was $700 for transducer and display.  even the display had an output to send to recorder

RE: How to accurately measure rotation?

You might want to explore RVDTs, an angular version of the more common LVDT position transducer and which uses the same basic principles.

As far as I know Penny & Giles, Moog, Schaevitz, maybe Yokogawa are still making them, and possibly a few aero industry suppliers too. They can form the basis of a very accurate measurement system when married up with suitable signal conditioning.
  

----------------------------------
  
If we learn from our mistakes I'm getting a great education!
 

RE: How to accurately measure rotation?

Hi richie888.  I'm in Ireland and used USDigital stuff on numerous ocasions.  Only a few days delivery.  

The Farnell one may low resolution and high friction.  Also USDigital has easy to use modules to convert encoder signals to what ever you want.  

Also take a look a Renishaw magnetic encoders.  Dont think you will get interference from edternal magnetic fields.

 

RE: How to accurately measure rotation?

The biggest question is still a resolution you are looking for. When you say you want to see if the IMU and measuring device hit the maximum angle at the "same time" that doesn't mean much.

If you use a rotary encoder that measures angle every tenth of a second and an IMU that is measuring every hundredth, reaching maximum angle at the "same time" is only accurate within 0.1 seconds. To test/calibrate the IMU to be as accurate as possible you need to have a device capable of a resolution greater than the IMU. Said another way, the IMU tells you the peak angle is at 3.56 seconds into the test and the encoder on the pendulum output says its sometime between 3.5 seconds and 3.6 seconds its not much of a calibration test.

Comprehension is not understanding. Understanding is not wisdom. And it is wisdom that gives us the ability to apply what we know, to our real world situations

RE: How to accurately measure rotation?

Quote (btrueblood):

So use two transducers at 90 deg.  Assumes the OP will be swinging his pendulum at fairly high angles. Within small angles, say +/- 15 degrees, the cosine error can darn near be neglected.
I'm pretty sure he mentioned 360° rotation. Also, since you don't know the accuracy he's willing to measure (since he doesn't even know so himself) there is no certain way to say whether or not the cosine error can be neglected.

2 transducers might solve this though, as you suggested.

It may be a lot cheaper then whatever he's looking into now.  

NX 7.5
Teamcenter 8

RE: How to accurately measure rotation?

Um, may have missed the 360 degree requirement.  I was thinking of a swinging pendulum, not a fully rotating device.

RE: How to accurately measure rotation?

use a stepper motor in reverse

RE: How to accurately measure rotation?

Rotary encoder is tough to beat, but you still need some equipment to log/count its output.  Also, you still have no idea what CPR you need.

Expanding on IFRs idea, you can use a motion-controller style stepper or servo motor to swing the IMU through a preprogrammed (and highly accurate if you do it right) motion profile.  These are either really expensive or not expensive at all, relatively speaking, because you haven't really specified what cost you are shooting for either.
 

Red Flag This Post

Please let us know here why this post is inappropriate. Reasons such as off-topic, duplicates, flames, illegal, vulgar, or students posting their homework.

Red Flag Submitted

Thank you for helping keep Eng-Tips Forums free from inappropriate posts.
The Eng-Tips staff will check this out and take appropriate action.

Reply To This Thread

Posting in the Eng-Tips forums is a member-only feature.

Click Here to join Eng-Tips and talk with other members!


Resources