SparkFun Forums

[andyfive]

Hi all,

I am looking for a way to measure small changes in distance within a defined space of 3 Meters. I have been looking at the specs for the HRLV-MaxSonar range of ultrasonic sensors which show that the resolution of these is 1mm. My issue is that with this resolution, my object will have moved outside of the 3 Meter space by the time the change in distance I which to measure has reached 1mm. My question is, is it possible to achieve a greater resolution than 1mm for distance measurement with any other form of sensor or combination of sensors (or indeed another ultrasonic sensor) or possibly increase the resolution of the HRLV-MaxSonar sensors? I would be looking for a resolution of better than 0.1mm over the 3 Meter range. I hope this is clear but please feel free to ask any questions for clarification.

Many Thanks,

Andy.

[Mee_n_Mac]

Seems you have 2 hard-to-satisfy requirements. A 0.1 mm resolution and some apparently quick update time. Does your usage necessitate a non-contact distance measurement ? How many K$s are you willing to spend on this ?

[MichaelN]

I agree with Mee_n_Mac - this is a difficult set of requirements. You could probably achieve that kind of resolution with ultrasonics, but it would take more fancy signal processing than what those simple units use. There is also the issue that ultrasonics have a fairly broad pattern, so depending on your physical layout it might be hard to get consistent results.

Very accurate commercial systems normally use lasers, I think maybe combining time-of-flight and phase to get resolution down to microns in some cases. Very expensive though.

[Duane Degn]

The ultrasound sensors work surprisingly well. Even the cheap $4 units on eBay are pretty accurate.
All ultrasound sensors are limited by the wavelength of the sound. A ultrasound sensor can not measure more precisely than the length of its wavelength (which I believe is one or two mm).

The medical ultrasound sensors use much higher frequencies (shorter wavelenths) so they can measure distances more precisely.

Parallax sells a Laser Range Finder that measures the location of a laser spot image with a small camera. The distance between the laser and the camera is one of the parameters that can limit its accuracy. It would be possible to increase accuracy of the LRF by moving the laser and camera father apart. This may require the laser to be angled a bit in order for it to be in the camera's field of view at some ranges. This kind of set up would probably have a limited range where you could measure the distance accurately.

If you were willing to use multiple lasers, you could have each laser positioned to maximize accuracy over a specific range and by selecting the appropriate laser (and algorithm to compute distance) you should be able to achieve a very accurate distance sensor over a greater distance than a single laser system.

Another possible option to having multiple lasers is to be able to aim a single laser very precisely. I think 0.1mm may be possible with such a system but it would be a very big difficult to build such a sytem IMO.

[MichaelN]

A ultrasound sensor can not measure more precisely than the length of its wavelength...

For the simple sensors, that's probably true, but you should be able to do much better by comparing the phase. Ultrasonic anemometers need to be able to detect the time of arrival with far more accuracy than 1 cycle in order to get useful resolution / accuracy.

(which I believe is one or two mm).

For a 40KHz sensor in air, the wavelength is about 8.5mm.

The medical ultrasound sensors use much higher frequencies (shorter wavelenths) so they can measure distances more precisely.

Yes, they operate in the megahertz range, but the speed of sound is much higher in tissue than in air, and this offsets some of the improvment in resolution. In any case, I'm confident such equipment uses a lot more sophistication than just time-of-arrival.

[Duane Degn]

For the simple sensors, that's probably true, but you should be able to do much better by comparing the phase. Ultrasonic anemometers need to be able to detect the time of arrival with far more accuracy than 1 cycle in order to get useful resolution / accuracy.

Thanks for the clarification and additional information.

I had computed the wavelength for an ultrasound sensor once, but apparently I didn't remember the figure very well.

[andyfive]

Hi Everyone & thank you all very much for your replies. I do need to the measurement to be non-contact. I am actually trying to determine how much spin effects the flight of an object. I have calculated that with a 0.1mm resolution, the object I am trying to measure will have had a lateral movement of 0.1mm at a distance of 1.84 meters given a spin rate 100rpm. I would ideally like to be able to determine a better resolution of spin rate than 100rpm but do not think any kind of ultrasonic sensor will have a measurement resolution to achieve this. I do not wish to use laser technology due to the high cost. I understand that there may be non contact ways of measuring spin-rate/angle directly which may be a better solution but do not know much about these. Does anyone have any suggestions?

Many Thanks,

Andy.

[fll-freak]

Just thinking out loud here...
Can you instrument the moving object? (IR or RF beacon, gyros, paint stripes, ...)
How big an object? A golf ball (my bet), a baseball, a beach ball, or something bigger?
Perhaps a camera system and do the data reduction post flight?
Could you apply a large voltage gradient across the test chamber, ground the flying object and see how much current flows? Kind of like a touch screen.
How about a similar technology to a Wacom graphics tablet (RF pulse).
What update rate do you need? A few Hz, or a few megaHz?
What is your budget? A few dollars, a few hundred, a few thousand, or more?

You still are fighting the math. 0.1mm across a 3 meter range is one part in 30,000. That means you need 15 bits of significance. Not impossible, but difficult on a hobbyist budget.

[Dorje]

If you're looking at the spin of something like a paintball or frisbee, and don't need real time data, you could use high speed photography and put a sharpie mark on the object.

Real time forms of non contact rotational speed measurement are typically light based, such as having a wheel with two holes in it. As the wheel spins, light passes through the hole and triggers a light sensor twice per rotation. Count the sensor triggers in a time period and you have the rpm. Similarly, a spinning object with no holes can have its rotational speed measured by putting a piece of reflective tape on it then counting the times a beam of light reflects off of that tape and back to a sensor.

If you don't need direct access to the data, but need/want it in real time, you can use a laser range finder. I'm a big fan of Bosch's line of products for this purpose. Low cost but very high quality. I've spoken to the company, and they have plans to release a sub-$100 model that has ports so the data can be ported to a computer.

My understanding is that you're looking at measuring something moving fast while spinning (moves 3m before the sensor you mentioned can measure a 1mm change), and that spin may cause lateral movement. Could you confirm this or clarify your goal so we can be more helpful please?

Edit: Wow, fll-freak, some of those are awesome ideas! Large capacitive chamber! Do ittttt!

[Mee_n_Mac]

Just thinking out loud here...
Can you instrument the moving object? (IR or RF beacon, gyros, paint stripes, ...)
How big an object? A golf ball (my bet), a baseball, a beach ball, or something bigger?
Perhaps a camera system and do the data reduction post flight?

I thinking you're onto a good idea here. Some photogrammetric method might work. But since we don't know anything about the test object, it's hard to come up with any concrete concepts. Is this thing in a lab or in the backyard or in orbit ?

[alex.forencich]

I am mulling over the design of a high accuracy, laser based distance sensor that could be able to measure to 0.1mm resolution or better, with several thousand measurements per second (raw, not averaged). The design is still on the drawing board, though, as it requires a good bit of RF circuitry.

If you're looking at measuring spin rate, what about paining 1/2 of said object black and the other half white, bouncing a CW laser off of it and measuring the pulse rate? That could give a very reliable indication of rotation rate. If you need to filter out background noise, the simplest way would be to get an IR laser at around 950nm, modulate it at 38 KHz, and then detect the returning signal with a standard 38 KHz IR remote receiver, such as the Vishay TSOP32338. The period coming out of that unit (assuming good SNR) will exactly correspond to the spin rate.

[andyfive]

Thanks again everyone for your replies. To answer some of the questions raised :

Can you instrument the moving object? (IR or RF beacon, gyros, paint stripes, ...)

In Theory, yes. I have looked at gyro/accelerometer solution but the G forces involved are too much for any commercially available devices to handle. Also, anything embedded in the ball would need to be small enough so as not to significanly effect the aerodynamic performance of the ball itself.

How big an object? A golf ball (my bet), a baseball, a beach ball, or something bigger?

The object I had in mind was initially a football but a golf ball would be interesting. However, any fast moving/spinning ball (e.g tennis ball) could be used to test any principles.

Perhaps a camera system and do the data reduction post flight?

I would like the system to operate in real-time or as close to real-time as possible. Not sure if a camera system can achieve this?

What is your budget? A few dollars, a few hundred, a few thousand, or more?

Budget would be ideally under 200 GBP

My understanding is that you're looking at measuring something moving fast while spinning (moves 3m before the sensor you mentioned can measure a 1mm change), and that spin may cause lateral movement. Could you confirm this or clarify your goal so we can be more helpful please?

Correct. If we take the golf ball as a base, It could be moving in excess of 150mph and potentially spinning up to 15000rpm. I would like to measure the spin at a point as close to the initial flight of the ball as practically possible to that I could then perform flight calculations based on the initial spin.

Is this thing in a lab or in the backyard or in orbit ?

I envisaged the system working in an outdoor environment with the ball being fired into a net approx. 3 meters away

what about paining 1/2 of said object black and the other half white, bouncing a CW laser off of it and measuring the pulse rate?

I see no reason why I could not mark the ball in some way to make the spin detection easier. I did however assume that a laser system would be too expensive for my budget but I may be incorrect here.

I may have a chance to visit a sensor show in the next few weeks so any suggestions as to what products to look out for and investigate further would be most appreciated

Many Thanks,

Andy.

[fll-freak]

I will go out on a limb and say that I do not think 200 GBP is going to get you a sensor system as accurate as you want especially if you want it in real time. Even if you were willing to accept an answer 20 seconds after the ball was hit it would be a tough nut to crack.

At 150MPH the ball would be in a video frame for only a few tenths of a second. Unless you had a high speed camera you would get just a few 60Hz frames. Not enough to properly and accurately measure the speed and spin.

"Westley: I mean, if we only had a wheelbarrow, that would be something."