Scanning mechanism

[unicycle1234]

Hi,

I'm working on a project that requires a scanning mechanism much like what you see in a photo copier. I intend to mount three CCD cameras (cover the cross section) on a moving frame that will sample the cross section as it moves across the object. The object will be put on a glass platform so the whole 360 degree can be scanned.

I'm pretty new to this type of thing, but I guess I'll need a programmable controller, servo, etc. Anyone can point me to the right website, company etc?

Thanks in advance.

Uni

[Chris Smith]

More information is needed here.

How big is the object.....
How fast does it move......
Continuous or non continuous.....
What sort of precision are we dealing with....
What sort of resolution do you expect.......
How far does it move in the operation.....
How often does this occur......
How hard is it to load, reuse, or use............
ETC

This can be extremely simple or very complex depending on what you actually need?

[haklesup]

Start with two Linear slide and a beam to connect them and you have the mechanical base for what you want. Next add a motor to move it. A stepper spinning a long worm gear (aka lead screw) is a good way, you can also use a linear stepper motor. Servo motors could also be used. Steppers make sense if the control signal is digital while servos work well with analog. Your project sounds PC based so the stepper makes sense, you should be able to get control software and hardware interface pretty easy.

These randomly selected sites covers much of it.

http://www.kerkmotion.com/products/slid ... erview.asp

http://www.nookindustries.com/slidesyst ... Series.cfm

[unicycle1234]

Thanks Chris and Hacklesup,

Regarding PC control, my goal is to not tie the device with a PC and software that needs to be installed etc. In fact, I would like to design a simple LCD with a start and a save button. I want to store the scanned image in a USB key, so the device is completely self-contained without its user messing with a PC etc.

More specifically.

How big is the object.....

It's about a foot long and half foot in the other dimensions.

How fast does it move......

I would like the scan to be done within 10 seconds

Continuous or non continuous.....

Continuous I guess for fast scan

What sort of precision are we dealing with....
What sort of resolution do you expect.......

Not sure yet, but as high as possible for the 10 second time.

How far does it move in the operation.....

Just enough to take the scan

How often does this occur......

Probably like a scanner in office. Not much.

How hard is it to load, reuse, or use............

Should be light, can be moved by a single person.

Thanks again.

Uni

[haklesup]

Mechanically speaking your scanner shouldn't be too much trouble. I would stick with the stepper as it can still be controled with a non PC driven circuit. There are plenty of linear slides of all sizes, with and without steppers and lead screws. A bridge between two of them to hold the camersa will probably require some custom machining unless you can find a setup already made.

That circuit would be quite sophisitcated to do without a PC (even with I suppose). It would need a micro controller, some local memory, a USB interface chip, A serial port to communicate with the stepper controller and uploading new code, a DSP chip, and some fancy firmware. The DSP is what you need to process all that video data so that you're not just storing a RAW video image on the USB stick (the file could be several Gb or larger if not pre-processed). Look for already written code libraries that come from the DSP vendor to help speed up development (another expense).

Now if you limit the file to an MPEG or AVI of the object and process it later in the PC it might be easier on the firmware and you will find more example circuits. You choose the balancing point for complexity, is it in the device or in the post processing on the PC. maybe you can find a stepper controller with a USB interface and eliminate the RS232.

An ethernet interface might eliminate the need for the USB and serial connections and would require less memory as you could stream the video to a remote PC for processing eliminating the DSP as well. On the downside, you need a local ethernet connection (wired or wireless)

[unicycle1234]

Thanks Hacklesup,

I like the suggestion of video signal, in fact it may make sense to transmit them back to host PC via wi-fi (or proprietary USB fob such as those used by baby monitor or security camera). This will make the scanner easier to build I suppose. Do you have any suggestion/pointer for where to get the camera (CCD), transmitter etc?

This really helps and thanks a lot.

Uni

[Chris Smith]

I would build mine like the common scanner, one pass, one time.

I would locate some linear slides with or with out the lead screw and stepper motor, set this up for a smooth pass on a table that is appropriate in size, and set up more than one camera and recorder for a full time 3-D photo shoot.

The speed of the plate is proportional to the camera resolution.

If it blurs slow down the plate or speed up the cameras.

Place your object on the plate and allow the plate to move under the cameras and lighting system which will have to be well lit in order not to shadow the object under inspection.

Several high wattage bulbs strategically placed should remove all shadowing for the cameras.

Also De-focused lasers can make a bright non shadowed lighting system

I have used them and microscopes to get rid of common shadows.

The whole mechanism can start the cameras, light up, start the video recorder, move from the upper stopper switch to the lower stopper switch and fully come to a halt all with out the need of any fancy micro controllers or timers.

The LCD may need a micro to show in English all of this for you, or simple use transistors, panel lights, FETS, SCRS etc with the use of limiting switches that can start and completely run the system and stop it all in sequence.

Another simple reverse switch mechanism can reset the whole mechanism back for another pass with out activating anything other than the motor in reverse.

The motor can use a set stepper control circuit set and covered complete with a pot for experimental variance. One tested it needs only to run the system at one speed from a switch there after. On/Off.

With out the LCD, everything can be simple on and off panel lights and switches, several switches over all, and limit switches at each end of the slide stroke.

Totally non micro controlled exactly like the scanner in the office, with or with out a LCD read out.

[stevech]

rent a laser or IR digitizer?

[Engineer1138]

If you're just moving a platform a fixed distance, at a fixed speed in one axis, why bother with steppers, servos, etc? Just use a regular DC motor with a gear train or belt reduction to drive it, then the only control you need is a limit switch at the end of travel

[unicycle1234]

To obtain an accurate point cloud from the scanned images obtained, I'll need very accurate location of the camera where the scanned image is sampled. I think the stepper motor will be handy for that purpose.

My question is now the following:
- since the motor speed will change (at least accelerate/decelerate), the time/frame information in the video signal is probably no use for surface reconstruction
- is there anyway to insert other information (such as step motor counter value) in the video frame before it's sent out, or I'm better off just send out the frame/location information seperately?

For calibration, I'll probably have some markers set on the background (the scanning happens in an enclosure). I think these markers may even be used for locating the camera for each frame so I don't need to send out the frame/location information afterall, and a straightforward DC motor may work.

Any comments?

[ecerfoglio]

If you allow some time / distance for the device to accelerate (and decelerate), I think that Engineer1138's idea will result in a smoother movement:

Just use a regular DC motor with a gear train or belt reduction to drive it, then the only control you need is a limit switch at the end of travel

As the cammeras (or platform) will be moving at constant speed, the time/frame information in the video signal will be usable for surface reconstruction. You only need a couple of "fixed references" (in the background) to compute the real speed of the system for calibration.

On the other hand, if you use a stepper to move the cammeras (or platform) you will have lots of vibration and you will have to add a "dampening time" between "snapshots", like

Move one step --> Allow vibration to damp --> adquire snapshot ---> next step

[Chris Smith]

Like all scanners, you need a few micro seconds and a 1/4 inch to start, level off, fully light, and then top out.

Any stepper motor set on auto will operate exactly like a motor with brushes after a few milii seconds, wont jitter or blur the camera, and do a nice job of a scan.

A nice fast ratio of gears will ensure the motor operates in the faster mode which removes the jitter. [fast speed motor]

The only question is what speed of the stepper and what frame rate of the camera are you using.

Most cameras use the 32 frame per second refresh, while the stepper motor will have several thousand pulses per second, meaning that as long as its not too slow or fast on the scanner, the camera should suffice very well.

The scan plate can always be slowed down or sped up by a few percent if the cameras don’t argue.

To locate any given timing point, do what all professionals do.

Mark zebra stipes on the plate with color coding and numbers for a reference so that you can do timing marks, scan marks, and resolution marks all based on the stripes and color coding.

A clock in the first few pictures of the frame experiment [or permanent] always helps but isn’t necessary.

100th of a second resolution is needed. [min]

The speed after the first few milii seconds [1/4 inch] will remain exactly constant, and NO jitter from the motor will be detected.

Any further feed back is individual, clocks, frame speed, etc.