Laser Inspection of Surfaces - Vibration Management !? 1

[collinsda]

Hi All,
hopefully someone will have experience of something like this !! - I am using a laser/camera displacement sensor to measure surface defects. The laser is accurate down to .25um so picks up quite a lot of the vibration as it is being moved over the surface (movement is achieved through an x-y table arrangement just like a machine table). Up until now this hasn't been a problem as the defects that I have been scanning have been big enough to be detected over the vibrational 'noise', however I am now trying to improve the resolution of the system. Can anyone advise me what is the best way to filter out the noise in the system WITHOUT buying expensive linear motion slides (outside dept budget I'm afraid). It seems to me that using a hi/low or bandpass filter might work well, but if anyone else has any ideas I'm open to suggestion.

Many thanks in advance,

DC.

 

[abeltio]

never heard from this before, just thinking out loud... could be your problem may be comparable to the background noise measurement?
Could you do a "dry run" without actually measuring any surface, get the value obtained as an offset and then run the sensor again (measuring) and deduct the offset?
Filtering up to your spec could do the trick but may you be masking one error with the other?
I could be totally wrong...
HTH
Saludos.
a.

 

[IRstuff]

If the vibration is coming from your measurement equipment, the only solution is to slow down the motion to reduce the noise content and to increase the amount of averaging you can do on a given position.

TTFN

 

[collinsda]

Abeltio/IRstuff,
thanks for the response. Ok, I will always have some degree of background noise, but at the moment it corresponds to an output below that of the lasers resolution which is fine. I do agree though that a reference surface would help somewhat and it is my intention to calibrate the system using this (ie, by scanning a flat surface with a finish higher than the resolution of the laser) - problem is, it's an open loop system and as a stand alone option its probably not the best idea although I will couple one with the other.
Slowing down the scan speed will more than likely reduce the noise yes, but surely if I operate at a constant scan rate (say using a simple harmonic or lissajous type motion) across the part I'm scanning, then the frequency of vibration will be pretty constant. If THAT was the case I could use a bandpass filter and make sure my sampling rate was lower than the frequency of vibration - which could also be a problem. Rather than slowing down the scan speed, I'd rather make the system more adaptive/intelligent.

 

[IRstuff]

If it's at particular frequencies, you can use a notch filter to reject only those frequencies. This would be better than a generic low-pass or band-pass.

It may also be possible to put your instrument on an isolation system that has a low enough natural frequency to reject the vibrations coming from the motion platform

TTFN

 

[Speedy]

This is just a thought but, sounds like you need one of those motion damper /isolators used in hand held cameras.

Also you could limit the scan to the middle of the path eliminating the high inertia efects at the ends of the travel.

For what it's worth....
Speedy

 

[collinsda]

Thanks lads....
At the moment I am using vibration damping but it still doesn't absorb all the vibration from the head during scanning, at least not at the resolution that I'm hoping to work from. That is a good point though to try and keep the scanning in the middle of the table to cut out the high intertia. Never heard of a notch filter before though - what exactly is it ??

 

[IRstuff]

Kind of depends on how you're processing the data; if you were collecting the data as if it were an actual profile in time, then you could apply a notch filter to remove a specific frequency.

A notch filter is essentially a narror, band-rejection filter, sort of the inverse of a band-pass filter.

TTFN

 

[collinsda]

Thanks IRStuff - now I understand - a notch filter is essentially a band stop filter right ??
Another thing I noticed on the system is that whoever set it up originally has grounded the digital ground (from the camera) directly to the analog ground. Couldn't this cause problems ?? Shouldn't I have a capacitor between the two ?? If so, how do I calculate what capacitor to use ??
Sorry for all the questions but this thing is a real can of worms.