aliced25 - 16-12-2010 at 07:35
Right,
Looking at what else has been going on in other similar threads, I've decided to throw this open.
I assume most people here have read the Scientific American Article on making a homemade Michelson Interferometer, they use dowels, a micrometer, a
lever and ball bearings to get the stage to move incredibly tiny distances in order to separate the sodium doublet.
Now, the problem with that is, anything like this is subject to physical issues relating to starting and stopping, there <s>is</s>
was no way to reduce friction to nothing so it would have been lumpy as fuck, which given the size of the steps and the
sensitivity of the measurement, would skew the results badly.
Thankfully, thanks to Messr Halbach & Co. we do now have the means at our disposal to render the stage frictionless. There are several papers on
how to levitate an X,Y stage using moving magnets (ie. attached to the stage), and stationary electromagnetic coils on the base (here too). There is even a few on using Halbach Linear arrays to keep the stage & the base separated (like this one), someone's even done a bloody PhD Thesis on it.
But, based upon the Seinfeld Episode with the NASA-Designed Pen, the one they had to spend millions to make because normal ball-point pens didn't work
in Space, while the Russians used a Pencil, I figured there was probably a simpler, less complex for the sake of complexity approach.
Let's say that our stage is 12.7mm wide (1/2") x 12.7cm long and I don't know, say the stroke is a foot long (12" or 300mm). To keep the stage in
place and levitating, we need to (1) keep it off the base (2) keep it off the right & left wall; and (3) keep it in position vertically. So we'll
need at least 4 magnets, smallish ones, N40 at least, for the base, each side and the top (ie. four on each frame). They can be small as, I mean, we
aren't going to be making this weigh several kg's are we?
So we'll need something like this - permanent magnets, diametrically magnetized, north on the bottom cylinders pointing directly at north on the top,
keeping the stage up & centrally located at the same time. I haven't worked out how to strap them in just yet, I'm thinking on it...
Attachment: IX.Optics.Heat.Electronics.5.Homemade.Interferometer.pdf (476kB)
This file has been downloaded 459 times
Attachment: interferometer.part1.skb (126kB)
This file has been downloaded 694 times
bquirky - 16-12-2010 at 08:09
well if you want to do it on the cheap..
if you can get your 'fine ajustment' range down to a few hundred microns there are plenty of solid things that you can "Bend" without encountering the
lumpyness that a traditional linear bearing are likley to produce.
But
The major problem with simpler setups is low frequency phase noise (stability). This is produced by such banal things as turbulence in the air ever so
slightly altering the path length. thermal drift if all the components of your system are not at precisely the same and constant temperature there
will be relitive motion between components. and the slightest vibration can cause chaos.
More complex interferometers get around some of these problems by using phase modulation. they will add a dynamic component to the system (like a
mirror mounted to a Pzeo stack) that is oscilated at a much higher freqency than the signal you are trying to detect this efectiley shifts your signal
up above DC away from many noise sources which can then be electricly or digitaly filtered to reveal your desired signal. there are other scheems
involving balanced detection. which splits your optical output signal across multiple detectors to isolate this or that phenominin.
There are other ways of using interferomiters that dont rely on physicaly moving the referance arm. it is posible to use a spectrometer insted of a
photodetector. the distance infomation in the sample arm becomes encoded in the frequency of the fringes of the spectrum (in the frequency domain)
thus it is not nesasary to physicly move a mirror vastly inproving speed and (in theory) sensitvity
a good explination can be found on wikipedia http://en.wikipedia.org/wiki/Optical_coherence_tomography#Sp...
aliced25 - 16-12-2010 at 16:30
Tell me more please, this sounds interesting - the major problem I'm having with designing a decent system is the moving parts, you say there is a way
around that? Obviously not using CCD/CMOS Sensors for the Mid-IR, so how am I to proceed? You've intrigued me.
Quite simply it looks like you are suggesting that birefringence could be utilized to build simplified, static FT Interferometers, I've suggested
similar projects before but been told that they were not really practicable in the infrared
[Edited on 17-12-2010 by aliced25]
phlogiston - 21-12-2010 at 16:24
I am not sure what your plan is exactly, but for accurately positioning a platform have you considered piezoelectric crystals? People have build ATM's
with these at home. In more professional settings, these can achieve atomic resolution (or even sub-atomic)
not_important - 21-12-2010 at 17:26
A problem with piezo and similar devices is that you need to move the mirror several wavelengths smoothly; the longer wavelengths for IR can make this
more difficult; then there's the issues bquirky brought up. There are, and you even referenced docs that mentioned such, designs that work with
vibrating arms - meaning the mirror rotates as well as translates, but I believe they take shaped mirrors rather than flat ones. And adding a higher
frequency modulation onto the basic movement isn't too difficult, some systems even use a pseudo-random sequence that is coherently detected, avoiding
most common background noise.
Optical materials can be a problem too. The decade over 2 to 20 um sees property changes in many materials, good achromatic performance across the
range isn't easy. Beam splitters and the like needs some care to avoid introducing artifacts or excess changes in beam intensity which can degrade
the performance of interference type interferometers by burying the reduced intensity wavelengths through exceeding the detector range. The wide
frequency range problem also makes birefringence more difficult
As an example, Fresnel rhomb retarders are used in wideband applications in the same fashion as quarter and half wave plates. Besides generally being
generally for the QUV to VNIR, they generally do not cover more than a 3:1 range at best.
BTW, pencils were out because then generate conductive dust. The Soviets had bigger rockets, used a lot of vacuum tube gear with higher level signals,
and weren't as worried about graphite in the circuitry. "simpler way" can often turn out to have complex qualifiers.