Quince - 13-8-2005 at 01:09
I'm putting together an X-ray machine to do tomography of small objects; my tube is a medical one, rotating anode, up to 300 mA /125 kVp.
I'm wondering if I could also use it to do crystallography. I'm not familiar with crystallography setups, so I'm hoping someone can
give me some idea of the appropriate beam geometry, flux density through the sample, and photon energy levels used. My idea is to photograph the
diffraction pattern on my scintillation screen and feed it into the computer.
Mr. Wizard - 13-8-2005 at 08:23
Sounds interesting. There was a special on TV the other night about using X-Ray diffraction to find the spiral nature of DNA. It seems main point of
the show was to show how these nasty male scientists had abused a talented Jewish woman who had actually made the best X-Ray photo of DNA, at the
time, but didn't recognize what it was. ?? These guys got a glimpse of the photo and figured it right out. How sneaky can you get? Bad bad boys!
Here is your Nobel Prize. Anyway in spite of the political point they did accidently manage some scientific tidbits that showed the beam was very
weak in the first experiments, which required very long exposure times; hundreds of hours. In a newer experiment they had a very intense beam focused
on a hair like crystal, which was in a cooling stream of very cold gas (fog or mist around the test crystal). So you might want to keep the beam very
narrow, even point-like, to get sharp patterns. The cooling might have been to slow molecular vibrations of the crystal or to keep it from getting
hot in the high powered beam. The only practical way I know of to get the beak narrow is with a pin hole in a lead or very dense metal plate: any
depleted Uranium in the scrap pile? Have you ever noticed the reflection off the round part of a chrome bumper casts a very sharp shadow, compared to
the fuzzy edged shadow cast by the sun?
unionised - 13-8-2005 at 11:36
I presume you know that's quite a robust Xray source you have there so we can take all the "don't zap yourself/ friends/ cat"
stuff as read.
Do you know anything about the spectrum of the tube's emision? The difraction patterns are much clearer if you have near monochromatic Xrays.
I also would think long and hard about what crystal to use (I don't think DNA would be an easy place to start).
Quince - 13-8-2005 at 19:05
Well, as to what crystals to use, I want to be able to run on the various products of my experiments to confirm I'm getting what I'm
expecting.
As to the cat, mine is 21 years old, so I don't think she has enough time left to develop problems as a result of a battery of X-rays. Indeed,
she's the perfect subject for my tomography experiments.
I didn't manage to find any good references on the details of crystallography setups.
unionised - 18-8-2005 at 12:34
"Indeed, she's the perfect subject for my tomography experiments. "
Is that what they call a cat scan?
I'm still not sure about the morality of this. Can you get hold of a dead cat to use instead?
Quince - 18-8-2005 at 15:44
It was a joke. I love that pussy.
Twospoons - 18-8-2005 at 15:50
For a test crystal try quartz. At the quartz crystal assembly factory I worked at for a while, the crystal axes were measured by x-ray diffraction
prior to cutting. I would expect there to be plenty of information around on x-ray diff for quartz, as its a common industrial process.
12AX7 - 18-8-2005 at 21:55
Doesn't quartz have a real complex structure? I doubt it matters much as long as it's regular as all crystals are, but something simpler
like sodium chloride or calcium fluoride could be a better test subject.
Tim
Mr. Wizard - 19-8-2005 at 05:48
Fused quartz would have a chaotic or complex structure, but crystalline quartz, as used in radio crystals, and in rock collections, are very orderly.
12AX7 - 19-8-2005 at 08:58
Yeah, but I mean the unit cell has like 60-some tetrahedral silica elements in it? There's a good reason it takes centuries to grow a crystal of
the stuff 
I suppose it would be better to ask, what's the diffraction pattern of cubic e.g. salt, vs. the quartz structure, vs. other lattices?
Tim
unionised - 24-8-2005 at 01:12
It didn't take centuries to grow the quartz crystal in my watch. On the other hand quartz does have a rather complex structure so I guess the
difraction pattern would be complicated too.
A rock collection might give you a nice selection of crystals to try. On the other hand, KCl will give about as simple a difraction pattern as you can
get and it crystalises quite well.
BTW, 125KV at 300mA is a lot of power, what are you going to plug it in to?
[Edited on 24-8-2005 by unionised]
Quince - 24-8-2005 at 01:22
That's not intended to be continuous, unless one was doing fluoroscopy (i.e. the tube would handle it, but I don't need it). I intend to
simply drive it by pulses from a MOSFET or IGBT half-bridge through my smaller transformer (from a dental machine), so I can get pulsed output.
Current limiting to avoid house breaker flipping is easy, just use the voltage drop across a very small resistance to turn on overload protection
transistors. A pulse is sufficient as I'm just going to use a digital camera to get the image from a scintillation screen into the computer. So
the only limit that a smaller power supply imposes is a limit on the duty cycle, and therefore the speed with which I can take a sequence of
projections for tomography purposes (though I'm guessing that the camera-computer transfer will be the more limiting here, even with a USB2
connection). I even got a nice large prism so I can have the camera on the side, out of the path of X-rays, in case some get through the
scintillation screen. I just have to build a nice stepped turntable, and I'll be making CAT scans of my cat
I've found commercial software that handles the numerics of crystallography, but nothing free. Any pointers?
[Edited on 24-8-2005 by Quince]
Lambda - 24-8-2005 at 01:35
Quince, you can puls opperate it by means of capacitor discharge. PWM (puls width modulation) may also be used, but becomes complex at these ratings.
Discharging a few capacitor banks in sequence, cumbersome, but easier maybe. They can be fired by rotating sparkgaps.
Twospoons - 24-8-2005 at 16:28
Looks like a good application for a Marx generator. You could simply use a NST for charging, size the caps for the energy, and stack 10 or so stages
to get 125kV.
The quartz suggestion was simply because its easy to obtain crystal slices with a predictable orientation, and the xray diffraction should be well
documented
Quince - 24-8-2005 at 16:35
Well, I already have an X-ray machine transformer... plus, sizing spark gaps is not fun (for me anyway).
BTW, why do algebraic tomography reconstruction papers use sums to compute the (discrete) line integrals, when absorption is an exponential
phenomenon? Shouldn't they be using multiplications? Or are they doing everything in the log domain?
My guess is something like this, so if anyone knows better, please correct me -- for a ray traversing discretized volume:
(I = intensity, _o = output, _s = source, t = transmissivity per discrete interval, dx = length of discrete volume element)
Standard constant absorption for one element:
I_o = I_s * exp(-a * dx)
Multiple elements (multiplications with same base, e, just sums exponents):
I_o = I_s * exp(-(t_1 * dx_1 + t_2 * dx_2+ ... + t_n * dx_n))
Natural log of both sides and assuming normalized I_s = 1:
-ln(I_o) = t_1 * dx_1 + t_2 * dx_2+ ... + t_n * dx_n
So I need to compute the LS for each projected image with pixels I_o before feeding into reconstruction algorithm.
Does this look right, or am I missing something?
[Edited on 25-8-2005 by Quince]
HammerOfLight - 29-5-2010 at 20:49
If your using old receiver tubes, they do produce x-rays, but they are of the soft variety, so your photo images are going to be a little blurry, if
you want sharper images, try E-BAY, you can get Russian x-ray tubes that have hot cathodes for high frequency x-rays for around $100 or less. Receiver
tubes are running in cold cathode mode, which means the electrons are interacting with some gases in the tube, producing soft x-rays, hot cathode
tubes are in a near perfect vacuum, and produces hard x-rays, the amount of power to the filament controls the intensity of the x-rays, the amount of
voltage to the anode controls the frequency.
Mr. Wizard - 30-5-2010 at 14:08
No doubt having a commercial quality X-ray tube would be nice, but for those who are suffering during these economic storms, we can turn to our junk
boxes and perhaps duplicate what this resourceful fellow has done. He claims you can generate huge megawatt pulses of x-rays from 100KeV into the MeV
range. He explains it much better than I could:
http://www.fineartradiography.com/hobbies/x-ray/vac.html
I have no connection with this site, and don't know him.
HammerOfLight - 30-5-2010 at 14:45
Oh yes, I have seen this gentleman's web site. You can also produce some x-rays using a small light bulb (nite lite, b type bulbs, etc.) and
connecting both parts of the socket to your HV source and gluing a piece of aluminum to the opposite side, careful or you'll burn a hole in the glass
after awhile, but the will produce 50-100 or more mR of soft x-rays. you can kinda turn it into a hot cathode and get more output for a little while
by putting the bulb on a rheostat, just barely turning it on until it glows orange, and then slowly and carefully ramp of the HV on the aluminum cap
glued to the end of the bulb, this way you accelerating the electrons boiling off the tungsten filament, make sure your HV is positive ionizing, not
negative. and start at around 10KV and work your way up, otherwise you could get arc over to the filament destroying it. another idea is to use a 12V
running light bulb from a vehicle and power the filament with a gel cell battery so an arc over wont destroy anything important.
phlogiston - 24-8-2010 at 11:25
You can even get x-rays from a unwinding a roll of tape in a vacuum.
http://www.wired.com/wiredscience/2008/10/video-the-scotc/
You haven't said much about what you goal is with this machine.
Do you want to use it to identify compounds? Solve structures? Of what? Inorganic compounds? proteins?
The requirements for your machine will depend on your goals. Also, are you sure the camera is linear? To solve structures, you are going to need
accurate intensity data for the various 'spots' that will appear on your screen.
If you want to solve structures the software/mathematics will be the real problem, the hardware is comparatively simple.
What material is the anode made of? You should try to obtain monochromatic x-rays. Also,
Intergalactic_Captain - 24-8-2010 at 14:14
I took a class on crystallography last fall, and let me tell you this - It is far more useful than given credit for, yet stupidly complex to
understand. The basic math can be done on (several sheets of) paper with 60 year old textbooks, but modern computer software has taken care of 99% of
that.
For a basic setup, you'll need a collimator and monochromator to bring your beam to something usable. Your tube will act as a "point source," meaning
that most of your light is at angles that you can't use - In school, our setup used a ~0.5mm beam of ~99% collimated beam - It may have been more or
less, but the idea is that you need a laser-like beam of the cleanest single-wavelength x-rays you can produce.
Beyond that, you'll need something to catch the refracted image - Either large-format film, x-ray film, or a huge and rediculously sensitive (and
expensive - had to wait for a grant to replace ours) CCD (and posphor screen in front of it). If you go the computer route, you can run and interpret
a spectrum in about a half hour with some basic knowledge and some rediculously powerful software. If you go film, basic structures can be narrowed
down by hand, though it's anything but easy. Either way, x-ray crystallography can be incredibly useful, provided you can grow a perfect single
crystal large enough (or small enough) to use. Powder diffraction is much easier, though much less useful for anything organic.
[Edited on 8-24-10 by Intergalactic_Captain]