| Weighty thoughts on weighing |
[Dec. 22nd, 2009|01:28 pm] |
I really need to rethink my approach to building a sensitive balance. As I said in a previous post I'm doubtful of success with using quartz crystals as sensors because, on the evidence I've gathered, they do not oscillate reliably under the kinds of loads I'd like to apply to them. I've also uncovered data suggesting that my next brilliant idea, determining the resonant frequency of a length of Nichrome wire under tension when built into a magnetostrictive oscillator, may not work so well because Nichrome exhibits bad "creep" under continuous loads. That is to say, when subjected to strain lower than the "yield point" (the quantity of strain that immediately produces permanent deformation in a substane), Nichrome slowly deforms anyway. Also I've perused G. W. Pierce's paper on magnetostrictive oscillators and learned that he only obtained strong oscillations from large masses of magnetostrictive metal, e.g. long rods of Monel metal. Small pieces of metal, Pierce stated, produced only feeble oscillations.
Have I mentioned, by the way, that winding coils is probably one of my least favorite activities in assembling electronic equipment? Also Pierce's circuits for higher-frequency magnetostrictive oscillators call for the use of a variable capacitor. The variable capacitor is, very possibly, my least favorite circuit element. I have only a few on hand and I don't even know what values they span. |
|
|
| Quartz microbalance experiments |
[Dec. 20th, 2009|09:22 am] |
Even though I broke the glass hook that I'd glued to a 2.048 MHz quartz crystal, I figured I could still test its ability to oscillate. The same circuits that didn't work for the crystal glued directly to a piece of glass, however, didn't work with this newer assembly either. The following circuit, however, worked (maybe) for both:
How the circuit works I'm not quite sure. It's using TTL logic in a decidedly non-logical manner and in building circuits like this I'm back to teenaged days of assembling things out of old TAB books ("101 Easy Digital Projects!" stuff) and hoping that they work like they say they do. One thing ought to be plain: without the crystal the circuit oscillates anyway. With the crystal in place, if the time constant of the RC sections is low enough, the oscillator locks onto the fundamental frequency of the crystal. If the RC time constant is too high the oscillator locks onto an overtone. I verified this by playing around with different intact crystals and RC values.
Both of the 2.048 MHz crystals I had altered oscillated in this circuit, but not where expected: rather than oscillating at a value somewhat lower than 2 MHz, as I expected, both crystals gave oscillations of about 2.2 MHz frequency. Why should this matter? Adding extra weight to a quartz crystal is supposed to lower its fundamental resonance frequency. If it's oscillating at a higher frequency it's probably because the oscillator has in fact locked onto a "spur". Spurs are caused by vibrational modes other than the desired "thickness shear" mode. They're always at higher frequencies than the fundamental resonance frequency and, unlike the fundamental mode which is dependent ideally only on the thickness of the quartz covered by the electrodes, spurious modes are unpredictably dependent on any number of variables. A chipped edge, for example, has little effect on the fundamental vibration but introduces or magnifies spurious modes.
I'm pretty sure that trying to use a crystal oscillating in a spurious mode as a weight sensor is not a good idea. There's no guarantee of a simple dependence of frequency on applied mass. If a slight flexure or change in temperature should just happen to decrease the energy level of some other spur that's coincidentally of similar frequency, the oscillator may "jump" to that new frequency. In fact it's now looking to me like a free-running oscillator is not the way to use a crystal modified in this way. Impulse response looks like the best bet: periodically shock the crystal with a very short and strong voltage spike, which sets the crystal oscillating briefly. The rapidly decaying voltage across the crystal as the transient oscillations die should, on frequency analysis, reveal the fundamental resonance frequency. But this is getting rather complicated for what was supposed to be a simple alternative. Now I can see why real quartz crystal microbalances are used mostly to monitor thin film depositions and surface adsorption and so forth: the extra load on the crystal is very small and symmetrically distributed, meaning that the crystal's vibrational modes are basically unchanged instead of being altered in unpredictable ways.
Frankly it's looking like other methods for measuring weight are more attractive. I had thought of measuring magnetostrictive vibrations in a length of Nichrome under tension from the applied weight but now I'm not so sure. The "yield point" of Nichrome wire is about 40,000 psi, worse than copper; hence an applied load of about 8 oz will permanently alter a length of 38 AWG Nichrome (actually there's no hard limit, but rather an arbitrary threshold of something like 1% deformation.) Worse, Nichrome apparently suffers from bad "creep" under continuous strain; that is, even at strains lower than the yield point the wire will slowly lengthen. One reason fused quartz was so popular in similar uses is that it exhibits almost no creep at all and its yield point coincides with its breaking strain, i.e. it is far more dimensionally stable than any metal. |
|
|
| People who work with glass fibers shouldn't, uh... |
[Dec. 18th, 2009|11:10 pm] |
Shouldn't be clumsy and thoughtless I guess. I was about to work with this assembly:
It's my second "draft" of an idea that I've mentioned previously, using a cheap quartz oscillator crystal to weight very small quantities by measuring the added weight's effect on the natural resonance frequency of the crystal. In my first attempt I merely cemented the flat side of a crystal to a glass sheet and tried to determine its resonance frequency with a few different oscillator circuits but the crystal refused to oscillate. For my next try, therefore, I decided I'd support the crystal by the edges and attach a small hook to it to hang the load off it. Some other in-construction pictures:
It's another 2.048 MHz surplus crystal. The chipped corners aren't my fault, actually; it was like that when I opened the case. The frequency of a quartz crystal's oscillation is determined almost completely only by the area covered by the electrodes so it quite possible that a slightly damaged crystal like this would still work just fine. It's cemented to an ordinary galvanized washer with little segments of glass capillary as standoffs. The hook is, or was, formed from another short length capillary drawn out into a thread then bent into a fine hook and cemented to the underside of the crystal. The adhesive used is ordinary cyanoacrylate "Super Glue". Its "shear modulus", i.e. the degree to which it deforms under shear stress, is fairly high although inferior to slow-curing epoxy cements. I surmised therefore that it would be less likely to dampen the shear vibrations of a quartz crystal.
I was preparing to test the assembly with an oscillator then hang bits of 38 AWG Nichrome wire off the hook to see if they produced detectible change in oscillation frequency. But while hooking everything up I broke off the end of the hook with a careless brush of the hand. Oops. You're really supposed to use fused quartz for such things but I'd need to get a MAPP gas torch and some fused quartz rod stock and maybe some other tools before doing even simple work with quartz fibers. I've also thought of running a thin tungsten wire through the capillary before drawing it out and working it. But maybe I should avoid the dangling hook model anyway.
To work the glass fiber, by the way, I had to resort to this wacky setup:
That's a BernzOmatic pocket torch with its gas flow set very low running to a medicine dropper tube. The tip of the dropper was held for a few seconds in a large torch flame until it melted almost shut. With further throttling of the gas flow by a clamp on the rubber hose I could get a flame about 1/8" high. Even this was rather too large for reliable working of glass fibers.
For the moment I'll cement or fuse another hook on and continue, but I think I need to start thinking of a different design. |
|
|
| Quantification |
[Dec. 16th, 2009|11:49 pm] |
One more test to go, the molecular spectroscopy exam. Let's hope that my at-home review touched upon the necessary areas. As I've complained before, the instructor has never given us any idea of what to expect on his tests--not in lecture, not in homework, not in class handouts.
I also filed my graduate application two days ago. What to do between now and the rejection letter I'm not sure. I wish I knew how long it would take.
With school almost over at least I get to start daydreaming about my mad science projects again. I still have an overall goal in mind that I described and started working on during the summer: to build a sort of inorganic analytical toolkit covering as many possible elements and other species as possible, preparing standards and reagents and devising and testing techniques. I was in the middle of working on tests for the alkali metals when fall arrived and I had to break my work off.
Since then I've been considering an entirely new approach--well, mostly new. Almost everything I've learned and done is "macro scale" chemistry. Roughly speaking, macroscale means dealing with gram quantities and volumes in tens or hundreds of milliliters. A step further down is "semimicro scale", maybe an order of magnitude smaller. I've done a little there on in the sense that any quick-and-dirty qualitative test carried out with a few drops of solution in a test tube or on a spot plate is "semimicro" analysis. But to do any quantitative analysis I've used conventional macro scale "wet" methods. And building up my own stockroom requires a lot of quantitative analysis. How else can I be sure that I've prepared reagents I can trust?
The trouble with macro scale methods is that, while they are simple, they are extremely wasteful. Let's say that I prepare 10 grams of (to pick an example) something that I hope is phthalic acid. This probably would have taken a fairly wasteful synthesis, too, since I would have been forced to use a suboptimal but safer and cheaper synthesis than usual. For the uses I'd put phthalic acid to I'd want to be sure it was of good quality. The standard melting-point method for checking the purity of an organic substance uses only a few grains of material but in the case of phthalic acid it's not much use because the substance does not have a clean melting point but instead decomposes over a wide range of reported temperatures. Hence I'd fall back on acid-base titration. I'd want to do three determinations to be on the safe side. With standard equipment--a 50 mL buret, 0.1M sodium hydroxide--three titrations might require at minimum about 0.6 g of material. Given that my current electronic balance is only of milligram precision I'd want to use a bigger sample for each titration to minimize error. What's more, preparing that 0.1M NaOH itself required yet more preparation of other chemicals and yet more titrations. It doesn't help that a solution of NaOH does not maintain a constant concentration over time without elaborate precautions so I can't just mix up a gallon of 0.1M NaOH, standardize it, and then forget about it until it runs out. The upshot? Just to figure out whether I prepared an acceptable yield of phthalic acid I need to waste a tenth of it--more, if I mess up a titration--not to mention wasting various amounts of other chemicals that require tedious and wasteful preparations of their own. It gets especially bad when (say) a redox titration is needed to make a good determination because the reagents needed for redox titrations are usually more numerous and more expensive.
There must be better ways and I think they lie in abandoning the conventional methods of quantitative analysis that I've put so much effort into. It's time to put away the burets and the volumetric flasks and the pH indicators and miniaturize. |
|
|
![[info]](http://l-stat.livejournal.com/img/userinfo.gif){kind=small}
monoceroset's journal
