Someone has no doubt done this, but I am interested in measuring very small changes in a xtal when it accumulates a surface material. I used simple versions in a vacuum chamber for thin film deposition, and they could tell a few Angstroms. But interested in a more sensitive change - say 100-1000X.
Without knowing much it sounds hard. Do you want 100 X the frequency resol ution of 100X the mass resolution. (freq. should go as square root of the mass.. so 100X mass resolution is 10^4 in frequency.) I'm guessing they al ready measure the frequency to a Hertz or so.. maybe finer? More resolutio n looks like counting for longer gate times. Can you count for 1000 second s? Well that's the brute force approach.
You'll likely be limited by crystal aging at that rate. Given a good SNR, you can measure the period to a very small fraction of a cycle with a sufficiently-good universal counter.
SAW devices are more surface-sensitive than bulk wave devices.
Cheers
Phil Hobbs
--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC
Optics, Electro-optics, Photonics, Analog Electronics
160 North State Road #203
Briarcliff Manor NY 10510
hobbs at electrooptical dot net
http://electrooptical.net
If your reference and your oscillator under test are good enough, yes. For instance, you can mix down to 1 Hz and measure the beat frequency with a reciprocal counter. It's quite feasible to get a 9-figure measurement of the 1 Hz, again if the SNR is good enough, and 6 figures is a piece of cake.
Cheers
Phil Hobbs
--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC
Optics, Electro-optics, Photonics, Analog Electronics
160 North State Road #203
Briarcliff Manor NY 10510
hobbs at electrooptical dot net
http://electrooptical.net
t in 10^6? one pico second? Sounds hard to me.. but I'm kinda slow :^)
Ahh allow me to be a bit "thick". But if you're mixing down to 1 Hz, then you're "measuring" for one second. It's sorta like counting 1 MHz with a o ne second gate, No? Though I'll admit you can slice that second a bit fine r and get a few more sig figs.
Right. You get six figures free by subtracting. By running the reciprocal counter faster or by doing triggered ramp stuff a la John Larkin, you can get a bunch more figures than that, provided that the noise is low enough.
It's Shannon's theorem again. More SNR -> more bits per second -> more measurement precision.
Cheers
Phil Hobbs
Cheers
Phil Hobbs
--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC
Optics, Electro-optics, Photonics, Analog Electronics
160 North State Road #203
Briarcliff Manor NY 10510
hobbs at electrooptical dot net
http://electrooptical.net
To be clearer, by mixing down to 1 Hz in advance, 1-MHz reciprocal counter can measure the period of a 1-MHz carrier to 1 part in 10^12 in one second. You don't have to make a two-week measurement.
Cheers
Phil Hobbs
--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC
Optics, Electro-optics, Photonics, Analog Electronics
160 North State Road #203
Briarcliff Manor NY 10510
hobbs at electrooptical dot net
http://electrooptical.net
Alas, I'm thinking of regular-old quartz clock oscillators - maybe the SNR is NOT good enough. If you go with square-wave drive, especially, the crystal is guaranteed not well coupled to the oscillator, so there's jitter. And, the usual (logic-gate-style) oscillator circuit will amplify that jitter.
Probably the crystal oscillator will have to be a sinewave type, to get the best precision, and that means some kind of master/slave oscillator circuit - it can get complicated. That's unfortunate, because most of the good solutions are prebuilt crystal-and-oscillator combos, you aren't going to be able to swap in some convenient unpackaged quartz of your own.
Temperature will be an issue. An AT or SC-cut crystal can be run, preferably temperature regulated, to operate at its turning-point temperature. A bad setup could drift a decent fraction of a PPM per degree C. Of course you need a good oscillator circuit and good power supplies, too.
As guys have noted, you can heterodyne it against some nearby, stable oscillator and measure the period of the difference. That needs good analog circuits and filtering and stuff. Oscillators that are close in frequency will really, really want to lock, so good isolation and shielding will be needed.
I guess it also matters where on the crystal the deposition is allowed to settle. Some lopsided masked pattern maybe?
--
John Larkin Highland Technology, Inc
jlarkin att highlandtechnology dott com
http://www.highlandtechnology.com
On Tuesday, March 25, 2014 2:27:04 PM UTC-7, John Larkin wrote: [about precisely measuring frequency offset for quartz crystals, as part of a scheme to measure surface mass accumulation]
There are lots of available tuning-fork crystals; the ends are plated (and the plating thickness was used to tune the frequency), so you're guaranteed a good sensitivity to extra mass when you use those rocks. Adhesion to the plated ends is key.
hen you're "measuring" for one second. It's sorta like counting 1 MHz with a one second gate, No? Though I'll admit you can slice that second a bit finer and get a few more sig figs.
I'm not sure what a reciprocal counter is, but driving home I was thinking about "my" one second gate counter. I could count the time after the 1 sec ond gate ends til the next signal transition came along... that sorta looks like the reciprocal... or remainder. And gives me some more resolution.
Locking oscillators: Neat! Can you say anything about what causes the locking? I mean, I can have a bunch of coupled oscillators, if their resonant frequencies are near each other they all can couple and I get a bunch of "normal" modes. (One for each oscillator)... ohh and locking is just the fact that I end up driving one of those normal modes.
(I think I answered my own question.. but I'll post anyway) George H.
A reciprocal counter is just your universal counter in period mode--the input signal is the gate, and it counts the clock. In real life they count N clocks in M input periods, and report a frequency of N/M.
Cheers
Phil Hobbs
--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC
Optics, Electro-optics, Photonics, Analog Electronics
160 North State Road #203
Briarcliff Manor NY 10510
hobbs at electrooptical dot net
http://electrooptical.net
Search on "butler oscillator". I'm not sure if it's still the gold standard, but back when I had delusions of being an RF guy, it was the way to go for high precision crystal circuits.
--
Tim Wescott
Wescott Design Services
http://www.wescottdesign.com
Electrostatic, electromagnetic, or just any kind of E&M including harmonics, and particularly with crystals, unbalanced acoustic coupling comes into play as well.
Anything that provides coupling will exhibit a pulling effect, where at a given phase shift between oscillators, there is a tendancy, a pulling force, that tends to draw them together or push them apart. Physics is the same as a free-rotating pendulum (i.e., a rotor, but with external acceleration), where it speeds up and slows down some incremental amount each cycle.
If you look at the product of two lightly coupled oscillators, you see the phase locks and skips; rather than a sine wave, it's kind of rampy and pulsey. Here's a sample from an early Theremin I made:
formatting link
the output is AC coupled, so you'll have to imagine the flat part ramping. Instead it kind of recovers and flattens out. But it definitely kind of lets-go and flops 270 degrees or whatever before locking again, for a moment.
A newer Theremin of mine (documented here:
formatting link
) has so little coupling that I can't even tell if it's locking or not. There's a very narrow band, with the oscillators at the same frequency (give or take fractions), which seems to be disturbed as much by 1/f noise (electrical sources, ambient convection) as actual locking behavior. Needless to say, it belts out the bass...
For direct thickness measurement of optically transparent films an ellipseometer is used. It can do angstroms thickness. It can be used at any laser wavelength.
Just in case you really wanted a thickness measurement.
part in 10^6? one pico second? Sounds hard to me.. but I'm kinda slow :^)
figures
then you're "measuring" for one second. It's sorta like counting 1 MHz with a one second gate, No? Though I'll admit you can slice that second a bit finer and get a few more sig figs.
I get the nature of your question a bit. I once saw a frequency counter that would reliably produce 9 significant figures of frequency on a 10 MHz signal a few times a second. The manufacture's literature said it also measured the fractional cycle error within the sample gate. As P. H. said, it requires a magnificent SNR. Most especially phase noise.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.