A similar method would be multiplying both the reference frequency and the device under test frequencies several times, before mixing them down and then analyzing the difference. A few classical (balanced) frequency doublers in each chain should do.
I have used this to test some local oscillators in some 1200 MHz converters, so the crystal frequency is multiplied by 10 to 100 times to end up in the same ballpark area. After some additional frequency translation in an SSB receiver, the output was an audio difference tone around 300-1000 Hz.
When tapping the PCB with a pen, the microphonics was easily observable as a variation in the audio tone. I was quite a puzzled, the non-boxed oscillator run nicely on the table. However, each time I looked into the circuit, there was a frequency variation every 2 to 3 seconds. Did I have an evil eye ?
When I momentary hold my breath, the tone variation sopped :-) Thus, the oscillator was quite temperature sensitive. Assuming I could detect the tone variation by 10 Hz, this simple method would give 0.01 ppm resolution for short time variations.
Instead of your ear, you could use your frequency counters, oscilloscopes or RTAs to analyze the characteristics of the variations.
You could count the number of cycles of one oscillator between when it comes into exact synchronism with the other oscillator and when it does it again. All you need is a pair of very accurate zero-crossing detectors, a bit of logic and a counter; no need for any sort of mixer.
--
~ Adrian Tuddenham ~
(Remove the ".invalid"s and add ".co.uk" to reply)
www.poppyrecords.co.uk
When I was in the radio monitoring, we used the phase creep between the reference signal and the signal to be measured to get sub-Hz resolution in tolerable time.
Sure, that's how real universal counters work. The mixer was mostly a thought experiment.
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 waveforms are nice sine waves and your SNR is high, that works great. You can do similar things with triggered ramps, which is how high resolution digital delay generators usually work. (I think JL sometimes uses instant-on LC oscillators instead.)
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
On Wednesday, March 26, 2014 4:54:22 AM UTC-7, Adrian Tuddenham wrote: [about a variant on beat-frequency measurement]
If you are working with square waves, this can be done with a two-bit up/down counter; reset, count up with source A, down with source B (edge-triggered), and the count will only reach 2 when A is faster than B, and near-exactly in phase. It will only reach -2 when B is faster than A, and near-exactly in phase.
Thanks, everyone, for the responses. I can't say I understand the reciprocal counting approach - I had just seen a circuit by Jim Williams, and lacking interest, had not allocated any space in my personal neuron budget for it.
My interest is to find a sensitive detector for an "electronic nose." These have many interesting uses, including medical diagnostics. A non-invasive diagnostic via breath is an interesting thing, and there is a lot of interest among researchers, as evidenced by a number of conferences and books about.
So you may ask, why re-invent the wheel? It's already passe!
Well, in comes Murphy's Law and various laws of practicality. Much of the work by academic researchers is geared to be "new, better, and different." So work on nano particles, biosensors, etc. is not practical, according to my design budget.
I myself look at a possible design in this way, which I'm sure is basically the same as how you engineers think, but I'll language it anyway. I see a engineering problem, and I see risk. The risk is usually associated with complexity, but sometimes could be simple things I am just ignorant of and should have known, but didn't. That aside, I ask myself, where do I want the complexity to be? Where do I want to push it? I want to push it where there are few surprises, unknowns, and instabilities.
So, in the e-nose work, I therefore get rid of all "bleeding edge" work - about
97% of it all. Crystal oscillators attract, because they are stable. (like a rock). Pushing complexity into electronics, proven SS devices, software good, since that's mature tek. Material science, too-arcane optics, chemfets, etc not so good.
So that bit of common sense is a simplifying assumption. Another is what I'll call the "pattern recognition paradigm." What this says is that, for problems like the e-Nose, you don't need as much determinism on the front end detector as a system w/o the AI. Here AI=pattern recognition=PCA=ANN=multivariate analysis, choose your poison.
So in the case of an e-Nose, a specific absorber need not be deterministic, as long as it's somewhat different from other absorbers for a molecule of interest. And that's good news, since a polymer film is a fairly hum-drum technology with few surprises.
This difference between deterministic vs. hum-drum sensors is an issue with NIR vs. mid-IR sensing. All the fundamental resonances are down in the mid-IR, > 3 microns. The NIR signature s "an-harmonic," indistinct and at best has a few higher harmonics of what's evident at 7 microns. But optics at 7 microns is a nightmare compared to 1-2 microns, in the engineering sense mentioned previously. So for some cases, pattern recognition on the NIR fingerprint band has proven effective and relatively easy.
Back to the e-Nose, I got interested in this area when I learned the air force was working on an e-nose to detect when a pilot was too fatigued to operate - a common problem in combat. Well, using a detector to tell something about the mental state of the pilot, is a little science-fictiony, let's be honest. So I researched a little further, and discovered the medical researchers were detecting cancer, TB, diabetes by waving an e-nose in front of a patient. This is a bit unbelievable.
Where this goes from here, it's too soon to tell, but I'll state in a few sentences, since this is a long post. These are IMPLICATIONS or hypotheses, not proven facts. 1st, a non-invasive diagnostic to detect a complex disease state is a big break through in practice.. 2nd, the systemic nature of the detected signal may provide direct evidence for a systemic theory of cancer and other diseases. 3rd, detection of disease states might best be done with detectors in conjunction with the AI modes mentioned. In sum, this could be the camel's nose of a paradigm shift in medicine.
Of course, medical advances are not synonymous with availability. I myself use life extension techniques discovered a few decades ago, but the public is unaware of. Thanks for listening, and I apologize for the lenth of this, as well as stating the obvious.
No pun intended, but have you ever sawed off the top case of a SAW chip? Can they be immersed in a solvent? Surface functionalization with silanes or titanates is a decent way to get selectivity along with repeatability of a surface coating.
What happens at the atomic level to make certain molecules like or dislike a surface? The nice thing thing about a statistically-trained chemometric system is you don't need to know.
Cush R, Cronin J M, Stewart W J, Maule C H, Molloy J and Goddard N J, Biosensor Bioelectron. 8 347-53 (1993).
That was the sensor my 1996 paper
A W Sloman, Paul Buggs, James Molloy and Douglas Stewart, Meas. Sci. Technol. vol. 7 pages 1653-1664 (1996).
talked about temperature stabilising to better than 0.1C - because 4000 of the right molecules grabbed by a monoclonal antibody made as much difference to the refractive index of the optical layer of interest as a 0.1C temperature change ...
People have came up with cuter ways of sensing the refractive index (which is one of the reasons that particular machine isn't made or sold any more).
Knowing what's going on can still be quite useful, all the same.
surface? The nice thing thing about a statistically-trained chemometric system is that you don't need to know.
Knowing what's going on can still be quite useful, all the same.
-- Bill Sloman, Sydney
Ah, a classical physicist! Regarding the molecular-scale interactions, organic chemists spend a lot of time in front of white boards trying to figure out something they call "steric hindrance" and "transition states." They are fairly successful in their "mechanisms," but when you have detailed interactions on a surface, among a collection of moieties and absorbing sites....You are forced to proceed empirically to a large extent.
In fact, this blind empiricism, while not giving Oz-scientists "soft and fuzzies," (sorry couldn't resist), has huge strengths in some situations.
Please allow me to describe a "blind computer" which is biological in nature, yet more powerful than Silicon for what it does. Suppose you want to create a blood test for TB. (picked randomly) There is a technique to get an antibody expressed on the surface of a phage virus, called "phage display." To refresh, a "phage" is a "lunar landing module" virus that feeds on bacteria. An antibody is a recognition protein for specific binding to something.
So you make a library of a range of these phages with different Ab's on them. Then make two columns. One is a column with immobilized diseased blood. The other is a column with healthy blood.
So you pass the library through the diseased column. Some of the phage chimeras will stick. You collect those. Then you pass them through the "healthy" column. What you collect is Ab's that indicate for disease, but NOT for healthy. The next thing you do is to re-infect a bunch of bacteria with these phages, and make a new phage library, based on this sub-set. And round and around, until the SNR - binding constant is optimum.
And it even gets better - what you emerge with is an optimum antibody for your test. Want a ton of it? Just turn off the library randomization and infect a
100 kilo batch of e coli bacteria. Presto! You have a system which incorporates AI and manufactures itself when done! 100 lbs. of Ab 12 hours later.
The algorithm is a simple selection method, sometimes called the Darwinian algorithm. It appears in software as well as other venues such as this. The needed knowledge is pushed off into gene-jockeying, a proven area. Want to benefit from high SNR diagnostics? These use a virus, so the FDA takes a dim view.
ike a surface? The nice thing thing about a statistically-trained chemometr ic system is that you don't need to know.
In fact I've got a Ph.D. in physical chemistry, which I got back before mon oclonal antibodies had become popular.
Some of my theoretical chemist colleagues were imagining modelling protein structures in enough detail to envisage how enzymes work, but it has taken quite a while for that particular vision to be realised.
Max Perutz had worked out the structure of haemoglobin in 1959, but it took him until 1970 to work out the structures of oxy- and deoxy- hemoglobin wh ich eventually let him work out in detail how the haemoglobin molecule tran sports oxygen around the body, and releases it where it is needed.
formatting link
t
fairly successful in their "mechanisms," but when you have detailed interac tions on a surface, among a collection of moieties and absorbing sites....Y ou are forced to proceed empirically to a large extent.
It's certainly the path of least resistance. "Steric hindrance" was a conce pt that was well established back in the 1960's, when I got my Ph.D. The co ncept of the "transition state" was also well-established back then, and my Ph.D. thesis includes the results of a partition function for a simple tra nsition state that I'd calculated for myself, with a lot of help from my fr iends and the literature.
Of course. But it doesn't exclude a more detailed understanding of what's g oing on.
n structures in enough detail to envisage how enzymes work, but it has take n quite a while for that particular vision to be realised.
***Yes, I'm not about to try a deterministic investigation of what does wha t at the atomic level. I'm interested in a workable product, and my point is tha t reductionist science of PhD chemistry type doesnt help you here much.
ok him until 1970 to work out the structures of oxy- and deoxy- hemoglobin which eventually let him work out in detail how the haemoglobin molecule tr ansports oxygen around the body, and releases it where it is needed.
*********************************actually you have missed the point. Tell me, what is the driving force f or the structure of Hemoglobin? (a test)
*******************************
SNIP
d
s.
going on.
ature,
vague generality...
***sorry you didn't understand it. It's a very powerful and precise techniq ue used by combinatorial chemists, and the point is, it doesn't require ANY detailed knowledge of the details of the binding interaction. None. Irrelev ant.
In fact, quantum physics says that neither you, me, nor Max Perutz can go b eyond certain limits of detail regarding the bonding. I'm sorry to break th at news to you.
**************** And ...I apologize in advance for what may seem a somewhat insulting questi on but one which I ask of of all physical biochemists to see if they know abou t the force determining a protein shape. Know what it is?
************** jb
I went on to become an electronic engineer, which isn't a common consequenc e of completing a Ph.D. in experimental physical chemistry. Whether this co unts as a decline or a recovery is debatable, but it's not a issue that I'm going to bother to debate.
ein structures in enough detail to envisage how enzymes work, but it has ta ken quite a while for that particular vision to be realised.
hat at the atomic level. I'm interested in a workable product, and my point is that reductionist science of PhD chemistry type doesn't help you here m uch.
Since your interest is in synthesising (and I don't mean just building your own compound) something that works, reductionist logic isn't going to be m uch use to you. A nuts and bolts understanding of some of the processes inv olved isn't exactly reductionist - science is quite as much about put thing s together as it is about analysing them in ever final detail.
took him until 1970 to work out the structures of oxy- and deoxy- hemoglobi n which eventually let him work out in detail how the haemoglobin molecule transports oxygen around the body, and releases it where it is needed.
for the structure of Hemoglobin? (a test)
You are claiming that I'm embracing reductionism, and now you want me to pu ll out a single "driving force"?
From my - relatively brief - time as an inorganic chemist, I can tell you t hat the haemoglobin molecule embeds four iron atoms in a rather complicated cage structure - the phrase "transition metal complex" comes to mind. Oxyg en and CO2 bond to the iron atoms, and the process of forming the bond betw een the iron atoms and the O2/CO2 slightly changes the shape of the cage in a way that makes the molecule more useful in it's job in the body. I've re ad - and think I understood - more complicated explanations some time ago, but I'm not going to bother getting re-educated to pass some half-baked "te st".
and fuzzies," (sorry couldn't resist), has huge strengths in some situatio ns.
's going on.
nature, yet more powerful than Silicon for what it does.
ique used by combinatorial chemists, and the point is, it doesn't require A NY detailed knowledge of the details of the binding interaction. None. Ir relevant.
It may not require it, but more detailed knowledge is likely to be useful, and certainly isn't irrelevant.
beyond certain limits of detail regarding the bonding. I'm sorry to break that news to you.
It doesn't come as a surprise. Heisenberg made the fundamental point quite a while ago, and even back in the 1960's chemists were scornful of physicis ts' attempts to annex chemistry as a branch of applied physics. Chemists we re doing molecular orbital calculations back then, but they had to be prett y crude, and today's more powerful computers aren't powerful enough to take us beyond the Hartree-Fock simplifications. Hartee published in 1927 and F ock in 1930, so that limit has been known for some time.
tion
out
Only one force? I've never been any kind of biochemist, but that does strik e me as a dumb question. The first constrain on the shape of protein - or a ny other molecule - is the lengths of the chemical bonds that hold it toget her, and the angles between the bonds at each atom. For a complex molecule like a protein, these leave a lot of possible conformations, which are pare d back - in the first instance - by the crude non-bonding interactions usua lly labelled as steric hinderance. You can't put two atoms in the same spac e.
The next set of constraints are the "hydrogen bonds" which are "non-bonding " only in the sense that they are weaker than regular chemical bonds, and t here are other non-bonding interactions which help stabilise particular con formations. I've played with "Foldit" but I've not dug in deep enough to ha ve any well-developed ideas about how it works.
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.