Check frequency
in [Basics]
|
From: John Fields on 19 Dec 2009 16:43 On Fri, 18 Dec 2009 17:39:35 +1100, "Phil Allison" <phil_a(a)tpg.com.au> wrote: > >"whit3rd" > ClueLess > John Fields > >> >For this application an oscilloscope will be useless and you >> > should use a frequency counter. >> >> Thank you very much got your very detailed suggestions, I will >> look for a frequency counter and try as you said. > >Er... no, that's not right. If you have a suitable reference >frequency source, you can watch the pretty Lissajous figure >on an X/Y oscilloscope (no dual channel, no counter) and if >it goes through a cycle in 100 seconds, with 0.1 second >stopwatch accuracy, you've just made a frequency >measurement with .001 Hz accuracy. For your >32 kHz crystal, that's a second per year kind of accuracy. > >A frequency counter is a convenient and quick solution, but >it is NOT required nor is it superior in accuracy. > >** Ask JF where the OP can get an affordable counter that >reads a frequency in Hz to seven decimal places - as in his > " 32768.XXXXXXX Hz ". --- Well, even you should have figured out that the XXXXXXX was figurative for the OP's requirements and was chosen to reflect _any_ accuracy he might require. --- >His magic 10 turn crystal PU loop need to be patented too. --- Build an oscillator and try it, loudmouth. I would, but I'm fresh out of watch crystals. Anyway, just for grins I put together a 30 turn probe and held it close to a 10k resistor with a 32768Hz TTL square wave going through it. I could see the signal, just barely, against fluorescent background noise, but the counter liked the noise better. JF
From: whit3rd on 19 Dec 2009 16:45 On Dec 19, 12:29 pm, John Fields <jfie...(a)austininstruments.com> wrote: > On Thu, 17 Dec 2009 21:31:35 -0800 (PST), whit3rd <whit...(a)gmail.com> > wrote: > >>... If you have a suitable reference > >frequency source, > Just what might that be? > >you can watch the pretty Lissajous figure > >on an X/Y oscilloscope (no dual channel, no counter) and if > >it goes through a cycle in 100 seconds, with 0.1 second > >stopwatch accuracy, you've just made a frequency > >measurement with .001 Hz accuracy. > With the crystal running at about 32768 Hz, how do you propose to make > one cycle last for 100 seconds? Look up 'Lissajous'. The proposed experiment is to use a frequency-shifting transformation, i.e. to look at a beat frequency against a known frequency standard. The standard would (best) be an atomic fountain or (more practical) rubidium-beam or (more common) aged calibrated crystal clock (not AT crystal, more likely GC or other exotic type). A common frequency counter might have an ovenized oscillator, but not 'reference-standard' precision. If one were able to calibrate a warmed-up counter for a day or so against NNTP time, it might suffice for seven-figure accuracy. Or, it might not. > If, as you've stated, you have a 100s timebase which is accurate to +/- > 0.1s, then it'll be accurate to one part in 1000, which is +/- 1000ppm. > > Then, since there are 60 * 60 * 24 * 365.25 = 31 557 600 seconds in a > year it'll be accurate - not to "a second per year kind of accuracy" - > but to 3156 seconds = 52.6 minutes per year kind of accuracy. No, that confuses the accuracy of the small-frequency beat with the accuracy of the larger frequency that one has applied (X-axis) to the reference (Y-axis). They aren't proportional. 'ppm' measurement of one doesn't tell the 'ppm' value of the other. The timebase of a common frequency counter IS a hard limit, in the sense of your 'ppm' calculation, to the accuracy of its measurements.
From: John Fields on 19 Dec 2009 19:13 On Sat, 19 Dec 2009 13:45:36 -0800 (PST), whit3rd <whit3rd(a)gmail.com> wrote: >On Dec 19, 12:29�pm, John Fields <jfie...(a)austininstruments.com> >wrote: >> On Thu, 17 Dec 2009 21:31:35 -0800 (PST), whit3rd <whit...(a)gmail.com> >> wrote: >> >>>... �If you have a suitable reference >> >frequency source, > >> Just what might that be? > >> >you can watch the pretty Lissajous figure >> >on an X/Y oscilloscope (no dual channel, no counter) and if >> >it goes through a cycle in 100 seconds, with 0.1 second >> >stopwatch accuracy, you've just made a frequency >> >measurement with .001 Hz accuracy. � > >> With the crystal running at about 32768 Hz, how do you propose to make >> one cycle last for 100 seconds? > >Look up 'Lissajous'. --- Why would you snip, from my post,: "Furthermore, Lissajous figures are generated by signals applied to the X and Y axes of the scope, so how do you propose to see anything meaningful with, say, Y running at 32768 Hz and X running at 100 seconds?" and then insert that: "Look up 'Lissajous'" crack? --- >The proposed experiment is to use >a frequency-shifting transformation, i.e. to look at a beat >frequency against a known frequency standard. >The standard would (best) be an atomic fountain or >(more practical) rubidium-beam or (more common) aged >calibrated crystal clock (not AT crystal, more likely GC >or other exotic type). A common frequency counter >might have an ovenized oscillator, but not 'reference-standard' >precision. If one were able to calibrate a warmed-up >counter for a day or so against NNTP time, it might >suffice for seven-figure accuracy. Or, it might not. --- If you've been following this thread, you should realize that what the OP wants to do is trim a microcontroller crystal to, probably, 32768.0 Hz or, at most I'd assume, to 32768.00 Hz. Also he has no test equipment and was considering using an oscilloscope VS a frequency counter. Using an oscilloscope and the methodology you suggest would certainly bog him down and require him to acquire metrology skills and equipment which, for his purposes, would be totally unwarranted, so what's the point of your diatribe? --- >> If, as you've stated, you have a 100s timebase which is accurate to +/- >> 0.1s, then it'll be accurate to one part in 1000, which is +/- 1000ppm. >> >> Then, since there are 60 * 60 * 24 * 365.25 = 31 557 600 seconds in a >> year it'll be accurate - not to "a second per year kind of accuracy" - >> but to 3156 seconds = 52.6 minutes per year kind of accuracy. > >No, that confuses the accuracy of the small-frequency beat >with the accuracy of the larger frequency that one has >applied (X-axis) to the reference (Y-axis). They aren't proportional. >'ppm' measurement of one doesn't tell the 'ppm' value of the other. --- Had you earlier more clearly described the scheme you had in mind, that would have fallen out immediately. However, the scheme you describe (which is as old as the hills, by the way) is hardly practical to implement from the OP's point of view; in my opinion. --- >The timebase of a common frequency counter IS a hard limit, >in the sense of your 'ppm' calculation, to the accuracy of its >measurements. --- Indeed, but looking over your post again, you mentioned nothing about the accuracy of the reference, but instead stated that a 100 second measurement window accurate to 0.1 Hz would yield 0.001 Hz accuracy, which is clearly not true. How can you reconcile that? JF
From: Phil Allison on 19 Dec 2009 19:49 "John Fields, Autistic " >>> With the crystal running at about 32768 Hz, how do you propose to make >>> one cycle last for 100 seconds? >> >>Look up 'Lissajous'. > > --- > Why would you snip, from my post,: > > "Furthermore, Lissajous figures are generated by signals applied to the > X and Y axes of the scope, so how do you propose to see anything > meaningful with, say, Y running at 32768 Hz and X running at 100 > seconds?" > > and then insert that: "Look up 'Lissajous'" crack? > --- ** Errr - because you repeatedly failed to comprehend how Lissajous patterns are commonly used to compare two frequencies ?? Is the great JF completely ignorant of a such a simple technique ??? Must be so - everything he says screams it. > Indeed, but looking over your post again, you mentioned nothing about > the accuracy of the reference, but instead stated that a 100 second > measurement window accurate to 0.1 Hz would yield 0.001 Hz accuracy, > which is clearly not true. Here is the idea: " If you have a suitable reference frequency source, you can watch the pretty Lissajous figure on an X/Y oscilloscope (no dual channel, no counter) and if it goes through a cycle in 100 seconds, with 0.1 second stopwatch accuracy, you've just made a frequency measurement with 0.001 Hz accuracy. For your 32 kHz crystal, that's a second per year kind of accuracy. A frequency counter is a convenient and quick solution, but it is NOT required nor is it superior in accuracy. " ** Fraid it is all true..... The demented JF has forgotten how to think analogue. Cos he has forgotten how to think at all. ..... Phil
From: Jasen Betts on 20 Dec 2009 03:25
On 2009-12-19, John Fields <jfields(a)austininstruments.com> wrote: > I would, but I'm fresh out of watch crystals. > > Anyway, just for grins I put together a 30 turn probe and held it close > to a 10k resistor with a 32768Hz TTL square wave going through it. > > I could see the signal, just barely, against fluorescent background > noise, but the counter liked the noise better. use a tuned probe. 32768Hz crystals work good for probes. |