Oh, I'm all for cannibalizing the commercial clocks, what with their pretuned ferrite loops, 60kHz receiver and demodulator, as a cost-and-time-effective method. But if someone did want to roll their own...
I have custom ordered tuning-fork style crystals down to 12kHz, and they were way more than the mass-produced $1 jobbies! I think $30-$50 each in onesies. $1 is really cheap in comparison.
In that case I'd probably try to do it with a PLL stabilized Q-multiplier instead of a crystal filter. Should work just fine at
60kHz. Or a conversion scheme that puts the IF in the low kHz range where it can be done with active or switched capacitor filters. Still have to stabilize the oscillator though. Just as a proof of concept...
True, $1 is cheap. Custom crystals are less and less popular. Some of the companies I used way back when are no longer there or aren't doing it anymore unless you buy thousands.
To help us out, it would be best if you describe your method of testing the resonance frequencies and the accuracy of your frequency meter/counter.
A -2 Hz "error" in frequency is about 33 PPM (Parts Per Million) or 0.0033 %. That seems to be within manufacturer's stated tolerance.
For what it's worth, the spectral occupancy needed by the WWVB signal is roughly 5 Hz. That is good enough to demodulate the AM of WWVB and still preserve the (relative) sharpness of the digital amplitude transitions for purposes of obtaining the correct time of day. Modulation on WWVB is roughly 30% AM at 1 second periodicity.
In my TRF receiver for 60 KHz, the carrier is extracted by amplifying the filtered signal and applying it to an over-driven MC1350P which acts as a limiter. Outside of the (relatively) broad selectivity of the tuned loop (Q roughly 45) and an interstage L-C tuned coupling, the final filter is simply two ECS crystals in series with a small capacitor to ground at the series connection point. The capacitor value was arrived at by "cut and try" substitution, much quicker than trying to calculate everything after an elaborate crystal measurement exercise. :-)
The final selectivity is narrow enough to eliminate most of the LF hash around the spectrum, especially the 4th harmonics of the TV set horizontal sweep frequency. That should work equally well on non- limiting demodulation to get the time-of-day data. [without the DSP supplied by the microcontrollers in the radio clocks...we have two commercial units in the house for that]
Measuring the exact crystal resonance frequency is NOT a simple exercise at 60 KHz. I would suggest looking closer at the Digi-Key links for technical data direct from the manufacturer. Those are found on the Digi-Key final part-number page just below the electronic catalog page PDF link. Manufacturer's data yields the parallel capacitance, maximum series resonance crystal equivalent resistance, and either the equivalent series inductance or the equivalent series capacitance. Digi-Key is excellent in their links to manufacturer's data in my estimation.
The Digi-Key pages says 100 ppm (+/- 6 Hz @ 60 Khz). The Epson web page says they are photolithography-finished and at least one Epson data sheet says the standard frequency tolerance is
20 ppm. Possibly they have no problem hitting 20 ppm and the actual tolerance is much better than that (but not over temperature).
20 ppm is still 1.2 Hz wide and the resonance probably is sharper than that (Q > 50K?).
This measuring project started when I tried to use the Epson crystal data to calculate what the series resonant frequency would be of the 60 Khz parallel specified crystals. After much mucking around with various numbers I decided to measure it.
And you're right, I've been ignoring calibration. Here's how I'm measuring the resonance:
I have a homebrew LC VCO running at 6 Mhz. It's full frequency range is about 5.99180 to 6.0053 Mhz. This is divided by 100 (two 74LS90's) and then low pass filtered resulting in a sine wave around 60 Khz.
The signal level is attenuated via 10k/1k resistors and then fed through the crystal with a 10k load on the other side. There's some additional loading from the x100 gain amplifier and then into a scope.
The circuit around the xctl looks like: .1 10k from-e-follower-lowpass--||----/\/\/\/\/----+---| xctl |----+---> to x100 amp < >
< < gnd gnd
I can see a noise widened trace on the scope plus some switching spikes/artifacts. As I tune the VCO, the noise trace is flat except at one specific frequency, which is about 1 Hz at most wide where the noise band becomes a sine wave.
I'm measuring the frequency of the 6 Mhz VCO with a Ramsey C-125 frequency counter. It's a standard ICM7216D counter with a cheap 10 Mhz crystal as the time base. It's uncalibrated (other than the factory, not sure of the date, possibly in the 70s?).
Ok, how to calibrate the frequency counter?
And how stable is the frequency counter?
I'm living in a cloud of RF noise, plus computers. In addition the frequency counter is a real RF noise generator too (multiplexed LEDs in addition to the counting circuitry).
By moving the counter and short wave radio to a different room I managed to hear the 2nd harmonic of a 5 Mhz crystal oscillator on 10 Mhz with WWV. It sounded like the beat frequency was lower in frequency than the 100 Hz WWV modulation pulses. So an upper bound of +/- 100 Hz at 10 Mhz would put the upper bound on the frequency counter of 10 ppm.
I'm not really happy with this calibration, I'll have to see what I can do to improve it.
Right now, I'm building a WWVB receiver for my University Thesis Project. I've looked around quite a lot and it's hard to find any chips on sale. I tried to contact the ATMEL people, but they said the chip was obsolete. Then I tried with Evox-Rifa (the north american distribuitor of MicroAnalogSystems) and asked for some chips, they say that they only had for production (big quantities) but they sent me two samples of their MAS9180 receiver chip. It is good to projects due to it comes in SDBIL package not as SMT device. Just do some research on the MAS9180 and you will find some information. Another site were i found a LOT of components and they were very heedful when you ask for some samples. There i asked for 5 of their CM6005 receiver (very similar to the MAS9180, but in SMT package), 3
- 60Khz pretunned antennas and 3 - 60Khz xtals. And I had the devices on my hands 1 week after that, from Germany. The web page is
formatting link
You can order samples and the best of all... FREE. I'm using the MAS9180 receiver with the antennas and the xtal from c-max. Right now my main problem is to decode the signal. I want to acquire the signal via the parallel port, and Im doing the software in assembly to use and old PII CPU for the project.
Zero-beat to WWV is hard to do better than 50Hz by ear. Mechanical aids will get you a little better but then at the few Hz level you hit variations in carrier due to ionospheric variation.
For a few hundred $, HP Z3801A's are available on the surplus market. They're a 10MHz OCXO locked to GPS. Short term Allan variation is
10^-12 or better over 1-100 seconds. Many lab counters and a lot of ham radio frequency counters will happily accept the 10MHz reference that the Z3801A makes.
Other telecom-related GPS-locked OCXO's/rubidium oscillators are available on the surplus market too, some make telco-related reference frequencies like 1.544MHz or 19.6608MHz which can be used to calibrate on.
I read in sci.electronics.design that Tim Shoppa wrote (in ) about 'wwvb receiver chip needed', on Tue, 26 Apr 2005:
Have you tried a Lissajou display?
--
Regards, John Woodgate, OOO - Own Opinions Only.
There are two sides to every question, except
'What is a Moebius strip?'
http://www.jmwa.demon.co.uk Also see http://www.isce.org.uk
The test setup is shown below. The input signal was produced by my homebrew function generator implemented with an AD9833. The clock for the 9833 is an 10 MHz SG-615B which has a specified frequency stability of 100 PPM.
Check out the latest Atlanticon proceedings, or the upcoming issue of the Homebrewer for a circuit to allow you to see zero beat to within a fraction of a Hz.
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.