After constructing my Elecraft K2/100 transceiver in 2003, I found myself in need of an accurate frequency reference to calibrate the transceiver's internal 4 MHz reference oscillator. It would have been a simple matter to compare the 5th harmonic of the oscillator against radio station WWV's 20 MHz carrier using a communications receiver, but 20 MHz reception was non-existent at the time due to low sunspot numbers and generally poor summertime HF propagation conditions. Thoughts began to turn toward designing a reliable, high quality frequency reference for the K2 and other projects to avoid this problem in the future.
Later that year, the ARRL announced plans to conduct a Frequency Measuring Test. Thoughts of participating in this event provided additional motivation to develop a high accuracy frequency standard against which off-air signals could be measured.
My frequency measurement methodology has evolved since 2003 from using a combination of mostly commercially manufactured radio equipment and instrumentation to one that uses hardware that is completely self engineered. More recently, this arrangement has been used to study the behavior of the Earth's Ionosphere by examining the Doppler shift produced through the motions of the refractive layer responsible for long distance HF signal propagation.
Soon after completing the K2/100, experiments involving a multi-turn resonant loop antenna, 40 dB gain preamplifier, and oscilloscope revealed that reception of radio station WWVB was indeed possible in New Jersey without too much difficulty.
|
|
After further research, the design for a carrier phase tracking receiver for WWVB began to take shape. This would take the form of a specialized narrowband receiver that electronically steers or disciplines a 10 MHz VCTCXO into phase alignment with WWVB's atomically controlled carrier. Synchronous demodulation and correlation decoding techniques were employed to reliably demodulate WWVB's amplitude shift keying under the influence of noise. Processing and decoding of WWVB's time code is performed through a Microchip PIC16F88 microcontroller that (among other things) functions as a real-time clock, controls 60 kHz RF phase shift circuitry that compensates for WWVB's hourly phase signature, provides UTC date, time, UT1 correction, and PLL error voltage information to the user through an LCD display, and sends time and date information in the form of a serial data stream for accurately synchronizing a data logging PC to the current time. Harmonics and sub-harmonics of the 10 MHz oscillator derived through binary division provide a wealth of local reference signals for frequency measurement and calibration purposes.
On-air signals whose carrier frequency is to be measured are received using a quadrature phasing (imaging rejecting) direct conversion receiver. The receiver employs a commutating (switching) mixer driven by the I and Q outputs of a digital phase shift network that is driven by a local oscillator operating at close to four times the frequency of the incoming signal. The receiver has a 100 Hz bandwidth centered on an audio output frequency of 1000 Hz. Gain and selectivity is obtained through the use of cascaded multiple feedback second order bandpass filters designed around low-noise operational amplifiers. Unique to this receiver design is the fact that the front end RF amplifier and local oscillator circuitry are separate from the receiver enclosure.
Receiver local oscillator injection comes from an MPF-102 based Hartley VFO that is manually tuned through several variable capacitors. The oscillator's frequency is also under the influence of a varactor diode that permits a small amount of electronic tuning.
The signal to be measured is tuned by first resonating the front end of the receiver to the desired frequency. The VFO is then carefully adjusted until the signal to be measured is demodulated as a 1000 Hz tone. The 1000 Hz tone is mixed along with a 1000 Hz reference from the frequency standard in a phase comparator. The comparator's output voltage is then introduced into the bias voltage of the VFO's varactor diode with the amplitude and polarity necessary to electronically steer the receiver's tuning toward a point where it becomes "locked" in phase with that of the incoming signal. In this state, the audio output of the receiver is driven to exactly 1000 Hz despite any minor VFO or on-air frequency drift that may be present. As an aid to manual tuning, the comparator's output voltage is monitored on an analog meter to ensure that the VFO is maintained near the center of the receiving system's "lock range".
This phase locked loop feedback arrangement produces a carrier phase tracking function similar to that in the WWVB receiver. The difference being that in this case, the receiver's local oscillator is disciplined against the phase of an on-air signal whose frequency is unknown. Measuring the frequency of the receiver's phase locked VFO and performing some simple math yields the frequency of the unknown signal to a very high degree of accuracy.
The receiver's local oscillator frequency is measured using a 10 digit digital frequency counter. In operation, a sample of the VFO's signal is gated into a string of 74HC4040 ripple counters, the first being a member of the Fairchild 74VHC very high speed CMOS family. The gating period and all the subsequent controlling logic is under the control of the WWVB-disciplined frequency standard.
After a precise gating period has passed, the count accumulated in the ripple counter string is sampled and latched in a series of parallel-to-serial shift registers. The data is then transferred out of the shift registers in a synchronous serial fashion to a PIC16F88 microcontroller. The microcontroller assembles the 32-bit data stream into four 8-bit words. It then performs a binary to BCD conversion of the data, and sends it on to a 10 digit LED display board as well as an RS-232 serial port for data collection and logging to a Linux-based PC.
Gating periods of 100 seconds, 10 seconds, 1 second, and 1 millisecond are available. Front panel controls allow the frequency measurement process to be halted or restarted at any time.
While the receiver's VFO is phase locked to the carrier of the unknown signal, a series of frequency counter measurements are made. Since the receiver's digital phase shift network divides the external VFO frequency by a factor of four, and since the VFO is tuned to demodulate the incoming signal as a 1000 Hz tone using USB phasing (the local oscillator frequency being placed below that of the incoming signal), the following relationship applies:
Frequency of on-air signal = (VFO Frequency / 4) + 1000 Hz
If LSB phasing is used, 1000 Hz is subtracted, rather than added to the (VFO Frequency / 4) term.
Example: Suppose the receiver is configured for USB phasing, and the frequency counter reads 13,316,000.00 Hz when the VFO is phase locked to the incoming carrier. What is the frequency of the unknown signal? Answer: 3,330,000.00 Hz.
Linux-based frequency logging software was developed to not only log a series of frequency measurements over time, but to also automatically compensate for the harmonic and 1 kHz offset relationship between the unknown RF and the VFO frequencies. Routines were developed to average a sequence of readings (to reduce the inherent +/- 1 LSD uncertainty in the frequency counting process), and to export logged data to gnuplot for visually examining frequency trends over time.
One of the advantages of this approach to on-air frequency measurement is that resolution is increased by virtue of the receiver's local oscillator operating at a harmonic of the on-air frequency and measured prior to any frequency division. If the frequency counter is configured for a gate period of 100 seconds, it can resolve the VFO's frequency down to 0.01 Hz. Since the VFO is operating at 4 times the unknown signal's frequency, this resolution (level of certainty) is increased to 0.0025 Hz. If lower on-air frequencies are to be measured, binary frequency dividers are inserted between the VFO and the receiver while the frequency counter continues to measure the VFO frequency directly. Therefore, with the VFO operating at 14,064,000.00 Hz and the VFO's frequency divided 4 times above and beyond that in the receiver itself, the frequency of the unknown signal becomes 880,000.00000 Hz with a resolution of 0.000625 Hz.
That's a resolution of less than one part per billion!
The carrier of radio station WCBS appears to be referenced to a GPS standard by virtue of the following readings taken within groundwave coverage of the station. Note that these readings were taken throughout WWVB's phase signature period (HH:10:00 to HH:15:00) with no apparent effect on the readings. (That observation speaks very highly of the phase signature compensation circuitry in the frequency standard.)
WCBS - New York, NY - 880 kHz Sun Jan 06 16:09:37 2008 UTC: 880000.000000 Sun Jan 06 16:11:27 2008 UTC: 880000.000000 Sun Jan 06 16:13:17 2008 UTC: 880000.000625 Sun Jan 06 16:15:07 2008 UTC: 880000.000000 Sun Jan 06 16:16:57 2008 UTC: 880000.000625 Sun Jan 06 16:18:47 2008 UTC: 880000.000000 Sun Jan 06 16:20:37 2008 UTC: 880000.000625 Sun Jan 06 16:22:27 2008 UTC: 880000.000625 |
The slight inconsistency in the readings is due to the +/- 1 LSD error inherent in the digital frequency counting process. This can be reduced through averaging a number or readings over time, the error being reduced by the square root of the reciprocal of the number of periods measured.
WFAN - New York, NY - 660 kHz: Sun Jan 06 17:25:36 2008 UTC: 660000.000000 Sun Jan 06 17:27:26 2008 UTC: 660000.000000 Sun Jan 06 17:29:16 2008 UTC: 660000.000000 Sun Jan 06 17:31:06 2008 UTC: 660000.000000 Sun Jan 06 17:32:56 2008 UTC: 659999.999375 Sun Jan 06 17:34:46 2008 UTC: 660000.000000 Sun Jan 06 17:36:36 2008 UTC: 660000.000000 Sun Jan 06 17:38:26 2008 UTC: 660000.000000 Sun Jan 06 17:40:16 2008 UTC: 660000.000625 |
The second table lists measurements taken of another local GPS-referenced radio station carrier. In this sequence of readings, the LSD count uncertainty quickly averages out to zero.
Changes in the effective height of the Earth's Ionosphere, variations in layer density, and Faraday rotation among other issues cause skywave propagated radio signals to be returned to Earth on frequencies slightly different than those transmitted. Individuals involved in FMT work find this a frustrating phenomenon. As the following plot reveals, it is often fascinating to study the effects Doppler shift produced by ionospheric instability poses to radio reception. It also provides an illustration of the challenges an FMTer is up against when trying to make accurate readings of distant radio signals.
In this plot, the atomically controlled carrier frequency of radio station CHU in Ottawa, Canada was plotted as a function of time, with each reading averaged over a 100 second period. While the frequency of the received carrier was fairly accurate and stable for about an hour and a half into this study, an oscillation began to develop at the 6000 second point that persisted for several hours.
There are two gaps in this plot. The first occurred because data gathering was suspended briefly to participate in a Frequency Measuring Test sponsored by K5CM (where by 80-meter reading turned out to be 0.078 Hz low). The second gap occurred because the test arrangement was left unattended for a period of time while the VFO drifted to the point of losing phase lock with CHU's carrier. Despite these brief gaps in data, the oscillatory pattern is unmistakable.
Doppler shift can be used to determine the effective velocity between a transmitter and receiver in the following manner:
Velocity = (Doppler Shift / Transmitted Frequency) x Speed of Light
Picking a convenient data point, CHU's carrier was found to be approximately 0.15 Hz high at 10000 seconds into the data collection period. Plugging some numbers into the above equation we find the velocity between the transmitter and receiver to be:
Velocity = (0.15 Hz / 3330000 Hz) x 3x10e8 meters per second = 13.5 meters per second
or about 30.2 miles per hour. Since the Doppler shift was positive, if we assume CHU's signal arrived at the receiver in one hop (and the transmitting and receiving sites are stationary), then the height of the reflective layer was decreasing by half this amount, or about 15 miles per hour.
The situation is far more complex than this, but it is an interesting exercise, nonetheless, to try to visualize the undulations taking place in the ionosphere at the time of this study.
Another experiment was performed to discover what effect, if any, antenna polarization might have on the Doppler shifts observed on ionospherically propagated radio signals.
During this test, the carrier frequency of radio station CHU was measured using a 0.08 wavelength vertically polarized resonant loop antenna, and a half-wavelength horizontally polarized dipole antenna. Note that CHU transmits using vertical polarization.
Since it was impossible to make frequency measurements through both antennas simultaneously using a single receiver and frequency counter combination, a logic control signal was tapped from the frequency counter and used to drive a relay that alternately toggled the receiver between each antenna after making each frequency measurement. The following plots attempt to illustrate how received frequency accuracy was influenced by antenna polarization.
There were times when the frequency measurements were very similar, and others when they were quite different. Most of the time the Doppler Shift trended toward the same direction, but this was not always the case. Faraday rotation taking place in the ionosphere is believed to be responsible for these effects.
One striking observation made throughout this experiment that is not evident in the frequency measurement data is that the signal received from CHU experienced far greater signal fading during horizontal dipole reception times than what was observed using vertical polarization. In fact, it was clearly apparent which antenna was being used during the course of this experiment simply by observing the aural quality of the signal being received at the time.
Further information related to FMT topics can be found at the following links:
No animals were harmed, nor any Micro$oft products used in the creation or distribution of this page.
|
John A. Magliacane Amateur Radio Operator: KD2BD Open Source Software Developer Internet Advocate Since 1987 Linux Advocate Since 1994 |
