Frequency Measurement

INTRODUCTION

In this section, we look at advanced techniques for those interested in maintaining their own precision references, in calibration of standards using these references, and in precise on-air frequency measurement.

Along the way, you'll discover techniques useful for Frequency Measuring Contests, receiver and oscillator drift performance measurement, and propagation monitoring.

The main tools of the frequency measuring expert are an Accurate Standard, a very stable Communications Receiver and a computer with sound card, equipped with Spectrogram Software. A good frequency counter, a phase comparator, and an oscilloscope also come in handy. Enormous amounts of patience and good documentation of results are also important requirements!

ACCURATE STANDARD

Without an accurate traceable Standard to make measurements against, frequency measurement with any attempt at accuracy is not possible. The quality of the Standard depends on the level of accuracy required. For casual use, say for 80m Frequency Measuring Contests, a good TCXO will suffice. With care, you can keep a TCXO within 1 part in 10⁷, provided you operate it in a modest temperature environment. For the most exacting measurements, a very good OCXO or GPS disciplined oscillator is necessary.

The most useful Standard will have outputs on convenient frequencies, preferably on all Amateur bands (achieved using dividers or harmonics), and preferably on a "round" frequency to allow direct comparison with as many oscillators as possible. Direct comparison is better than two-step measurement because the accuracy of other devices (transfer oscillator, receiver etc) does not come into the equation.

The 5MHz double-oven OCXO is the best Standard the average ham could hope for. Outputs at 1MHz, 100kHz and 1Hz are readily achieved through division, and it is best to mount the OCXO and all the dividers in a well shielded box so that the signal level can be controlled for comparison with weak signals. Very good OCXOs can be found in surplus equipment - old telecoms gear, test equipment, and especially redundant "Transit" satellite navigation receivers. These can be had for a song since this older satellite navigation service has been replaced by GPS.

The Standard needs to be operated continuously, preferably battery backed, and should be checked at least monthly. The easiest way to do this is with a phase comparator, but a TV set can also be used. See below.

COMMUNICATIONS RECEIVER

The Harris RF-505A Receiver

The two main criteria required of a good receiver for frequency measurement are Stability (lack of drift and wobbles) and Repeatability - the ability to set it to an expected frequency setting, or return to a previous setting. Most modern digital transceivers and receivers allow you to set the frequency in specific steps, say 100Hz or 10Hz steps, so you can set a specific frequency from the dial, but there are a few traps:

• Some transceivers (notable the Kenwood TS-430) can be set in precise 10Hz steps, but not from the dial. The dial has hysteresis caused in the rotary encoder, so if you tune to a frequency displayed as a 10Hz step, you may be several Hz in error. If you tune with the microphone up-down buttons however, you will have the exact steps you need. Another thing about the TS-430S (and some other transceivers) is that you may need to modify the display to provide 10Hz readout.

• Many of the newer Direct Digital Synthesis receivers are able to step in very small steps (even 1Hz or less), but don't actually step in precise 1Hz or 10Hz steps, so the frequency may be in error by 0.5Hz or worse. You may not know which way the error is, and it may vary as you move up the band. Typically the error will get larger and larger (is 0.5Hz large?) and then suddenly be the opposite direction and get smaller again. If the problem is worse than 0.5Hz, forget using the receiver for direct measurements. Such receivers are generally very stable (especially with any high stability option), and usually single reference derived.

• Some receivers and transceivers use different band oscillators for each band, and a different oscillator for the BFO on USB from LSB. These receivers will potentially have systematic errors that mean that calibration performed on one band and one sideband will not be transferrable to other bands or BFO settings. The receiver will need to be calibrated on each band for direct measurements.

Indirect Measurement
The average receiver is a reliable tool for indirect measurements, i.e. for recovering frequency differences which are compared directly, rather than by measuring with the receiver. For example, to adjust an oscillator to 3600kHz, beat it with a Standard harmonic or sub-harmonic at 3600kHz, and measure the difference between the two using a spectrogram. Even modest receivers are useful in this application. This can be done either using an AM receiver and the spectrogram to measure the beat note (provided the two signals are of similar strength), or better, using an SSB mode with the receiver offset so both sources produce about a 1kHz note, and measuring or aligning the two signals using a spectrogram.

Direct Measurement
Direct frequency measurement using a receiver is quite another matter! It requires the receiver itself to be calibrated and traceable, with perfect repeatability. It will preferably use a single reference source for all oscillators, and maintain the same PPM accuracy throughout its receiving range. Its calibration can be checked or adjusted against a Standard using a spectrogram, and can be checked throughout its whole range in the same way, to ensure that the calibration is correct. The adjustment is best done at the highest frequency possible, since this is where the PPM error translates to the largest difference in Hz.

For direct measurement, the author uses two high quality receivers:

• A Harris RF-505A (single reference using double oven OCXO, 1 part in 10⁸) which tunes in exact 100Hz steps, using rotary switches and PLL synthesis (picture above).
• A Kenwood TKM-707 (single reference single oven OCXO, 1 part in 10⁷) which uses PLL synthesis, tunes in 100Hz steps with 10Hz clarifier steps and an LCD display which shows 100Hz steps. This is operated with the clarifier off.

The Kenwood TKM-707 Marine Transceiver

Effort put into improvement of receiver stability and accuracy is well worth while, although some receivers are extremely difficult to tame. The author has worked long and hard with an otherwise excellent Icom IC-R71 with 10Hz steps, and has still not achieved acceptable stability, although repeatability and fast warmup are good.

SPECTROGRAMS

The Spectrogram is a graphical computer program designed to display frequency on one axis (usually vertical) and time on the other, with the third axis (brightness) indicating signal strength. Most of the programs operate in real time using input digitized from audio by a PC sound card. Some use data from an external digitizing source, such as a DSP unit. The most useful Spectrogram programs also offer an amplitude versus frequency display (spectrum display) and digital signal strength and frequency measurements for the strongest frequency component detected. In this way frequency resolution to 0.01Hz can be measured.

The Spectrogram uses an FFT technique, and the best types for this application use narrow-band low sampling rate with windowing, allowing frequency spans as low as 5Hz or less to be viewed without loss of resolution. The dynamic range of sound cards and FFT software is very good, enabling frequency measurements to be made on weak signals, and comparison of signals of quite difference signal strengths. Some Spectrogram programs are optimized for different purposes (full audio range, high speed, or very weak signals) and are not always as useful for frequency measurement purposes. The programs described below are recommended specifically for their frequency measurement capability.

Spectran
Written by Alberto I2PHD and Vittorio IK2CZL, this is probably the most universally useful program. Used at the lowest available sampling rate and with manual gain adjustment, Spectran does a good job if the gain adjustments are made with care. Offset frequency display is available, and cursor readings of frequency and amplitude are provided. Spectran includes an excellent spectrum display with wide dynamic range (see below).

Spectran's excellent spectrum display with peak measurement
(Note display shows 1Hz sidebands and easily 70dB dynamic range)

For Doppler monitoring, Spectran works very well, although setting brightness and contrast optimally takes some skill and patience. The next image, thanks to Dave ZL1MR, is an example of NVIS Pedersen Rays and Backscatter, received at a range of 15km. The transmission was from ZL1BPU (40mW on 3840kHz) and this recording represents about two hours across sunset. The weak slightly curved but stable line is the ground wave, bent by receiver warmup drift. The two Pedersen Rays from the ionosphere are clearly evident and much stronger, and spectacular clouds of backscatter are also visible. If you look closely you'll also see brief 4Hz sidebands every 10 minutes. These are caused by the 10WPM Morse ID on the transmission (who says a Morse transmission has no bandwidth?)

ZL1MR's recording of ZL1BPU on 3940kHz

Spectran is also useful for general signal monitoring and narrow drift measurements, and includes a preset NDB mode. The calibrations on the display are excellent. With the AGC on, the signals tend to be "washed out", losing detail and making comparison of close signals difficult. It also will not run slow enough for useful drift measurement of less stable oscillators. Download Spectran from www.weaksignals.com.

Spectran in Doppler mode
(click on the picture for a closer view of the settings)

For Doppler monitoring, set Spectran at 8000 or 5512 samples/sec and use 0.031Hz resolution or better. Set Auto-brightness (Auto brig.) off, use the low gain setting and keep the Brightness and especially Contrast close to the minumum settings.

Argo
Argo, written by the same team as Spectran, is designed to detect very weak signals, and is not useful in this application as the signals always lack detail and are too broad. It is incredibly sensitive, and makes good meteorgrams however! See www.weaksignals.com.

An ARGO meteorgram recorded on 17675kHz (Radio New Zealand carrier)

Spectrum Laboratory
By Wolfgang Buescher DL4YHF, Spectrum Laboratory is very complex to use, but is very capable for a wide range of applications. To use it well you must understand FFT technology well, and must use a fast computer (over 500MHz preferably), with plenty of memory. It has the advantages of user-settable presets with offset and gain calibration. The cursor readings are three dimensional - frequency, amplitude, and time. Horizontal and vertical time scales are available. See the DL4YHF website.

Easygram
Uses an FFT engine (SPECTRUM.DLL) written by R.S. Horne, while the user interface is by Petr OK1FIG. Easygram is fine for many Amateur applications, and will do reasonably narrow spectrograms, but its use for frequency measurement is limited to coarser measurements due to lack of resolution. With adjustable speed and resolution it is also one of the better programs for drift measurement. The program has user definable settings, a frequency indicating cursor, and different display sizes depending on settings, which are very helpful, but it is not as easy to use as other programs. See the Easygram website.

Easygram in Doppler mode

EVM DSP Programs
Two excellent programs are offered by Peter G3PLX for the Motorola DSP56002EVM development board. These programs digitize the data with an external DSP unit, and send the data to a PC display application via a serial port. There are two particular advantages to this approach - very fine spectrograms with high resolution, and independence from the sound card (means you can run two spectrograms at the same time). For frequency comparison the Windows EVMDOP program is excellent. It offers fine resolution and a narrow 2.5Hz span. For wider applications, such as drift measurement, the DOS EVMSPEC program is excellent. These programs are available here: EVMDOP.ZIP EVMDOP.ZIP and EVMSPEC.ZIP.

A typically beautiful EVMDOP spectrogram (WWV backscatter on 10MHz plus local OCXO Reference)

PHASE COMPARATOR

The Phase Detector is used to measure the phase difference between signals, and from this and the time taken for the phase to change by a certain amount, frequency difference can be determined with great accuracy and high resolution. With very high precision signals there is really no other convenient method.

The phase comparator works best with local signals, but there are techniques to be described that allow comparisons to be made using off-air received References. The simplest phase comparator technique uses an oscilloscope to compare two oscillators in X-Y fashion. This technique isn't very helpful if one of the oscillators isn't sinusoidal in output, or there is a large frequency ratio between the frequencies compared, and the technique of triggering the oscilloscope from one signal and observing the drift on the other is then preferable. With this technique, frequency differences of 0.1Hz are measurable without effort.

A simple extension of this technique can be used to calibrate a Reference against a TV signal from a source known to use a Rubidium standard or better. Divide the reference signal down to 15625Hz (1MHz÷64) and feed a sample of this signal into the TV video, or even just into the antenna socket along with the TV signal. With the level adjusted correctly, two vertical lines will be visible on the picture - a black one, and a white one, about half the picture width apart. Place a sticker on the screen by one of these marks, and time how long it takes for the same mark to drift past, off the screen and back to the same point from the other side. If it takes one minute, the error in the reference is about 1 part in 10⁶; 10 minutes, 1 in 10⁷.

For higher resolution measurements, a purpose-built phase detector is the best answer. The traditional technique is to use a digital phase detector or analog mixer (depending whether the signals are digital or sinusoidal) which operates a meter or preferably a chart recorder from the output. Very small phase changes and frequency differences can be discerned by using 24 hour monitoring. Nowadays, the micro controller makes this a whole lot easier.

The author has developed a micro - based phase detector that fits inside a Standard Oscillator unit, allowing it to be calibrated and monitored with ease. The micro is clocked by the Standard Oscillator, and divides it down to provide 1kHz, 50Hz and 1Hz reference signals, and in addition, accepts a composite video signal from a TV receiver or video recorder. The vertical sync (PAL, 50Hz) is extracted and used to sample the phase of the local Standard with a high degree of precision. This phase measurement (a number 0 to 4999 for a 5MHz Standard) is transmitted via a serial data message for monitoring on a PC. Since the micro also keeps and transmits UTC time, this is also available for monitoring on the PC. The matching PC software displays phase over 10 hours or more, so drift can be monitored, and computes the frequency offset with accuracy that increases with time, to a resolution of one part in 10¹⁰. The Standard Oscillator is not locked or disturbed in any way, so its superb short-term stability is retained.

It so happens that the hardware design for this unit is almost identical to the Precision Transmitter design - you just leave off most of the switches, the VCXO components and the modulator and transmitter section. The ERROR output is used to drive a 100uA meter.

Reference Comparator Schematic

The author's PC Reference Phase Display

A Reference can be adjusted using this design in a matter of minutes, calibrated to the highest accuracy in a matter of 20 minutes, and its ageing rate checked easily on a regular basis, all without the use of a calculator. Long term monitoring also gives a good indication of the reliability of the TV network Reference and the stability of the local Reference oscillator.

REFERENCE SOURCES

In many parts of the world, HF and LF transmissions are available which allow reasonably good comparisons of local oscillators to be made - with care. The LF transmissions are very stable, and allow oscillators to be phase locked directly to the reference.

For convenience and portability, the commercial world now uses GPS as the principal source of accurate timing, including frequency reference. Although not easy to lock to, the 1pps (1Hz) pulses from specialized GPS receivers provide high accuracy, typically ±1us, and when integrated over time, frequency accuracy of 1 in 10¹⁰ is achievable. The NMEA (serial data) time messages from consumer GPS units are not suitable for frequency measurement or control.

In smaller or less developed countries, gaining access to a suitable quality reference for calibration or phase lock can be quite a problem, since there are no conventional transmissions available. The remaining choices are GPS (expensive) and TV signals (few are suitable). Without such resources the options are indeed limited.

HF Transmissions
Transmissions from WWV and the like have long been the mainstay of Amateur frequency checks. Unfortunately the signal quality for this purpose degrades quickly as the distance from the transmitter increases. The biggest problem is Doppler, where the received carrier frequency varies with time as a result of changes in the ionospheric refractive index. The picture below shows what happened to VNG (now QRT) on 16MHz.

16MHz VNG transmission

The straight line is a local GPS disciplined Reference - the wobbly line (two lines in places!) which varies by as much as ±2Hz or more (about 1 in 10⁷) is the VNG signal received at a range of about 2500km. The faint lines above and below are the second tick sidebands of the VNG signal.

LF Transmissions
WWVB (USA 60kHz), GBR (UK 60kHz), DCF77 (Germany 77.5kHz), JJY (Japan, 40 and 60kHz) and others offer high quality time signals (seconds ticks) along with very precise carrier frequencies. It is rather a challenge to phase lock to these signals, because the transmissions are keyed, but good results can be achieved. Some other LF signals such as European broadcasting stations also offer high carrier frequency accuracy and high transmitter power, making their signals very useful for phase locking. The BBC transmission on 198 kHz and Radio France on 162kHz are examples. Omega navigation signals (now QRT) and Alpha (Russian) navigation transmissions on VLF are less useful because they are keyed only with short pulses.

GPS
The GPS navigation system transmits very precise time from each satellite. This time is used by the GPS receiver to make precise measurement of satellite transmissions, in order to determine position. Not only is the time available to the user, but in some GPS units, notably the "OEM" bare circuit board type, a digital signal at 1Hz or 10kHz is also available. It is simple to phase lock to the 10kHz signal, but the accuracy is generally limited. It is rather more of a challenge to lock to the 1Hz "tick", but the results can be very good (a few parts in 10¹⁰). There are Amateur designs for GPS disciplined oscillators, such as that by Brooks Shera W5OJM or James Miller G3RUH. A similar unit was built by the author.

TV Transmissions
Most television broadcasting networks use a simple OCXO reference (or worse) to define the timing of their transmissions. The equipment will generally tolerate accuracies of 1 part in 10⁶, so although this is not a problem to normal TV reception, it renders the transmissions useless as a frequency reference.

However, some networks (notably TV1, TV2 in New Zealand and the ABC in Australia), use a Rubidium Standard, and so the timing of the signals generated can be very useful. In New Zealand the transmitted signal has better than 1 part in 10⁸ performance, and is monitored and reported regularly by the Government Measurement Standards Laboratory.

The PAL colour subcarrier (4433618.75Hz) can be extracted from a TV receiver and used to calibrate a frequency counter, but the best signals for phase locking are the horizontal sync (15625Hz) and vertical sync (50Hz). In NTSC countries, the frequencies are 3579545Hz, 15750Hz and 60Hz. The author prefers the vertical sync (50Hz) as it is a convenient factor of 1kHz and appears to offer better noise rejection than the line frequency. In addition, the frame sync is without the serration pulses present on the line sync output from most sync separators.

The author offers a small phase-locked transmitter design which allows precision transmissions to be made on 160 - 30m bands using TV phase lock.

---

Copyright © M. Greenman 1997-2005. All rights reserved. Contact the author before using any of this material.  HOME