Frequency Measurement

MEASURING FREQUENCY

In past years, Amateurs had no great need to know exactly where their transmitters were transmitting, or their receivers receiving, provided they could hear each other and stay within the band edges. Stability was not much of an issue either, so long as the signal could be followed. Equipment was either "crystal controlled" (and therefore always right!), or if a VFO was used, some sort of calibrator was used when operating close to band edges. Heterodyne frequency meters were sometimes used, as were simple crystal marker oscillators, GDOs and wavemeters, while digital frequency counters only existed in laboratories.

Now, there is a much greater need for both accuracy and stability. High frequency band occupancy is high, especially in the narrow-band digital sections; the newer (WARC) bands are quite narrow, with band edges that are not on "round 100kHz" points, and there is greater interest in narrow-band and weak signal operation, especially on LF and VHF/UHF. For weak-signal skeds on 70cm, frequencies need to be known to 10Hz or better in order to find a signal - and that's about one part in 50 million!

This trend in increasing need for precision over the years is illustrated in the above graph. As an approximation, the level of frequency precision required for the cutting edge of Amateur Radio has arguably increased almost one power of 10 per decade since the 1950s!

Before we go any further, we need to define some technical terms related to frequency measurement and accuracy.

DEFINITION OF TERMS

Accuracy

Accuracy describes how well the equipment is calibrated. Accuracy is of lesser importance than stability, as there is no point in calibrating equipment carefully if it won't be in the same place next time you check! If the equipment is stable but inaccurate, you can at least write down the calibration error and subtract it from your readings. See the picture below which illustrates accuracy and stability.

Ageing

All oscillators change frequency slowly with time. Crystals that vibrate mechanically will have molecular level changes that minutely affect the oscillation frequency, so even if nothing else changes (oscillator supply voltage, capacitance, temperature), the crystal will slowly change frequency, usually increasing in frequency. Old oscillators are frequently better in this regard as ageing is fastest with new oscillators. Caesium and Rubidium Standards are used commercially because they have very low ageing rates.

Calibration

The process of referring (or transferring) a Reference to a Standard, for the purpose of determining the offset in the Standard from the Reference. The Standard might be adjusted, or at higher accuracies, since adjustments become quite difficult, the offset might simply be noted and taken into account during subsequent measurements. A "Transfer Standard" is a calibrated Standard that is portable, and is taken to other Standards for calibration. The alternative method is to receive the Reference (typically by radio), and perform the calibration remotely. This is what Amateurs can most easily do.

OCXO

A crystal oscillator in an oven - Oven Controlled Xtal Oscillator. Crystals designed for use in ovens have a thermal characteristic that has a plateau at the oven temperature, hence they are very stable, provided the oven temperature is stable. Operating temperatures are typically 70 - 80°C, so above even the highest ambient temperatures. The best OCXOs are the "double oven" type, with the oscillator and crystal in an inner oven, with proportional control, and the whole inside an outer oven. Even the cheaper OCXOs can benefit from good thermal lagging provided by a polystyrene cover. OCXOs have slow warmup and use more power than other types. OCXO stabilities can be as high as 1 in 10⁸, with ageing rates as low as 1 in 10⁹ per day, so make a good choice for an Amateur primary reference.

Offset

The difference between a Standard (or equipment being measured) and a Reference is called the offset. This is also usually quoted in ppm, and represents the current working error correction due to all error sources. You need to work this out just before, and just after making any critical frequency measurements, and use the mean to correct the readings.

PPM

Frequency accuracy, stability and ageing are typically quoted in "ppm" (parts per million). This is because invariably these factors are dependent not on any error in Hz, but in the amount of error per Hz. While this could be expressed as a percentage, the very small numbers that would result are not convenient. Hence, the error is typically expressed in parts per million. An error of 1ppm is the same as 0.0001% or 1 part in 10⁶. Reference sources are quoted with errors as small as 0.000001ppm, or 1 part in 10¹². Typically the accuracy, stability and ageing rate per week are numerically similar - for example, an 0.1ppm stability TCXO will have a calibrated accuracy of around 0.1ppm and an ageing rate of less than 0.1ppm/week.

Reference

A device with higher (and traceable) inherent accuracy than the Standards it calibrates. Invariably the Reference will be traceable to a national or international standard, and the Reference must have at least one power of 10 better performance (accuracy, stability) than the devices it calibrates. If the reference is received by radio, it's received version must have the necessary stability.

Spectrogram

A widely used technique among precision frequency enthusiasts, which allows the frequency of oscillators and other signals to be compared over time. The Spectrogram uses a Fast Fourier Transform (FFT) mathematical technique to analyse sound, typically from a PC sound card, to generate a graph of frequency (vertical axis) against time (horizontal axis), with signal strength as brightness. The image below is a high quality spectrogram, captured using EVMDOP by G3PLX and a Motrorola DSP56002EVM DSP processor.

Stability

Stability is a measure of how well the equipment stays where you put it. It is defined for crystals and oscillators as the change in operating frequency over the quoted temperature range. Modern synthesized rigs rely on one or more crystal oscillators, and so (in comparison to free-running oscillators) are reasonably stable. The best equipment refers all frequency generation to a single high stability oscillator, chosen for the best possible results. A stable rig will move very little during warm-up, and will not change frequency much when the weather is warm or cold. It can be very irritating when monitoring precise signals if the receiver drifts when you open the shack door! See the picture below which illustrates accuracy and stability.

Standard

A calibrated oscillator, used to measure other oscillators. Normally the Standard will be on a "round" frequency such as 1, 5 or 10MHz. The Standard may be built into equipment such as a frequency counter or signal generator, or in a separate box simply used as a signal source. The Standard needs of course to have suitable stability, low ageing, and to be regularly calibrated.

TCXO

Temperature Compensated Crystal Oscillator - a type with superior performance to a normal crystal oscillator, since the temperature variations are corrected by a thermistor and varicap (variable capacitance) diode. Some of the better units use quite sophisticated control methods. TCXOs use much less power than OCXOs, so are good for portable equipment, but the stability is more modest - 1 in 10⁶ to about 2 in10⁷.

Variance

Variance is a measure of how likely the oscillator is where you think it is at any given time. Oscillators always have some noise, and variance is a statistical measurement of this performance limitation. Variance is usually quoted as the Allen Variance, which is a measurement involving the Standard Deviation of the oscillator, and always has the measurement time period (Tau) quoted. Good oscillators will have low drift (ageing) and low noise (variance).

VCXO

Voltage Controlled Crystal Oscillator. Used in phase-locked devices such as the ZL1BPU Exciter, and other devices where a crystal oscillator is "steered" to follow another standard (the technical term is "disciplined"). A varicap (variable capacitance) diode is used to alter the crystal frequency very slightly. Some TCXOs and OCXOs have voltage control, and so offer better performance.

Illustrating Accuracy, Stability and Variance

The above picture principally illustrates the difference between Accuracy and Stability. In this picture, the warmup characteristic of a good receiver is illustrated over the first 100 minutes from cold switch-on. The receiver is a Kenwood TKM-707, with single reference oscillator, a single oven OCXO at 30.3Mhz.

The receiver was tuned on USB to 14.999000MHz, and receiving the third harmonic of a 5MHz precision reference at 15.000000MHz. Hence the receiver output was an audio tone at about 1000Hz. Once warmed up, the line has an offset from 15.000Mhz of about -1Hz, which is the Accuracy (about -6 in 10⁸). Note the variation in the line - after warmup there are slight wiggles - the peak-to-peak value is the Stability, about 0.5Hz, or 3 in 10⁸. Once the thermal and power supply factors are removed, all that causes the frequency to be unpredicatble is the effect of system noise - the Variance, which is of the order of 1 in 10⁸ in 1000 sec. Only the very best receivers do better than the one illustrated.

Note also the warmup time is about 20 minutes. The faint clouds in the background are from WWVH in Hawaii, also operating on 15MHz.

MEASUREMENTS IN THE SHACK

These days, digital frequency counters are to be found in most shacks. With a good counter, it is the work of but a moment to check the frequency of an oscillator. Of course the counter also needs regular calibration, but we'll deal with that later.

But what to do if you don't have a counter? Well you could of course build a counter, but there are other possibilities if you have a general coverage receiver and a crystal calibrator, or a general coverage receiver with digital readout. These techniques also apply to measuring signals on the air, and are covered in the next section. It must be said however, that these techniques do not generally give high quality results because of receiver limitations. In particular, receivers with analog dials do not easily yield good results, unless the dial is very well marked and carefully linearized (the Collins 51J family for example).

The days of the heterodyne frequency meter, Lecher lines, the GDO and the absorption wavemeter are long gone!

MEASUREMENTS ON THE AIR

Using a Calibrator
Tune in the signal to be checked (on a local signal you could tune to a harmonic if necessary). Note the receiver frequency reading that gives zero beat in SSB mode. Then tune in the nearest calibrator marker, tune to zero beat, and note the frequency discrepancy from what you expect. Subtract this from the original reading. This technique, used with care, will yield results of a few Hz on 80m (1 part in 10⁶). Many Frequency Measuring contesters still use this technique, or a variation of it.

Using a Receiver with Digital Readout
Check the calibration of the receiver against a calibrated source (reference oscillator or on-air signal). You may need to ensure that the measurement is made on the same band as the signal to be measured, if the receiver uses different offset oscillators on each band. Note the receiver offset for zero beat. Then tune in the signal, and subtract the offset from the measured frequency. This technique can also yield results of a few Hz on 80m (1 part in 10⁶).

Using a Receiver and Audio Measurement
Both the above techniques can be improved by using a fixed audio frequency instead of zero beat. This is because the exact zero is hard to determine. You don't need an especially accurate audio reference, so long as it is stable between readings. A 1kHz tone from a calibrator, or even a note from a musical instrument will do. Best of all, use a tone and an oscilloscope, or a computer. Perform the measurements as described, but adjust until the received beat tone in SSB is the same as the reference tone (make sure you use the same sideband each time). With an oscilloscope, trigger the trace off the reference tone, and adjust the receiver until the trace is stable, and shows the same number of cycles across the screen as the reference (or use an X-Y mode and tune for an elliptical pattern). This technique can yield results of a about 1Hz on 80m (well under 1 part in 10⁶). The limitation will be reading the receiver dial. With many modern transceivers, the tuning can be made to go in 10 or 100Hz steps, which makes these readings easier.

You can also use a cheap counter (or a computer with sound card rigged to measure audio frequencies) to measure or set the audio frequency.

More Sophisticated Techniques
This isn't the place to describe super-accurate measurements. For these you need to see the Frequency Measurements for Experts page.

CALIBRATING A FREQUENCY COUNTER

If you don't have a precision reference at least one power of 10 better than the quoted accuracy and stability of the counter, don't even consider adjusting the counter yourself! It is possible to adjust a counter by measuring the 4.433619Mhz oscillator in a TV receiver, but you must use a TV station known to use a Rubidium Standard, and the risk of frying yourself inside the TV set is high!

The best solution is to take the counter to a precision reference for calibration, or to borrow a precision reference. If you live in the greater Auckland area, you can borrow a quality Transfer Standard (portable precision reference) from ZL1BPU for 24 hours at no charge. It has an output at 5.000Mhz and is battery operated. If you live elsewhere, ask at your local radio club. Commercial radio and cellular phone technicians usually have access to a good reference. You can also build a TV-referred local standard with 2 in 10⁸ performance fairly inexpensively.

Adjusting the counter to a precision reference is simplicity itself! Set the counter and precision reference running for at least eight hours, use the counter to count the reference using the longest gate time available (highest precision), and then adjust the counter reference oscillator until the display reads all zeros, or just flickers between 999999 and 000000.

If there is no available adjustment, or you are unable or unwilling to adjust the oscillator in the counter, note the error and calculate the error in PPM. Write this value on a sticker on top of the counter, and add the date. This way you will know the offset to use in measurements, and the ageing trend next time you calibrate. Even if you do adjust the oscillator, it is a good idea to add a sticker with the calibration information. Calibration should be performed at least every year, and perhaps more frequently during the first year of operation for a new reference or counter.

---

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