Single Sampler Heterodyne Method for Wideband Frequency Measurement

Some previous frequency measurement apparatus have utilized a single sampler. See, for example, U.S. Pat. No. 3,984,770 issued on Oct. 5, 1976 to David C. Chu and entitled "Frequency Measurement Using Phase Continuous Frequencies Switching". These single sampler systems are much to be desired over the prior art dual sampler systems such as that shown in U.S. Pat. No. 3,836,758 issued on Sept. 17, 1974 to Richard F. Schneider and Arthur R. Bloedorn and entitled "Wide Frequency Range Counter System Utilizing Automatic Range Searching and Loop Compensation". This is because significant cost savings can be achieved by eliminating one of the samplers required.

However, previous single sampler systems have typically utilized techniques wherein the single sampler is alternately driven by signals of two different preselected frequencies. The disadvantage of this technique is that changes in the input signal frequency due to frequency modulation or drift between the time the input signal is sampled using the signal of the first frequency and the time that the input signal is sampled using the signal of the second frequency may result in errors in the calculation of the harmonic number and sideband of the input frequency.

In accordance with the preferred embodiment of the present invention, a signal of unknown frequency is input into a sampler. The sampler is driven by a sampling signal which is caused to vary in an irregular or random manner by a control signal. The IF output of the sampler is gated to a counter using synchronous detection by the control signal. The correlation of the modulation applied to the sampling signal with the resultant frequency modulation of the IF signal cancels out frequency modulation of the input signal which may be present. Using the IF frequency value from the counter, one can compute N and the sidebands, since the IF frequency will vary in the same way as the sampling frequency, except for a scale factor which is equal to the harmonic number. If the frequency of the input signal is also varying, the effect of this variation on the measurement accuracy will be greatly diminished by the correlation process .

FIG. 1 shows a basic block diagram of a single sampler system driven by a deterministic or random signal for use in frequency measurement.

FIG. 2 is a block diagram of a single sampler system using two frequency counters and two distinct driving frequencies.

Referring now to FIG. 1, there is shown a block diagram of a single sampler system for frequency measurement. An input signal of unknown frequency is input on a line 110 to a sampler (or harmonic mixer) 115. The sampling (or mixing) signal on line 118 is provided by a variable frequency oscillator 120. The frequency of the signal on line 118 varies in a random manner in response to the signal on line 125, which is provided by signal generator 130. Signal generator 130 may provide a deterministic or a truly random signal on line 125 or simply one that varies in a pseudo-random or irregular manner.

The frequency of the signal on line 116, of frequency f.sub.IF, will vary in response to the signal on line 118. However, the IF frequency variation will be related by a scale factor equal to the harmonic number, i.e., .+-. N. This relationship is explained below;

where f.sub.0 is the average local oscillator frequency of the signal on line 118; f.sub.x is the average frequency of the input signal; .DELTA.f.sub.0 (t) is the applied random variation in response to the random or deterministic signal on line 125; and .DELTA.f.sub.x (t) is the variation if any in the frequency of the input signal. Then we have for the upper and lower sideband cases:

By computing the cross correlation between the applied frequency variation .DELTA.f.sub.0 (t) with the resultant .DELTA.f.sub.IF (t), we have

If we make .DELTA.f.sub.0 (t) independent or uncorrelated with .DELTA.f.sub.x (t), then the last term of C can be made negligible, and hence can be disregarded. This can be done by using a random variation of the frequency f.sub.0.

Defining R = Corr (.DELTA.f.sub.0 (t), .DELTA.f.sub.0 (t)) and with the last term eliminated, we can evaluate N as:

(round means round to nearest integer of the exact value computed.)

The fact that N is an integer number has been used to cancel any errors in the determination of the harmonic number N. Furthermore, the sign of N indicates which sideband is present.

Referring again to FIG. 1, there is shown the block diagram of the preferred embodiment. Sampler 115 provides the signal on a line 116 in response to the input signal on line 110 and the sampling signal on line 118. IF amplifier 135 amplifies and filters the signal to produce the IF frequency signal on line 137.

The tuning control signal on the line 121 is adjusted so that the IF amplifier output signal on line 137 is approximately in the center of the IF amplifier frequency band when an input signal is present on line 110. When an input signal is being sampled, signal generator 130 provides a random or pseudo-random control signal on line 125 which causes variable frequency oscillator 120 to vary the sampling signal on line 118 in a random or pseudo-random manner. Also, gate 140 is enabled by the control signal on 131 and the IF frequency of the signal on line 137 is measured by a counter 145. Furthermore, the random or pseudo-random signal on line 125 from signal generator 130 is used to drive a correlator 150 to correlate the IF amplifier output on line 137 with the signal on line 125. Since the IF frequency on line 137 varies in the same manner as the signal on line 118, except for a scale factor, .+-. N, one can determine both the harmonic number and the sideband present from the signal on the line 151, as described more fully above. Computing and display circuits 160 may then compute f.sub.x from the harmonic number and sideband information from the signal on line 151 and the measured IF frequency from counter 145. This is possible using the formula:

in FIG. 2 there is shown a variation of the basic circuitry shown in FIG. 1. The circuit of FIG. 2 uses two frequency counters to replace correlator 150. Instead of the signal on line 118 varying randomly over a range of frequencies, it is switched between two distinct frequencies, f.sub.1 and f.sub.2 in a random or pseudo-random manner. The IF frequency on line 137 is then synchronously detected and measured by frequency counters 201 and 202. As in the system of FIG. 1, a measurement control signal on line 131 enables the counting of the IF frequency. Knowing f.sub.1 and f.sub.2 and f.sub.IF1 from counter 201 and f.sub.IF2 from counter 202, one may now calculate f.sub.x per the following formula: ##EQU1##

For greater accuracy, an additional measurement may be made. Because the IF frequencies are being chopped between the times that frequency counter 201 and frequency counter 202 are enabled, there will be a count error of .+-. one count each time the counters are switched. This error may be cumulative and will not necessarily cancel out in every case. Therefore, after f.sub.IF1 and f.sub.IF2 have been determined using frequency counters 201 and 202, one may disable the random control signal on line 125 and cause f.sub.1 to be gated to line 118 for a predetermined period. During this time, one can enable either counter 201 or 202 to count the IF frequency f.sub.IF0. Using this method, a more accurate IF frequency would be obtained, since there would be no possibility of cumulative .+-. one count errors and the frequency f.sub.x would be calculated as follows: ##EQU2##

Note that the result of the division of (f.sub.IF1 -f.sub.IF2) by (f.sub.1 -f.sub.2) is the harmonic number with a small error which is primarily caused by modulation present in the input signal on line 110 which is not cancelled out using the technique of the present invention. However, this residual error will not interfere with the accuracy of the measurement, unless it is quite large, since the harmonic number is always an integer. Therefore, the calculated value is simply rounded to the nearest integer, and the small error which may be present is of no effect in the determination of f.sub.x.