This note describes a greatly improved method for measuring cross modulation (X-MOD), composite triple beat (CTB), and composite second order (CSO) distortions. Besides providing greater accuracy this method is insensitive to the phases of the components of composite distortions. Linear detectors such as used in spectrum analyzers are phase sensitive and as a result their measurements may vary several dB over a period of several seconds. A more complete description along with a theoretical analysis of this method is covered in MATRIX TECHNICAL NOTE MTN-110.

The traditional method for making (CTB) and (CSO) measurements uses an RF spectrum analyzer for these measurements.

Because these signals have a bandwidth of approximately 20 KHz, there is a lower limit to the distortion that can be measured. This is determined by the weakest signal level and the system noise figure at the weakest signal point. For accurate measurements the measured distortion should be 5 to 10 dB above the noise floor. Corrections are sometimes possible for the case where the measurements are as little as 3 dB above the noise floor but at these levels other errors such as detector non-linearity and stability of the noise levels enter into the picture. Noise floors in the order of -90 dB are possible with some care. High gain devices present the more difficult problems because their input signals are necessarily small.

The MATRIX method of making these measurements overcomes some of these limitations by modulating the carriers with a low frequency and measuring the detected distortion in a narrow frequency band. The result is that noise floors of -110 dB are quite practical. The limitation of this measurement is no longer the RF bandwidth but rather the final bandwidth and of course the time the narrow bandwidth imposes on the measurement. Measurements of distortion in the order of -100 dB take about 10 seconds. Naturally, distortions further from the noise floor can be made in much less time.

The MATRIX method of making these measurements measures the true carrier to distortion power ratio. Spectrum analyzers measure noise and noise-like signals with an error of 2.5 to 3 dB (with the true power being greater than the measured). This error is a result of the log display scale. This is the correction factor used when making carrier-to-noise ratio measurements. By convention, this correction is not used when making carrier to (CTB) or (CSO) measurements.

Since the MATRIX method measures true carrier-to-distortion power, +2.5 to +3 dB must be added to these measurements to provide agreement with traditional methods.

Instrument settings are described for each step in the tests. The setup block diagram is shown in figure 1.

The Multiple Frequency Signal Generator must provide a test signal containing distortion products much smaller than those to be measured. Each carrier must be individually controlled and must be adjusted as described below for the distortion tests.

The RF spectrum analyzer must have an IF output in the range of 3 to 30 MHz. This is then demodulated in the SQ1 Mean Square Module. Special Mean Square Modules are available for frequencies up to 400 MHz.

The FFT spectrum analyzer is used as a narrow band filter to reduce noise and allows for accurate amplitude measurements.

Note: BOLD LETTERING indicates "hard keys" on the RF spectrum analyzer and FFT analyzer. Settings on these instruments not indicated do not affect measurements.

Generator Settings:                Required Carriers on 100% modulate all carriers at 15.75 KHz

Bandpass Filter Setting:        Set to frequency where measurement is to be made.

FREQUENCY

CENTER FREQUENCY        Set to frequency where measurement is to be made.

SPAN                                        0 HZ

RESOLUTION BW                 1 MHz Preferred, 100KHz Minimum

AMPLITUDE

REF LEVEL                            0 dBm

UNITS MENU

dBVrms

RETURN

WINDOW MENU

FLATTOP

RETURN

Top Ref

(ENTER) 20

dBV rms

Y/Div

(ENTER) 20

dB

INPUT RANGE

(ENTER) 10

dBV

MARKER WIDTH

(TOGGLE TO) NORM

MARKER SEEKS

(TOGGLE TO) MAX

GRID DIV/SCRN

(TOGGLE TO) 10

SPAN

(ENTER) 1.56

KHz

CENTER FREQ

(ENTER) 15.750

KHz

Adjust manual control to bring marker rails to center screen.

NOTE: If desired the span may be reduced. The span setting determines the resolution bandwidth and so the noise floor. It also determines the time it takes to make the measurement. The measurement time is displayed as "Acq time" on the FFT screen. For 1.56 KHz span this is .25 sec. For a 48.75 Hz span this time is 8 sec.

You should have a signal in the middle of the screen with a marker at the peak. The magnitude of the marker should be about -10 dBVrms. The signal level output of the squaring circuit must be less than -5 dBVrms for linear operation. Adjustment of the RF spectrum analyzer may be required if the output of the squaring circuit exceeds -5 dBVrms.

(MARKER)>REF                   Y should = 0 dB NOTE: Level must be less than -5 dBVrms

Generator setting:                   Measurement channel to CW

INPUT RANGE

dBV

Overload light should be off, if not adjust "input range" higher until overload indicator light is off.

READ Marker value "Y", (Y is usually a negative number)