Thursday, February 28, 2008

FM Transmitter harmonic measurements

Making FM transmitter output harmonic measurements
In this discussion, the transmitter consists of the transmitter plus any external harmonic filter and its directional coupler used for power output measurements. Measurements must be taken of the complete system to produce meaningful results.

The transmitter must be connected to a dummy load
You cannot make meaningful harmonic measurements by sampling inside the transmission line when the transmitter is connected to an antenna. This is because the antenna provides little or no load for harmonic voltages; therefore they can be anything at all, not controlled by terminating impedance. In fact, depending upon the distance between the antenna reflection and the sampling point, the observed harmonic voltages can exceed the fundamental voltage.

Since antennas used for FM broadcast are resonant, they will always attenuate harmonics. They function as band-pass filters. Therefore, if the transmitter is clean when operating into a resistive load, its radiated signal will be at least as clean having been through the band-pass filter created by the antenna when measured within the primary lobe of the antenna. This does not, however, mean that a received signal will be free from harmonics because many antennas may produce fundamental frequency nulls in their patterns without attenuating harmonics. In other words, if perhaps an aircraft receiver flies through an antenna null, harmonics at that location may appear to exceed those permissible by the Federal Communications Commission. Therefore, it is possible that a FM station may require additional harmonic attenuation to comply with the Commission's requirements even though the equipment by itself shows full compliance. Such problems may require field intensity measurements to resolve.

Sampling the signals
You use a capacitive probe in the transmission line. These have a known 6dB/octave 20dB/decade attenuation factor provable with simple math, and accepted by the FCC. This will appear to increase the "gain" at each harmonic according to the chart below. Then, after you establish a reference level of the fundamental, you attenuate the fundamental with a trap so that your spectrum analyzer is not producing the observed harmonics. You can readily calibrate your trap with a signal generator and your spectrum analyzer.

A simple trap that works well consists of about 4 turns of #14 wire 3/8” in diameter and a 15-100pF compression trimmer. You connect this parallel-tuned circuit between two BNC connectors in a small metal mini-box. You align the trimmer screw over a hole between the two BNC connectors. You can tune the trap to the low side (tighten the trimmer) when you set your spectrum analyzer reference level. Then you can loosen the trimmer to attenuate the fundamental by about 20dB. You can now remove 20dB of attenuation from the spectrum analyzer input to see the harmonics. Assume the second harmonic was shown at -60dB

It is really 86dB down, i.e., 6dB "gain" from the capacitor, and 20dB "gain" from removing 20dB of input attenuation. During calibration of your trap, you will find that it affects harmonics only about 1 or 2dB. A common error is to add 6dB of adjustment for each harmonic. This is incorrect; the third harmonic is not the second octave! The forth harmonic is!

Six decibel per octave chart
Fundamental 0dB
2nd Harmonic 6dB 3rd Harmonic 9.5dB 4th Harmonic 12.04dB
5th Harmonic 13.9dB 6th Harmonic 15.6dB 7th Harmonic 16.9dB
8th Harmonic 18.1dB 9th Harmonic 19.1dB 10th Harmonic 20dB

Capacitive probes
You use a capacitive probe because it automatically increases the level of the signals that you are measuring, improving the resolution of the test setup. Furthermore, other sampling methods such as directional couplers result in additional ripples in the frequency response that must be calibrated out. Dummy loads that provide a power sample, i.e., 20dB pad loads, result in additional measurement anomalies that also need to be calibrated. You can make a well-behaved capacitive probe by using an old insert (slug) that plugs into your thru-line power meter. You remove its components and drill a hole to contain a BNC connector. Solder a penny to a piece of #14 wire with the other end soldered to the center pin of the BNC connector. The penny should be placed even with the transmission-line side of the slug. You need to use an old-fashioned penny, one made from copper. Modern ones do not solder and seem to be made from a plated composite material. During your calibration, you will find that the penny functions as a good capacitor out to at least 10 times FM band frequencies.

Be sure to prepare a chart so you don’t need to calibrate your test setup every time you use it. Also make sure that the input of the spectrum analyzer has a resistive attenuator switched in every time you use it. The capacitive probe will only function properly if it is terminated in a resistor. A properly terminated coaxial line is a resistor, however if the input pad is entirely switched out, the line may be terminated by only the mixer diode, producing some strange results.

To produce meaningful results with guaranteed accuracy, you should substitute a signal generator for the transmitter and calibrate the system. Although the signal generator must have a known or constant output, its actual output level is not important. The entire test setup should be calibrated using the signal generator. You should observe that the voltage measured with the capacitive probe truly increases in amplitude by 6dB (doubles) for each octave. If this is not the case, then the fault in the test setup must be corrected before any other tests are made. Be sure to prepare a chart so you don’t need to calibrate your test setup every time you use it.

The author has performed measurements for FCC Type Acceptance for over 100 transmitters during a span of over 25 years. This does not constitute legal advice but only explains how and why it used to be performed by broadcast equipment manufacturers.