Noise Source Diagram

$25 to $50, and performs well. Every VHFer I know has one, with most of them waiting for a noise source to be usable. Earlier tube models, like the AIL 74 and the HP 340 and 342, have problems with drift and heat, but they can also do the job.

Another alternative is to build a noise figure meter.5

Using the Noise-Figure Meter

I'll describe the basic procedure using the model 75; others are similar, but the more complex instruments will require studying the instruction manual.

Input to almost all noise-figure meters is at 30 MHz, so a frequency converter is required (some instruments have internal frequency converters; except for the HP 8970, I'd avoid using this feature). Most ham converters with a 28-MHz IF work fine, unless the preamp being measured is so narrowband that a megahertz or two changes the NF. The input is fairly broadband, so LO leakage or any other stray signals can upset the measurement—this has been a source of frustration for many users. There are two solutions: a filter (30-MHz low-pass TVI filters are often sufficient) or a tuned amplifier at 30 MHz. Since a fair amount of gain is required in front of the noise figure meter, an amplifier is usually required anyway.

A noise source (which we will discuss in detail later I is connected to the rear of the instrument by a coax cable: a BNC connector marked DIODE GATE provides +28 V for a solid-state noise source, and high-voltage leads for a gas-tube noise source are also available on many versions. The noise-figure meter switches the noise source on and off. The noise output coax connector of the noise source is connected to the receiver input.

The model 75 has four function pushbuttons: OFF, ON, AUTO, and CAL. The OFF and ON positions are for manual measurements: OFF displays the detector output with the noise source turned off, and ON displays the detector output with the noise source turned on. If all is working, there should be more output in the ON position, and a step attenuator in the IF line may be used to determine the change in output, or Y-factor, to sanity-check our results. The knob marked GAIN is used to get the meter reading to a desirable part of the scale in the OFF and ON positions only; it has no effect on automatic measurements.

The AUTO position causes the instrument to turn the noise source on and off at about a 400-Hz rate and to calculate the NF from the detected change in noise. The model 75 has a large green light near the meter which indicates that the input level is high enough for proper operation—add gain until the light comes on. Then the meter should indicate a noise figure, but not a meaningful one, since we must first set the ENRdB using the CAL position. The lower scale on the meter is marked for from 14.5 to 16.5 dB of ENR; adjust the CAL ADJ knob until the reading in the CAL position matches the ENR of the noise source.

If the ENR of your noise source is outside the marked range, read the section below on homebrew noise sources.

Now that we have calibrated the meter for the ENR of the noise source, we may read the noise figure directly in the AUTO position. Before we believe it, a few sanity checks are in order:

•Manually measure the Y-factor and calculate noise figure.

• Insert a known attenuator between the noise source and preamp— the NF should increase by exactly the attenuation added.

• Measure something with a known noise figure (known means measured elsewhere; a manufacturers claim is not necessarily enough).

Finally, too much gain in the system may also cause trouble if the total noise power exceeds the level that an amplifier stage can handle without gain compression. Gain compression will be greater in the on state, so the detected Y-factor will be reduced, resulting in erroneously high indicated NF. The Gaussian distribution of the noise means that an amplifier must be able to handle 12 dB more than the average noise level without compression. One case where this is a problem is with a microwave transverter to a VHF or UHF IF followed by another converter to the 30-MHz noise-figure meter, for too much total gain. I always place a step attenuator between the transverter and the converter which I adjust until I can both add and subtract attenuation without changing the indicated noise figure.

One final precaution: noise-figure meters have a very slow time constant, as long as 10 seconds for some of the older models, to smooth out the ran

• A- - NEC 68119 #2 X- - NEC 85619 *— Noise/Com NC302L 9- - NC302L + chip att.

Fig 6—ENR of several versions of the noise source of Fig 5.

dom nature of noise. If you are using the noise-figure meter to "tweak" a receiver, tune very slowly!

Sky-Noise Measurement

Another way to measure noise figure at microwave frequencies is by measurement of sky noise and ground noise.3'6 Sky noise is very low, around 6 K at 10 GHz, for instance, and ground noise is due to the ground temperature, around 290 K, so the difference is nearly 290 K. At microwave frequencies we can use a manageable antenna that is sharp enough that almost no ground noise is received, even in sidelobes, when the antenna is pointed at a high elevation. A long horn would be a good antenna choice.

The antenna is pointed alternately at clear sky overhead, away from the sun or any obstruction, and at the ground. The difference in noise output is the Y-factor; since we know both noise temperatures, the receiver noise temperature is calculated using the ratio) and Ecl I2-

The latest version of my microwave antenna program, HDLANT21, will make this calculation.3 Since the measured Y-factor will be relatively small, this measurement will only be accurate for relatively low noise figures. On the other hand, they are the most difficult to measure accurately using other techniques.

A system for measuring sun noise was described by Charlie, G3WDG,

Noise Source Diagram

Fig 8—Schematic diagram of the homebrew noise source using a chip-resistor attenuator. C1 and C2 are microwave chip capacitors, ATC or equivalent.

that also works well for measuring noise figure from sky noise.7 He built a 144-MHz amplifier with moderate bandwidth using MMICs and helical filters that amplifies the transverter output to drive a surplus RF power meter. The newer solid-state power meters are stable enough to detect and display small changes in noise level, and the response is slow enough to smooth out flicker. Since my 10-GHz system has an IF output at 432 MHz, duplicating Charlie's amplifier would not work. In the junk box I found some surplus broadband amplifiers and a couple of interdigital filters, which I combined to provide high gain with bandwidth of a few megahertz, arranged as shown in Fig 3.1 found that roughly 60 dB of gain after the transverter was required to get a reasonable level on the power meter, while the G3WDG system has somewhat narrower bandwidth so more gain is required.

Several precautions are necessary:

• Peak noise power must not exceed the level that any amplifier stage can handle without gain compression. Amplifiers with broadband noise output suffer gain compression at levels lower than found with signals, so be sure the amplifier compression point is at least 12 dB higher than the indicated average noise power.

• Make sure no stray signals appear within the filter passband.

• Foliage and other obstructions add thermal noise that obscures the cold sky reading.

• Low-noise amplifiers are typically very sensitive to input mismatch, so the antenna must present a low VSWR to the preamp.

A noise-figure meter could also be used as the indicator for the sky-noise measurement, but a calibrated attenuator would be needed to determine the Y-factor. Using different equipment gives us an independent check of noise figure so we may have more confidence in our measurements.

W2IMU suggested that the same technique could be used for a large dish at lower frequencies.8 With the dish pointing at clear sky, the feedhorn is pointing at the reflector, which shields it from the ground noise so it only sees the sky noise. If the feedhorn is then removed and pointed at the ground, it will see the ground noise.

Noise-figure meters are convenient, but if you don't have one, the equipment for measuring sun and sky noise could also be used indoors with a noise source. The only complication is that the Y-factor could be much larger, pushing the limits of amplifier and power meter dynamic range.

Noise Sources

The simplest noise source is simply a heated resistor—if we know the temperature of the resistor, we can calculate exactly how much noise it is generating. If we then change the temperature, the noise output will change by a known amount. This would work if we could find a resistor with good RF properties whose value does not change with temperature, an unlikely combination. There are commercial units, called hot-cold, noise sources, with two calibrated resistors at different temperatures and low VSWR. Typically, one resistor is cooled by liquid nitrogen to 77.3 K (the boiling point of nitrogen), while the other is heated by boiling water to 100° C, or 373.2 K. The preamp is connected to first one resistor, then the other; the difference in noise output is the Y-factor.

Since the boiling point of pure liquids is accurately known, this type of noise generator can provide very accurate measurements. However, they are inconvenient to use, since the receiver must be connected directly to alternate resistors (the loss in an RF switch would significantly reduce the

Fig 7—The homebrew noise source of Fig 8.

0 0

Post a comment