144 432 903 1296 5760 10368 Frequency (MHz)

Fig 11—ENR of the PC-board noise sources (Fig 10).

144 432 903 1296 2304 3456 5760 10368 Frequency (MHz)

Fig 12—Measured VSWR of the PC-board noise sources (Fig 10).

Enhance Miniature Construction with Optical Feedback

Many microwave construction projects, and most modern equipment, use extremely small surface-mount components, Working with these parts requires steady hands and good vision. And as we get older, our vision usually deteriorates—I got my first bifocals last year.

I'm convinced that the key to working with tiny parts is to see them well. When I built the first noise source with the tiny 0402-size (1 x 0.5 mm) chip resistors, it was frustrating trying to get the resistors soldered where I wanted them. After that experience, I was on the lookout for a surplus stereo microscope and finally located one at a reasonable price. These microscopes are commonly used for microelectronics assembly work, providing moderate magnification at a long working distance.

After I set up the microscope on my workbench with adequate illumination, I was ready to build some more noise sources. Now the tiny chip resistors were clearly visible, and I was able to hold them in place with tweezers while soldering them exactly where I wanted them. On another project, I wanted a clearance hole in the ground plane around a hole drilled though a Teflon PC board. The hole is 0.025 inches in diameter, and I was able to cut an octagon around it with a hobby knife; the length of each side of the octagon is about the same as the hole diameter. Then I lifted the unwanted copper with the point of the knife. Magnification makes miniature work feel precise and easy instead of clumsy and frustrating.

What the microscope does is add gain to the feedback loop from the eyes to the hand. Our hands are never perfectly steady, but adding this feedback steadies them under the microscope, as the brain takes input from the eyes and automatically compensates (after a bit of practice).

A microscope is an elegant solution for very small parts, but any optical magnification helps. I have also used magnifiers, jewelers loupes and "drugstore" reading glasses. If the reading glasses are stronger than you need, they will provide additional magnification; just don't try walking around wearing them.

Other aids to miniature work are tweezers, fine-point soldering irons and lots of light. When an object is magnified, proportionally more light is required for the same apparent brightness. Tweezers help in holding small objects—I prefer the curved #7 style Swiss tweezers, of stainless steel so solder won't stick. Finally, a temperature-controlled soldering iron prevents overheating, which can destroy the solder pads on surface-mount components; 700° F tips are hot enough. All the tools I've mentioned came from hamfests, surplus places or flea markets, at reasonable prices.

So, even if you think that microwave project with tiny parts is beyond your capability, use a magnifier and give it a try. I'll bet you surprise yourself.—N1BWT

11, was higher at 10 GHz than the hand-cut one and reasonably flat with frequency—and consistent from unit to unit. The VSWR, however, was still high at 10 GHz, as shown in Fig 12. It is difficult to make a really good coax-to-microstrip transition at 10 GHz! Ordinarily, in an amplifier, we simply tune out the slight mismatch as part of the tuning procedure, but broadband tuning is much more difficult. As a final improvement, I dug up some 5-dB SMA attenuators from the swap session at Microwave Update last year. Adding one of these to the worst unit in Fig 12 reduced the VSWR to 1.18 at 10 GHz (below 1.10 at lower frequencies) and the ENR to 5.0 dB. This performance is every bit as good as a very expensive commercial noise source, lacking only NTIS-traceable calibration.

Noise-Source Alignment

The only alignment requirement for a solid-state noise source is to set the diode current; the current is always set at the highest frequency of interest. A noise figure meter must be set up with converters, etc, for the highest frequency at which the noise source might be used and set to display the detector output (OFF position on a model 75). Then voltage from a variable dc power supply is applied to the noise diode through the 1-kP current-limiting resistor. The detector output should increase as the voltage (diode current) increases, reach a peak, then decrease slightly. The optimum current is the one that produces peak output at the highest frequency (I set mine at 10 GHz). The optimum current for the NC302L noise sources was between 7 and 12 mA. Then additional resistance must be added in series with the current-limiting resistor so that the peak output occurs with 28 volts applied, so that the noise source may be driven by the noise-figure meter. Once the proper resistor is determined and added, the dc end of the noise source is connected to the diode output of the noise-figure meter and the meter function is set to ON. This should produce the same detector output as the power supply.

Then the meter function is set to AUTO and the meter should produce some noise-figure indication, but not yet a calibrated one. However, it is good enough to tune up preamps—a lower noise figure is always better, even if you don't know how low it is.

Noise-Source Calibration

Much of the high price of commercial noise sources pays for the NTIS-traceable calibration. Building a noise source only solves part of the problem—now we need to calibrate it.

The basic calibration technique is to measure something with a known noise figure using the new noise source, then calculate what ENR would produce the indicated noise figure.

Fortunately, the calculation is a simple one involving only addition and subtraction; no fancy computer program required. Simply subtract the indicated noise figure, NFinciicato(j, from the known noise figure, NFactua], and add the difference to the ENR for which the meter was calibrated, ENRca|:

ENR(miLse smircel = ENRca| + (NFactual - NFindicated> Eq 15

This procedure must be repeated at each frequency of interest; at least once for each ham band should be fine for amateur use.

The known noise figure is best found by making the measurement with a calibrated noise source, then substituting the new noise source so there is little opportunity for anything to change. Next best would be a sky noise measurement on a preamp. Least accurate would be to measure a preamp at a VHF conference or other remote location, then bring it home and measure it, hoping that nothing rattled loose on the way. If you can't borrow a calibrated noise source, it would be better to take your noise source elsewhere and calibrate it. Perhaps we could measure noise sources as well as preamps at some of these events

Using the Noise Source

Now that the ENR of the noise source has been calibrated, the noise-figure meter calibration must be adjusted to match. However, the model 75 in the CAL position has only 2 dB of adjustment range marked on the meter scale. Older instruments have no adjustment at all. However, we can just turn around the equation we used to calculate the ENR and calculate the NF instead:

NFactual = NFindicated + (ENR(noise source) " ENRcal) Eq 16

There is a short cut. My noise source has an ENR around 12 dB, so I set the CAL ADJ in the CAL position as if the ENR were exactly 3 dB higher, then subtract 3 dB from the reading. Even easier, the meter has a +3DB position on the ADD TO NOISE FIGURE switch. Using that position, I can read the meter starting at 0 dB. Any ENR difference from 15 dB that matches one of the meter scales would also work—rather than an involved explanation, I'd urge you to do the noise figure calculations, then try the switch positions and see what works best for quick readout.

Reminder: Noise-figure meters have a very slow time constant, as long as 10 seconds for some of the older models, to smooth out the random nature of noise. Tune slowly!

Don't despair if the ENR of your noise source is much less than 15 dB. The optimum ENR is about 1.5 dB higher than the noise figure being measured.1 The fact that today's solid-state noise sources have an ENR around 5 dB rather than the 15 dB of 20 years ago shows how much receivers have improved.


The value of noise-figure measurement capability is to help us all to "hear" better. A good noise source is an essential part of this capability. Accurate calibration is not necessary but helps us to know whether our receivers are as good as they could be.


'Pettai. R., Noise in Receiving Systems, Wiley, 1984.

2Graves, M. B., WR0I, "Computerized Radio Star Calibration Program," Proceedings of the 27th Conference of the Central States VHF Society, ARRL, 1993, pp 19-25.

3Wade, P., N1BWT, "More on Parabolic Dish Antennas," QEX, December 1995, pp 14-22. The HDLANT21 program may be downloaded from

4Pastori, W. E., "Direct-Reading Measurement of Receiver-Noise Parameters," Microwave Journal, April 1973, pp 17-22.

5Bertelsmeier, R., DJ9BV, and Fischer, H., DF7VX, "Construction of a Precision Noise Figure Measuring System," DUBUS Technik 3, DUBUS, 1992, pp 106-144.

6Fasching, H., OE5JFL, "Noise Figure Measurement using Standard Antennas," DUBUS Technik 4, DUBUS, 1995, pp 23-25.

7Suckling, C,, G3WDG, "144 MHz wideband noise amplifier," DUBUS 2/1995, DUBUS, 1995 pp 5-8.

8Turrin, R., W2IMU, "Method for Estimating Receiver Noise Temperature," Crawford Hill Technical Note #20, September 1986.

9Britain, K., WA5VJB, "10 GHz Noise Source," Microwave Update "87, ARRL, 1987, p 63.

10Wade, P., N1BWT, and Horsefield, S., NR1E, "Homebrew Solid State Noise Sources," Proceedings of the 1992 (18th) Eastern VHF/UHF Conference, ARRL, 1992.

11Sabin. W. E., W0IYH, "A Calibrated Noise Source for Amateur Radio," QST, May 1994, pp 37-40.

12Valley, G. E., Jr., and Wallman, H., Vacuum Tube Amplifiers, MIT Radiation Laboratory Series, McGraw-Hill, 1948.

13Wade, P., N1BWT, "Building VHF Power Attenuators," QEX, April 1994, pp. 28-29. The PAD.EXE program may be downloaded from

14PC boards are available from Down East Microwave. , o 5

in m

0 0

Post a comment