An Example of Gear for the 145GHz Amateur Band

Brian Justin, WA1ZMS

I. Introduction
II. Understanding the Difficulties
III. Propagation
IV. Antennas
V. Phase Locking
VI. Conclusion
VII. Acknowledgements
VIII. References

I. INTRODUCTION

As interest in amateur radio millimeter-wave operation grows, it falls upon the shoulders of today's operators on those bands to document our efforts in order to further others' interest for tomorrow. To that point, this paper is a presentation of the author's current work on the 145 GHz amateur band. Although several expensive and hard-to-obtain subassemblies are used in the documented design, the information is presented as an attempt to foster others' thoughts, ideas, and conversations around the design of such millimeter wave gear. What seems unobtainable today may very well be common place in just a few years as more commercial surplus millimeter-wave equipment appears on the used market and at flea markets.

II. UNDERSTANDING THE DIFFICULTIES

Link Budgets

Before anyone can even begin to realistically build gear for bands such as 145 GHz, it's helpful to have a good understanding of the types of problems that could be encountered. Following the experience gained from previous work on 47 and 75 GHz (1), one realizes that atmospheric losses must be dealt with if any type of DX is to be obtained, i.e.: >5km. Overcoming these losses is a matter of having sufficient transmitter power and a low enough receiver noise figure. This may seem over simplistic, but before starting a six-month project such as building gear for a millimeter-wave band, it would be nice to know if all of your hard work is going to pay off or if your best DX will be <500 meters! This is true, of course, unless your goal is to just 'be on the band'. Nevertheless, in either case, it would be nice to know if your design is meeting your goals or if you are missing out in working the next grid by only 2 dB.

A thorough and complete link budget analysis must be done in order to understand the magnitude of the problem. When first starting the analysis, some basic assumptions must be made. Those include such items as power output and receiver noise figure. The other items needed include projected values such as anticipated receiver bandwidth, antenna gain and atmospheric losses. With just the above-mentioned five items, it becomes a simple matter of "doing the math" to arrive at a value of the maximum distance over which a QSO could be made or what weather conditions are needed to work between two given sites.

The initial values of transmitter power and receive noise figure in the link budget calculations can be 'seed' numbers that are assumed to be practical and obtainable for the type of gear that is planned to be built. Refer to the figure below for an example of link budget calculations, using such 'seed' numbers.

The result of the link budget calculations will be the number of decibels of loss between the transmitter and receiver that can be tolerated if a QSO is to be made. This total loss number must then be proportioned between the free space loss and the atmospheric loss. The free space loss follows the well-known formula (2):

Free Space Loss = 96.58 + 20 Log (f) + 20 Log (d),
Where: f = frequency in GHz and d = distance in km

For 145 GHz, the equation can be reduced to:

Free Space Loss = 139.8 + 20 Log (d)

The details of atmospheric losses are discussed below in Section III. However, keep in mind that the free space loss is only a function of distance for a given band and remains fixed at all times. That fixed value must be subtracted from the link budget total, which then yields the maximum atmospheric loss that can be tolerated. The allowable atmospheric loss must then be divided by the distance, yielding the loss per km that must be achieved based on weather conditions.

For example:

Link budget total loss = 165 dB (a randomly selected number)

Free space loss (for 10 Km path) = 159.8 dB

Allowable atmospheric loss = 165 dB - 159.8 dB = 5.2 dB

Allowable loss per Km = 5.2 dB / 10 km = 0.52 dB/km

Reading from graph in Section III,
Outside weather dew point must be < ~19 degs. F for S/N = 0 dB

If in the above example the dew point was much lower than 19 degrees F, we could either accept a better signal to noise ratio or we could extend the distance and reduce the signal to noise ratio.

Locating 145 GHz parts

By far the most difficult task of building gear for the upper millimeter-wave bands has to be the procurement of parts that will function on the bands. On a band such as 145 GHz, coaxial-based multipliers and mixers are rarely used in the commercial arena, so the likelihood of finding such devices at flea markets is very rare indeed. The use of waveguide devices is most common, and although the price will be higher than most 47 GHz parts, they can be found if you take time to look. Ideal waveguide sizes for the 145 GHz band would be WR-8 through WR-4. However, the application of lower frequency assemblies should not be ruled out. For example, a diode detector mount designed for 38 GHz will still provide mixer function at 75 and 145 GHz. The conversion loss will be high and the input waveguide will be overmoded, but it WILL work and can provide the basis for a simple station.

In the stations designed by the author, a pair of modern, state-of-the-art, harmonic multipliers was located and provided very good performance. The mechanical design along with the associated varactor diodes were donated to the author by the Univ. of Virginia. These stations may be hard to reproduce because of the somewhat limited availability of similar multipliers (short of purchasing them new), but the ideas and concepts of the station design are very applicable to all of the millimeter-wave bands and others' projects as well.

III. PROPAGATION

Atmospheric Loss

With regard to the millimeter-wave bands, the atmosphere can be thought of as "the great equalizer". In other words, no matter how much power you may manage to generate and no matter how low your receiver noise figure may go, you will soon reach a point where the atmospheric losses will add up such that a QSO can't be had no matter what else you may try.

This is not to say that modes such as PSK-31 or QRSS CW shouldn't be tried to extend DX records, but every ham will eventually run into the atmospheric "brick wall" of millimeter-wave DX and be forced to re-think his/her approach to the bands.

Gas Losses

The major sources of atmospheric losses that we will encounter on the amateur millimeter-wave bands are oxygen and water vapor. Since these two gasses cannot be eliminated anywhere on the earth's surface, we must understand their effects and use the constantly changing weather patterns in the atmosphere to our advantage.

Since air pressure can be a direct indicator of the amount of oxygen in the air and dew point is an indicator of the amount of water vapor in the air, these two values are our prime indices of millimeter-wave conditions. The expected effect from oxygen losses versus a change in atmospheric pressure, will have a lesser effect than water vapor when looking at total losses in the amateur radio millimeter-wave bands. This is due to our bands, with the exception of 119 GHz, being sufficiently far from the major oxygen resonance lines.

For example, the range of atmospheric pressures that we may experience can vary from around 950 to 1050 millibars. For a fixed value of water vapor, say 7.5g/m3, and a fixed temperature of 15 degrees C, the total atmospheric losses would vary from .92 dB/km at 950 millibars to 1.0 dB/km at 1050 millibars, delta of 0.08 dB.

If we now hold the pressure constant at 998 millibars, and the temperature constant at 15 degrees C again, and change only the water vapor concentrations from around 1g/m3 to around 4.5 g/m3, we see a change in total atmospheric losses from .16 dB/km at ~1 g/m3 to .72 dB/km at ~4.5 g/m3, or a delta of 0.56 dB.

We can see from these above examples that a typical change in water vapor values will have a greater effect on 145 GHz path loss than would a typical change in atmospheric pressure. The effect of a change in barometric pressure would be more noticeable if the water vapor concentrations were already low, such as on a very dry winter day. The ideal weather for millimeter-wave propagation, in general, is a dry, low-pressure system. However, since nature typically provides us with wet weather in low-pressure systems, the best that we can do would be to conduct our millimeter-wave QSOs on high mountains during very dry days.

Not all millimeter-wave bands experience the same weighting ratio of water-to-oxygen losses as depicted in the above example. In an extreme case of low water vapor, the total atmospheric losses at 145 GHz are less than those at 75 GHz. The primary reason is the existence of an oxygen resonance line around 60GHz. A similar water resonance line exists near 183 GHz. In the final analysis, the total loss is a complex function of frequency, water vapor concentrations, and atmospheric pressure.

Below is a simple graph that was created using atmospheric data supplied by MIT, based on the literature (3), and verified through experiments by the author. The graph shows the expected total atmospheric losses vs. dew point for the 145 GHz amateur band at a barometric pressure of 1013 millibars.

Locating Weather Data

Once you know what type of weather data you need (i.e.: dew point) in order to determine existing propagation conditions or to predict future conditions, you need to find a reliable source of such data. Current dew point conditions can be obtained from most local NWS weather sources and dew point forecasts can be found on the Web at:

http://wwwagwx.ca.uky.edu/ldm-images/ngm_dewp_ani.gif

IV. ANTENNAS

Typical antennas used on the millimeter-wave bands are the horn, the parabolic and the lens. The horn and the parabolic are the most common, although a horn and lens combination have been proposed and successfully used on the lower, 10 GHz band.(4) While a horn antenna may be easy to design and construct, the basic parabolic antenna offers an attractive, simple to use design and with many surplus 30cm dishes to be found at flea markets, it's an ideal choice for mm-wave work. Although the surface finish becomes a critical factor, it is the dish's overall curvature and its deviation from a true parabola that become increasingly important. An RMS surface tolerance error on the order of 0.2mm will cause a given dish to exhibit about 6dB of gain reduction from the idea dish at 145 GHz.(5) It therefore can be seen that efforts to find an ideally smooth, optically correct, dish antenna should not be on the top of the list when starting out the construction of a millimeter-wave station.

To that point, the two parabolic dishes used in the author's stations were surplus ones that had been found at flea markets. One was originally used for 23 GHz commercial TV links and the second was a military surplus millimeter-wave dish. Both dish feeds were designed to be Cassegrain sub-reflector based but flat plate sub-reflectors were used in the initial tests. Later, true hyperbolic sub-reflectors were designed with the aid of G7MRF's spreadsheet based software.(6)

A simple open-ended section of WR-8 was initially tried as the feed for the sub-reflector, but it was found that the pattern of the complete dish antenna suffered from poor gain and had many side lobes. This problem was characterized in HFSS when an open-ended piece of WR-8 was used as a simple horn itself in free space. As the ratio of waveguide wall edge to wavelength increases, it becomes apparent that the open-ended guide begins to look increasingly like an open slot in a sheet of metal. Thus with standard sized waveguide wall construction, radiation caused by surface currents on the waveguide edge contribute to a poor pattern with many side lobes.

Dual-mode W2IMU horns were then constructed to allow for proper sub-reflector illumination.(7) The dual-mode horns were designed using HDL_ANT software.(8) The construction was simplified by grinding the end of a carbide drill bit to the required angle to allow the forming of the taper section of the dual mode feed. The horn was then fabricated by simply plunging the drill bit into a section of brass dowel on a lathe. The small diameter section of the horn was made by drilling the proper sized circular waveguide hole with the lathe. The driven end of the dual mode horn had a lip formed on it, to allow the entire horn to be slipped over the end of a piece of WR-8 and held in place by a small set screw. The resulting step transition between the WR-8 and the circular guide of the dual-mode feed was modeled in HFSS and was found to be an acceptable ~20dB S11 match.

V. PHASE LOCKING

By far, one of the most challenging and interesting tasks in the design of the author's particular 145GHz station was the phase-locked loop for the Gunn oscillator. The output of this locked Gunn is used to drive the X2 transmit multiplier, which is a copy of a design of a 160GHz multiplier made at the Univ. of Virginia.(9)

A commercially-obtained 72.5GHz Gunn oscillator functions as the VCO block within the PLL. The particular Gunn source does not have a varactor tuning diode, so the oscillator must be electronically tuned using voltage pushing of the Gunn diode's supply voltage. Although this method of tuning and phase locking has been described in prior amateur projects, the method appears to be used by very few as its application can require detailed PLL design and analysis. (10)

Detailing the intricate design evolution of the stations is beyond the scope of this paper; however, a very important point to mention is the integration of the PLL loop filter into the Gunn's voltage regulator. In concept, a traditional PLL would place the loop filter just after the phase detector. In a voltage- pushed design, the output of that filter would then drive an adjustable voltage regulator which supplies DC power to the Gunn diode. Since the voltage regulator is after the loop filter, any noise created in the regulator would easily modulate the Gunn diode and result in a very noisy source. The idea of placing a passive loop filter after the regulator was dismissed due to the problems associated with having high DC current flowing through the filter. With the integration of the regulator and loop filter, any noise created by the regulator would also be filtered by the low pass action of the filter. A complete and detailed ADS simulation of the loop was made with the assistance of Dr. William Overstreet, K4AJ. This allowed for optimization of the closed loop and the resulting phase noise to provide a signal that could be used for narrowband CW work.

Since the PLL was designed to be locked anytime that DC power was applied, it was decided to use FSK-CW modulation by dithering a frequency-trim adjustment voltage on the reference oscillator. The resulting frequency shift at 145 GHz was only a few hundred Hertz, but was able to be copied without too much difficulty.

The transmit frequencies of the two stations were offset from one another to create a difference IF frequency that was monitored with an ICOM R-7000 receiver.

Below is a block diagram of one of the 145 GHz stations in its final design.

VI. CONCLUSION

Two of the above-described stations were constructed during the summer of 2000 and were used in several grid expeditions during the following months. A new North American DX record was established on November 6th, 2000 at a distance of 34 km over a line-of-sight path.(11) At that time, oven- controlled oscillators were being used rather than rubidium standards, as shown in the above station block diagram. This created a difficult-to-follow frequency drift, which remained no matter how much thermal insulation was placed around the oscillators. The drift made the signals very tricky to keep centered in the 1 KHz receiver pass band. This indicated that a more stable reference source was needed and for future contacts a rubidium oscillator was incorporated into each station.

With the better oscillators in place, the dish feeds were then further improved by the addition of the dual-mode horns and the DX record was bettered to 61 km on January 1st, 2001 over a line-of-sight path.(12) At that point the 0 dB S/N ratio limit had been reached and no better DX could be achieved, even on the driest of days, in the area of southwest Virginia.

Future improvements may include an optimization of the mixer conversion losses. Since in this design the X2 multiplier serves as a harmonic mixer for the receiver, its conversion loss and resulting noise figure are rather high. The multiplier assembly was never intended to be used as a receive mixer, so no consideration was given to that application during its design phase at the Univ. of Virginia. By careful adjustment of the mixer's DC bias point and LO drive power, however, it is hoped that a lower conversion loss will be achieved.

VII. ACKNOWLEDGEMENTS

The author wishes to thank the following people for their assistance in the project, encouragements and help during the grid expeditions: R. Frey, WA2AAU; the members of the Mount Greylock Expeditionary Force; W. Overstreet, K4AJ; G. Howell, WA4RTS; P. Lascell, W4WWQ; J. Price, N4QWF; Drs. T. Crowe, D. Porterfield, K. Hui, of the Univ. of Virginia; T. Dawson, machinist; and D. Wheeler of Harmonix Corp.

VIII. REFERENCES

B. Justin, WA1ZMS, "World's First VUCC on 75 GHz", Microwave Update Proceedings, ARRL, 1999.
M. Van Valkenburg, "Reference Data For Radio Engineers", Chapt. 33, p.22, Prentice Hall, 1993
J. E. Allnutt, "Satellite-to-Ground Radiowave Propagation", London Peregrinus, 1989.
P. Wade, N1BWT, "Practical Microwave Antennas", UHF/Microwave Projects Manual, Vol. II, ARRL, 1997.
R. Johnson, "Antenna Engineering Handbook", Chapt. 17, p.37, McGraw Hill, 1993.
M. Farmer, G7MRF, "Large Dish Cassegrain Development Using CAD & Spreadsheet For Millimetric Bands & Practical Implementation", Microwave Update Proceedings, Mt. Airy VHF Radio Club, Inc., 2000.
R. Turin, W2IMU, "Dual Mode Small Aperture Antennas", IEEE Trans. Antennas & Propagation, Vol. AP-15, pp.307-308, March 1967.
P. Wade, W1GHZ, HDL_ANT Software Ver. 3b3, http://www.w1ghz.cx, 2001
D. Porterfield, "Millimeter-wave Planar Varactor Frequency Doublers", Ph. D. dissertation, University of Virginia, Aug. 1998.
M. Dixon, G3PRF, "Microwave Handbook, Vol. III", Radio Society of Great Britain, 1992.
E. Pocock, W3EP, "The World Above 50MHz", QST, ARRL, p.86, January 2001.
E. Pocock, W3EP, "The World Above 50MHz", QST, ARRL, p.101, March 2001.