Paths Selected

For completion of VUCC on 75GHz, 5 grids squares must be contacted from a single, fixed, primary location. These five grids for a rover station should be optically line of site to the primary location in order minimize the path losses and to aid in antenna pointing. It should be noted that the radio horizon is in fact some 33% farther than the optical horizon due to tropospheric refraction of the RF signal. Terrain blocked, non line-of-site paths might be possible but would likely require anomalous propagation modes such as ducting and/or knife edge diffraction. With the goal being a timely achievement of VUCC, the "paths" of least resistance were selected; those being ones in direct line-of-site. Since line-of-site paths allow for rather accurate loss calculations, the amount of transmitter power, receiver sensitivity and antenna gain needed to communicate over that path can be easily determined.

In addition to the usual free space losses, atmospheric losses must be accounted for when operating on the millimeter wave bands. These additional losses due to water and oxygen absorption are rather high on long paths and can easily equal or exceed the free space loss. This is the root of the challenge of the millimeter wave bands. As the operating frequency is increased, the amount of water absorption increases. An interesting note is that there is a slight drop in oxygen absorption on 75GHz compared to that of 47GHz, but despite that difference, water vapor is still the major loss factor at 75GHz.⁽⁴⁾

Although the water vapor losses on 75GHz are close to twice that of 47GHz, it was decided to use the same five sites for this VUCC attempt that were used for the first 47GHz VUCC. Three of the 5 paths have distances of around 60Km, one is a "short putt" of only 25Km while the fifth grid is at a distance of 114Km. This fifth grid contact ties the current world distance record for the 75GHz band held by DK4GD and HB9MIO.⁽⁵⁾

Refer to figures 1 through 5 (the path figures are not available on the web page as they are printouts from a software package, if we get them scanned in they will find their way here - ed.) for the path profiles for each of the five grids. Propagation and terrain modeling software were used to generate the profiles. All available propagation software packages known ignore man made obstacles as well as vegetative growths when determining microwave path profiles. Therefore, several attempts at finding a site clear from trees and other obstructions were needed. Many mountain tops and vistas may have clear openings, but it is less frequent that there is clearing in the direction that is desired. Each site was selected based on its location within a particular grid and verification of a clear path was made by prior site visitations.

If computer generated path profiles were not available, one could have resorted to the more traditional method of using 4/3-earth profile graphs to manually plot the profiles. GPS receivers and topographical maps are indispensable tools for determining potential sites.

Once a potential site has been located and its distance to the primary site calculated, path loss calculations should be made. In addition to the free space loss, the additional losses incurred due to the atmosphere absorption must be included in the total loss summation.

a = 32.45 + 20Log(f) + 20Log(d) + d·( g _wo + g_oo)

where,

a = total path loss in dB
f = frequency in MHz,
d = distance in Km,
g_wo= water loss in dB/Km,
g_oo= oxygen loss in dB/Km.

EQ 1 - Millimeter wave path loss

The loss due to oxygen absorption remains primarily a constant value for a particular frequency regardless of altitude, up to about 5Km where the air starts to become noticeably thin. This simplification of oxygen loss holds true for most domestic paths with the exception of the Rocky Mountain region.

Loss due to water vapor however is another issue. Different air masses can hold varying amounts of water vapor and thus will have an effect on the total water loss. The absolute humidity or the dew point is the primary indicator of water vapor in a given air mass. Barometric pressure also has an effect but since its contribution is rather small, it was ignored during the loss calculations for the five aforementioned paths. Assuming a stable, non-turbulent air mass, the absolute humidity remains a constant despite the fact that the relative humidity may change from day to night. Temperature changes can occur, but if nothing disturbs the air mass, its absolute humidity will remain constant. Thus the time of day or temperature has little direct influence on the atmospheric losses in a stable air mass. Some diurnal variation was noticed on some test paths, but further investigation is required to document and quantify this effect.

Below is a simple table that can be used to determine the atmospheric loss due to water vapor at various values of temperature and relative humidity for the 75GHz amateur band.

RELATIVE HUMIDITY

TEMP	20%	30%	40%	50%	60%	70%	80%	90%	100%
0/32	0.022	0.034	0.043	0.054	0.065	0.076	0.087	0.098	0.109
5/41	0.030	0.046	0.061	0.077	0.093	0.108	0.123	0.138	0.154
0/50	0.042	0.064	0.083	0.106	0.127	0.149	0.169	0.191	0.212
5/59	0.057	0.087	0.116	0.144	0.174	0.202	0.231	0.261	0.290
0/68	0.078	0.117	0.156	0.195	0.235	0.273	0.313	0.351	0.391
5/77	0.105	0.156	0.209	0.261	0.313	0.365	0.418	0.470	0.523
0/86	0.137	0.206	0.274	0.343	0.412	0.481	0.549	0.618	0.686
5/95	0.179	0.269	0.359	0.449	0.538	0.627	0.721	0.814	0.901

TABLE 1 - Water vapor loss in dB/Km at 76GHz

Equipment

The frequency multiplication method of signal generation was selected over fundamental generation (ie: Gunn oscillators) for two reasons. The first being the ease application of narrow band modulation through on/off keying of one of the multiplier stages. Narrow band modulation has the distinct advantage of allowing the use of narrow band IF filters in the receiver. The narrower the IF filter, the less noise power there is in the receiver. The lower noise power has the effect of increasing the signal to noise ratio. Thus for a given received signal such as 1K0A1A (CW), the signal to noise ratio would be 22.5dB higher than if 180K0F3E (WBFM) were used.

The second reason for selecting frequency multiplication was to take advantage of existing building blocks already in hand and their resulting good frequency stability. Several surplus 12GHz sources were available as were some 37GHz waveguide multiplier assemblies. Gunn diode oscillators do exist for 75GHz but none could be procured through surplus or located at amateur flea markets.

Some experimental equipment previously built by the author used WBFM and exploited the harmonics from 25 and 37GHz Gunn diode oscillators. These stations had limited communication range due to the relatively low power levels (ie: 100uW) and the use of wide band modulation necessitating a wide IF bandwidth, but they did allow for the design, construction and testing of parabolic antennas suitable for use on the band regardless of the modulation method or power.

The completed narrow band stations each start with a crystal controlled 5th overtone oscillators. Oscillators of these types have lower phase noise when compared to fundamental crystal oscillators. This oscillator is temperature compensated to about 5ppm. The oscillator is then phase locked to a higher stability ovenized reference oscillator. ⁽⁶⁾ The result is a crystal oscillator with .03 ppm stability at the desired frequency. The high stability is desired for two reasons. First, it will minimize the long term, absolute frequency error which makes tuning the signal difficult. The second is to minimize the short term frequency error so as to keep the signal within the pass band of the receiver. Since 1ppm stability is the equivalent of 75KHz of frequency error at 75GHz, it can be seen how difficult it is to keep the signal within the 1KHz bandwidth of the IF receiver if non-compensated oscillators are used. The .03ppm stability used here results in 2.25KHz of frequency error. The lock time for the oscillator pair is on the order of 20 seconds after the ovenized oscillator has reached operating temperature.

The stabilized crystal oscillator signal is then used as the reference frequency for a 12.6GHz Frequency West type phase locked loop assembly. This PLL assembly phase locks a 1.26GHz power oscillator to the incoming reference signal. The 1.26GHz signal is then used to drive a step recovery diode harmonic multiplier. The resulting 10th harmonic of the power oscillator is filtered and appears at the output of the assembly at about +13dBm.

The 12.6GHz signal is then delivered via coaxial cable to a times-three multiplier. This multiplier is an active device and produces at its output, in waveguide, the third harmonic of the input signal. The resulting 37.8GHz signal is about +18dBm.

That 37.8GHz signal is then feed through a wave guide circulator to protect and isolate the times-three multiplier from load changes and the poor input return loss of the following stage.

The final and most critical stage is a times-two multiplier. This multiplier is based on a design created by Dr. David Porterfield (of Virginia Millimeter Wave) for the completion of his doctoral thesis at the University of Virginia.⁽⁷⁾ The design is split block in nature and uses an SB13T1 planar Schottky GaAs multiplier diode mounted on a quartz microstrip substrate. Incoming RF is feed to the diode assembly via WR-19 waveguide and the resulting second harmonic is coupled off via WR-15 waveguide. The key to this very successful design is the optimization of the diode's imbedding impedance to minimize the conversion loss of the multiplier. Extensive attention to detail and precision machining are needed to construct such a multiplier. The design of the multiplier was donated by Dr. Porterfield, and the GaAs diodes were donated by Dr. Tom Crowe of the Semiconductor Device Laboratories of the University of Virginia. Construction of the multipliers were made by the author with extensive help from Mr. Kai Hui of the UVA Receiver Laboratory. The completed multipliers delivered +12dBm at 75.6GHz with a +18dBm, 37.8GHz drive signal applied.

In one station, the times-two multiplier is also used as a sub-harmonic mixer for receive. In this application, the diode bias port serves double duty as an IF port as well. The resulting receiver noise figure is around 15dB as measured by the conversion loss in the mixer. The second station's noise figure using the same method, was originally over 30dB. Therefore an alternative approach was tried. A 3 port W-band mixer was on hand and was tried out. The original tuning frequency of the mixer was unknown, but the mixer resulted in a conversion loss of 12dB at 75.6GHz and thus was placed into service.

Below is a table showing the resulting frequency plan used in each station. Both stations use ICOM R-7000 receivers as tunable IFs. The final IF frequency is around 257MHz.

	Crystal Osc.	X14	X10	X3	X2
Station #1	90.004MHz	1260.056MHz	12.60056GHz	37.80168GHz	75.60336GHz
Station #2	90.310MHz	1264.340MHz	12.64340GHz	37.93020GHz	75.86040GHz
IF Freq: 75.86040GHz - 75.60336GHz = 257.04MHz

Table 2 - Station frequency plan

Antennas

The antennas used for the all contacts were one foot parabolic dishes originally used for terrestrial 24GHz commercial links. Although the dish "true-ness" may be in question, and may compromise the gain at 75GHz, they seemed sufficiently useful. Each dish has an F/D of 0.3. This makes the construction of a feed a bit more difficult than for a shallower dish. Dishes with this low an F/D, cause the feed placement to become very critical.(8) The construction of a feed that provides the proper E and H field illumination of the dish is also critical. The low F/D dish does not lend itself to easy illumination. Although other dishes would have been easier to feed, I elected to use what was on hand at the time and see what results could be obtained.

Since a standard gain horn would not provide the proper illumination, the simpler approach of using a flat plate Cassegrain sub-reflector was tried first. A piece of rectangular brass hobby stock was found to be the right dimensions to act as a slightly reduced height WR-15 guide. Since the hobby stock cost is lower than that of true WR-15, several inexpensive experiments were made.

The hobby brass was used as a flush, open ended feed with a two inch diameter flat plate Cassegrain sub-reflector mounted in front of the open end. Attempts were made to construct a true hyperbolic sub-reflector but the difficulties in creating the proper machining drawings and equations lead me to stick with the original flat plate.

Each antenna is mounted on a heavy duty camera tripod. With the use of a pan and tilt head, the tripod is easy to transport and allows for both independent azimuth and elevation adjustment. The antennas were sighted in over a short optical path so that rifle scopes mounted on each dish could be adjusted for proper aiming. Since an open guide feed is used, asymmetrical E and H field patterns are generated. The result is unequal beam widths in the horizontal and vertical planes however this does not greatly compromise antenna pointing.

The calculated beamwidth of a one foot parabolic dish at this frequency is on the order of 0.9 degrees and has a gain of 44dB assuming a 50% feed efficiency. In practice, efficiencies of 30% are more common unless a great attention to proper feed design is given.

In the field, the pointing of the antennas was made through the use of rifle scopes and the sighting of local land marks for paths that were beyond line-of-site. Additionally, a surveyor's compass was used. With good technique, the dish can be pointed to with in 0.5degs of the desired azimuth and elevation.

Earth curvature effects are rather noticeable on paths that are long (ie: >50Km) and with high mountain locations. For example, two equal height mountain tops 110km apart, require 0.5degs of down-tilt from the horizon. Such subtle effects become critical when pointing antennas which have 3dB beam widths on the order of a degree or two.

The Contacts

The first test contacts made were disappointing since signal margins were lower than expected. The antennas were suspect since their gain had not been measured or verified. A standard gain 18dBi horn antenna was located and comparison tests were made. It was determined that the dish antennas were some 12dB below the expected gain value. Careful empirical adjustment of the flat splash plate Cassegrain sub-reflector resulted in achieving improved gain. The plate was placed at an initially calculated location along the focal axis of the dish and was then adjusted in .125mm increments until the maximum possible gain was achieved. This gain was compared to that of a standard gain horn and was found to be within a few dB of what should be achievable with a 1 foot dish at 75GHz. Four of the five VUCC QSOs had margins in excess of 10dB. The last and furthest contact at 114Km yielded about 0dB signal margin. All of the contacts were made using CW.

Future

Equipment for use on the 75GHz band is very rare, and without suitable test equipment is very difficult to get operational. Lower frequency spectrum analyzers can be used along with external frequency mixers. 24GHz power meters can be used with tapered waveguide transitions to give a rough indication of power levels being generated.

As the millimeter wave bands find themselves being more commonly used for commercial applications such as anti-collision vehicle radar and radio astronomy, items will start to appear on the surplus market and will be an excellent source of amateur radio parts.

Over time it is hoped that the receiver noise figures can be reduced through tuning and that the antennas can be optimized for higher gain by the construction of better feeds.

Acknowledgements

The author wishes to thank and acknowledge the following people for their assistance in the conception, design, and construction of the equipment along with encouragement throughout the effort. Thanks go to: Mr. Dean Dixon, W9YRH and Bob McBrine of Millitech for the donation of a pair of circulators; Robert Mignard, N1DVC for the 38GHz multipliers; Doug Sharp, K2AD and G. P. Howell, WA4RTS for roving efforts; Dr. Tom Crowe of UVA for the donation of the SB13T1 GaAs diodes; Mr. Kai Hui of UVA for assistance in construction and testing; and Dr. David Porterfield of VA Millimeterwave for the design of the 76GHz multiplier.

References

⁽¹⁾ B. Atkins, "The New Frontier", QST, Dec. 1988, p.87.
⁽²⁾ K. Britain, "Microwave USA", DUBUS, Vol. 27, 3/98, pp. 42-43.
⁽³⁾ K. Britian, "Microwave USA", DUBUS, Vol. 28, 2/99, pp. 42-43.
⁽⁴⁾ T. Frey, Jr., "The Effects of the Atmosphere and Weather on the Performance of a mm-Wave Communications Link", Applied Microwave & Wireless, Feb. 1999, pp. 76-80
⁽⁵⁾ E. Pocock, "The World Above 50MHz", QST, Aug. 1999.
⁽⁶⁾ C. Houghton, K. Bane, "Phase Lock Control Circuit For Use With Brick Type Oscillators", ARRL UHF/Microwave Projects Manual Vol. I, 1994.
⁽⁷⁾ D. Porterfield, "Millimeter-wave Planar Varactor Frequency Doublers", Ph. D. dissertation, University of Virginia, Aug. 1998.
⁽⁸⁾ P. Wade, "Practical Microwave Antennas", ARRL UHF/Microwave Projects Manual Vol. I, 1994.

The world's first VUCC on 75GHz