Comment:  While the descriptions given on this page are specifically referenced to application with the VE2EMM Doppler II unit, they can generally be applied to other units as well.  Please note that this is not an official page by Jacques - see the disclaimer below.

Perhaps the most important piece of of the DF array is arguably the antenna itself.  For a system that uses an "electrically spun" antenna - like a "doppler DF" - this is certainly important.  Here are a few things to keep in mind when designing/building such an antenna array:

• Minimize switching losses.  Typically, PIN diodes are used to "hard-switch" antennas.  PIN diodes are specifically optimized to present minimum losses and IMD (distortion) when they are "on" and maximum isolation (loss) when they are "off."  While "switching" type diodes (such as a 1N914/1N4148) may be used, they actually perform rather poorly (in terms of loss and distortion) as compared with a good PIN diode.  As it turns out, the absolute "off" isolation of the PIN diode is somewhat less important as the signal from the antenna that is currently "on" will pretty much swamp out any leakage.
• All antennas in the array except the one being currently selected should be "invisible."  On some of the earlier system designs, the unused antennas were "shorted out" to increase signal isolation - that is, keep RF from getting in through those antennas not being used at that instant.  This had the problem, however, of making those unused antennas appear as parasitic elements - that is, as reflectors or directors.  Instead of being an array where a single omnidirectional antenna was being selected at any given instant, the array was a multi-lobed monster - anything but omnidirectional.  This effect can distort and confuse readings considerably.  The way around this is to put a PIN diode right at the feed of the antenna, or to simply use odd multiples of 1/2 wavelength (e.g. 1/2, 3/2, 5/2) of coax between the PIN switch and the antenna array to maintain a high impedance when "off."

A vehicle-mounted DF antenna array.  This array is on a plate that is attached to the vehicle's luggage rack.
Click on the image for a larger version.

For the antennas described here, only 2 meter operation was anticipated:  If conditions warrant the addition of capability of other bands, they will be added as necessary.

Also, to simplify construction, the switching systems only have a single PIN diode in the antenna switch:  The multiples of 1/2 wavelength are used to help make the antenna look "disconnected" when it is not being selected.  In all cases, it is important that all coax lengths (between the antenna and the switch) be equal.

A vehicle-mounted antenna array:

The goals of a vehicle-mounted antenna were:

• Relatively quick to install and remove.
• The ability to "stow" the whips and the ground plane elements to minimize profile and visibility.
  
• Attach to existing structures on the roof of the vehicle.
  
• Good performance.

A close-up view of the radial mount and one of the four whips in the "in-use" position.
Click on the image for a larger version.

In the case of this antenna array, the vehicle to which it was to be attached had an adjustable luggage rack.  The lateral bars of the rack were simply adjusted so that the plate was centered on the top of the vehicle and the plate is held in place by hook-and-loop (a.k.a. "Velcro" (tm)) straps.

The plate on which the array is constructed is 24 inches square, painted a dark color to blend in somewhat with the vehicle, and BNC bulkhead connectors (for the whips) are arranged in a 18 inch square centered on the plate.  To provide increased effective ground plane area, a total of eight 19 inch radials were added at the edges of the plate.

BNC connectors aren't ideal for long-term exposure to weather - as they are not inherently waterproof - but being that this antenna was intended for occasional use, it was decided that their application was sufficient.  If another style of connector is desired at a later date, they may be easily installed.

Note that an array of dipole antennas could also have been mounted on the vehicle - and they would likely perform a little bit better in terms of weak signal performance and bearing accuracy than an array of verticals.  Their use would, of course, complicate the mechanical arrangement and they are certainly less "stealthy" than four small verticals on the roof of a car...

The whip antennas - version 1:

The four whips were constructed using straightened, 14 gauge Copperweld (tm) wire, cut to an electrical 1/4 wavelength at the center of the 2 meter band.  The elements were soldered to the center pin of a crimp-type BNC connector (which was drilled-out to accommodate the diameter of the wire) and elements were then secured within the connector using first, clear "quick-set" epoxy, and then covered with a layer of quick-set gray epoxy.  (Note:  The lossiness of the metal filler in the gray epoxy is negligible in this application..) Although it is difficult to see from the picture, the bottom three inches of the whip is paralleled with another piece of wire (which is soldered to it) to provide additional stiffness.  The beginning of this paralleled section is just above where the ferrule of the BNC connector ends - and is embedded in epoxy.

Ideally, stainless steel whips would be used - but at the time of construction, four suitable (and identical) pieces could not be found.  If stainless is used, note that it typically requires a special acid-based flux (available at many hardware stores) to permit soldering.  At the end of these whips, a nice, large, rounded "solder blob" was carefully added both to prevent wind static from accumulating, and to blunt the end of the whip to reduce the eye hazard.

The whip antennas - version 2:

After several years of using the Copperweld (tm) whip antennas, I acquired several stainless steel antenna whips, all identical, that I used for whips and many of the same methods were used to mount these whips in the BNC connectors.  As before, RG-59 crimp-type coaxial cable connectors were used.  This time, instead of drilling out the center pin of the RG-59 connector to accomodate the whip's conductor, I used a short piece of copper wire - one end soldered into the center pin - to connect the center pin to the stainless steel whip.

One difficulty involved in using stainless steel is that soldering to it can be difficult.  For this, there are a number of special acid-based fluxes available that will etch the surface of the stainless steel and make it solderable.  Typically, these are liquid fluxes and one simply "wets" the surface to be soldered, waits a moment for the acid to work, and then, using a very hot iron, solders the connection using lead-free fluxless plumbing solder.  It can sometimes be difficult to find a liquid flux specifically designed for soldering to stainless steel (and it also works nicely on ordinary steel) and I have noticed that water-based fluxes used for plumbing often work as well, as does plain muriatic (hydrochloric) acid.

While ordinary, flux-bearing electrical solder can be used, it seems to me that the rosin flux can interfere with the flow of the solder itself.  Additionally, the typical lead-free solder used for plumbing is actually much stronger than the standard 60/40 tin/lead-based solder anyway.  When soldering, I simply use a standard temperature-controlled soldering iron turned up to maximum temperature (about 800 degrees F) and, using a tip wiped clean of ordinary solder and flux, make the connection, remembering to re-wet, clean, and re-wet the tip again with "ordinary" solder after I am done.

After the whip has been soldered to the center pin, it is carefully "snapped" into place.  Generally, I find it better to partially insert the center pin into the connector and then, with a fine pair of needlenose pliars, pull on the center pin until it "snaps" into place:  Doing this avoids the need of having to push the pin into place with the whip and the jumper wire - a process that can sometimes "kink" that short piece of jumper wire or, worse yet, break it off either at the center pin or at the solder joint.

Once the center pin has been installed, it is important to carefully center the whip inside the ferrule, making sure (with an ohmmeter) that the whip isn't touching the body of the connector and shorting it to ground.  Once this is done one may then backfill the spacing with epoxy to provide mechanical strength and ridgidity.  Experience has shown that a "filled" epoxy is generally superior in this application to a "clear" epoxy as it will hold up far better when exposed to the heat, cold, stresses and vibration that will be encountered.  While it is true that these "filled" epoxies are much lossier at RF than the clear types, the losses incurred here are minimal - but one would still probably want to avoid transmitting 50 watts through it.

Details of the mounting hardware of the radials.  Note the two holes in the circuit board material.  When in use - or being stowed - these holes, along with the #8 screw (on the left edge of the picture) prevent the radial from rotating in the wind.
Click on the image for a larger version.

The ground radials:

As mentioned before, ground radials were added to increase the effective area of the plate's ground plane.  Even though this plate is in relatively close proximity to the (metal) top of the vehicle, it is desirable to make the ground plane as large as possible to provide a uniform plane at the base of the antennas.

The ground radials shown are constructed using straightened 8 gauge Copperweld (tm) wire.  In this case, where 19 inch radials were desired, a 36 inch piece was bent into an "L" at the halfway point and then soldered to a 2.5 inch square piece of glass-epoxy circuit board material.   A 5/16" brass washer was then soldered to the "wire" side of the circuit board as shown in the picture) and a hole was drilled through to accommodate a 5/16" bolt, and another brass washer was soldered to that side, aligned with the hole.

A 5/16" hole was drilled near the corner of the antenna ground plane plate and a 5/16" stainless bolt (1-1/4" long) was installed using split lockwashers and a locknut.  The circuit board material was then placed over the bolt, followed by a split washer and a fender washer and the entire assembly was secured using a brass wing nut.  Note:  The brass wing nut was used in favor of stainless in order to avoid possible galling of the hardware.

As this array is expected to be used at highway speeds, some additional hardware was needed to prevent the ground radials from coming loose and "swiveling" - so "locking pins" were added:

The radial assembly was swung out so that it was in its "operating" position and a hole was drilled through the corner of the circuit board material (near the bend of the radial, as seen in the lower-left portion of the picture) and through to the aluminum antenna plate.  A 1" long #8 stainless screw (along with a lockwashers) was then installed using threadlocker (e.g. blue Lock-Tite [tm].)  After this, the radial was swung around 180 degrees (to the "stow" position) and another hole was drilled to provide a "stowage" position.

As in the case of the whips, large solder "blobs" were carefully put at the end of these radials to minimize static accumulation and (more importantly) protect the eye(s) of someone passing by.  Additionally, the end should be coated with silicone or rubber to provide further eye protection and prevent scratching of the car's paint should they touch or rub against it.  One possibility is to use rubber "tool dip" to coat the ends of the rods.
 

A close-up view of the radial mount, showing the locking pin in the lower-right corner of the circuit board material.
Note that the hardware has been marked to help identify the antenna number - a useful feature both in re-assembly and when connecting cables..
The radial is shown in its "stow" position, with the two elements "inwards."
Click on the image for a larger version.

When the array is being set up - or stowed - one simply loosens the wing nuts, rotates the radials to the desired position - and tightens the wing nut again.  Using a pair of locking pliers, the top one or two threads of the 5/16" bolt were purposely "damaged" slightly so that the wing nut would never simply unscrew and fall off:  It will simply stop when it gets to the damaged threads - and prevent loss of hardware.

It is highly recommended that as the pieces are drilled, that they be marked - preferably with a stamp and/or an electric engraver - to identify where the hardware is to go.  This is not only useful when connecting the coaxial cables, but because the hardware is likely to fit only in the location it was drilled (unless one measured and drills very precisely) it aids in final assembly.

Using the array:

Before the connectors and other hardware is mounted to the plate, it would be a good idea to paint the plate a color to somewhat match your vehicle.  Also, it would be a good idea to mark the plate to identify the top and bottom, the front and back, as well as the number (sequentially) the antenna connections 1 through 4.  This may be done using a metal stamp, an electric engraver, or whatever other permanent means you devise.  If a metal stamp is used to mark the plate, rub some paint of a contrasting color into the stamping and then gently wipe it off - leaving that color inside the stamp - to make it easily visible.

As mentioned before, this array is held in place between two luggage racks - adjusted to a spacing slightly less than the size of the plate - using Velcro (tm) straps of the sort used to secure cables.  To accommodate these straps, slots were cut in the aluminum plate near the front and rear edges:  The straps are wrapped around the luggage rack bar and back through the slot, making for a very secure - yet easily removed mounting.

If your vehicle does not have a luggage rack or bars on top, one can mount some rubber feet to the bottom of the plate and hold it down with straps or elastic cords in much the same way a car-top cargo container would be secured.  All that is necessary is to provide enough space to connect and clear the coaxial cables on the bottom of the plate.

Once connected to the DF system, the normal calibration should be carried out - verifying that both the connections to and rotation of the antennas are correct.

The job of the antenna switching unit is to select only one of the four antennas in the array and connect it to the receiver and this is done, in this case, using MPN3404 PIN diodes.  These diodes are relatively inexpensive, have fairly low loss, have reasonable isolation, and are packaged in a TO-92 (transistor-like) case.  While perfectly well-suited for 2 meter operation and below, they may not be the ideal choice for 70cm - where other devices (in surface mount packages, etc.) may be better-suited.
 

The exterior view (top) and two interior views of the antenna switching unit.  The added driver board is on the left-hand side of the center picture.
Click on the image for a larger version.

A NOTE ABOUT THE CIRCUIT BOARD:

The etched circuit board shown in the picture was obtained from FAR circuits and is the "Roanoake Antenna Switch board."  Note that this board is sold with the Roanoake Doppler DF board and the Montreal Doppler II and is not listed by itself - you will have to ask about its availability if you want just this board.

As mentioned previously, the VE2EMM unit can drive the PIN diodes directly - but I was uncomfortable about presenting the I/O pin of the PIC microprocessor directly to the outside world - and have chosen to "isolate" it with a 1k resistor.  The addition of this resistor offers protection against short circuits, static discharges, and RFI, but it does make it necessary to use other circuitry to drive the PIN diode switches.

The '1488 Driver - Standard version:

In this case, I'd originally used a standard bipolar '1488 RS-232 line driver:  These chips are cheap, readily available, have bipolar output (as in being able to output both positive and negative voltages) and easy to use.  For the negative supply, some of the negative voltage from the RS-232 driver chip on the 'EMM board was siphoned off (through a 220 ohms resistor) and filtered with a 220 uF capacitor.  Under load, this supplies between 3 and 4 volts to the '1488 - but this is more than enough to present a negative voltage to the PIN diode to make certain that is is "off."  (Note that the data sheet for the MPN3404 shows that a reverse bias of 1.5 volts will result in a device capacitance of just under 1.5 pf, dropping to about 1.25 pf at 20 volts.

The '1488 Driver - CMOS version:

If available, it is recommended that the CMOS versions of the '1488 be used:  The lower quiescent current of the CMOS device (the "14C88") will allow a more negative voltage to be output.  After using the switcher for a while, I replaced the bipolar version of the  '1488 with the CMOS version and did some testing:  The "plain vanilla" '1488 presents about 3 volts (negative) to the PIN diodes when off and it loaded the V- supply down to about 4.8 volts, while the CMOS version didn't appreciably load the V- supply at all and presented nearly 9 volts across the PIN diodes.  The positive supply is simply the +12 volts used to power the unit.  Practically speaking, I did not notice any difference in performance between the two:  The only real difference is the amount of V- supply loading.

Circuit details:

The schematic shows how the board is wired.  The output of the PIC is sent to some simple inverters consisting of NPN transistors (possibly not necessary, but cheap and easy to do...) and the pull-up resistors internal to the '1488 are used.  The outputs of the '1488 go through some RF chokes, to the PIN diode, and the DC return consisting of a 1.5k resistor to ground.  PLEASE NOTE:  The CMOS version of the 1488 does not have internal pullup resistors, so external pullup resistors (10k) must be used as noted on the schematic!

The RF blocking choke consists of about 24-25 turns (covering most of the resistor body, a value of about 1.2 uH) of #30 enameled wire wound on a high ohmic-value 1/2 watt carbon composition resistor (100k or higher - I had some 150k 1/2 10% and 20% resistors in my resistor drawer) - and the turns were held in place with a drop of cyanoacrylate instant adhesive.  The actual voltage applied to the PIN diode results in a flow of 5-6 milliamps through the PIN diode - which is enough current to bias it to an effective "ON" resistance of less than one ohm according to the data sheets.

With this switcher configuration, voltage actually appears on the antenna itself, so it is important that there be no actual DC continuity in the antenna to ground.  If use of an antenna with a DC return is anticipated, one could add a capacitor to block this voltage.  Practically speaking, the '1488 driver will, itself, limit the current to a value that the PIN diode can handle, so there is little danger of damage to the diode or driver circuit under DC short conditions.

The entire switching circuit is housed in a small die-cast aluminum enclosure or aluminum minibox and the boards are mounted using metal standoffs.  If at all possible, an all-metal enclosure should be used.  If an all-metal box cannot be found, mount the connectors on small metal plate (a metal lid - if one comes with the box, or your own piece of metal or circuit board material cut to the appropriate size and shape.)  Short runs of RG-174 coax connect the board itself to the BNC antenna connectors:  It is important that each of the four runs to the switched antennas be of equal length - and that their lengths be taken into account when making the four coax cables connecting to the antenna.  Also, it has been helpful to put a blob of hot-melt glue where the coax cables connect to the connectors and circuit boards to prevent fatiguing and breaking of the fragile coax cable when the board is moved, flexing the coax repeatedly.
 

Schematic of the as-built antenna switcher.  (Drawing revised 12/04)
Click on the image for the full-sized version.

For ease-of-use, a 9-pin "D" connector was used to supply both power and the switching signals to the antenna switch box.  When completed, one should mark all connectors appropriately, and mark the box with the proper configuration of the DF controller:  If it is built as shown in the schematic, one would select "4+" (that is, 4 antennas, positive switching.)  My numbering of the antenna connectors resulted in the need to use counter-clockwise (CCW) rotation of the antenna array.  Finally, I soldered and extra PIN diode to the ground plane (not shown) so that I would have an extra handy should a field replacement be required.

Note:  The diodes suggested are of the MPN3404 type.  Note, however, that these diodes have a higher capacitance than some newer, smaller, and more modern devices and that an alternative device would be suggested if use on 70cm or higher is anticipated as the higher capacitance may compromise the "off" isolation (and thus the performance) somewhat - although the MPN3404 reportedly works OK. Sorry, but I don't have a specific recommendation for another diode, but if I build a 70cm version I'll post the details here.

Testing:

The 'EMM unit makes testing the antenna array pretty easy.  In the TEST menu, one may select one specific antenna to activate - and it is easy to verify that this antenna is activated due to the voltage appearing on the antenna whip itself.  Not only does this verify that the PIN diode is present and operating, it can be used to identify a specific antenna.  When the antenna is de-selected, one should see zero volts:  If a negative voltage is seen (more than a few millivolts) then the PIN diode is probably shorted.  (Typical "off" isolation is on the order of 20-25 db on 2 meters for a single, series PIN diode.  If no antennas are selected, this isolation may measure lower.)
 

"Why don't I see a 20-30 db drop when I select 'NONE' on the antenna test?" When testing the antenna array with all 4 antennas, you would think that, when monitoring an off-air signal, you would see 20-30 db drop when switching from any antenna to "NONE" - but you might not.  Why? When one antenna is selected (if an antenna is connected to that port, that is) the impedance at the common point (where all of the PIN diodes come together and then go to the receiver) is around 50 ohms, the same as the antenna. If you select the "NONE" setting, however, you have an open circuit that is NOT 50 ohms  and the signal being received is going to be that from leakage through the diodes.  If there is no 50 ohm load present to "swamp" out the leakage, the "off" attenuation may seem to be lower than it really is. The proper way to test isolation is to select, antenna #1 (for example) and terminate its connector with a 50 ohm load, thus simulating having at least one antenna being connected to the receiver:  In this way, the "common point" will now be at 50 ohms rather than some odd (probably high) impedance.  You would then apply the test signal to one or more of the other antenna jacks to measure the isolation.  Using a single MPN3404 as a switching diode as shown in the schematic, the "OFF" isolation will be 20-25 dB - more than enough for an FM-based system such as this. Note:  On the switcher depicted in the schematic, more current would flow through the PIN diode if a 50 ohm DC resistive load were connected, but that shouldn't affect measurements.

Another method of verifying antenna operation with the TEST function is to tune in a constant signal source.  When NONE is selected, the signal should be weaker - but the strength of the signal when antennas 1-4 are selected should be the same.  Doing this - and possibly connecting/disconnecting antenna cables, can help identify antenna cables and possible problems.  Note that many FM receivers (especially handie-talkies) will "peg" their S-meter on even on weak signals, so if no further S-meter deflection possible on an already-pegged meter, so you may have to add some attenuation between the switch box and the receiver in order to "un-peg" the reading.  Also note that if you are testing with a signal that is extremely "multipathy", you may, in fact, observe different signal strengths on different antennas:  This condition may be identified simply by moving the array (or the vehicle to which it is attached) around a few feet to see if things change.

Operational hints:

As mentioned previously, while mobile one typically calibrates the array so that "North" (zero degrees) is straight ahead of the vehicle.  While it is best that one calibrates on a signal that is quite some distance away - and from a source that is in the clear (that is, a line-of-sight transmitter on a mountaintop or a very tall tower) - one can also "make do" with another person standing in front of the vehicle (a fairly long distance away) with an HT on low power (beware multiplath! - do this only in a very large, uncluttered parking lot or open area) - or with someone some distance ahead of you while driving down a straight, uncluttered rural road.

One should also verify that the antenna rotation (CW or CCW) is set properly.  This is done by first setting "north" as being straight ahead, and then turning the car to the right to see that the signal veers to the "west" and then turning the car to the left to see that it veers to the "east."  If the signal does the opposite, try reversing the direction.  If the direction seems to vary randomly, verify that all four antennas are working, are connected in proper order, and that you aren't actually in an area where the signal being monitored isn't multipathy.

As part of normal operation, one should occasionally re-check the calibration.  Turning a 360 degree circle in a parking lot and verifying full rotation of the bearing is also advisable occasionally - to do a "sanity check" if nothing else.  Finally, if you are in "hot pursuit" of a target (a bunny or an "evil-doer") it might be good, at the first opportunity, to check the calibration again - noting the observed offset in your recorded observations so that previous (and possibly subsequent) readings may be corrected as appropriate at a later date.

It is recommended that you keep a few extra PIN diodes around.  Once in a great while, I'll inexplicably find a blown diode on the switcher:  I have no idea why it might happen but if it does happen, your DFing activities are suddenly over.  I have soldered several extra PIN diodes to the ground plane of mobile switcher (I took the picture before adding them...) so that they will be handy if I need to do a field replacement.  Because both leads of each diode are soldered to the ground, they are perfectly safe from harm.

Lastly, if you are using a transceiver for the DF readings, make absolutely certain that you disable transmit.  Some radios have means to specifically disable the transmit, while one may simply have to disconnect the microphone for others.  The PIN diodes will likely be able to handle low power on an HT for a short period, but they certainly wouldn't take kindly to a burst of 50 watts or so from a mobile rig!

The four-dipole antenna array - before final completion.
Click on the image for a larger version.

If one has a DF system that can be hauled around in the car, chances are that it sits around in a closet or on a shelf most of the time doing nothing.  Generally speaking, a mobile antenna array system is not well-suited for fixed-location applications for one or more of the following reasons:

• Often, the mobile antenna array isn't weatherized to the extent required for continuous outdoor exposure.  On a vehicle, it can withstand occasional rain/snow, but the connectors/antennas/hardware may not last prolonged UV, weather, and temperature extremes.
• A large plate - or even a mag-mount array may not be convenient to install in a clear location.  If one does so, it may not be convenient to retrieve it when one needs it for mobile operation.
• If one is installing a DF system in a fixed location, it is desirable to minimize sources of errors:  This usually means that one takes a bit more care in designing, building, and installing the array.

• Using 1/2 wave dipoles instead of vertical antennas.  A 1/2 wave dipole doesn't need a ground plane, and it has a bit larger "aperture" than a vertical antenna to be able to intercept signals.
• If a ground plane is used, one must elevate it sufficiently high to clear nearby obstacles.  This can be very awkward to do - whereas elevating dipoles is pretty easy to accomplish.
• In many locations, the ground plane may accumulate snow.  Having a dipole array (and thus, no ground plane) handily avoids this problem.

Another view of the antenna array.  (The elements aren't really bent - it's the wide-angle lens...)
The PVC portions of the structure have been painted gray for UV protection.
Click on the image for a larger version.

Constructing the array:

When constructing the dipole array, several things need to be kept in mind:

• The antennas need to be arranged in a perfect squared that is less than 1/4 wavelength (typically 0.22 wavelengths) on a side - and this spacing needs to be maintained as the structure moves in the wind.
• The supporting structure should be non-metallic if possible - or consisting of as little metal as practical.
• It should be adequately sturdy to handle expected wind loads and weather conditions as well as prolonged exposure to the elements.

Instead, I devised the scheme shown in the picture - using plenty of elbows and Tee connectors - to provide the 4-way dipole support plus the vertical support.  When constructing a structure such as this, it is best if one "dry fits" all of the pieces together - making sure that the dipoles will be arranged in the appropriately-sized perfect square - prior to gluing the lot together.

The dipoles themselves are constructed of 1/2" copper water pipe.  The far end of each pipe is capped, and the "inside" ends of the dipole (where the coaxial cable is connected) are soldered to threaded bushings that screw into the threaded "tees" on the ends of the support pipes.  A small pigtail of #12 wire was soldered to the threaded bushing and the length of the pigtails should be such that they nearly meet in the center of the tee.  Note that the length of the 1/2 dipole should be measured end-to-end (including caps and "pigtail") once it is screwed into the tee. This length should be about 38-1/2 inches or so for operation on 2 meters.  Note that the precise length isn't particularly important, as these dipoles are fairly broadbanded.  What is important is that all four of them should be as identical to each other as possible!  (Note:  A small 1/16" drain hole should be drilled in the bottom cap of the dipole to prevent moisture accumulation.)

Once the bushings were screwed into the tee, the coaxial cable was then soldered to the #12 pigtail through the open end of the tee and dipole was marked to identify which element was grounded and which was connected to the center of the coax - and the coax was checked for continuity.  After this check, the coax connection inside the tee was buried in clear silicone sealant to prevent moisture ingress and to provide additional mechanical strength.  Remember that once the Tee with the dipoles is glued into place (after having made certain that all dipoles are oriented the same way - with grounded side up - or down - for all four) the antenna really cannot be disassembled and repaired easily - so do it right the first time!
 

The antenna array's switching controller - at the base of the mast.  This also shows where the cables emerge from the mast.  Although BNC connectors are used for the 4 antennas, they have experienced no weather-related effects being mounted as shown.
Click on the image for a larger version.

One of the trickier aspects of construction is fishing the coax through and around the elbows/tees that make up the support structure.  This was done as the pipes and tees were assembled by fishing small rope (or string) through the pipe.  If this is not done during assembly one will have to be very clever in figuring out how to get the cables around the bends and through the tees as required, or one may be able to tie fishing line to a bolt and tip the assembly to get the bolt to fall through the holes and slide through the pipes.  Once the entire thing is assembled, one uses the ropes/string to pull the coaxial cable through.  A note:  It is best to use a small coaxial cable such as RG-58 for this - and to make sure that the coax runs are as symmetrical as possible.

After the antenna support was completely assembled, the white PVC portions were painted with gray paint.  This was done not only to match the color of the ABS pipe, but to protect the PVC from the effects of UV exposure.

The vertical mast portion of the array is about 3 feet long, as which point there is another Tee - followed by another two feet of mast.  As shown on the picture, there is another right-angle (pointing down) forming a "weatherhead" through which the cables emerge and the enclosure containing the antenna switches are mounted below this point.

As in the mobile-mounted array, there are no PIN diodes at the antennas themselves:  Had this been done, it would have been prudent to make this antenna easily repairable so that the PIN diodes at the dipole element could be easily replaced.  As it is built, one would probably have to saw the array apart to get at the dipole's connections.  With no PIN diodes at the dipoles, it is imperative that the coax length not only be of exactly equal length, but that they be integer multiples of 1/2 wavelength long, electrically - yet be as short as practical.
 

The DF array as installed on the roof.  It has been in service since the summer of 2003 and has withstood several 70+ MPH winds.
Click on the image for a larger version.

Installation:

When originally installed, the center of the array was only about 4 feet above the (metal) roof.  While the array seemed to work ok, bearing accuracy was only within 10-15 degrees or so - with the error magnitude varying depending on direction.  In order to remedy this, the array was raised up another 6 feet using a piece of galvanized 1-1/2" conduit, putting the center of the array at about 10 feet above the roof - or more than one wavelength.  The effect of raising the array was to improve accuracy to better than +-5 degrees for those locations that were line-of-sight to the array as well as improving weak-signal performance.

Because my house has a metal roof and I didn't want to punch holes in it, I decided to use a ballast mount of the sort used for installing small satellite antennas.  As the name implies, this is a metal frame held down by weight - in this case, cement blocks.  Mats are used to protect the roofing from damage from the blocks and the metal frame, and the mount was modified to provide a plumb (vertical) support on the pitched roof.

Since installation, in mid 2003, I have done zero maintenance on the antenna and the array has experienced 70+ mph (>110 kph) winds on several occasions and has suffered no damage - despite swinging around in the breeze a bit more than I'd like.  At some point, I may slide a hardwood dowel inside the ABS mast to provide increased strength and rigidity.

The inside of the antenna switching controller.  From left-to-right:  The diode drive circuit, the PIN diode switch, and the preamplifier.
Click on the image for a larger version.

The "base" antenna switching unit:

The "base" antenna switching unit is very similar to that of the one used mobile in many respects:

• Only a single PIN diode is used to select each antenna.
• Multiples of 1/2 wavelength coax are used connect each antenna to maintain high impedance when one particular antenna is "disconnected."

• The control unit needed to withstand prolonged outdoor exposure.
• I didn't want to run a negative supply to the antenna controller.
• A preamplifier was added to overcome coax losses.  From the roof of the house to the receiver may be quite a long run.

The drive circuitry is fairly straightforward:  An LM324 is used as a comparator/driver, taking the inputted drive signals from the 'EMM box, converting the voltage levels, and driving the PIN diodes.  (A green LED was used to provide a 2.1 volt reference for the comparator.)

Note:  The LM324 was chosen not only because it was cheap and available, but because its output can swing to within 0.1 volts or so of the negative (or ground) rail.  If one substitutes another op-amp, make certain that it, too, can swing very close to the negative rail - or, better yet, both rails!  An amplifier like the TL084 is not suitable in this application.

Because a negative supply was not used, and because it is necessary to put a negative voltage across the PIN diode to place it firmly in an "off" state and reduce its capacitance, a "floating ground" was used on the PIN diode:  Much like on the mobile antenna system, the DC return to the antenna is "grounded" via a 1.5k resistor.  In this case, however, that 1.5k resistor is not tied to DC ground but is bypassed at RF, connected to the "return" line of all of the other PIN diodes, and biased at about 1.9 volts using a red LED.  When any one of the PIN diodes is turned on, the current flow charges up the bypass capacitors (plus an electrolytic) on the "common" line to the LED voltage.  When a PIN diode is off, however, the LM324 drives the voltage to nearly zero volts, thus providing a "negative" voltage across the PIN diode:  Because at least one LED is always on, the 1.9 volt bias is always available when antenna rotation is occurring.  With proper bypassing at each PIN diode/resistor, there is no significant RF coupling occurring on this common line.

The output of the PIN diode switch is applied to a simple grounded-gate JFET preamplifier that uses an MPF102.  There is nothing at all special about this preamp:  Its gain is fairly low - only about 12 dB or so.  The MPF102 isn't a very low-noise device, with its noise figure typically in the 3-4 dB range at 2 meters.  Its sole purpose in life is to provide a little bit of bandpass filtering (owing to the input and output networks on the amplifier) and (most importantly) overcome the losses in the coaxial cable between the antenna switch unit and the receiver.  One nice thing about the MPF102 preamp is that it is quite rugged:  It has already withstood several accidental "keyups" with a 10 watt transmitter without damage.  (Having 50 feet of RG-58 between the transmitter and the preamp and its associated losses is also helpful in preventing damage...)  A lower-noise transistor such as the J310 or U310 could be used instead, but there would be no improvement in performance while the rotation (antenna switching) is occurring - although weak-signal performance may be improved slightly when rotation is stopped and a "test" antenna is selected

This (or any) preamp cannot be expected to do much to improve the signals after passing through an rotating antenna array as the PIN diode action (during antenna rotation) tends to create a fair amount of low-level noise on its own, negating the need of a very low-noise preamplifier as well as masking extremely weak signals.  It should also go without saying that placing preamplifiers in front of the PIN diode switching is not a good idea as the phase of the RF signal will be affected by its tuning and make useless any readings that might be obtained.

Preliminary versions of the schematics are now online.  These have been drawn from my notes, but have not been double-checked against the as-built unit.  Please note:  There are a few specific details that I need to fill in (by opening up the rooftop unit and checking) mostly having to do with a few specific resistor values as well as the precise details having to do with the tap positions on the preamplifier's input/output coils - but none of these details are extremely critical, anyway.  I believe that the schematic is very "close" and will work well if built as described.

• Switch driver circuit -  This shows a the PIN diode driver circuit and some of the additional circuitry.  Note that the Op-Amp comparator and PIN diode circuit is replicated so that there are a total of four of them - one for each of the antennas (although there is no reason why 6 or 8 couldn't be used if an antenna array was constructed thusly.)
• JFET Amplifier -  This is a simple "post-switcher" amplifier that is used solely to overcome the losses incurred by the coaxial cable from the antenna array.  As mentioned, it does not have very high gain or stellar performance in terms of noise figure.  Note that due to its tuned input and output stages, it does have a limited bandwidth.  While its response is flat within 2-3 dB over the 2 meter amateur band, its gain approaches unity in the mid 150 MHz range.
  

Note:  Neither the author or UARC officially endorse any vendors mentioned above.  The level and satisfaction of performance of any of the above circuits is largely based on the skill and experience of the operator.  Your mileage may vary.
 

Do you have any questions on this or other DF-related topics?  Go here.

Return to the KA7OEI ARDF Page.
 

This page updated on 20101223