ANTENNAS FOR LOW POWER APPLICATIONS
By Kent Smith
Introduction:
There seems to be little information on compact antenna design for the low power wireless field. Goodantenna design is required to realize good range performance. A good antenna requires it to be the righttype for the application. It also must be matched and tuned to the transmitter and receiver. To get the bestresults, a designer should have an idea about how the antenna works, and what the important designconsiderations are. This paper should help to achieve effective antenna design.
Terminology:
Wavelength - Important for determination of antenna length, this is the distance that the radio
wave travels during one complete cycle of the wave. This length is inversely proportional to the frequency and may be calculated by: wavelength in cm = 30,000 / frequency in MHz.Groundplane - A solid conductive area that is an important part of RF design techniques. These are usually used in transmitter and receiver circuits. An example is where most of the traces will be routed on the topside of the board, and the bottom will be a mostly solid copper area. The groundplane helps to reduce stray reactances and radiation. Of course, the antenna line needs to run away from the groundplane.
dB (decibel) - A logarithmic scale used to show power gain or loss in an RF circuit. +3 dB is twice the power, while -3 dB is one half. It takes 6 dB to double or halve the radiating distance, due to the inverse square law.
The Basic Antenna and how it Works.
An antenna can be defined as any wire, or conductor, that carries a pulsing or alternating current. Such acurrent will generate an electromagnetic field around the wire and that field will pulse and vary as theelectric current does. If another wire is placed nearby, the electromagnetic field lines that cross this wirewill induce an electric current that is a copy of the original current, only weaker. If the wire is relativelylong, in terms of wavelength, it will radiate much of that field over long distances.
The simplest antenna is the “whip”. This is a quarter
wavelength wire that stands above a groundplane. The
most common examples are found on automobiles and are
used for broadcast radio, CB and amateur radio, and even
for cellular phones. This design goes back to the 1890's
when Marconi set out to prove that radio signals could
travel long distances. To be successful, he had to stretch a
long wire above the ground. Due to the low frequencies,
thus a long wavelength, the wire had to be long. He also
found that the wire worked better when it was high above
ground.
All antennas, like any electronic component, have at least two connection points. In the case of the whip,there must be a connection to a ground, even if the groundplane area is nothing more than circuit traces anda battery. The whip and groundplane combine to form a complete circuit. The electromagnetic field is set upbetween the whip and the ground plane, with current flowing through the field, thus completing the circuit.Ideally, a groundplane should spread out at least a quarter wavelength, or more, around the base of thewhip. The groundplane can be made smaller, but it will affect the performance of the whip antenna. Thegroundplane area must be considered when designing an antenna.
A quarter-wave whip is not a compact antenna. At 1 MHz, in the AM Broadcast band, one quarter of thewavelength is about 246 feet, or 75 meters. At 100 MHz, in the FM Broadcast Band, it is nearly 30 inches(75 cm). This dimension continues to shrink at higher frequencies, being nearly 3 inches (7.5 cm) at1000 MHz. A simple formula for the quarter-wave (in cm) is: 7500 divided by the freq. (in MHz), or forinches: 2952 / freq. (in MHz). This formula is only a starting point since the length may actually be shorterif: the whip is overly thick or wide, has any kind of coating, or is not fed close to ground. It may need to belonger if the ground plane is too small.
The length of the antenna should be measured from the point where it leaves close proximity to ground, orfrom the transmitter output. If a whip is mounted on a box, and connected to the transmitter with plain wire,that wire becomes part of the antenna! To avoid mistuning the antenna, coaxial cable should be used toconnect to an external antenna. On a circuit board, the equivalent to coax is a trace that runs over agroundplane (groundplane on the backside). The above are examples of transmission lines, whose purposeis to efficiently transfer power from one place to another with minimum loss. Do not try to run an antennaline too close to ground, it becomes more of a transmission line than an antenna. Fortunately for those whoneed a small remote device, a transmission line left open-ended will radiate some energy.
Antenna Characteristics:
Gain:
An antenna that radiates poorly has low “gain”. Antenna gain is a measure of how strongly the antennaradiates compared to a reference antenna, such as a dipole. A dipole is similar to a whip, but thegroundplane is replaced with another quarter-wave wire. Overall performance is about the same. An antennathat is 6 dB less than a dipole is -6 dBd. This antenna would offer one half the range, or distance, of thedipole. Compact antennas are often less efficient than a dipole, and therefore, tend to have negative gain.Radiation Pattern:
Radiation is maximum when broadside, or perpendicular to a wire, so a vertical whip is ideal forcommunication in any direction except straight up. The radiation “pattern”, perpendicular to the whip, canbe described as omnidirectional. There is a "null", or signal minimum, at the end of the whip. With a lessthan ideal antenna, such as a bent or tilted whip, this null may move and partly disappear. It is important toknow the radiation pattern of the antenna, in order to insure that a null is not present in the desired directionof communication.
Polarization:
It is important that other antennas in the same communication system be oriented in the same way, that is,have the same polarization. A horizontally polarized antenna will not usually communicate very effectivelywith a vertical whip. In the real environment, metal objects and the ground will cause reflections, and maycause both horizontal and vertical polarized signals to be present.
Impedance:
Another important consideration is how well a transmitter can transfer power into an antenna. If the antennatuning circuit on a transmitter (or receiver) is designed for a 50 ohm load, the antenna should, of course,have an impedance near 50 ohms for best results. A whip over a flat groundplane has an impedance near35 ohms, which is close enough. The impedance changes if the whip is mistuned or bent down, or if a handor other object is placed close to it. The impedance becomes lower as the antenna is bent closer to ground.When the whip is tilted 45 degrees, the impedance is less than 20 ohms. When the whip is bent horizontal to
one-tenth of a wavelength above ground, the impedance approaches 10 ohms. The resulting impedancemismatch, a 5:1 ratio (VSWR) will contribute an additional loss of 2.6 dB.
Printed Circuit Whip, or “Stub”
The whip can be made as a trace on a printed circuit board (PCB). This is very practical at frequencies over800 MHz. At lower frequencies, a full size whip may be too long, even when wrapped around a fewcorners. The length of the whip should be 10 to 20% shorter than the calculation, depending on thedielectric and the thickness of the board. In most cases, 15% shorter is close enough. If the unit is to behand held, the antenna can be made a little shorter, to compensate for the effect of the hand.
At 916 MHz, a trace that is 2.25 inches (57 mm) long
will provide a reasonable impedance when hand
effects are included. Keep the antenna trace awayPrinted Open Stub:
from other circuitry and ground, a quarter of an inch
(6 mm), or more. Non-ground circuit traces may be916.5 MHz
seen by the antenna as part of the ground system, and
RF voltages can be induced on nearby traces.Our sample PCB Stub is shown in the drawing at
right. The overall size of the board and ground is notcritical. The radiation pattern is omnidirectional, with
a gain of -8 to -12 dBd, when the board is horizontal.Polarization is horizontal. If the whip did not run
parallel to ground, the gain would be higher,
however, two sharp nulls would be present. If the
board were oriented vertically, with the antenna
above the groundplane, the polarization would be
vertical. The antenna would have an omnidirectional
pattern with -8 dBd of gain.0
270
Radiation Pattern of Open Stub Antenna
(916.5 MHz)
Compact Antennas:
The Short Whip
A simple alternative to the whip is to make it shorter than aquarter wavelength and add an inductor near the base of thewhip to compensate for the resulting capacitive reactance. Theinductor can be made by coiling up part of the whip itself. Thistype of antenna can have performance nearly equal to that of afull size whip.
RFM uses such a design for the wire antennas that are suppliedwith our demonstration boards. Details of the design can befound in the HX/RX portion of the Product Data Book. TheRFM short whip is optimized for under-sized groundplanes.When tested on the edge of a small board, gain was only 3 to4 dB less than a full sized whip and groundplane.
270
Loaded Whip Antenna (434 MHz)
270
The Short PCB Stub
One big advantage for the short whip is that it can be a traceon a PCB, with a chip inductor used to tune out thecapacitive reactance of the antenna. If the trace runs parallelto ground, the real part of the antenna impedance will beapproximately 10 ohms. In a hand-held unit, the impedancewill be raised substantially through hand effects. For a tenthwavelength strip on a board with hand effects included, theantenna has a capacitive reactance of about 150 ohms. At433.9 MHz, this would require a 56 nH inductor to cancelShort Stub: 433.9 MHzthe capacitive reactance of the 2.7 inch (70 mm) long line.The radiation pattern will be fairly omnidirectional, with ashallow null along one axis. The polarization is roughlyparallel with the edge of the board. Tuning is not extremelycritical, small variations in inductor value or antenna lengthwill not have a great effect on performance. Our sampledesigns, at 433.9 and 916 MHz, resulted in maximum gainsof between -12.5 to -14 dBd off the side of the board. Thenull dipped down to about -26 dBd. This is moreomnidirectional than some other designs, and hand effectswill help to reduce the null depth.
The key to this design is to keep resistive losses low, usewide traces (if a PCB trace), and good quality inductors.Adjust the inductor value for maximum output in theenvironment that it will be used. Gain can be improved bymaking the whip longer and thus reducing inductance. But,in some cases, it may be better to shorten the trace and addinductance rather than to run the antenna close to othercircuit board traces.
270
Short Stub (916 MHz)Short Stub: 916.5 MHz27 nH
The Spiral
Another way to shorten a whip is to coil it up to form aflattened coil of wire. It can be a trace printed onto acircuit board. On a board, the length of the trace is a littleshorter than a quarter wavelength. The antenna must nothave a groundplane directly under it, and should occupy aclear end of the board. For example, start with a six inchlong thin trace wrapped in a 0.75 inch (19 mm) squarearea, then trim a little of the length until it resonates at433.9 MHz.
Antenna gain and impedance will vary with the size of thegroundplane. Our 433.9 MHz version had a fairly smallgroundplane area of 17 sq. cm, while the 916 MHzversion had a quarter-wave long ground. The 433.9 MHzantenna had a maximum gain of -10.5 dBd, with a smallnull of -24 dBd. The 916 MHz antenna had a gain of-5 dBd max. Comparable gain is also seen when lookingat the board face-on.
This antenna does not give circular polarization; thepolarization is parallel to the long edge of the board. Aswith a stub, when the board is oriented vertically, it isvertically polarized and omnidirectional. This antenna ismore easily detuned by a hand, which makes it lesssuitable for hand-held remotes.
Spiral: 433.9 MHz
270Spiral Antenna (434 MHz)
The Helical (Coil)
This is similar to a spiral that is not flattened. Start with apiece of wire that is 2 or 3 times longer than a whip andwind it into a coil. The number of turns on the coil willdepend on wire size, coil diameter, and turn spacing. Thecoil will need to be cut to resonate, and can be fine tunedby spreading or compressing the length of the coil. If thecoil is wound tightly enough, it may be shorter than one-tenth of a wavelength. This antenna tunes sharply,requiring care in tuning. The real part of the antennaimpedance is less than 20 ohms, and depends on the sizeof the coil and its orientation to ground.
For 433.9 MHz, we wound 14 turns of 22 gauge wirearound a 0.25 inch (6 mm) form. When tuned, it’s lengthwas just under one inch. The proximity of this coil toground makes a big difference in performance. When thecoil runs near and parallel to ground, maximum gain isonly -18 dBd. When the loose end of the coil was pulledaway from ground, as shown in the alternate versiondrawing, gain increased to -5.5 dBd, and the null becamedeeper.
The big problem with this antenna is the mechanicalconstruction and it's bulky size. It can be easily de-tunedby nearby objects, including a hand, so it may not be goodfor hand-held use.Helical:stretch to tune0
270Helical Antenna (434 MHz)
“Chip” Antenna
The latest entry into the antenna field is the tiny “chip” antenna.They are surface mount devices that are typically 8 by 5 by
2.5 mm, making them the smallest design available. They may befound for frequencies less than 300 MHz and up to 2500 MHz.These antennas are similar to whips in behavior, only muchsmaller. If an antenna can be reduced in size, while maintainingefficiency, bandwidth will be reduced. So these devices have a verynarrow bandwidth and must be made to the exact frequency.
These devices are very groundplane dependant. As a result, theyare easily detuned by hand effects, the wrong size groundplane, oreven the wrong thickness and dielectric of the board. The chipantenna must be used according to the manufacturer’srecommendations.
For 433.9 MHz, we mounted a chip on a 5 inch long board andobtained a maximum gain of -10 dBd. Not bad when you considerthat the spiral has equal gain, but consumes five times as much areaon the board. The 916 MHz version did better with a 2.6 inch longgroundplane for a maximum gain of -3.2 dBd. The polarization isparallel to the long axis of the chip, so maximum radiation isperpendicular to the long axis. There is a deep null (nearly 40 dB!)looking at each end of the chip. This would be a big problem if anomnidirectional pattern is required from a horizontal circuit board.When the board is vertical, the pattern is omnidirectional.
270
Chip Antenna (434 MHz)“Chip” Ant.: 433.9 MHz
The loop is entirely different from a whip, in that bothends of the antenna are terminated. In this case, the endthat is opposite the transmitter (or receiver) is grounded.
A capacitor is used to tune the antenna to a realimpedance, instead of a coil. An advantage of a loop isthat it is not easily detuned by hand effects, although theimpedance may still vary. The loop can be made small,does not require a groundplane, and takes no more spacethan a short whip. For these reasons, loops are verycommon in hand-held devices.
There are some disadvantages. Small loop antennas havea reputation for poor gain. A small loop will have a verynarrow bandwidth. This makes tuning extremely critical.Tuning is often done with a variable capacitor, which addsto the cost, both parts and labor. If the loop is largeenough, it may be practical to use a non-variablecapacitor. This requires careful adjustment in engineeringstages, to ensure that it is properly tuned with a standardvalue capacitor.
Our example loop antenna covers a 12 by 35 mm area onthe end of a board. It is tuned to 433.9 MHz with avariable capacitor. This antenna is very omnidirectional,but had a gain of only -18 dBd. A larger loop should haveimproved gain.Loop: 433.9 MHz90
12060
30
1800210330
240
270300
This is an unusual design that looks like a loop, butrequires no direct grounding. It is comparable to a loop inSemi-Loop: 433.9 MHzperformance, and can be adjusted to present a non-reactive load. This antenna uses a trace that runs all theway around the edge of a small PCB. The far (open) endis capacitively coupled, through the board, back to thetransmitter end of the antenna. The antenna is resonatedby varying the length of the short overlapping line.Tuning is not very critical. Hand effects will improve theimpedance, with little effect on tuning. Polarization isparallel to the PCB, and the pattern is omnidirectional.Our design had a gain of -15 dBd at 433.9 MHz. Thisdesign works very well for hand-held devices.
As with any other designs, this antenna should not run tooclose to ground. For this design, the transmitter and othercircuitry, including battery, should be grouped around thecenter of the board, leaving the antenna in the clear. Thecircumference of the board needs to be well under one-quarter wavelength. We have had good results with acircumference of about 0.15 wavelength, and a line widthof 1 to 1.5 mm, when used in the 400 MHz region. If thedesign is used on a thinner board, the 5 mm overlap willneed to be shortened.
270
Semi-Loop Antenna
434 MHz5 mm.060 inch thick FR4
Modified Dipole Antenna
A dipole can be shortened somewhat by bending the wire
or line back on itself, but not too close to itself. We built a
version on a PCB, shown at right. This antenna has almost
the same performance as a full size dipole, but is more
compact. The thickness and dielectric constant of the
board will affect the tuning, so the length may need to be
adjusted.
This type of antenna is an attractive solution where space
allows. However, a dipole should not be located close to a
large metal area or groundplane. The groundplane will
become part of the antenna, and performance will suffer.
Like the normal dipole, the radiation pattern shows deep
nulls and good gain. The impedance is a little lower, but
still near 50 ohms. Like many of the previous antennas,
radiation from the face of the board is just as strong as
from the long edge.
90
12060
15030
1800210330
240300
2703mm linewidth
Common in radar systems and/or on aircraft, a variation of the slot antenna may have potential above800 MHz. A quarter-wave slot is cut into a metal sheet or unetched PCB, and if enough area is available,will provide omnidirectional coverage. Our sample antenna at 916 MHz required a 75 mm long PCB. Thelength of the slot was 59.5 mm for 0.060 inch (1.5 mm) thick FR4. A different thickness or dielectric willrequire changing the length of the slot. One end of the slot must be left open. The slot was fed near theclosed end, in this case 4 mm from the end. The feedpoint impedance can be adjusted by moving the feedtoward or away from the closed end. Tuning is somewhat critical.
When the board is horizontal, the pattern is omnidirectional around the edge of the board, thus horizontallypolarized. We also see omnidirectional coverage when the board is vertical (with the slot horizontal). In thiscase, polarization is vertical! It may not make sense, but a horizontal slot is equivalent to a vertical whip inthis case. Gain is -4.5 to -6 dBd. The feed can be a trace on the backside of the board, with a via used tomake connection with the top of the board near the slot.
90
12060
30
1800210330
240
270300
The Patch antenna is a very low profile design, which consists of a round or rectangular patch of metal veryclose to a groundplane. The patch is usually printed on a circuit board and can be made as part of theenclosure. Antenna coverage is in any direction above the groundplane, or a hemispherical area. The patchantenna does require a substantial amount of area on a PCB, which makes it more practical above 800 MHz.It has a narrow bandwidth so care must be taken to tune the size of the patch carefully. It is sensitive to thethickness and dielectric constant of the PCB and small variations will mistune the patch completely. It isalso sensitive to coatings, but not extremely sensitive to hand effects.
A practical example for 916 MHz can fit into an area only 30 by 40 mm. The patch size is 27 mm wide by38 mm long for a board thickness of 0.060 inch. A thinner board or higher dielectric can require cutting theantenna a little shorter. About one-tenth of an inch of board space should be left around any ungroundededge of the patch. One edge of the patch should be grounded with multiple PCB vias. The antenna is fedwith a line crossing through the grounded edge to the 50 ohm point on the patch, or by a transmission linecoming up through the bottom of the PCB. The 50 ohm point is about 13 mm away from ground on ourexample patch. The 50 ohm point for any design can be found by moving the vias toward or away from thegrounded edge. The farther the feed is away from the ground vias, the higher the impedance will be.
This type of patch is not a full-size, half-wavelength patch, so performance is not as good as a larger sizepatch. A full-size patch has no grounded edge, so vias are not required. Our example rectangular patch has again of -8 dBd. Placing the board against a larger sheet of metal will improve the gain by another 4 dB. Ifthe antenna is made wider than one inch, up to about 3 inches wide, a few more dB can be gained.Polarization is perpendicular to the grounded edge. Gain is good in almost any direction where the patchcan be seen, but drops rapidly when looking at the edge of the board.
The trapezoidal version allows for less length so that it can fit into smaller spaces. Patterns and behavior arethe same, but the gain is a little lower. We measured about -12 dBd maximum, on a 40 by 90 mm board.
Rectangle Patch: 916.5 MHz
270Trapazoidal Patch over a
Small Ground Plane
(916.5 MHz)
270Trapazoidal Patch over a
Large Ground Plane
(916.5 MHz)
Enclosures
An antenna should not be located inside a conductive, or metal enclosure. Care should be taken to keep theantenna away from metal surfaces. If a conductive area is large in terms of wavelength (one half wave ormore), it can act as a reflector and cause the antenna to not radiate in some directions. If a metal box is usedfor an enclosure, an external antenna is required.
Testing and Tuning
Antennas may seem to be a mystical art. Unlike many electronic devices, any change in nearby materials ordimensions can affect antenna performance. Trying to build a published design does not guarantee results.Testing an antenna design is necessary, tuning is usually required, and there are pitfalls along the way.
A network analyzer is normally used to test the impedance or VSWR of the antenna. Some antennas thathave an impedance near 50 ohms can be tuned by looking at return loss or a VSWR display. Lowimpedance antennas may require the use of a Smith Chart display to get good results. In this case, theantenna should be tuned to a point near the pure resistance line.
There are other options, such as a spectrum analyzer with a tracking generator, that can be used with adirectional coupler. The coupler will feed power to the antenna while feeding the reflected power from theantenna back to the analyzer. The coupler must have an isolation between the Generator and RF Input Portof 20 dB or more. Calibration is done by noting the power readings with a 50 ohm load connected and thenunconnected. Using this technique, “return loss” can be measured. If the antenna is near 50 ohms, the returnloss back to the RF input port will be high, due to the antenna absorbing most of the power. A good antennawill show as a dip on the screen at the correct frequency. A dip of only 3 or 4 dB (about a 5:1 VSWR) isnormal for a low impedance antenna measured on a 50 ohm analyzer. A dip of 9 dB (about 2:1) or moreindicates a well-matched antenna in a 50 ohm system. If the dip is not centered at the right frequency, theantenna length or tuning needs to be adjusted.
Antenna measurements of any kind are tricky since the antenna is affected by nearby objects, including thesize and shape of the circuit board, and even by the cable connections to the network analyzer. Pass yourhand close to the antenna and the dip should move around a little. If it does not, the antenna may not beconnected properly. Antennas that are ground plane sensitive may see all additional wires as an extension ofthat ground. Try wrapping your hand around the cable that goes to the analyzer. If the measurement changesmuch, you may need to try a different tactic. One possibility to minimize RF currents on the cable is to put afew good high frequency ferrite toroids or some absorptive material over the cable.
The best way to fine tune a remote transmitter antenna is by using the transmitter itself. Put an antenna on aspectrum analyzer and try to keep other large metal objects out of the way. Find a place to locate thetransmitter that is away from metal and a few feet away from the analyzer. Always locate the transmitter inthe exact same spot when testing. If you have a desk that is wood, mark it’s position with a pencil or tape. Ifhand held, hold it in your hand just above the marking on the desk. Be sure to position your hand, and therest of your body, the same way during each test. Take a reading of the power level, and tune the antenna toachieve maximum radiated power. The same thing can be done for a receiver. Transmit a signal to it, andadjust the antenna to receive the lowest signal level from the generator.
Common problems with antennas usually involve insufficient free space around the antenna. The antennacannot run close to ground or any other trace without effecting the antenna performance. This includestraces on the other side of the board, batteries, or any other metal object.
Receiver performance can be degraded by digital circuits. Digital switching is very fast and creates highfrequency noise that can cause interference. Keep receiving antennas away from digital circuit traces. Try tokeep digital traces short, and run them over a groundplane to help confine the electromagnetic field that isgenerated by the digital pulses. If an external antenna is used, then use a coaxial cable.
A transmission line for G-10 material that is .06 inch thick requires a trace width of a tenth of an inch , halfof that for a .03 inch thick board. This results in a 50 ohm transmission line that will carry RF withminimum loss and interference.
High static voltages may damage sensitive semiconductors or SAWs. We recommend placing an inductorbetween the antenna and ground to short out any static voltages. For the 400 MHz region, a value near200 nH is a good choice. At 916 MHz, a more appropriate value may be 100 nH.
Acknowledgments
The author would like to thank John Anthes, Harry Boling, and Jeff Koch for their assistance in thepreparation of this paper.