Chip antennas? vs?

choice of a wire antenna or an on board chip antenna (very convenient) both at the same price.

using one.

The wire will be omnidirectional normal to the axis of the wire. I believe it will be better than the chip antenna. IIRC, the chip antenna is not truly omnidirectional. The only truly omnidirectional antenna of which I am aware is an isotropic antenna which is imaginary.

Do you have a cell phone?

Cheers! Rich

In general (and to a very simplistic approximation), any antenna which is shorter than 1/4 wavelength is going to need some form of "loading" in order to resonate properly. Since the loading and impedance-matching components are real rather than theoretically- perfect, they will add some electrical loss to the antenna system... some portion of the transmit power will be lost as heat.

Again, to a first approximation, the shorter the loaded antenna, the higher the losses.

As an example, small hand-held radios often use inductively-loaded "rubber duck" antennas. a 6" rubber duck may be as much as 10 dB worse, in efficiency and sensitivity, than a 19" quarter-wave whip. That's 90% of your transmit power going away as heat, and a similar loss of incoming signals (resulting in reduced receiver sensitivity).

I don't know the specific loss figures for popular chip antennas. You'd need to look at the data sheet and look at the "in-system" gain figures (i.e. those that account for losses from the PC board, etc.).

You may see complex antenna radiation patterns shown, and may run into two different ways of describing the gain. "dBd" is gain as compared to a half-wave dipole (the commonest "reference antenna"). "dBi" is gain as compared to a theoretically-perfect "isotropic" antenna (one which radiated power equally in all directions).

A real antenna's gain figure measured in dBi is always going to be higher than the gain measured in dBd... 3.02 dB difference if I recall correctly.

WiFi routers that don't have external antennas, are most often using PCB stripline antennas... simply traces etched onto the PC board. These are less expensive than using "chip" antennas, as long as you have enough PCB space to put them.

No antenna is "truly omnidirectional" in the perfect sense (e.g. exactly equal power radiated in all directions in the sphere). It's physically impossible to achieve. There are always compromises in the radiation pattern... you get to pick which compromise best suits your application.

El 19-09-11 21:13, mkr5000 escribió:

choice of a wire antenna or an on board chip antenna (very convenient) both at the same price.

using one.

Statistically spoken, the wire will be better, but this doesn't help you, as there is lots of variation. I saw (measured) chip antennas that impressed me, except for the radiation pattern.

Basically an unobstructed wire radiates most of the power perpendicular to the wire. So when you place it vertically, you have omni directional behavior in the horizontal plane. Remember that other objects (metal and plastic) in the vicinity of the antenna affect the radiation pattern. Of course metal affects more.

If you have space and don't have means to verify the radiation pattern, I would go for the "wire" version (highest probability of success).

There is no info on performance in the datasheet?

choice of a wire antenna or an on board chip antenna (very convenient) both at the same price.

using one.

With some dielectric chip antennas (don't know if this is what you meant) you also had to be very careful with the PCB board layout, as it also contributes to radiation. Other "on board antennas" are a "proper" loop antenna variant, already sufficiently well matched and performing more or less as as the wire. In theory, under ideal conditions, the wire may perform better, but in practice, the incomplete ground plane, the enclosure, nearby objects or cables feeding the board etc. seriously degrade performance.

Pere

Thanks! I stand, corrected (he said, rising from his chair).

In the real world a simple wire antenna nearly always works better than a chip antenna.

For a small antenna there is a three-way tradeoff between efficiency, bandwidth and size. So you can have a small efficient antenna with narrow bandwidth, but as soon as the environment changes the antenna detunes and the performance drops badly. A 1/4 -wave monopole or a half-wave dipole is fairly tolerant of changes in the immediate environment and has a wide bandwidth. It is also usually very efficient.

Furthermore it is usually possible to get a good match to 50 Ohms with a 1/4 wave monopole just by adjusting the length - there is seldom a need for any matching components.

If you have the space it is the best choice.

John

How about a PCB trace? The problem with a wire is that it needs mounting and its shape is hard to control.

A pcb trace can work well as a monopole so long as it is kept well away from the ground plane and other conductors (except at the feed point of course).

A trace which is at right angles to the edge of the ground plane in a copper-free area is nearly as good as a wire monopole. It needs to be a little shorter because of the dielectric loading. There is more loss than with real wire.

Don't make it too narrow as this increases the losses. Some means of measuring the return loss such as a network analyzer or spectrum analyzer with tracking generator and directional coupler is very helpful. There is often an interaction with the ground plane dimensions, so lengths do not always work out quite as expeected. Think of the empty space immediately around the antenna as an essential part of it and avoid putting conductive things there. Even non-conductors such as plastic housings can have some effect if they are close to the un-driven end.

John

Hello John,

And this is where it goes wrong frequently (and I am sure you know that too).

People measure the antenna separately (without the actual ground plane and casing present), or they connect the cable to the antenna input and forget that the cable is now part of the electromagnetic structure.

In case of a large ground plane (that is not in high impedance resonance), common mode cable effect will be (very) limited, but with a small ground plane, or a ground plane with inconvenient size, common mode current at the cable towards the VNA may have large effect.

Yes. It is instructive to place multiple ferrite absorbers on the cable to the VNA. Sometimes it makes a big difference, other times very little.

I usually design an SMA footprint into the layout close to the antenna feed point with the feeder track passing straight through it and cut a hole in a prototype board to allow the SMA and test cable to be attached.

Bonding the (semi-rigid) cable to the pcb with copper foil over most of the board area and then blocking surface currents in the cable with ferrite beyond the boundary of the board usually works well.

John