Bluetooth PCB inverted F antenna design

[sensors]

I've been trying to design an omnidirectional PCB inverted-F antenna for a bluetooth project. The aim with the design is to ensure the best range possible with the space I have; Ideally an average gain of >1dBi. I've come up with a design that I think may work, but I really am struggling to verify if it will perform as required.

Here is the design layout (the tail connects to ground, just doesn't look like it here):


I know antenna design is not a trivial task, but will this antenna work to my standards? if not, what might need modifying to make it perform?

Are there any alternative antennas out there that may fit my performance and spatial requirements? A chip antenna may be a possibility, but it seems many of them need mounted vertically and I don't have the space for that. I looked at a TI PCB IFA design but it's just slightly too large for the ~34mm x 5.5mm I have available.

[T3sl4co1l]

Your requirements are conflicting:
Omnidirectional, yet high gain ("best range possible").

The best you can hope for is high efficiency, so that power isn't being lost into, say, the FR4 (which isn't the greatest dielectric at radio frequencies).  If that's always true (or good enough), gain and directionality are inversely proportional.

There's no such thing as a true omnidirectional antenna (which relates to the "Hairy Ball Theorem" in mathematics, interestingly enough), so you'll have to settle for dipole, or dipole-like, which gives about 2dBi (i.e., 2dB above isotropic, assuming it were possible).

If you need more range than this provides, you must sacrifice directionality (and size!).

So if this is all within your expectations, great.

So on to the much more important question -- will it work?

If the antenna is tuned poorly, it may not work at all!  You need it to resonate very close to the operating frequency.  This looks to be a 1/4 wave type, so its bandwidth should be okay*, and it won't need to be cut terribly accurately to the frequency.  You should be able to solve the rest by substituting parts in the antenna matching circuit (all the L, C and R parts on that side of the chip, I suppose).

*Which also implies it could be smaller -- you still get a dipole radiation pattern, but the physical size can be smaller.  Small antennas couple more poorly to space, so they need to be tuned more precisely, and need to be even more efficient, to achieve the same gain.  The smallest size antenna is one that just barely covers the ~2MHz bandwidth that BlueTooth requires, but this would be very precise to actually manufacture.  Still, this explains how most IoT radios can be so small.

Oh, and you may want to add more stitching vias near those matching components.

Final FYI: if you're just making one-offs, it's probably "who cares".  But if you want to sell this thing, you'll need CE certification for an intentional radiator.  That's expensive, and it's a very good reason why radio modules are so common these days.  The module itself (with integral antenna, or a connector suitable for a limited range of appropriate antennas that can be attached to it) is CE certified, so that the rest of your product doesn't need to be.

Tim

[dcarr]

Really two options here:

1. Build a prototype and test the performance
2. Model the antenna and PCB in a full wave EM simulator that supports far field radiation patterns

The former is probably faster, easier and likely cheaper but won't give you a quantitative answer (ie: gain = 1.2dB), but will tell you "it works".
The later option will give you an antenna pattern, but you'll need to invest significant effort to get a realistic model (and quite possibly significant cash to pay for the software).

If you do go the EM simulator route, take a look at Sonnet (expensive), OpenEMS or EmGine (last two free).

David

[Ammar]

Finite difference time domain modeling is the way to go here. I have access to CST microwave studio, which works a treat.

Otherwise you can try building the antenna and testing it. Depends what equipment you have available though.

[pigrew]

TI has some other reference design that (while I have not measured them), may work. Take a look at http://www.ti.com/lit/an/swra117d/swra117d.pdf (AN043). It has a meandering F-antenna pattern. Though, it still may be slightly too large. That antenna is 15.2 x 5.7 mm. Perhaps you can find an extra 200 um by moving the ground plane down a bit?

But, this antenna may not be as good as the one you currently have. I would have to simulate (or just test it out) to see. I have no experience with antenna design, so I would tend towards using an already existing design.

TI's app note AN058 lists most of TI's designs.

Quickly browsing chip antennas, there are options like Johanson's 2450AT43B100E, which only needs 4.87mm.


-Nathan