Question about this PCB antenna

I'm not an expert on RF design. So I'm going off of this reference design to use this chip.

So the guide, pages 5-7 (copied below) seems pretty straight forward. Cut out a 5mm x 10mm opening in the board, run your traces to the antenna pins on the chip.

My question is, does the outside size not matter or am I missing it? If this is to scale it looks like it's about 2.5 mm wider than the opening on both sides in their drawing. If it is flexible what does it's size effect?

And in addition, does anyone have a guide they can suggest for PCB antennas that would cover ones that work for this kind of RFID? They are all 860-960 MHz chips.

In case anyone found this looking for similar devices, the two 860-960 MHz chips with I2C I have found that I've been able to find in stock (though not in many places) have been the LXMS2HACNF-165 from Murata mentioned in this thread, and the SL3S4011FHK/SL3S4021FHK from NXP. If anyone knows of more, please let me know. These both seem to be deprecated.

Here is the datasheet in question

With dimensions like 5 mm from the chip out to the ground plane area, this "antenna" is electrically short, and being arranged the way it is, my guess is that the type of antenna they are working with here is a loop, since the entire RF path here is electrically pretty short ( about one eighth wavelength, roughly) and both conductors are brought around in a loop shape.

This kind of antenna is usually fairly narrow bandwidth. With chips like this it is the entire pcb that will be radiating and so the larger the overall ground plane is, the more effective it will be as an antenna system. However, the ground plane dimensions are probably not terribly critical to the impedance matching problem that they are trying to solve with this layout, otherwise they would certainly have specified the overall board/groundplane size. From my experience making pcb 800 MHz antennas, you will find that performance improves the larger you make the ground plane, at least up until it is one half wavelength in either dimension. At that dimension, the entire board is efficiently resonating and you tend to maximize the antenna efficiency and gain. Most applications make the board size whatever is needed to fit the product design and then just accept whatever radiation performance they get. But stick to their recommended layout to optimize the impedance match as I'm sure they've put a good deal of work into it. If it is operating as a loop, then the outer dimensions of the pcb won' t make any difference to the impedance match, but if you have a chance to make a larger board, go ahead and do it. If not, don't sweat it. The reason for this is that the majority of the RF current is flowing along the inside edge of that opening and not much is flowing on the outside edge of the pcb.


If the board dimensions are completely aribtrary, do yourself an favor and set one side to 120mm. Oh, and make sure the ground plane is present all the way around the perimeter of the board. This would be helpful.
(note 120 mm is one half wavelength at 850MHz times the velocity factor of the pcb which is roughly 0.7)

Thanks, that was a lot of new information for the size of the post. I knew that wavelength was the speed of light over frequency, c/f, but I hadn't heard of velocity factor. My usual area of work is microcontrollers and power electronics.

So I'm curious then about this other reference design I used. I'm taking a shotgun approach here because I'm not totally sure what will work best and ordering PCBs is comparatively cheap.


Like you mentioned it's a loop. The design notes said to make sure the legs are symmetrical if at all possible (not sure why that matters to be honest). So following what you mentioned, c/850MHz * .7 = 247mm. Why is the top 80 mm? If you follow the outer loop of this board you get 275mm, and the inner loop is about 60mm, which I suppose is close to 1/4 wavelength. It's the inner loop that matters right?

Just to make sure my assumptions are right, because I don't really know what I'm doing. There should be no ground plane behind the antenna, and the antenna should only be copper on one side?

The last question is easier, so that one first. Yes, there should be no groundplane behind the antenna and the antenna should only be copper on one side.
As for the other reference design, well.... you know what, all this effort that goes into calling one antenna a loop and another a dipole can be a bit misleading because you can change a design in a few siimple ways and make a loop into a dipole just with dimensions. This new reference strikes me as one of those cases. Here we have a dimension around the inner ground plane edge of 62 mm, which works out to be electrically larger and roughly one quarter wavelength. So, you can call it a loop, or you can call it an electrically short slot dipole, which I'm less familiar with. The textbook names come about as a result of the current distribution on the conductor which in this case is not purely loop nor purely slot dipole, and perhaps a bit inbetween. Anyways, The fact that they dimension the outside of the board implies that not as much current is constrained to the ground plane opening and more is flowing on the outer edge, so that current contributes more to the overall impedance match when compared to the previous loop structure. Just a guess.

Most antenna designers use EM simulators nowadays and stray a long ways away from classic structures like dipoles and such. Open up a phone and you may not recognize a classicly shaped antenna and even calling one a PIFA derivative is perhaps not even accurate. But I digress. Are you going to build both designs and see how they work?

Yes, I'm producing both. I do a lot of proof of concept contract work and this job is mostly in my wheelhouse but designing RFID antennas is not. So in order to make sure I get something working in time the cost of having several boards made is not a big deal. I do have an Othermill PCB mill that I was considering using to test boards even faster, but a simulator may be a good idea as well. I've downloaded EZNEC and am working on figuring it out, if you have any recommendations I'd be glad to hear them, I'm proficient at MATLAB, C/C++, C# and python so even if it requires some coding or scripting I'll do fine.

I really appreciate the help, by the way.

I was also looking at designs of RFID sticker antennas in the same range (860-960 MHz) for inspiration. One like this looks good for the space I need to fit it in, but I'm not sure how to figure out exactly how big it should be. For now I've ordered a bunch of these stickers (they're super cheap) and I was considering even just cutting the tiny chip out of the middle and soldering on a UMCC that I can attach to an adapter board I made for the chip. If that's not too wonky of an idea.

That's a wild pattern but we can see that there is a balanced structure where the ends are not folded back to the start, so this began its design life as a dipole, then got top loaded with those larger pads at the ends, then got meandered to shorten it. That loop-like shape that is embedded at the center, that may be a structure to match the impedances since many of these chip designs assume they are driving a loop and have their output impedances set up that way.

I've done a fair number of simulations with EZNEC and enjoy using it, but it has some severe limitations so can't be used in all situations. At my last job we were using CST Microwave suite the most. CST, unlike EZNEC is a 3D simulator that understands planar conductors and 3d shapes (and any sheet or volume current flow) while EZNEC requires that you simplify your model to wire elements only. So, you can get useful info out of EZNEC, but first you have to boil your design down to interconnected wires, and this requires knowledge or intuition about where the current is flowing so that your wire approximation is close enough to be useful. I've done some EZNEC models of antennas on vehicles for example where the vehicle can be modelled by drawing an outline of it with wires and filling in the large planar sections with a simple criss-cross or very simple grid of wires. The complexity depends on frequency(well wavelength to be more accurate).
The nice thing about EZNEC (like many programs that use NEC cores) is that its fairly easy to learn quickly and do useful things with. CST on the other hand has a steeper longer learning curve. Not to mention a huge difference in price to buy the package!
In your case, if you use EZNEC on your reference designs, will you be able to interpret the results? It will tell you things like impedance and radiation performance and give you a feel for current flow. Can you understand all these things?

This begs the larger question, which is, how will you know when your antenna is working? Are you able to measure radiation performance accurately? I doubt it because even RF engineers struggle with this. Will you be able to measure how close your impedance match is? That's not easy and requires a network analyzer or similar instrumentation. Or will you just do the best range test that you can and leave it at that. That's what most people do who are simply applying one of these RFID chips to their product. That may be a good enough approach that doesn't demand you understand exactly how the antenna is working, and in this case you may be wasting your time with a simulator.

in post #1, a 1/2 wave dipole is formed by the two traces along the top edge.

So the antenna came in and it works, about as well as the reference antennas on the sticker tags. Now I just need to get that I2C read to work, which shouldn't be too bad if it's like most I2C memory.



So I'm going to keep prototyping with this. But I was wondering, if my goal is a long read distance in a smaller size, what would you say is a good type of antenna to try? This RFID is in the 860-960 MHz band, for my region in practice we're centered around 915 mhz. I don't have an exact size goal but I have to fit it into an existing product, under 2 square inches would be best.

I would try the other antenna that you put in this thread, the first one. You can fit that into your board size without too much trouble.

Effectively, this is a balanced loop antenna. your layout shows 'ground plane' that is actually part of the loop. the ground plane on the other side of the board.
I might question the dimensions, seems a tad short and you can't have ground plane behind the dipole.
Your sticker antenna would be a better design, should well fit in your size requirement. Try those as a template and work from there.
Most of my antennas in that band and just above are J-POLE in nature, single ended, they are worth a couple DB gain.