Hey
Anyone know the key difference between ceramic antenna and stamped metal antenna? Both goes under SMD/chip antenna. From what I understand the ceramic ones can be smaller and are easier to manufacture, while the metal ones are a bit better?
“Chip” ceramic antennae generally have poorer range than PCB trace or stamped metal MIFA / IFA designs. This Cypress app note gives a good overview:
2 Likes
With antennas, almost all the time bigger means better. It has less to do with the type of construction, and more to do with the size of the elements. There are many chip antennas that are actually quite large related to center frequency.
Some ceramic antennas are just PIFA type with a high dielectric to get miniaturization. Most stamped metal antennas are of the PIFA topology.
However, most ceramic antennas are really just matched inductive loads. In many cases, I’ll use a ceramic chip antenna at a high frequency (e.g. 3 GHz) and extend a trace from it to get a lower frequency (e.g. 1575 MHz). The circuit for this is just like a monopole trace with inductor and capacitor components for size reduction (usually bandwidth reduction, too), but instead of two-three components it’s one component, the price break is sometimes better, and there’s less time spent fooling around on the VNA.
Here’s a nice page with chip antennas and stamped metal antennas. IMO, Ethertronics (now AVX) makes the best chip antennas in the business, but they are usually not no simple to use as, say Johanson. It’s important to know what is the topology of a chip antenna before you integrate it.
4 Likes
Thanks both of you! This made things a lot more clearer. I just got a full feature ESP32 and that had a stamped metal antenna, while most seem to have PCB trace. Then I fell down the antenna rabbit hole and this ceramic appeared. What I am working on is to build something that controls multiple LED strips from scratch, so will need antennas.
Would also recommend everyone who don’t know much about antennas but are interested to read " AN91445 Antenna Design and RF Layout Guidelines.pdf" in this link. It is very compact and explain most stuff I have been wondering about regarding antennas.
I was looking for a ceramic/chip antenna fo a 2G/NB-IoT/M1 application, they all seems to require huge ground planes (>100mm). I wander how could be possible to integrate 2G/NB-IoT/M1 in a small IoT device obtaining reasonable RF performances…or should I give up the 700-900MHz range?
100mm ground plane is usually recommended for 900 MHz monopole antennas. You can put cleverly-placed slots in the ground plane to reduce the size somewhat, but it’s not going to make 100mm into 50mm without some losses to bandwidth or efficiency.
With some antennas it’s also possible to extend the ground plane with an inductive load (i.e. a long, meandering or spiral trace), which will decrease bandwidth but keep efficiency. The difficulty you have is that you need to support a spectrum that you have no control over.
The reality, though, is that cellular is not a low-power technology, regardless of what the marketers will tell you, so you’ll probably want an 100mm battery anyway. The ground plane can be above the battery.
1 Like
I do agree that a big battery would be needed, bu in this case a 30x60mm battery will be enough. The device has to last 4 months and send data every 3 days. Therefore I was looking for something to be integrated in that space. Would stamped metal antenna require less ground plane or they have the same requirement?
I don’t know exactly which stamped metal antenna you’re looking at, but most likely it will have the same ground plane issues. I recommend using an inductive antenna rather than a capacitive antenna (slot, PIFA), because in my opinion it’s easier to add controlled inductance to a ground plane than controlled capacitance.
You should build the device with the reduced ground plane size you are targeting. See if it makes any difference to your application (it might not be an issue). Then you can experiment with getting some extra efficiency in the low bands. This gets complicated to explain on a forum, but we can revisit it when the time comes.
1 Like
Do you have a sample design for this, maybe just a screen capture of the PCB layout and a couple of notes? I think I understand what you mean, but it would be interesting to see how it comes out in practice.
I agree with @jpnorair’s comments. On a project several years ago I tried several types of chip antennas. I matched them with a pi network using the VNA. It turned out my meandering trace antennas were better.
I made four boards with different antenna trace geometries. I just took a guess at different lengths and different pitches of zig-zagging.
For each of the boards I did these steps:
- I through each on on the VNA and cut away Cu trace until the lowest magnitude reflections were at the desired frequency.
- I matched the modified antenna with a pi network.
- I went to a field with few nearby objects to create multipath reflections, transmitted with a given power and measured RSSI. I did this several times and over several nearby frequencies and averaged the results, to mitigate the effects of multipath.
From the data from those boards, I found some were significantly better than others. It wasn’t the ones I would have guessed. Some sort of FEA simulation might reveal why, but my trial-and-error approach let me to antennas significantly better than chip antennas.
You might think chip antennas are easier, but you have to impedance match them. Even if you follow the recommended layout precisely, it won’t be matched.
The most important thing for any antenna I’ve made work is that the entire length of metal of the antenna and the PCB its on be > 1/2 wavelength. You wan the trace to be >1/4 wavelength, but it’s not necessary. The whole thing being > 1/2 wavelength was necessary. I could impedance match to a rock, which is how I came to think of chip antennas, but I could never make anything work without the whole board + any metal attached to it being > 1/2 wavelength.
2 Likes
Thanks for the tips! The 1/2 wavelength rule of thumb makes sense if you think about it but I hadn’t seen it expressed like this, very valuable. The reason I’m looking at chip is that I’d like to keep everything as small as possible and the traces tend to use quite some space. For miniaturization the chip or helical structures like
are attractive (I’m looking at Wifi 2.4Ghz antennas).
Something I don’t see described much is the sensitivity of the various antennas to their surroundings. I like to make small processor modules that I use in a variety of projects and so the enclosure and what’s around the processor module varies. That makes me want the “most robust” antenna that can tolerate the most crap around it without falling apart more than the antenna with the best max performance…
1 Like
If you have a VNA, then it’s pretty easy to impedance match with a grounded line and a tapped feed. No need for a T-filter or Pi-filter. I’ll keep a series capacitor on the feed to compensate for enclosure parallel capacitance, and that’s it.
I don’t know of a parameter for this. I call it “de-tuning”. I think any antenna needs to be matched for its particular board. I find the same geometry on a different board, even with a chip antenna, has a different impedance.
When I did a body-worn device, I tuned it to be somewhat near by body, so that it was decently tuned sitting by itself or close to the body, while the optimal position would be near the body. I did not do any testing to prove the “happy-medium” tuning is superior to tuning for sitting on the bench or clipped onto the user.
I don’t know exactly the term to describe an antenna’s susceptibility to external near-field affects, but that’s what is happening.
You can minimize this by using an antenna design that has a small near-field. Often that is opposite to performance, though. There are some hacks to workaround this, for example, ground-backed PIFA and microstrip antennas aren’t greatly affected by what’s behind the ground plane. The newer (and more exotic) hack is to use metamaterial resonators to alter the near-field. Or, you can try an antenna that already implements a magnetic dipole metamaterial, such as one of the iMD chip antennas from Ethertronics/AVX.
You can see the basic idea in many of Johanson’s product datasheets. They describe a reference design, but you don’t need to follow it exactly. Often, better performance can be achieved with a slightly different design.
Here’s a board from a bunch of years ago, maybe 2013. It is using a 490 MHz Johanson antenna, but the frequency of the solution is 433. C26 is a series tuning cap. C101 is usually a 0 Ohm jumper, if I recall. This antenna has excellent 50 Ohm tuning, also if I recall, a bandwidth around 20 MHz (5%), and decent efficiency (50%) in its use-case. The device was an iPhone accessory, so the ground plane of the phone gave the antenna a good balanced load.
1 Like
Thanks for taking the time to post this, very illustrative, makes sense.