Today Sigfox is releasing a really good document about antenna design. This pretty document can be found following this link (available on July 7th). This document from TI also contains really well detailed information about the alchemy of antennas and the different kind of product you can find. From Silicon Labs you also have an interesting development kit for 868 antennas with performance published in this application note.

This is to say antenna is one of the most important part of a 868 object, as for Sigfox as for LoRa where the problem is basically the same. I discussed with different sigfox object designed and every-time the antenna was a big project issue. Some has payed about 10-15k€ to get a custom optimized design.

I never experiment a custom design but I made test with various antennas and the purpose of this post will be to give you my feedback on the different solutions.

Antenna is critical as it can give you gain or loss to the initial power you send from the transceiver. Power is given in dB and on the 868MHz band you are allowed to emit with +14dB of power. The SigFox / LoRa sensitivity is about -130dB / -142dB. It gives you a budget link about 140-156dB.

Regarding this you could say … I’m not regarding 5 or 10 dB but you do not have to forget the dB scale is not linear : power is doubling every +3dB … For this reason at least 1dB of loss on the emitter side is crucial for your application.

That said, I’m also the first to say : don’t be stupid, if you do not have enough money to get the custom antenna you need to save all your dB and make a U0 class device, if your application is external and bigcity focussed … save your money and go to market with a U1 even U2 certification, there is no big risk.

On the other hand after making different test, I know you can obtain really acceptable result with standard products and eventually by adding some transmission power from the transceiver to balance the antenna loss. For making this the new version of the TD1208 (The TD1208R) allow a +16dB transmission. It have a large impact on power consumption (+10mA) but it can save 2dB and make your device working better for a low price impact.

So … back to the antennas … as the best way is to use a performant antenna to preserve energy and ensure the best radio performance.

In my different project I used different antennas, some good, some really bad. I’ll not comment the really bad as we can always tell it was caused by my global design… Because what you have to understand is antenna is a point but the box you use, the place you put the object have a large impact on radio performance. Any antenna have to be adapted function of the box (also location of the object). This tuning request expensive equipment and specific skills to use it. Making a good tuning can save most of the DB and is a really important phase in the design.

One of the first antenna I used was a 1/4 wave big SMA antenna. You can find many standard on Mouser an other component providers :

The one you see here is a Connit version, it looks like the standard one but this one has been tuned to be centered on the SigFox center frequency. This is an advantages in the sigfox design (compared to LoRa) : the center frequency is fixed and the bandwidth is really thin so you can design the antenna specifically for the network. LoRa center frequency depends on operator choice and vary from a place to another the antenna performance will vary and will have to be as large as possible.

This antenna is a reference for me for Sigfox design and a way for me to compare antennas is to measure RSSI from the different basestation with this antenna and compare with RSSI obtained in the same period with the tested antenna.

This antenna have a big problem : it is really big and do not fit all the object you want to create.

For this reason I tested many antennas to find a small and efficient one. One of the best I tested was the Anaren antenna :

Anaren 868 antenna

This antenna is a simple net and it has been really efficient until you twist it. When twisted you quickly loss 10dB or more but when keep straight I obtained really good result for a good price.

In most of my design it was complex to keep this antenna straight especially in small form factor design. I took a look to more compact and solid antenna. From Taoglas I found an interesting antenna :

Taoglas PC81 antenna 

It is a PCB antenna where the antenna part is about 1cm square. The form factor is compact and a 3M glue makes it easy to attach on a small box. The performance was not really good in the different test I make. I keep it on some external application where I dod not had to complain but the loss was about 10dB.

This make me to look at ceramic antennas, I’ve tested some (really small) but really not working well losing more than 10dB. I’m actually using the Vishay A01 in different design where I’ve good a good radio performance.

Vishay A2 antenna

This antenna is small, easy to solder (reflow still mandatory), not expensive and gives good results. Its compact size allow to use it in small objects.

I also experiment some more curious design : from the sigfox hacker community I received this surprising antenna :

custom antenna

I was not really confident but in my test, compared to the reference antenna (connit) it had a loss limited to 6dB with no tuning circuit. No so bad.

Now you have an idea on how antenna science is a mystery and you see it can take many different form factor, form factor is not a good indicator of antenna quality and you know object packaging have a large impact on performance … You can read the SigFox really good white-paper about antenna design : even if you will not build the perfect antenna you will have a better understanding of how it works and what to not miss to make a well designed object.