NFC Performance: It’s All In The Antenna

NFC tags are a frequent target for experimentation, whether simply by using an app on a mobile phone to interrogate or write to tags, by incorporating them in projects by means of an off-the-shelf module, or by designing a project using them from scratch. Yet they’re not always easy to get right, and can often give disappointing results. This article will attempt to demystify what is probably the most likely avenue for an NFC project to have poor performance, the pickup coil antenna in the reader itself.

The tags contain chips that are energised through the RF field that provides enough power for them to start up, at which point they can communicate with a host computer for whatever their purpose is.

"NFC" stands for "Near Field Communication", in which data can be exchanged between physically proximate devices without their being physically connected. Both reader and tag achieve this through an antenna, which takes the form of a flat coil and a capacitor that together make a resonant tuned circuit. The reader sends out pulses of RF which is maintained once an answer is received from a card, and thus communication can be established until the card is out of the reader's range.

For the majority of tags likely to be experimented by Hackaday readers the RF frequency is 13.56 MHz, and the RF emissions are supposed to be in the magnetic field plane rather than the electric field. There's nothing complex about the antennas, indeed it's easy enough to make one yourself by winding a suitable coil and tuning it with a small variable capacitor. The RF properties of the antenna can be explored with instruments as simple as a signal generator and an oscilloscope, or if you’re a radio amateur old enough to have picked one up, a dip meter. For the purposes of this article I’m using a NanoVNA because of its extreme convenience, and I’ve set it to measure SWR on port 1 with a sweep between 10 MHz and 20 MHz. I’m loosely coupling it to the NFC antennas I’m testing by means of an RF pickup coil, one turn of wire about 10mm diameter soldered to a coaxial connector and secured with a bit of glue. When I place the pickup coil over an NFC tag, I’m rewarded with a sharp peak on the VNA from infinity down to near 1:1 SWR. This works well with most reader coils and with lower power NFC tags that simply contain a memory chip, but my VNA doesn't provide enough energy to measure those tags with higher power integrated circuits such as bank cards, a public transport card, or my passport.

Immediately, the VNA pinpoints one of the problems inherent to mass-produced NFCs, that the resonant frequency is rarely exactly on 13.56 MHz. In writing this article I found that both cards and readers appear to resonate anywhere between 13.5 and 15 MHz, with the majority being measured at about 14 MHz. In practice most readers provide more than enough energy so the tag can still be energised despite the resulting inefficiency, but for any NFC tag system to work at maximum efficiency it should have both reader and tag adjusted to resonate at the 13.56MHz frequency of communication.

Most tags, and the cheapest reader modules, have very little effort put in to tuning them to resonance, but one of the more interesting tags I examined for this piece, a bank card subjected to a teardown by a hackerspace friend, shows a very clever approach to automated tuning. A bank card is a standard chip card made from two laminated layers of plastic, with the chip contacts appearing in the front face. Upon dismantling it can be seen that the chip and its contacts are on a small piece of plastic about 10 mm by 10 mm that can be lifted clear of the card.

This module can be read by a card reader, but only when it is placed directly on the antenna rather than with any part of the whole card in proximity to the reader as would happen in a shop. To ensure the small chip module can be energised by a reader over the whole surface of the card, the rear half of the card is a printed circuit board that is simply a tuned circuit with a large coil and an ingenious variable capacitor made from a row of small PCB plates. The coil is half-and-half round the edge of the card and closely round the chip, allowing it to pick up the field over a large area and couple the resulting energy closely into the chip. It's tuned during manufacture by cutting a trace connecting the capacitors, at a guess this will be an automated process. Measuring its resonance it turns out to be a little higher than 13.56 MHz, but since that measurement was made on a dismantled card with no chip in place it's likely that the resonant point will have been moved upwards.

Turning to the readers, the more expensive devices have a built-in variable capacitor and will have been factory-tuned to 13.56 MHz, while the cheap modules normally have a fixed capacitor and resonate at a higher frequency. Experience with these cheaper modules suggests that they will usually interact with the simpler cards such as the ubiquitous MiFare Classic, but that they are unable to provide enough energy to power the smarter cards such as the MiFare DESfire tags. Adjusting the antenna on the module for resonance at 13.56 MHz improves the efficiency to the extent that the higher-power tags can be read, for example in the picture is a cheap reader module prepared by a hackerspace friend. He used an RF pickup coil and an oscilloscope to measure the amplitude of the 13.56 MHz carrier, and adjusted the tuned circuit until a point of maximum amplitude had been reached. In this case he wound his own coil and removed wire from it turn by turn to find the maximum, but the same result could just as easily be done with the PCB coil and a small trimmer capacitor. This cheap reader now works with DESfire cards that previously required a far more expensive module, making the process well worth the effort.

So while much of the technological magic in an NFC tag lies in its digital electronic package it's worth remembering that making it all work is still a firmly analogue antenna. A bit of old-fashioned RF tweaking work with your ‘scope and a signal generator can transform their performance for the better.