In Part 1 we went over the significance of non-volatile memory in an RFID tag, and talked about the different forms of radiation resistance. We’re interested in having not just the electronic chip survive exposure to radiation, but the data stored in memory as well. Now here we are in Part 2 where I’ll explain how flash memory works and why it doesn’t like radiation. The Secret Sauce is coming, but you’ll have to wait for Part 3 for that.
So… non-volatile memory. We know from Part 1 that means the stored data hangs around after power is removed, which is a crucial feature for a passive RFID chip. But how does that work? Nearly all RFID chips use a memory technology known as EEPROM (electrically erasable programmable read-only memory), more commonly known as flash memory. To describe how flash memory works I’m going to start with an analogy.
Imagine you’re out in your backyard and you need to leave a message for someone, and all you have on hand is a sleeve of red solo cups (this happens to me all the time). You decide to lay out the cups on your picnic table in a rectangular array, and fill some of the cups with water to represent ones, and leave others empty, representing zeros. That gives you a binary representation of your message, which is exactly how we do it in the RFID chip. Even if you splash some water around so that all the cups are wet, it’s still obvious which ones are nearly full, and which ones are basically empty. That “noise margin” is another benefit of digital technology.
The cups and water analogy is exactly how flash memory works, except that the cups are built using circuit storage elements known as capacitors and the water is replaced by electrons. Using switches, we can put a high voltage across the capacitor thus charging it with electrons. Those switches can also be used to drain the capacitor of its electrons, resulting in an “empty cup”. A comparator circuit is used to effectively measure how many electrons are stored on the capacitor. As with the cups, a great deal of accuracy is not needed here – the comparator only needs to distinguish between mostly full and basically empty. It’s output is binary: zero or one.
In a perfect world, the electrons stored in the capacitors would last forever. But one look at any TV show featuring a Kardashian will tell you the world is not perfect. Capacitors leak. Returning to our cup analogy, if you left the message out on the table for a long time, the water in the cups would eventually evaporate. It might take several months, but eventually your tabletop memory would be storing only zeros, thus corrupting your data.
Now imagine you’ve had great success with your solo cup messaging scheme, so now you want to start leaving even longer messages. You replace the cups with tiny thimbles that are so small they only fit one drop of water. Compared to the couple dozen solo cups, you can fit about one hundred thousand thimbles on your picnic table. Now you’ve got some real data! Of course, discerning between a zero and a one is much trickier and your message only lasts for a couple weeks before the water drops evaporate. On a hot sunny day, your message is only good for an hour or so.
And so it is with flash memory. Capacitors leak for many different reasons, and while the leakage is small enough to allow for decent non-volatile memories to be built, that data is only reliable for 5 years or so. As with the cup and water scenario, leakage is exacerbated by high temperatures, so much so that data stored in flash will only last for a few months at high temperatures. And flash memories have certainly been made denser over the years, allowing for larger amounts of data to fit on the chip, but the result is that those capacitors don’t store nearly as much charge when they are full, so it’s a lot more difficult to tell the difference between a one and a zero. And as with the thimbles, since there’s a smaller number of electrons stored when the capacitor is full, it doesn’t take as long for them to leak off and look like a zero. For RFID chips that need to store large amounts of data for 30 years or more, it’s just not going to happen using flash memory.
Now let’s consider radiation. Those electrons that get stored on the capacitors are atomic particles. Radiation is also made up of atomic particles, but instead of electrons they are alpha, beta or gamma particles. Still, they are particles none the less and they tend to interact with the electron particles and make a mess of things. In our cup and water analogy, radiation can be thought of as a thunderstorm. Some of the rain will fill the cups, turning zeros into ones. The rain is coming down hard enough that it knocks some of the cups over, turning ones into zeros. It generally makes a mess of things. We see a similar effect with flash memory in the presence of radiation – even at small doses stored data gets corrupted. At doses large enough to achieve a sterilization effect, data is completely obliterated.
Armed with an understanding of how flash memory works, hopefully now you can see why in Part 1 I said that most people think a passive RFID chip that can survive radiation is simply impossible. So how does Tego do it? That’s the Secret Sauce and for that you’ll have to wait for Part 3, but here’s a bit of foreshadowing. Flash memory is an electronic memory, meaning it is built out of components that never change, but the electrons that flow through those components cause various behaviors. Instead of an electronic memory, the Tego chips use a mechanical memory. That’s right, the electronic chip that communicates with the reader uses a mechanical memory.