Jon is one of my favorite historians / writers in the semiconductor space! His videos are great but they're just recaps of his weekly substack (which is also excellent) asianometry.substack.com
I've worked with EPC RFID tags professionally - those are the tags that can be read at a distance of a few meters, and that can (sometimes) be brought for a few cents. It's an interesting technology, but it's never going to replace barcodes en masse, in the retail space.
One major issue is the read rate - if you're getting a 99% read rate your RFID setup is uncommonly good, but in retail failing to charge for 1% of items isn't sustainable.
When RFID tags are used for marathon timing, they'll often give each runner two tags, and set up two or three reader portals at the start/finish line, because using a single tag and a single reader portal just doesn't give a good enough read rate.
The technology is also exceptionally difficult to debug - the signal from the reader is orders of magnitude more powerful than the reply from the tags, and they both transmit at the same time. So even if you've got professional wireless engineers with $50k spectrum analysers, they'll be able to tell you the reader is operating, but won't be able to measure the strength of the tag's responses.
There are of course an endless stream of things to fiddle with - different reader antennas, multiple reader antennas, different tag antenna shapes, different tag types for sticking to different materials, tags that work better in certain orientations, tags that are designed for use on metal, tags that are designed for use on metal but with a 2mm foam pad for backing, and so on. Of course the famous two cent RFID tags are the very most basic ones.
And you can get crossover between reader portals - the readers are of course very sensitive to the returned signal, so if you have two portals side by side the left portal can power a tag passing through it, and the response can be strong enough for left and right portals to both detect it. You can imagine how this would be a problem for supermarket checkouts.
Not to mention the fact a two cent RFID tag costs too much to be put onto a 30 cent can of beans, which might only have a three cent profit margin.
The real benefit of RFID tags is in being able to track high-value/high-markup items like razor blades throughout the entire supply chain. When the razor blade factory ships out X packs in Y boxes on Z pallets, you can count every packet out of the factory and into the distribution hub, then out of the distribution hub and so on, ensuring none go missing. Read failures are less of a problem here, because if a pallet shows up as missing an item you can stop it and scan it again. The tags are also used in things like hotel towels which are constantly being sent between the hotel and the laundry provider, to make sure shipments contain the number of towels they're supposed to contain.
I recall the example used in many white papers being the scanning of a pallet of soda cans inside of an open truck or shipping container. All of the multi-path interference would've made it one of the hardest environments I can image for getting reliable reads.
With some care, you can see the tag replies with a 10eur rtl-sdr, but the rest if the software stack does not exist for free.
CISC will sell you a license to rent the software, but you need to provide your own rtl-sdr stick.
Voyantic (now Impinj) has their own software that used an Airspy sdr, but it was end-of-lifed after Impinj bought them.
And then for just cimparing tag performance, there is Voyantics Tagformance tester, but that's for comparing tag performance, not debugging something.
But yeah, really annoying to see anything on a nornal swept spectrum analyzer.
Maybe this is a dumb question but why does it require a 50k$ spectrum analyzer to see the returned signal from the tag? The frequencies involved aren't that high (<1 GHz) and I don't get the impression that the symbol rate is particularly high. Is it below the noise floor for low cost SAs or something?
You can solve the crossover problem by energizing tags using different frequency bands. Some RFID manufacturers use this strategy to eliminate crosstalk between readers in adjacent vehicle lanes.
Active RFID tags do exist, but they're basically just a wireless transponder with a battery. Satellite tracking probably exists but it would need a fair amount of power and a non-noisy RF environment.
My favorite thing about GPS is all satellites broadcast on the same carrier frequencies and the received signal at the surface of the earth is several dB below the noise floor. Which you can do with complicated signal encoding involving known values at the receiver and a really slow overall bitrate on top of a much faster carrier.
There's more to it: they increased the overall accuracy (and I guess implicitly the bitrate?) over time much further than was originally intended, by more clever electronics and math on the receiver.
