It's important to understand the difference between UHF RFID (this technology) vs. others like BLE.
UHF RFID tags harvest all of their operating power from a reader's transmitted RF signal. Multi-access is handled at the protocol layer via probabilistic time-division multiplexing. The tag doesn't actually "transmit" -- rather, it is a cooperative radar target that reflects RF power back to the reader (sort of like an SOS mirror) by intelligently, electronically changing its radar cross section. The tags contain a single IC, and the antennas (for commercial ID-only tags) are roll-to-roll printed... to the tune of Billions of tags per year. The net effect: Tags cost around $0.10 each in volume. The read range is practically limited to around 20 feet, as dictated by FCC limits on the reader's RF output power (1Watt). Nominally, gen2 UHF RFID tags operate at bitrates up to ~500kbps, but there are some specialized tags that can transmit at 100Mbps using just a few microwatts (we used them to transmit dragonfly neural signals during flight!).
Meanwhile, BLE technologies can do push notifications, mesh networking, and the like. The beacon (tag) may contain a full TX-RX transceiver. However, due to the tight spectral requirements of their transmissions, they generally require an external (off-IC) crystal oscillator. The net effect is that BLE beacons are more costly -- on the order of $2-3 ea. This means that they are much less likely to be used to tag everyday items (like a carton of milk) or deployed in the billions for outdoor monitoring.
Source: I'm an expert on these systems as applied to robotics, healthcare, and remote sensing [1,2,3,4]. I'm giving a talk on them next week at IEEE RFID / RFID Live. Oh, and you can also buy WISP-like tags from a company called Farsens.
Thank you for the Farsens link! Really interesting RFID tags.
Out of curiosity, do you know any good sources for programmable, or better yet, open source UHF RFID readers? I've gone through lots of options but none seem quite flexible enough.
The two most-common commercial options are Impinj and ThingMagic readers. Neither are open-source (hardware or firmware) -- they just have APIs. If you go back to the old ThingMagic M5e (rather than newer M6e), there are high-level open source libraries that I wrote:
You can buy single-chip readers from AMS (eg. on DigiKey) and build your own reader -- not a light undertaking due to RF PCB layouts, but manageable. I've also heard of people using the USRP (GNU Radio) to build gen2-compliant UHF RFID readers, but I believe they require the PCI-bus variant to adhere to protocol timing standards.
Fantastic! I appreciate the heads-up. The open source ThingMagic drivers would be very handy.
It has crossed my mind to try integrating a single-chip reader but I'm unsure of what the FCC requirements would be for selling it as part of a commercial product. It's a little surprising that there hasn't been a kickstarter type of project to build an open source reader chip and get it FCC spec'ed.
The AMS chip itself is probably FCC compliant, since the output power is so low (10-13dBM, IIRC). The problem arises when you add an external PA to get the full 30dBm output power permitted by the FCC; then you need to re-certify.
When I clicked through, what I was hoping for was an e-ink tablet without the battery, where you have to hold it close to an inductive charging surface each time you flip the page.
Such a thing sounds stupid at first blush, but in combination with ubiquitous inductive charging mats (imagine them on e.g. every desk in an office), it could make for extremely cheap, thin and light e-ink devices—to the point that they start being more like documents than readers: flimsy laminated 'paper' sheets that happen to be able to update themselves.
Heck, if you didn't expect to use them outside of the installation (if they were for displaying internal manifests in a warehouse, or patient charts in a hospital, etc.) then you could externalize all the "document reader" electronics, and just make the device into a dumb framebuffer that hardware in the mat can push images to. (Though you might not need to do this; we can fit some pretty complex logic inside e.g. a credit card, so we might be able to fit all the storage+firmware chips for a full e-reader inside a piece of paper without that making it less piece-of-paper-y.)
But I want my house to listen and sense me. The advantages far outweigh worries. Besides what makes Big Brother isn't the existence of surveillance it is the lack of control over it, who uses it, and how. So fight for open source, transparency, end of advertising/marketing. Not against technology.
I want my house to know what room I'm in (for lights, environment, music following, to adjust temperture when I'm active or too hot, to answer alexa like queries, to announce alarms reminders, to tell me the kitchen is hot and smokey cause I forgot beans on stove, to give me alert someone pulled into my drive, stopped in front of my house, to call 911 if my heart stops.
Is there any practical limit to how much energy a device like this can buffer before springing into action? I'm just curious if it would be possible to quickly leech enough to power a peripheral like a solenoid, or at least ensure enough energy is available to complete whatever computation the device is intended to perform
(Vaguely imagining a completely unpowered electronic lock, and a handheld 'key' that I guess would contain a fairly beefy battery)
> Vaguely imagining a completely unpowered electronic lock, and a handheld 'key' that I guess would contain a fairly beefy battery
That's a good idea, actually. I dunno about this technology specifically, but there are wireless power standards that can transmit more power, e.g. Qi can provide 5W at a up to 4cm according to Wikipedia. That should be plenty suitable for a lock I'd think.
It'd be even cooler if cell phones could transmit power wirelessly, but AFAIK phones with inductive charging hardware can only receive. If they added the ability to transmit, suddenly your phone's battery could become a sort of universal power source for talking to various passive devices like locks and stuff.
Though, this is slightly different from what's described in the article. If I'm understanding it correctly, this thing is harvesting its power from a standard RFID signal, so you don't need any special power transmission hardware. No way would you be able to get enough power out of it to power any kind of mechanical actuator for a lock.
True, but an electronic lock has lots of advantages over a mechanical one. Especially for a business, you can trivially grant and revoke keys as well as track access. I feel like in principle they could be more secure too as all but the fanciest of locks are pretty trivial to pick, though that's predicated on somebody building an electronic lock without hardware/software/cryptographic flaws (good luck).
We looked into this for a while for some of our products and even filed some ip (such a waste of time). We were doing it with thermopiles on our vents and using a buck and boost. It worked for heating well, maybe well enough for cooling, but was price prohibitive (would up retail 50% at least). https://flair.co
The linked article claims the current commercially available batteries have been stagnant, not the research, which is quite active but is only expected to bear fruit in the medium term.
I'd rather have one of these that has a tiny, lithium-ion battery that it barely uses. Makes much more sense with the same applications. When done, you either plug it into a charger or (more homebrew) take the battery out to plug it into a charger. Regardless, I see a tiny battery getting better results than sacrifices made to be "powerless."
Now, what's the smallest PC you all know that can run for a long time and with some useful work on a tiny, lithium-ion battery? That would be interesting.
I was going to ask that, but for me it doesn't have to be a PC. We work on pretty low power electronics at work, but it is very hard to figure out what the best combination for practical hobby electronics would be. So like what kind of MCU, radio and display would consume the lowest power while still practical enough for me to have fun with. That would mean the MCU has to be 10+ mhz and the display needs to be at least file of a smaller smartphone. eInk would work.
It could be combined right? Very tiny low power lithium ion with inductive loading in some way that is efficient and easy enough.
Forget fitness taker, smart watch and tiny little things. I'm thinking thin smartphones. If one of these can be powered by multiple RFID signal sources. Do the math for required power signal strength and get working around the limitations.
That will not work out. Even ignoring the elephant in the room that is powering the display, a smartphone communicates wirelessly over hundreds of meters.
You will never get that much power from RFID readers in any practical setup (if you're McGyver, a setup with thousands of readers, if not way more, and a few square meters of antenna might 'work', for a loose definition of the word)
Wow, they could not use a different acronym from WISP? I guess wireless ISPs are loosing business to Fiber etc, but still. From a quick google search, it would be better to call it WSIP.
UHF RFID tags harvest all of their operating power from a reader's transmitted RF signal. Multi-access is handled at the protocol layer via probabilistic time-division multiplexing. The tag doesn't actually "transmit" -- rather, it is a cooperative radar target that reflects RF power back to the reader (sort of like an SOS mirror) by intelligently, electronically changing its radar cross section. The tags contain a single IC, and the antennas (for commercial ID-only tags) are roll-to-roll printed... to the tune of Billions of tags per year. The net effect: Tags cost around $0.10 each in volume. The read range is practically limited to around 20 feet, as dictated by FCC limits on the reader's RF output power (1Watt). Nominally, gen2 UHF RFID tags operate at bitrates up to ~500kbps, but there are some specialized tags that can transmit at 100Mbps using just a few microwatts (we used them to transmit dragonfly neural signals during flight!).
Meanwhile, BLE technologies can do push notifications, mesh networking, and the like. The beacon (tag) may contain a full TX-RX transceiver. However, due to the tight spectral requirements of their transmissions, they generally require an external (off-IC) crystal oscillator. The net effect is that BLE beacons are more costly -- on the order of $2-3 ea. This means that they are much less likely to be used to tag everyday items (like a carton of milk) or deployed in the billions for outdoor monitoring.
Source: I'm an expert on these systems as applied to robotics, healthcare, and remote sensing [1,2,3,4]. I'm giving a talk on them next week at IEEE RFID / RFID Live. Oh, and you can also buy WISP-like tags from a company called Farsens.
[1] http://www.travisdeyle.com
[2] http://www.travisdeyle.com/research.html
[3] http://www.hizook.com/blog/2015/08/10/mobile-robots-and-long...
[4] http://www.wired.com/2013/06/dragonfly-backpack-neuron/