A few weeks ago a new device entered the ever-widening bluetooth beacon market on Kickstarter. Unlike the competition, the new iFind from WeTag Inc. promises that it can offer all of the features of a standard bluetooth beacon without ever needing a battery charge or replacement. This bold feature is relatively unexplained leading to a large amount of online skepticism. In an attempt to cut through the controversy, we were lucky enough to conduct a short interview with WeTag’s CTO Paul McArthur. While his response did help us to attach more concrete numbers to the company’s claims, we are still unconvinced that iFind is a viable product.
Like its name implies, the iFind is a device designed to help you find items you have misplaced. Using the associated smartphone app, the device can be remotely triggered to emit a beep which helps the user locate the tag and whatever is attached to it. Similarly, the iFind can be activated by shaking to perform the reverse and help the user find his or her phone.
Unlike competing device Tile, WeTag claims that the iFind beacon utilizes radio frequency energy harvesting to provide power to the coin-sized device giving it a relatively infinite lifespan without ever needing to be charged. Unlike wireless charging systems such as your phone’s Qi case or your electric toothbrush which utilize low frequency near-field inductive charging, true RF energy harvesting plucks stray radio signals out of the air far away from the transmitter and converts them into useable electrical energy.
One of the simplest examples of such a system is also one of the earliest wireless devices of any kind. The Crystal Radio, is a simple radio receiver that requires no battery or external power source. In this device, the energy contained in the radio signals broadcasted from a local station is used to power a very weak filter just enough to drive a small speaker and play music or other radio programs. Because there is so little power in a typical radio signal, these devices suffer from very weak audio output and fell from popularity in the early 1920s when louder and more robust externally powered tube radios hit the market.
Driving a speaker requires a fairly large amount of power, but thanks to modern advancements in extremely low-power integrated circuits, many device designers are finding more practical applications for RF “energy harvesting” that don’t require nearly as much. While you probably won’t get enough power to charge your smartphone, Nokia demonstrated the ability to harvest 5mW of power from ambient signals back in 2009 with hopes of some day keeping a phone in standby indefinitely. Batteryless wireless sensors have also been demonstrated though they too have not yet reached mass production. Unlike your car’s radio which plays music continuously, these devices can sit in standby and collect and store energy over a period of time before they are required to use that energy to transmit or receive a small amount of information.
There are many challenges to be overcome in the burgeoning field of RF harvesting. Devices requiring a lot of power require larger antennas. Smaller handheld devices need very efficient power conversion to turn what little energy they do absorb into a form that can be used. Unlike the radio receiver in your TV, RF harvesters must be designed to pick up multiple stations simultaneously to make the most out of what is available.
Many of these problems have been solved in self-contained systems such as in the familiar RFID tags used to catch thieves at Old Navy or open the door to your office, but the biggest challenge facing ambient RF harvesting is dealing with the fact that many RF signals from familiar wireless devices are not transmitted with the intent of being used as a power source. For example, a properly configured wireless router will only transmit as much power as is necessary to cover its users. Transmitting any harder can cause interference with neighboring networks in crowded environments where the number of networks has surpassed the number of available frequencies. This is one of the reasons the FCC limits the maximum transmit power of these devices (other reasons include general safety though that doesn’t come into play until the signals get very powerful; see “microwave ovens” for more).
Note that when your laptop’s WiFi signal drops to zero bars, its antenna isn’t absorbing enough energy to make sense of the incoming signal even with all of its sophisticated amplification and signal processing. Actually using a signal this weak to power a circuit is an entirely different story.
iF in doubt…
With all of the challenges outlined above, it’s no surprise that the internet met the iFind campaign with skepticism. Hackaday posted a general discussion topic about the viability of such a solution; among the comments were concerns that a bluetooth beacon requires more power than is available to a device as small as the iFind. While a proprietary sensor network may be able to store up energy for hours before transmitting any information, a device adhering to the guidelines of the bluetooth specification cannot wait so long between consecutive transmissions. Other complaints involved the general vagueness of the project regarding its proprietary “power bank” and the secret sauce that allows it to perform as well as claimed.
In an attempt to better understand what the iFind team is trying to do, we reached out to their CTO Paul McArthur with some questions about the project. We focused primarily on technical details which I will cover later, but we found his responses regarding the team’s business strategy to be informative:
By my math, a 27x32mm solar panel can expect to have around 800mW of power incident on it in direct sunlight which is many thousands of times what is required to power a bluetooth beacon. Even a poor solar panel used under man-made light would be able to derive enough power from that to keep a simple BLE beacon running. Why did you choose RF energy harvesting over solar power?
The solar panels are a good source of energy and can be used to recharge the power bank. The main difficulty was in the manufacturing process for this device. We had several different iterations of the housing due to reliability reasons (we can discuss further if you like) and attaching a solar panel only aggravated this. The main intent from the onset of the project was to design the hardware so that it can be modular and use different sources. We had done a lot of research on the past on the EM and thought this was where we would have the best initial results for a product release.
Being able to power a device off of ambient RF harvesting is no small feat and yet you’re using it to power a bluetooth beacon *and* an accelerometer. By my estimates, an accelerometer such as the LIS3DH will draw somewhere on the order of 30mAh a year in a standby “wake-on-shake” state which is not a trivial amount. With so many other devices in the ultra-low-power market (wristwatches, calculators, temperature sensors, etc), why did your team choose a Bluetooth beacon as your first application?
The business model has several different devices in the development cycle, due to the modular design. The beacon market is relatively new and not saturated so it was seen as a good place to build a brand. The accelerometer is there due to the results of three extensive market research studies that we conducted. People rated the reverse search feature as the number three feature, after cost and size. We do not know why but they did.
We often question the motivation behind some of the more bizarre projects we’ve covered, but Paul was able to provide concise and reasonable replies. With the “Why” out of the way, we quickly joined the Hackaday crowd in investigating the How.
It wasn’t until the iFind team published a technical brief on June 1st that the controversy really heated up. While the team had already specified that the device will not work in rural environments where few RF transmitters are used, they never made it clear exactly what the minimum RF requirements were. In Sunday’s technical brief, the team specified that the device is most efficient when input power is around 10dBm which is “typical for home Wifi”.
To address this statement, it’s important to discuss what dBm means. The standard unit for measuring power (energy delivered over time) is the Watt. A 1HP motor outputs 745 watts of power while a 75 Watt light bulb draws…75 Watts. When one measurement is just ten times another, the Watt is a pretty reasonable unit to use. In the world of radio though, power levels can easily vary from 100,000 Watt to 0.0000000001. Because of this, RF power is often measured in dBm which is a decibel or logarithmic scale based around 1 milliwatt (0.001W) of power. On this scale, 10dBm equates 0.01 Watt while 0dBm is 0.001. Every 10 dBm marks a tenfold change in power. In this way, that long number shown above equates to -70dBm or 0.1 nanowatts. While dBm is convenient for dealing with large and small numbers, it can also be very misleading. 40dBm is not twice as much as 20dBm. It’s actually 100 times as much. For this reason, I will be sticking to Watts whenever convenient in this discussion.
Part 15 of the FCC rules limits the amount of power that can be transmitted over the Industrial Scientific and Medical (ISM) radio bands. These are the frequencies that include the 2.4Ghz and 5.8Ghz used by WiFi. The laws prohibit WiFi transmitters from transmitting over 1Watt (30dBm) of power assuming that that power is transmitted isotropically (equally in all directions). The effective power can be increased with directional antennas, but considering that the iFind is meant to be used in mobile environments, we will only be focusing on the ideal isotropic example.
When it comes to receivers, things get a little more complicated. Depending on their shape, their size, and how well they are tuned, receivers can have wildly varying ranges of power received from the same source. A general rule of thumb for the best case scenario is to determine how much RF energy is actually passing through the receiver. This is called flux and it varies with how far the receiver is from the transmitter. As a metaphor, consider a partially inflated balloon covered in dots. Pressing a quarter against the balloon will cover some of the dots. Doing the same after inflating the balloon more will cover up fewer dots as they will have moved farther apart. In this example, the transmitter is at the center of the balloon while the quarter is our receiver. The number of dots covered by the quarter represents the amount of power received. Again, antenna design is a lot more complicated than this and the effective size of a receiver can be larger or smaller than its actual size, but it’s a good place to start.
So the question to be answered here is whether or not 10dBm (0.01W) is a reasonable amount of power for an iFind to capture from conventional Wifi. The dimensions of the iFind are given as 27mm x 32mm or 8.64 square centimeters. To figure out how much power is available to the iFind, we need to calculate the power per square centimeter available and multiply it by 8.64. If a WiFi transmitter is outputting 1W, its power per square centimeter at a particular distance can be calculated by dividing 1W by the surface area of a sphere with that radius. At 20 feet away, this gives us a flux of just over 0.2 microwatts per square centimeter and a maximum possible power received of 1.7 microwatts.
This is a very small amount of power. Roughly a half the power drawn from a standard digital watch which never has to fire up up a power hungry wireless transmitter. Elsewhere in this technical brief was a table outlining exactly the amount of current required for each of the operational modes.
The assumption is that when the device is harvesting more energy than it’s using, it’s storing the extra for later use. In order for the device to operate continuously, it needs to on average harvest at least the same amount of power that it is currently using. Without information on the operating voltage of the circuit, the current measurements don’t tell us much, but I was able to confirm with Paul that the power bank of the device operates at around 700mV and that the one second interval rope mode requires 275 microwatts of power to operate normally.
A lot of people are asking about the minimum power required in “every day conditions”. I’ve made an attempt to calculate the minimum average RF power required to keep the device functioning. Please take a look at tell me if I am wrong. You state that the power bank can power the device for 18 days while it is in 1-second Rope mode which draws an average of 36uA. This means that your power bank can store at least 15.5mAh. At 10dBm (10mW) input with 70% conversion efficiency, it is drawing 7mW of power. If this is enough to charge a 15.5mAh power bank in 1.5 hours (assuming no losses anywhere else), your power bank must be operating at an average of around 700mV. 36uA at 700mV means a continuous power draw of 25uW to keep rope mode working. This means that the minimum input power needed over RF to guarantee at least continuous functionality (without any energy being stored) is -9dBm. -9dBm is 125uW and when captured at 20% efficiency means that 25uW makes it into the device. Is this correct? if not, do you know the minimum RF power required?
The power bank calculations as you have stated are correct. The bank does have an equivalent of around 15.5 mAHr. The charging graph was for a tag that was only enabled in sleep mode. The firmware will not let the tag operate until a predefined level is reached. In operational mode to maintain a rope of 0.036mA, without any draw from the bank needs 110uW (Maximum: you missed some residuals) to run, so 275uW at the input with an RF/ DC efficiency (40%) worst case. That translates to -5.6 dBm at the input.
So while the report was full of claims about charging the device in 1.5 hours at 10mW, the critical number is that the device needs to receive an average of .275mW to operate in one second rope mode. I did a little math to try and see how close to an RF transmitter the iFind would need to be placed to operate in this mode:
You say that the device needs 275uW of power to operate continuously in Rope mode. Assuming you have an antenna of 27x32mm (best case considering the dimensions of your device), you will need a energy flux of 318mW per square meter. Assuming a case where only a single wireless access point is present (similar to your test conditions as you described them) and assuming that this access point has the maximum allowable output of 1000mW with a omnidirectional antenna, to get 318mW per square meter, you would need to stand within approximately 50cm of the antenna. Does this distance match up with your test configuration?
Testing was done from 1cm to 10m measuring the signal strength. This was then correlated to the measured output.
This is a little troubling. The campaign makes it clear that the iFind will not operate when it is a remote location (The Sahara Desert is often given as an example), but the amount of RF required to keep the device functioning is very high. 50cm from a 1000mW transmitter is already asking a lot, but most conventional WiFi routers don’t even output at that level. Even routers that do output that level only do so as required to transmit data. Short burst of data can take just milliseconds, and even when streaming 1080p YouTube videos over a 54Mbps connection, the transmitter is only transmitting an average of 10% of the time. The fact that they bothered to test power received from 1cm away from the transmitter raises even more questions about what they consider to be a typical use case.
Of course, there are other energy sources than a single access point. Their antenna apparently picks up two separate bands of RF. One in the ISM band and another somewhere in the TV/Radio broadcast bands. Anyone who has attempted to pick up a television radio broadcast knows how big the equipment is and how finicky the connection can be. With this in mind, it’s hard to believe that a device as small and as mobile as the iFind could draw power from these sources with any reasonable amount of efficiency. It’s especially vexing considering that unlike the ISM band which has a wavelength of just over a centimeter, broadcast frequencies can have wavelengths several meters in length which makes it harder for small devices to pick up efficiently
Paul clarified for me that the two plots shown on their update page represent the ISM and radio broadcast bands (not sure which is which), and as you can see, they show similar levels of efficiency:
While this level of efficiency is indeed impressive, in order for a 100 kilowatt FM transmitter to provide the 10+ dBm where the iFind is most efficient, the 8.64 square centimeter device would need to be placed well within 100 feet of the transmitter. Just keeping the device alive would require standing within 518 feet of the transmitter. Even factoring in the large number of transmitters spread throughout a home and a city, the numbers just don’t add up.
iFind your lack of faith… appropriate
Paul and his team have been very receptive to my questioning, but try as I might, I still can’t find any way to validate what vague and implied level of performance they appear to be advertising. While they claim that the device will never need to be charged, they do admit that that won’t always be the case, and the inclusion of a battery meter leads one to believe that it’s something that the user must need to at least worry about a little.
Especially troubling is the inability or unwillingness of the team to simply list what a typical use case is. Many backers asked the question “will this work in my house” and didn’t get very good responses:
These are the kinds of questions that any salesperson should be able to answer about their product. It’s one of the first steps of proving product viability. You take a survey of average conditions and you test if it will work. If they cannot with confidence state that the device will work in what is a very reasonable description of a home environment, how exactly do they intend to please their backers with what they deliver?
Furthermore, their only product demonstration is obviously a fake made clear by the lack of bluetooth on the phone they are supposedly connecting to.
Their video also contains a clip of someone hooking up an accelerometer breakout board that can be found on Sparkfun:
The operating voltage of this part is 1.5V higher than their power bank can provide, and it draws more current in sleep mode (3uA) than their entire circuit reportedly requires.
These details lead me to believe that the iFind prototypes shown are non-functional and are instead just form factor placeholders similar to what we saw with TellSpec. The accelerometer choice is still odd though. While it’s possible that it was used temporarily for an earlier prototype, it seems like a waste of time to start development using a part that cannot function in the final version. There’s also the possibility that it was a neat looking off-the-shelf component added to spruce up a video which otherwise would contain absolutely zero content regarding the electrical hardware of the device.
Can they do it?
The team has not demonstrated their product working, though they have promised vaguely that they will submit it to independent testing “in the up coming weeks”:
Submitting a Kickstarter product to an independent testing group prior to the completion of the campaign is very uncommon, but then there is a lot uncommon with how this team has handled their campaign. If the product was currently working, they would have demonstrated that. If it wasn’t working, they probably would not have so freely provided such incriminating information. My personal explanation is that the device is functional but not quite as amazing in practice as many of the backers may have assumed, and the currently functioning prototype is too ugly and bulky to show off.
Based on my discussion with Paul and what little is explained in the marketing campaign, here’s how I think iFind will perform. The device will harvest some amount of power passively over RF. While this will not be enough to charge the device from dead, it will help to extend the 18 day lifespan of the internal battery maybe an extra few days or so. When the battery gets to some predefined low level, the user will be notified through the app and need to take some sort of action. As Paul explained:
The device will not run continuously without some sort of charge, be it residual or a “charge mechanism”. We do not claim that it will run perpetually. Even in sleep mode in the middle of the Sahara it will run out. If need be, the user will be warned and have to decide the solution from instructions we will provide..
While again the definition of “Sahara” is vague, the team is very comfortable admitting that the user will need to take some action. I can see a user needing to place their iFind on top of their wireless router for a few hours or near their phone overnight. This might be okay for your keys, but if you stuck your iFind to the TV remote, I can imagine it being a pain. I guess the real question is whether or not this experience is better than using similar products like Tile. The iFind is still small, cheap, and its lack of charging port and removable battery makes it much more robust mechanically. Still, as a passive device that’s supposed to live on your keys or your pet, it seems like it requires a little more maintenance than many backers will want to provide.
Sure, you never have to pay for a battery, but is that better than Tile which needs a new battery only once a year?