UWB: The Tech Behind Apple AirTags

Apple, earlier this week finally launched AirTags after a lot of speculation over the last couple of years. For those of you who don’t know, these are small discs that you attach to your belongings and “Tag” them digitally. It’s designed to help you find these items in case you misplace or lose them. My interest in the AirTag is because of a cool tech called Ultra Wide Band(UWB) which enables all the magic of finding these tags in an indoor environment precisely and with a very high degree of accuracy. I wanted to take you through the details of UWB tech, how it works and how Apple may have found a killer product in this small device. 

AirTags are meant to be attached to items that you consider valuable. Starting with the specs, it’s a small 32mm(1.3″) diameter circular disc that weighs in at 11g. It contains a 3V CR2032 battery which is replaceable(Replaceable batteries are welcome change these days in the tech industry!) and a built-in speaker which you can ping from your phone to make it beep to let you know where it is.  It contains an accelerometer to sense motion. The core tech consists of 3 RF technologies built into this small form factor. Those are Bluetooth Low Energy(BLE) for proximity finding of the tag, Near-Field Communication(NFC) for tapping the phone to get the owners info from the tag in case it’s lost and an Ultra Wide Band (UWB) mode for precision location finding which is made possible by Apple’s new custom U1 chip. Since BLE and NFC are techs that are widely discussed, today I would like to focus on UWB.

UWB Technology

The UWB refers to radio signals which have relatively high bandwidth. US Federal Communication Commission aka FCC classifies signals with bandwidth greater than 500MHz or have a fractional bandwidth (i.e) Bandwidth/Centre frequency) of 20% as UWB. Whereas EU classifies any device as UWB if it exceeds a bandwidth of 50MHz. The important takeaway is that UWB signals have significantly higher bandwidth than most narrowband signals used in communication. The frequency range allotted for UWBs by the FCC is between 3.1GHz to 10.6GHz.

Although it has a higher bandwidth, the permissible power transmission limit allowed by FCC part 15 over this wide frequency range is only -41.3dBm/MHz as shown in the figure below. The power needed is considerably lower compared with other communication signals of WiFi, Bluetooth or GPS. This low power feature is what is allowing AirTag to run up to a year with a small battery.

Emitted 
Signal 
Power 
-41 dBNMHz 
1.6 1.9 
2.4 
802.11b 
Cordless Phones 
Microwave Owns 
3.1 
5 
802.110 
"Part 15 Limit" 
UWB 
Spectrum 
IC6 
Frequency (GHz)
Frequency Spectrum of UWB along with other signals.
Image Courtesy: S. Schwartz and J. Bobier

Power limits of the UWB signals in the other ranges are also very low as shown in the frequency table below so that it doesn’t interfere with any other signals.

Frequency Range (MHz)Radiated Power Limit (dBm)
960 – 1610 -75.3
1610 – 1990 -53.3
1990 – 3100 -51.3
3100 – 10600 -41.3
Above 10600 -51.3

So far we know that UWB signals have high bandwidth and low transmission power. It transmits information in terms of extremely narrow pulses(Less than 2ns) in a defined time interval. It’s an encoded communication taking 32 and 128 pulses to encode a single bit of information. As pulse width is very narrow, it is a high-speed form of communication with an upper limit on data rate at 27Mbps. This is much higher than other close range transfers of Bluetooth which maxes out around 2-3Mbps in the best-case scenarios. So UWBs can be used as a reliable technique for high-speed data transfer with a low power requirement(Think Mobile to Mobile File transfers). But low power essentially means that the range of these signals is not that large.

UWB systems are utilised for localisation. For ranging it utilises the time of flight of signals. Assume there is a UWB tag in a room with UWB sources placed at 3-4 points on the room. Now assuming all the UWB sources are time synchronised with one another and it pings the UWB tag. Based on the response from the tag, you will be able to pinpoint the location of the tag with a very high degree of accuracy with the time of flight calculations.  I am not going to go into detail on the math/algorithm side of estimating position. Please refer here if you want to have long read on the different ranging types and how the algorithm works. Also, check out US Patent No US10171129 from Apple on how pulse shape information can be used by UWB receivers to improve range accuracies.

The AirTag- iPhone Ecosystem

The AirTags are supposed to be used exclusively with Apple iPhones. From iPhone 11, they have added the same U1 chip present in the AirTag on to iPhones. Courtesy of the awesome iPhone 12 teardown pics from iFixit we can see the USI Module(marked in Blue) which contains the U1 chip on the mainboard along with the A14 bionic iPhone Processor (marked in red)

iPhone 12 teardown

From the TechInsights teardown of the iPhone 11 Pro, you can see the internals of the module and U1 chip. It uses a 16nm FinFET Semiconductor technology fabricated by TSMC. I couldn’t find many details on the Qorvo chip on there. It doesn’t show up on the Qorvo product listing, must have been a custom RF IC job for Apple.

Tech 
- ns 一 gh 
80
Internals of the UWB Module containing the U1 chip on Apple iPhone 11

The rest of the details of the U1 chip can be found from the FCC filings of iPhone 12 and AirTags. (PS: These documents are freely accessible on FCC website and it’s one of my favourite goto places to reverse engineer and learn the internals of products)

Since I don’t have my hands on an AirTag yet and I couldn’t find any teardowns of it on the internet this early(I will update if I do get pics later on). The rest of the things mentioned below are my deductions based on FCC filings and common sense. Internal pics of AirTags are not available on FCC yet as companies usually give a request to FCC to keep certain files confidential for 180days from when you get approved. After 180 days, FCC mandates that these be made public.

U1 chip handles the UWB communication for Apple. If you go through the filed RF testing report, you can see that the U1 chip uses two-channel frequencies at 6.5GHz and 8GHz with 4 signal configurations available for each channel setting. There is an integral patch antenna with -1.6dBi Gain @6.5GHz and -0.6dBi @8GHz. Apart from this, I am sure there will be a Bluetooth antenna and a passive NFC antenna on there. (PS: If you read through the report in detail, you will understand the FCC testing steps done to get an FCC ID for AirTags. Testing is done till 40GHz to see if there are unintended radiations till those frequencies)

Going through the internal iPhone teardown photos(this was released only a couple of weeks ago on FCC) Part 1 and 2, you can clearly see all the marked sections of the chipsets and the antennas used for UWB. I have shared snapshots below for iPhone 12 and iPhone 11 UWB antennas from FCC documents. You can clearly see that there are 3 UWB antennas in the iPhone.  You will see the same antenna cut-outs in the metal back for iPhone 11 to accommodate the antennas in the Xray view.

How I think Precision finding works in the AirTag-iPhone combo is that the iPhone acts as a UWB source and it receives info from the AirTags via these three antennas. The difference in time of the signal and the angle of arrival from the 3 antennas positioned at 3 different spots is what is making the precision location finding possible to an accuracy of few centimetres.

A2410, A2411, A2412, A2413 
UWBO, UWBI, ı_JW33, ANT8/UVVB2, 
ANT6 antenna fed- 
Front 
iPhone 12 
A2160, A2216, A2217 
I_JWB LJO, U 1, 03 antenna feed- Front 
iPhonc I I
iPhone Multiple UWB Antennas
X-Ray view of the back panel of an iPhone with slot cutouts in metal for UWB antennas

U1 chips seem to be able to communicate with other U1 chips only. I hope Apple opens this up to be compatible with other UWB chips as well. Also, it seems UWB is not open for the developer community to develop 3rd party solutions. Documentation for Nearby Interaction seems to have come up though. I suppose it will be opened up soon.

Application Use cases

Now that the tech behind UWB is clear. Where can these low power high bandwidth systems be used? Use cases seem to be plenty for UWB  in asset tracking, social distancing, payments, hands-free access control, presence-based detection systems etc. The key application is in the person/object localisation space. This opens up a plethora of use case scenarios. So much so that Apple has gone on patent many of them like Automated Access control(Patent No. US10285013, US20190135229A1) for cars and any doors. Imagine walking up to any door and it automatically opens up with UWB in your phone. There are many automotive players working on this solution. (And yes this solution is possible currently via BLE too!).

Imagine you are in a large store and you use UWB to navigate indoor to a predesignated item in a store which you found (Think about massive stores like Ikea, Walmart) online. Since UWB is fast, it’s currently used in NFL on shoulder pads of players to help in real-time fast position updates to your screen and it also helps automatic camera track the players.  UWBs are going to play an important role in the future IoT tracking device space like AirTags or even in local location-based advertising as you walk by a shop. I am sure in a year’s time, you will find AR-based apps on UWB wherein you will just open up your phone camera on a scene and the item you are looking for will be shown on your screen. (Good for messy homes).

UWB vs Bluetooth

There are other competitors like the Tile who have been around for quite a while now. They use BLE for tagging and finding parts, but the accuracy will nowhere be closer to what UWB provides. BLE 4 and 5 just uses the strength of the signal(RSSI) to approximate distances. UWB has high bandwidth(>500MHz) compared to BLE(2MHz) and this improves UWB’s multipath performance significantly. Assume a closed room, with multiple UWB transceivers, multipath interference means that a receiver will get signals directly from the transmitter as well as signals which got bounced off from the walls of the room. The way to solve it is to use multiple frequencies for transmission so that chances of multipath fading at both these frequencies at the same time and position is very small. UWB’s large frequency bandwidth just fits perfectly in this use case over Bluetooth.

UWB vs Bluetooth 5.1

To be honest, this is where I am really not sure which is better. The latest Bluetooth 5.1 spec brings in the direction-finding which is based on Angle of Arrival(AoA) and Angle of Departure(AoD) of received signals similar to UWB. With a BLE antenna array, you will be able to pinpoint the location of an object. How accurate these are can only be determined when you try both these out in real life. There haven’t been any proven publication demonstrating the accuracy of direction finding by BLE 5.1. I am assuming UWB holds still a slight edge over BLE in multipath fading due to its higher bandwidth. I think accuracy will be slightly higher, but time will tell.

UWB might win out on its fight with BLE 5.1 just because of the fact that Apple has decided to push UWB in its phones. As per the usual trend, the rest of the industry will follow Apple with no questions asked. Both iPhone 11 and 12 and the latest Apple watch contains the UWB enabled U1 chip so it’s clear where Apple is putting its money.

AirTag vs Tile

Ob\ed in Chine

Although I like Tile as a product, I am sure Tile would need to reinvent itself as Apple has come into the space with the same price points. Tile has filed an AntiTrust Lawsuit against Apple. Tile just would not be able to compete with the iPhone ecosystem as a tracker app. There will be far more people using iPhones than folks who would have Tile App installed on their phones. The tracking of items works better if there are more people using the tech anyway. Tile would probably have to move away from their BLE tech too as UWB is gaining traction.

Conclusion

As an Android user, I would personally love it if Apple opens up the U1 chip to communicate with other UWB chips. Decawave(Bought by Qorvo) and NXP are the major other players making UWB ICs. The NXP chips are the ones used in the latest Samsung Galaxy Note20 Ultra to give it UWB capabilities. But Samsung is using UWB for faster file sharing and as a smart door access card for now. The good thing about this feature is that Samsung will support other Android phones with UWB in the future. So that’s a great step in the right direction. Samsung also has its Smart Tags lineup which is similar to Tile’s using BLE. Given they have UWB chips now on their flagship devices they may also soon switch to UWB for tags as well.
Post Edit: One of my readers mentioned that Samsung is planning to sell SmartTags+ which is BLE + UWB tag very soon.

In terms of standardisation for the UWB tech, there seem to be two consortiums that have been created. UWB Alliance and FiRa Consortium. Both groups have powerful members with UWB Alliance taking members mainly from automotive space like Hyundai and Kia, Bosch and Analog Devices, whereas Fira Consortium is consumer tech-heavy with Apple, Samsung, NXP and Qualcomm. Last year these two groups came together on a joint agreement(which is nice for the tech industry as such) in which the UWB Alliance will focus on matters arising from the promotion and safeguarding of UWB technology and advancing updates of the UWB rules in the United States and the EU. The FiRa Consortium will focus on UWB use cases specification, IEEE 802.15.4z interoperability and certification.

To conclude, UWB tech seems to hold so much promise and the future looks bright for it especially when Apple to pushing it along. AirTags for me will solidify the UWB adoption market for the future. UWB tech on Phones will soon be a norm for most flagship devices. Also, a fun fact to sign off on, Apple AirTag is their cheapest gadget in terms of price yet. Cheaper even than their silicone watch straps. Now let that sink in. 🙂 

Cheers 😀

If you liked the post, Share it with your friends!

Christmas LED Light Teardown

Background

Right from childhood, I was often fascinated by those flashy LED lights which everyone used to put up on Christmas, Diwali or in festive celebrations in my town during Onam. I always used to wonder how these LED strings used to work. “How were they able to get it to sequentially blink”? I tried to search online, but couldn’t really find a schematic for it and whatever I could find, had a wrong schematic.

Last week, something reminded me of these lights and I had to figure it out once and for all as to how these worked and why they were this cheap(I could buy long LED strings with different programmed lighting pattern for less than a dollar).

Teardown

I went down to the supermarket and got a string of 28 LEDs(Multicolor), they were on a discount sale for INR 50(0.7$).

It plugs into your normal 240V or 120V AC power plug in your home. It has 8 pre-programmed modes which can be cycled by a push button.

The build quality, as you would expect, is not that great, but for less than a dollar, that’s all you can hope for. It’s a dangerous circuit with no protection from the AC main lines. Be extremely careful trying to play around with these circuits while it’s turned ON.(I got a couple of mildly nasty shocks myself) This device, is as cheap as it can get. Just think about the manufacturing costs of these devices, if they are retailing for a dollar.

It says CE certified, but I really dont think so…

It has 28, 3mm diameter LEDs and contains red, blue, white, green LEDs stringed together in a sequence(More on that, a bit later on).

Right, let’s take it apart and get down to the electronics. It consists of a single side PCB with bare minimum components.

It contains a COB or Chip-on-Board(Probably a cheap micro/ or an ASIC).

Most of you would have seen this as a black blob(like the picture above) in cheap electronic devices like calculators and toys. That black stuff is epoxy resin. These are similar to your normal ICs devoid of the external package. It only consists of the raw semiconductor die. The semiconductor die is soldered by an automated machine to the PCB via extremely thin wires. This process is called Wire-Bonding. A blob of black epoxy is poured on top of this wirebond to protect it. If you want to learn more about this, checkout Sparkfun’s blog post on how chip-on-board devices are made from a factory in China. It’s a fascinating read.

Schematic

I did some reverse engineering by tracing out the circuit and following was the schematic I was able to come up with.

As you can see, it uses a couple of 3-pin SCRs(Silicon Controlled Rectifiers). SCR part number is PCR406(Datasheet). SCRs are usually used as a switch in AC circuits. For those of you who don’t know what an SCR is or have forgotten about from your engineering device classes read on, I will very briefly explain its working. For more detailed explanation checkout this link.

SCR is an unidirectional, four layer p-n-p-n device
as shown below. It consists of an Anode, Cathode and a Gate.

The gate controls the follow of current between the Anode and Cathode. There are different ways of turning ON a gate, we are focused on particular mode, wherein there is a positive voltage being applied to the gate with respect to cathode. When that happens, that junction is forward biased and there is a current flow. This current is called as the Latching current(6mA for PCR406 from its datasheet). Once the gate is turned ON, if there is a positive voltage at the Anode side, a current can actually flow through the SCR. Now even if the Gate pulse is turned OFF, current continues to flow till the point when the current through the SCR drops below a value called as Holding current(5mA for PCR406 from its datasheet). If you are still confused as to how it works, check out this video.

Now, let’s go over the rest of the circuit. There are 28 LEDs split into 2 arms of 14 LEDs each(LAx and LBx). First 4-5 LEDs on each arm have its own current limiting resistor of value 5.1kΩ(Resistor is soldered on to the leg of the LED and heat-shrinked with a sleeve). Each arm is controlled by an SCR whose gates(G1 and G2) are driven by the Chip on Board. During the positive cycle of the AC waveform, a positive voltage pulse from the COB(once the latching current is reached) turns ON the SCR and a current flows in a particular arm. Now the current continues to flow in this arm till the input 230V AC waveform moves to the negative half and the SCR will be turned OFF. The timing triggers of the SCRs from the COB are during the positive half of the AC waveform as it conducts only in the positive half cycle.

Now, how does the LED strip give out all these crazy blinking patterns if they are connected only as 2 arms? There lies the beauty of this design, the LEDs are not wired one after the other in a single arm in appearance although the schematic-wise it’s like that. Meaning, the two arms are intertwined with each other visually.
Following the convention in the schematic above, LEDs are wired as follows,

LA1 –> LB1 –> LA2–> LB2–> …. LA14–> LB14

(Please note that the two arms are not connected in series, it’s just that the final arrangement is intertwined so that when one arm is active(when the gate is turned ON), it looks like just the alternate LEDs are lit up, although, say, LA1, LA2, LA3, … LA14 are all ON at the same time)

Now control the timing of turning ON section of these SCRs with a controller and you can get those fancy LED sequences along with Fade effect (depending on which point in the positive AC wave you trigger) on the SCR’s gate.

The push button is connected probably to one of the I/O ports of the COB and it tells the controller to cycle through the prerecorded LED sequences. The resistor, capacitor and the diode path(R1–>C1–>D1) helps provide a DC voltage across the power pins of the COB. I am still unclear about the function of the resistor R2(1MΩ) which is connected between the one of the AC terminals and the COB. Maybe it’s a pull down resistor? (If you know what it is for, do let me know in the comments below. POST EDIT: Check comments. Its used as a reference resistor).

Well, this design is scalable to any number of LEDs and any number of patterns. For really crazy patterns, all you need is more parallel LED arms and each of those arms being controlled by individual SCRs.

So that’s it. Hope you learned something new today. You have to hand it to the Chinese, to make something so cheap. You can complain all you want, but it does take some sort of skill and innovation to come up with something absolutely dirt cheap(Even though it’s by cutting corners)

Do let me know in comments or via email at amaldev.000@gmail.com if there are any issues with the schematic or the working principle. Feedback is welcome.

If you enjoyed this one, you might enjoy my following posts as well on
How to electronically track your Currency Notes
Hacking Indian Electronic Voting Machines
A Smart Chair: For those Lazy Workaholics
OR
you can sign up for the spam free newsletter which will allow you to get update mails whenever I write a new tech article. ?

Till next time,
Happy Hacking…

Post Script

I didn’t have oscilloscope lying around while I was doing this teardown, it might have been great to see some working waveforms. Maybe I should simulate this circuit online in EasyEDA. If there is enough interest, I might go ahead and do it in my free time.

If you liked the post, Share it with your friends!