"Tile Pro" Tracker - Impressive Range

[jumpingfool]

I have recently purchased the newly released "Tile Pro" bluetooth tracker after noticing a number of reviews online were giving good reports in terms of the BLE range.

The device is using the TC35678 from TI connected to an RF amplifier that I've been unable to identify marked X8 OZ4D. It also has a pretty funky looking PCB antenna - you can see it in the FCC report here:

fccid.io/.../Internal-Photos-3495407

They claim battery life of 1 year on a CR2032.

Under test the line of sight range from this device is nearly double that I can achieve from the NRF52 DK. Is this a function of the amplifier, the antenna design or potentially both?

If I move my design to the 52840 with +8db output can I reasonably expect to achieve similar performance in terms of range and battery life, or will an RF amplifier need to be incorporated into the design in order to achieve this?

[ovrebekk]

Hi

Was the output power of the nRF52832 +4dBm when you compared it, or was it lower?

I strongly doubt the antenna of the Tile module is better than the one on the DK, due to it's smaller size. If they did a good job tuning it (which I assume they did, judging by the range) it should be comparable.

If the range is double this is most likely caused by higher output power, and double the range equals roughly 6dB higher output power.

To summarize, if the Tile was double the range of the nRF52DK at +4dBm, then I doubt the nRF52840 at +8dBm will be able to beat it, but it should get you close.

As a final note, the TC35678 appears to be a Toshiba part. I wasn't aware they were in the BLE space, but these days everyone is.

Best regards
Torbjørn Øvrebekk

[jumpingfool]

Hi Torbjorn - really appreciate your help.

I managed to get hold of a PCA10056 today and do some experimenting. I built the proximity demo with the TX power set to 8db, flashed the board and then did some comparisons against the Tile board.

With the antennas on both boards optimally positioned it is a very close race, with the Tile giving roughly another 10-15 feet of additional range over the PCA10056. My guess is the RF amp on the Tile board is probably boosting the signal to 10db which is giving it a slight edge. Impressive they've managed to get the power budget down so low if that's the case though - any idea what manufacturer the "X8 OZ4D" part comes from?

However, it is a different story when the boards are rotated and positioned such that the antenna is at 90 degrees to the receiver. Here the PCA10056 drops out at about half the distance of the previous test while the Tile board shows almost no impact.

Given the application of the Tile board as a key finder where orientation to the receiver is an unknown clearly the funky looking antenna is optimised for this use case. Do Nordic offer a keyfinder reference design with a similarly optimised antenna, or if not could we engage you to assist with optimising our design?

[endnode]

Btw. there are some great nRF52 based beacons which are having line of sight >1000m (pure beacon ADV packets @+4dBm Tx Power visible on normal smart phone with <-100dBm sensitivity)

[jumpingfool]

Thanks for the heads up - the video is certainly impressive. Just ordered a few of these to try out.

Do you have one of these boards? How do you find performance in a real world setting?

I guess my question would be, if this line of sight range is possible at +4db using the nRF52832 why does the PCA10040 only seem give a fraction of that (tested with iPhone 8)? Presumably this can only be down to a difference in the antenna? Have the two boards been tuned differently and if so how?

[endnode]

Yes, I have couple of Ruuvi tags and yes, I've picked the tag from that tower so I can say that nRF52832 with Tx Power set to +4dBm have such performance. Indeed once you have any obstacles (it's enough to walk much closer to the tower but among the trees in the wood on that hill) the story is very different, you will see much lower RSSI or complete drop of packets.

And yes, the antenna design is what can make magic. It's big difference if you have experienced RF engineer who is able to run simulation (in expensive software) before finalizing the PCB and testing then prototypes and adjusting tuning based on actual PCBs received from the factory or if you just get some stock rectangle and put some textbook antenna to it. I'm sure Nordic have this kind of expertise but it's question if it was used or even desired in case of their dev kits.