Estimote stickers dropping like flies :-(


#1

Hi everyone. For a project with the museum “Buddenbrookhaus” here in Lübeck, Germany, we were working together with five students, about 15 years old each, to create two exhibits for an exhibition. We, some guys from a local hacker- & makerspace were supporting these students to implement their great and crazy ideas. In the end, one of the exhibits involved 21 estimote stickers for an IoT suitcase. (I’ll try to make an English translation of our blog article describing the project in the next few days)

The first experience with the first box of ten estimote stickers was great, in that it was relatively easy to start and get going with the nodejs library “node-bleacon” on github, by Sandeep Mistry. Unfortunately, quickly we discovered, that out of these ten stickers, three ones were not working at all or only after some squishing… The Estimote support was very helpful to send us some replacement stickers very quickly, together with the second batch of ten stickers (as we needed at least twenty for the project).

That was in April. The exhibition opened at the 12th of June with finally everything up and running. Until then, the number of malfunctioning stickers had increased from three to five, we had to replace two more stickers about one week before the grant opening. So we ordered a third batch of ten stickers, just to have some stickers ready for any quick replacements.

Last week I got an email (24th of June), that one more sticker misbehaved. I went through the list of 21 stickers that should be there remotely and even found two of the 21 stickers gone. I came over to replace them today - and by that found out that from Friday until Tuesday, so just within four days, the number had increased from 2 out of 21 to 5 out of 21 not working anymore… Now I’m running out of replacement stickers…

The remaining stickers in the replacement box seem to have some issues, too: One of them beaconing at 100ms interval even though not in motion. One beacon sent gyro values x=0, y=0, z=0 once in motion, and only send correct values when still (battery starting to die, unable to power both the gyro chip and bluetooth at high rates, maybe?). Btw, I’ve never used the iPhone app or changed any beaconing intervals. I mostly used node-bleacon, sometimes the Linux “hcitool lescan” tool to check whether a sticker seemed to work and just a few times I checked with the Android app.

For the first batch with the three broken ones, we had opened up one of the stickers: It was clearly a hardware defect (will send a picture the next days): The hard solder blob of the battery holder had made two tiny but big enough rips into the flexible circuit board to create a loose connection.

For the others that died over time, it seems to me like the batteries are draining too quickly. Are there any known issues with a sticker firmware bug, resulting in getting locked to a 100ms interval? Or could it be possible that node-bleacon sends some messages to the stickers which might increase the battery usage?

Sorry for the quick write up, I’ll try to describe the project and setup more thoroughly the next days and ill try to gather more information, like the battery and beaconing values via the official Android app and will compare that to their actual behaviour.

If you have any recommendations what else we could/should check, then please let us know.

PS: The estimote stickers are mostly still. Only sometimes visitors are moving an object with an estimote sticker inside from a table into suitcase.

PPS: The Buddenbrookhaus had planned to lend the overall exhibition to other museums in Germany after this special finishes in January. But for the IoT suitcase, that won’t work out with the current issue of regularly broken estimote stickers and overall maintance work…

PPPS: We all love Estimote, the museum as well as the people from our hacker- & makerspace are totally hooked with the potential use cases of these and love to do more with them. And we are loving the availability of open source software for this hardware. So let us know how we can help to improve them, what to code & test, etc. :slight_smile:


Passive vs. Active Bluetooth LE Scanning
#2

As promised, here are the pictures of the broken sticker we opened:

As you can see, the big soldering blob for the battery socket has an ugly cut next to it.

For the battery life, I’m currently trying to investigate whether it might be due to how node-bleacon scans (or whether it is a sticker firmware or hardware issue):


Raspberry Pi can catch Nearable packet?
#3

Thanks SO MUCH for your detailed report. Terribly sorry for the trouble, and let me assure you we’re doing our best to address most of the reported issues.

Yesterday we submitted an update to our Estimote iOS app which, when it hits the App Store in a couple days, will allow you to update your stickers to firmware 1.3.0. Among other new things introduced in 1.3.0, we fixed a bug that was sometimes causing stickers to enter a weird loop that would drain the battery very quickly. Another thing we fixed is, stickers will now continue working even if some of their non-criticial hardware components fail, and the new app will also automatically report any hardware/firmware errors to Estimote Cloud, so that we can better diagnose what’s going on.

So I would recommend, when you get your new replacements, update their firmware right away (they will most likely still come with an older version), and let us know how it goes.


#4

@T_X

Do let us know how it goes with the new firmware 1.3.0, whenever you receive the replacements. :slight_smile: I’ve been quite hesitant about replacing my dead stickers as I wasn’t sure that the ‘loop’ problem will be fixed yet. :frowning:

On a separate note, we decided to open up a few of the dead stickers, and found that there are some differences in the hardware - e.g. the ‘newer’ PCBs seem to be harder, while the ‘older’ PCBs were very soft, like cardboard material. Not sure if this makes any difference… will post some pictures later!


#5

Hi heypiotr, I know, especially firmware updates should be carefully tested before rolling them out, but just wanted to ask whether there are any news regarding the 1.3.0 release. We tried updating today, but only 1.1.0 showed up in the Estimote App of an iPad.

Btw., are there any release notes regarding firmware updates? Love to read them myself as a firmware developer :-).


#6

First update for now, starting with the good news: 20 replacements arrived last Monday and was able to have a look at them today. They are all running, with a battery status showing “High”. 10 of them are with and the other 10 without a casing, so it’s easy to reboot those or replace their batteries.

Bad news: One of them again has a motion status stuck on still. Even after rebooting by pulling the battery, it is still stuck on “still”. Therefore I can’t update the firmware or any settings on this one. Temperature sensor has sane values though.

Another bad news: Of the two spare, working replacements we still had available, I tried to update them last week. For one of them I was too late, it had died, too. For the other one, I successfully updated it to 1.3.0. Now, one week later, the same sticker is still advertising a battery status of “High”, but it is stuck at temperature showing zero degrees and an in-motion status of always “still/No”. I can’t adjust it’s settings anymore. This sticker has always stayed in the cardboard box, I showed it to another guy once within the last week, but despite from that did not actually move it since the update a week ago.

For the ten “stickers” without a casing, they came with a TX power of 4dB preconfigured. I wanted to adjust that 0dB like the other ten stickers, but looks like any TX power settings I try to set on the iOS app are not applied. Going back to a stickers overview it still shows 4dB. Rebooting/power-cycling it does not help either.

I am a little reluctant to spend another full day replacing the stickers from within the exhibiton items (it’ll need some sewing/cutting/glueing to hide them again and to keep them theft resistant). As there still seems to be a firmware bug with a sticker staying in one state (this time the “still” one). I will open up this sticker and powercycle it in the next days to confirm that it is a firmware bug and that it is not the gyro sensor that magically died.


#7

PS: For the other new, 19 stickers, they are all on 1.3.0 now. I updated their setting to 100ms interval as we need responsiveness when items are chosen/moved, but set them to in-motion-beacons-only mode. In the hope that maybe you “just” coded some stupid race condition between the periodic timer interrupt and the gyro interrupts.

(I’m kind of itching to see the firmware code, I have a little bit of Linux kernel coding experience and finding/fixing race conditions in its C code as well as embedded wifi router firmwares, but unfortunately the Estimote sticker firmware is not free software. So can’t help with that even if I wanted to :frowning: )


#8

PPS: But the new PCB looks great. The harder material of the new SB0 one seems more durable and probably should not be susceptible to the issue I had shown in the pictures above anymore, I think. Great improvement in that regard :-).


#9

Can you private-message me your Estimote Account email? Thanks to the extra reporting we added 1.3.0, I should be able to look out if there are any problems with the stickers that report wrong motion/temperature values.


#10

Hi @heypiotr, hope you got my PM last week. Anything you could spot in the logs on your side? Let me know if you need any further information or testing from our side.


#11

We’ve also had that problem with the developer kit. We’ve had it about 8 months and already 8 of the 10 stickers are dead