NaNs from Muse 2016 from Android SDK but not iOS SDK


#1

The Muse 2016 reports “Great” (1) electrode connections on both iOS and Android SDKs for all 4 electrodes.

However, straight from the SDK, my EEG readings from ANDROID SDK look like this:
01-25 21:04:03.370 7215-7215/com.D/EEG Packet: NaN 813.1135531135532 NaN NaN
01-25 21:04:03.370 7215-7215/com.D/EEG Packet: NaN 812.7106227106227 NaN NaN
01-25 21:04:03.370 7215-7215/com.D/EEG Packet: NaN 809.8901098901099 NaN NaN
01-25 21:04:03.380 7215-7215/com.D/EEG Packet: NaN 807.4725274725275 NaN NaN
01-25 21:04:03.380 7215-7215/com.D/EEG Packet: NaN 815.934065934066 NaN NaN
01-25 21:04:03.380 7215-7215/com.D/EEG Packet: NaN 839.7069597069598 NaN NaN
01-25 21:04:03.380 7215-7215/com.D/EEG Packet: NaN 815.934065934066 NaN NaN
01-25 21:04:03.380 7215-7215/com.D/EEG Packet: NaN 793.3699633699634 NaN NaN
01-25 21:04:03.390 7215-7215/com.D/EEG Packet: NaN 807.4725274725275 NaN NaN
01-25 21:04:03.390 7215-7215/com.D/EEG Packet: NaN 822.7838827838829 NaN NaN
01-25 21:04:03.390 7215-7215/com.D/EEG Packet: NaN 829.6336996336996 NaN NaN
01-25 21:04:03.390 7215-7215/com.D/EEG Packet: NaN 824.3956043956044 NaN NaN
01-25 21:04:03.390 7215-7215/com.W/MUSE: PACKET eeg counter discrepancy last:16912 this:16914
01-25 21:04:03.390 7215-7215/com.I/MUSE: PACKET emitting 12 x 1 eeg NaNs
01-25 21:04:03.390 7215-7215/com.D/EEG Packet: NaN NaN NaN NaN
01-25 21:04:03.390 7215-7215/com.D/EEG Packet: NaN NaN NaN NaN
01-25 21:04:03.390 7215-7215/com.D/EEG Packet: NaN NaN NaN NaN
01-25 21:04:03.390 7215-7215/com.D/EEG Packet: NaN NaN NaN NaN
01-25 21:04:03.390 7215-7215/com.D/EEG Packet: NaN NaN NaN NaN
01-25 21:04:03.390 7215-7215/com.D/EEG Packet: NaN NaN NaN NaN
01-25 21:04:03.390 7215-7215/com.D/EEG Packet: NaN NaN NaN NaN
01-25 21:04:03.390 7215-7215/com.D/EEG Packet: NaN NaN NaN NaN
01-25 21:04:03.390 7215-7215/com.D/EEG Packet: NaN NaN NaN NaN
01-25 21:04:03.390 7215-7215/com.D/EEG Packet: NaN NaN NaN NaN
01-25 21:04:03.390 7215-7215/com.D/EEG Packet: NaN NaN NaN NaN
01-25 21:04:03.390 7215-7215/com.D/EEG Packet: NaN NaN NaN NaN

My EEG readings from the same muse(s) on the iOS SDK all look great. For example:
902.967032967033 837.289377289377 246.190476190476 890.07326007326
916.263736263736 841.318681318681 342.893772893773 897.728937728938
917.069597069597 839.304029304029 273.589743589744 901.758241758242
909.413919413919 835.677655677656 160.769230769231 909.413919413919
909.010989010989 835.677655677656 352.564102564103 903.369963369963
919.48717948718 832.857142857143 287.692307692308 884.432234432235
913.443223443223 828.424908424908 275.604395604396 879.194139194139
896.520146520147 822.783882783883 354.175824175824 890.47619047619
885.641025641026 823.589743589744 350.14652014652 898.131868131868
893.296703296703 828.424908424908 464.175824175824 897.728937728938
898.534798534799 831.648351648352 268.754578754579 894.102564102564
890.879120879121 836.483516483516 205.494505494505 902.161172161172
879.194139194139 841.721611721612 446.849816849817 889.67032967033
880.0 844.542124542125 457.728937728938 882.417582417582
882.014652014652 844.945054945055 356.190476190476 891.282051282051
875.567765567766 850.586080586081 294.542124542125 865.091575091575

I have tried multiple Muse 2016’s, including a couple brand new ones, and they all report similar issues.

I have tried multiple Muse 2014’s, and they do NOT have the same issue. Android spits out good data (and iOS doesn’t because obviously it’s not supported).

This appears to be a serious bug in the 6.0.2 Android SDK. Please advise.


#2

I’ve seen results like this sent in from a few users of Muse Monitor. In all cases it was down to the Android device itself, as when they tested the same headband on a different Android device it worked ok.

It’s worth noting that all the users who had problems said that the Interaxon Calm app worked fine… but I think (IMHO) this is because it just drops NaN data and works with what it’s got, to preserve the relaxation experience.

So far these are the devices which have had NaN data problems:

  • L6 G6 (x3 users)
  • Samsung Galaxy Tab A (x1 user)
  • Amazon Kindle/Fire Tablet (x1 user, not sure which specific model. Only happened after applying the latest firmware update 5.6.0.1.)
  • Amazon Fire TV (x1 user - myself!)

When I got my third report of problems with an LG G6, I looked up my install stats on Google Play. As of December 2017, I had 20 users with the LG G6… so if those people haven’t contact me, I have to assume their phones are working ok.

This makes me think that rather than the whole device being inherently incompatible, there must just be a common manufacturing defect with the bluetooth chip in this phone.


#3

Super interesting. I’ve tried on a Galaxy Tab A (super old) and a brand new Galaxy S8+. Both show the issue.
On an original Google Pixel it looks like I don’t get NaNs (possibly 0’s instead).

Muse team, what’s the plan here?


#4

My wife has a Pixel so I’ll give that a test when she gets home.

Can you re-test with the v6.0.1 SDK? I didn’t bother updating to v6.0.2 as the release notes said it was mainly for Windows stability.


#5

I’ll definitely try that out and get back to you but I won’t get a chance to for a little while (possibly a couple days or weeks).

Would really love to hear from the Muse team on this too. It’s so prevalent they must have run into it.


#6

My wife’s Pixel is working fine with Muse Monitor on SDK v6.0.1 No NaN values.