GPS time synchronization is 2s off on some units

Hi all,

I am a lucky owner of 12 Audiomoth dev with connected GPS modules, and encountered a strange issue: Sometimes a device will start recording 2 seconds earlier than others. Obviously, when trying to locate bird sounds via multilateration, this is an annoying issue. Everything between the Moths on a deployment is identical, all share the same hardware and programming (an external mic, a 18650 lithium battery as power supply and obviously a connected GPS module).

And yet, within a deployment, one or two seemingly random units will consistently start exactly 2s earlier than all others (ie all recordings from that unit in the deployment are 2s off). The time difference is exactly 2s, basically to the ms (as far as I can tell). So the GPS time setting apparently works in principle, but the recording is still offset by 2s. The filename shows a wrong starting time, ie when it was supposed to start, not when it really did.

The affected Moths change apparently randomly between deployments, therefore it is not connected to individual Audiomoths. I did not test if this failure is related to individual GPS modules - I will do this as soon as I can, but I doubt it somehow (GPS is all about time, so a 2s difference is worlds apart).

I rather suspect some kind of software bug, 1) either when setting the clock during the GPS phase prior to the recording, or 2) while programming the recording times for the devices on the PC.

Has anyone had a similar experience?

There is an obvious workaround - to take out exactly 2s from each of the .wavs of an affected unit - but that is not really the way it is supposed to work...

A pic is attached, showing the audios of the 12 moths next to one another on a table: it is easy to spot the two moths that suffer from the 2s-offset-bug.