I would like to share some results as a continuation of an earlier post on energy consumption of SD cards at https://www.openacousticdevices.info/support/sd-card-support/u1-vs-u3-microsd-cards-field-data-contradicts-recommendations?origin=notification

We conduct an indoor experiment using Audiomoths with firmware version 1.11.0 in a recording window of 180 minutes. The sample rate is set at 48kHz, and the gain is set at medium. We enable sleep/record cyclic recording. We use daily folders for the generated WAV files. We combined two sets of different recordings and sleep durations with different SD cards and battery models to assess the effect of sleep durations, and the type of card and battery. These are:

Alcaline vs lithium

Lithium batteries do not extend the operating time beyond what alkaline batteries did when the recording schedule was intensive (recording each minute during the 180-minute window). Compare BAU2 vs BAU1.

Recording scheme

As it was expected, a less intensive recording scheme (recording one out of five minutes during the 180-minute window) allowed recorders to operate longer, with both type of batteries. Compare BAU1 vs BAU1_deg and BAU2 vs BAU2_deg.

SD card

Our slow card allowed recorders to operate longer than the fast cards did for both recording schemes: compare BAU2 vs BAU1_b (5’’ sleep, 55’’ recording), BAU1_deg vs BAU1deg_b (240’’ sleep, 60’’ recording).

As it happens, all the cards we tried were manufactured in Taiwan.

And maybe that is the crucial point.

Manufacturer

In another extensive deployment with audiomoths equipped with SanDisk Extreme 64GB cards and Duracell Pro alkaline batteries spread in around 100 outdoor sites in mainland Spain (recording window 150 minutes from 30 minutes before sunset, sample rate 48kHz, two-minute sleep and one-minute recording) we found the manufacturer country to have a consistent effect in the durability: those cards made in Malaysia and China were still operating after five months (these were manually stopped), while those from Taiwan interrupted operation in roughly half of that time (see graph below).