For sampling rates under 26 kHz, typical of sampling with the low-power channel, the standard A-size lithium cell should be sufficient to record for anywhere from 6 to 14 days depending on the ambient temperature and the extent to which auxiliary sensors are sampled.

For the fastest sampling rates, longevity may still be constrained by available storage. At 232 kHz sample rate, for example, 64 GB of storage capacity should be sufficient to record a maximum of 38 hours. However, the increased power demands at fast sample rates increase susceptibility to cold, battery contact resistance, and battery condition; thus even at fast sample rates power remains a concern.

This being said, a 128-GB Acousonde 3A tested at 464 kHz sampling rate in January 2012 filled 89 and 100 percent of storage (34 and 38 hours, respectively) in two tests running on a standard A-size lithium cell. A new battery was used for each test. The first test occurred at room temperature, the second at an average of 7.5°C (the better performance at colder temperature may reflect variability among new batteries). These tests indicate that the longevity of a 128-GB Acousonde sampling at 464 kHz is nearly balanced between storage-limited and power-limited. Thus the 64-GB configuration operating at fast sample rates is probably storage limited except in adverse ambient conditions.

Any Palm III or later PalmOS-compatible PDA is fine. We have tested the Palm III, IIIx, IIIxe, V, VII, m100, m125, m500, T|X, Tungsten T3, and Z22, as well as the Zire 21, 31, and 71; the Handspring Visor Deluxe and Visor Edge; and the Kyocera 7135 smart phone. In addition, customers have used a Palm Tungsten E2, a Palm Treo smart phone, and a Sony Clié. All have worked with the Acousonde and/or with its predecessor the Bioacoustic Probe. We are confident that any PalmOS-compatible PDA made by Palm, Handspring, or Sony will suffice.

Further, you do not need the cable or software for synchronizing the Palm with your personal computer, as the Acousonde can transmit its companion PalmOS application directly onto the Palm via infrared. If you have a rechargeable model you will of course need the charging cable (usually molded together with a wall-socket power adapter).

We recommend, if possible, using a Palm that operates from store-bought batteries rather than on an internal rechargeable battery. Store-bought batteries are more convenient in the field and there is no dedicated charger and cable to keep track of. Not only that, but the rechargeable battery is often the first thing to go wrong with a Palm as it ages. Models that operate on store-bought batteries include all models of the Palm III series, the Handspring Visor, Visor Deluxe, and Visor Neo, and all models of the Palm m100 series except the m130 (the m100, m105, and m125).

Cetacean Research Technology stocks Palm units specifically for sale to Acousonde customers. Customers have also found new Palms through Amazon.com and other retailers, and used Palms at swap meets and through online classified ads. A used Palm is perfectly adequate to operate the Acousonde; however you must test it. We have seen several used Palms with faulty batteries and touchscreens.

We are aware that, as a discontinued technology, Palms will become gradually harder to find. As of early 2016, however, Amazon.com still offers new Palms for sale, and we are confident that Palms will be easily obtainable for several years to come.

We regret that we are unable to offer advice more specific than this, or to confirm unconditionally that any specific PDA model not listed above will work.

In some circumstances we can preload a PalmOS handheld with the Acousonde firmware and mail it to the user, where it can be used to update one or more Acousonde units. Alternately, Acousonde units may be returned to us for firmware update.

• Two acoustic channels instead of one
• Low-power acoustic channel similar to Bioacoustic Probe, 9.28 kHz max bandwidth
• New high-frequency acoustic channel samples up to 464 kHz
• Acoustic gain choices now just 0 or 20 dB (dropped 10-dB choice)
• Anti-alias filters can be bypassed
• Future firmware will support "ping-pong" alternating sampling of both channels concurrently
• Tiltmeter (accelerometer) now 3D 10-bit (compared to 2D 16-bit)
• 3D compass provides new orientation sensing
• Real-time clock accurate to within 1 minute a year in normal temperatures
• Firmware supports easier and more precise time synchronization
• Up to 256 gigabytes per-deployment recording life (compared to 1 gigabyte)
• Full-speed USB data offloading (compared with standard infrared)
• New technology throughout offers room for future firmware/hardware improvements

Every time the Acousonde writes a new data file when sampling, it begins by writing a small amount of metadata. These metadata provide per-file calibration constants that account for all gain that is currently being applied. To put it another way, if you could compare the metadata for a file written with 0-dB gain and a file written with 20-dB gain, you would find differences. The 0-dB file would have one set of calibration constants, the 20-dB file another.

So there should be no need manually to account for the gain specified at recording time. It is already incorporated in the calibration and MATLAB file-import process.

Per-unit gain specifications and sensor sensitivities come pre-installed in each Acousonde. At build time, we verify each unit's gain specifications by recording an electronically-injected test signal of known amplitude with all possible gain paths. Then, at sample time as the data are saved, the Acousonde combines the user-selected gain and fixed A/D scaling with the known hydrophone sensitivity and places comprehensive overall calibration parameters in each data file's metadata header.

So as long as you import the data for analysis in a way that accesses and applies the per-file embedded calibration values (such as with the MATLAB script available in the downloads area) you will not need to calibrate the files independently.

We do not recommend calibrating the raw data samples yourself from a priori hardware parameters. Not only would you need to track the different sensitivities of each unit you use, you would also need to keep careful note of any user-selected gain associated with each data file and whether the anti-alias filter was bypassed or not (which can also affect gain). It is easier and less error-prone to take advantage of the Acousonde's built-in calibration system.

• Ensure that the Palm's batteries are fresh (replaceable batteries) or charged (rechargeable batteries);
• Ensure that the Palm's time is set correctly, which can be done directly on the Palm;
• Load the "Acousonde" commanding app onto the Palm via direct infrared transfer from the Acousonde itself; and
• Operate the "Acousonde" commanding app once it is on the Palm.

• Robust, inexpensive, handheld commanding unit for use in difficult field situations
• No cables
• No commanding sockets on Acousonde that could compromise reliability
• No prior pairing or other per-Acousonde setup necessary
• Select communication with desired Acousonde easily, even with several other Acousondes nearby
• Commanding app only needs to be maintained for one platform
• Commanding app can be loaded onto a Palm directly from an Acousonde in the field without need for cabling or any other equipment
• User does not need to sign up for a vendor's app "ecosystem"; in fact user does not need network connectivity at all

Nevertheless, we understand that Palms will gradually become harder to find and that eventually an alternative must be pursued. A stopgap measure may be simply to write a command-line app for a laptop computer mated with infrared hardware, and indeed we already have this set up for development and testing. If all Palms disappeared tomorrow we could quickly modify this software to replace the Palm functionality.

The high-frequency channel option, if installed at build time, can record at the hardware's maximum 464 kHz rate. The standard low-power channel cannot record faster than 154 kHz, but as this channel's anti-alias filter does not cut off higher than 9.2 kHz and the low-power hydrophone itself falls off above 30 kHz, the 154-kHz maximum sample rate does not impact usable bandwidth.

The low-power channel features an adjustable anti-alias filter that is automatically set to no higher than 0.38 times the sampling rate (it may be set lower; 0.38 times the sampling rate is the upper limit for the filter cutoff). This adjustable filter, like the sampling rate, is only capable of certain discrete cutoff frequencies. These are determined with the formula 9.28 kHz / {1, 2, 3...}; thus, 9.28 kHz, 4.64 kHz, 3.09 kHz, 2.32 kHz, 1.86 kHz, etc. For example, for a sample rate of 25 kHz on the low-power channel, signals above 9.28 kHz will be filtered out, but for a sample rate of 20 kHz, the anti-alias filter will be set to the next lower cutoff frequency, which is 4.64 kHz.

For the high-frequency channel, if installed, the bandwidth is determined by a fixed-frequency 6-pole linear-phase filter with a cutoff at 42 kHz (down by 22 dB at 100 kHz). If you do not sample the high-frequency channel at 232 kHz or above there may be aliasing of higher frequencies within the filter band into lower frequencies, which may or may not be acceptable depending on your recording environment.

The easiest way to incorporate higher frequencies on the low-power channel is to defeat the anti-alias filter at deployment time. This is only suitable if you are confident that all significant signal content will be at frequencies below one-half of the sample rate you choose (also at deployment time), otherwise your data will include aliased artifacts.

If you need to record frequencies above 9.28 kHz without aliasing, you will need to specify the high-frequency option (Option B003-HF) when you order and use that channel when recording. Remember that, by default, Option B003-HF incorporates a high-pass "pre-whitening" filter that attenuates frequencies below 10 kHz. If your application requires flat response from a few tens of Hz or less to frequencies over 10 kHz, then customized high-frequency characteristics will be necessary (Option B003-CF).

• the preamplifier input, set by the input shunt resistor and the hydrophone capacitance;
• the preamplifier input follower stage (10 Hz);
• the preamplifier gain stage; and
• the mainboard input (10 Hz).

The combination of these four high-pass filters results in a total filter response that high-passes at 22 Hz for the low-power channel and 10 kHz for the high-frequency channel for late 2011 units (earlier units had higher low-power-channel cutoff frequencies up to 38 Hz).

The Acousonde's high-frequency channel can apply a form of prewhitening by using a ceramic-and-shunt resistor cutoff frequency of 10 kHz. This filter attenuates incoming signals by 3 dB at 10 kHz, by 20 dB at 1 kHz, and by 40 dB at 100 Hz. Above 10 kHz the prewhitening filter has little effect. The low-power channel, on the other hand, is intended for calibrated general-purpose monitoring applications at lower frequencies and is not suitable for prewhitening.

Any incoming analog signal with a frequency above one-half the digital sampling rate (the "Nyquist frequency") will be aliased when digitized, and once the aliasing takes place, there is no way it can be undone in post-processing. Therefore digital recorders must filter out frequencies above the Nyquist frequency before digitization. However, if one is confident that one's recording environment contains no signals above the Nyquist frequency, or that any such signals are weak compared with the corresponding in-band sound levels, or if one wishes intentionally to alias incoming signals (as in the strobed-flywheel example above) then anti-alias filters can be omitted. The Acousonde provides optional anti-alias bypassing.

The other auxiliary channels, however — compass, temperature, pressure, and light — are not sampled simultaneously. These channels feed to a single high-resolution auxiliary analog-to-digital (A/D) converter through a multiplexer, and only one channel can be sampled at a time. Even at its fastest the A/D requires 6 milliseconds to acquire one sample; thus, compass Y is acquired 6 milliseconds after compass X, and compass Z another 6 milliseconds after that. At a 40-Hz sample rate, this means compass Z will be sampled halfway between samples of compass X! These data may require extra postprocessing to be useful in assessing very rapid changes in 3D orientation.

The cycle period must divide integrally into each 24-hour period; for example, 5 minutes, 15 minutes, and 60 minutes are all valid cycle periods, while 13 minutes and 57 minutes are not. The beginning of each cycle is aligned with the time of day. For example, 60-minute cycles start at the top of each hour. A cycle period cannot be longer than one day (1440 minutes).

The duty period specifies how many minutes of each cycle that acoustic recording will be active. For example, with a cycle period of 60 minutes and a duty period of 15 minutes, acoustic sampling will begin at the top of each hour and continue until a quarter after the hour (assuming zero-minute delay period; see below) at which time acoustic recording will cease until the top of the next hour.

The delay period specifies how long after the beginning of a new cycle to wait before beginning the duty period. For example, with a cycle period of 60 minutes, a duty period of 15 minutes, and a delay period of 30 minutes, recording would not begin until half past the hour and would continue until a quarter before the next hour. The delay period plus the duty period cannot be longer than the cycle period.

Fixing duty cycles to the time of day ensures that multiple units operating with the same duty-cycle parameters will record the same snapshot of time provided their clocks are synchronized, a technique known as synoptic recording.

Duty cycling applies only to acoustic recording. Once the sampling program begins, auxiliary sampling, if enabled, will be continuous.

We calibrate the pressure transducer (depth sensor) ourselves when we pressure-test each unit, and also estimate the effect of temperature on the depth sensor. Only the depth offset may need occasional recalibration to assure that the depth sensor reads zero at sea level and at sea-surface temperature. The depth offset may be "zeroed" via the Palm control interface.

The compass is calibrated at the factory to determine offset and scale values adequate to adjust the three magnetic axes for matching response; this is sometimes called "sphere-ing the ellipsoid". This can be redone by the user if there is concern that the compass sensor may have suffered a change in sensitivity, for example if exposed to a strong magnetic field that the built-in automatic degaussing cannot correct. We also recommend a further calibration exercise at your field location to maximize confidence in reconstructed animal tracks. A sample exercise for use with the "TrackPlot" track-reconstruction software is described here. Note that the compass calibration process does not support accurate measurements of absolute magnetic field strength.