Ecoobs Batcorder




The principle of the Ecoobs Batcorder is the same as the D500X. The unit is not made to use in the hand, but to be left in the field to pick it up later (no loudspeaker). This is the official website of the Batcorder.

batdetinterference

You can see that the microphone is at the end of an antenna-like rod. This design eliminates any problems with reflections. Also, because the membrane is small, there are no problems with phase differences along the surface of the membrane, see this link. Its sampling frequency is 500 kHz and its resolution 16 bits. The link here provides some more information. The system is able to reject non-bat ultrasonic sounds to prevent the memory from storing sounds of crickets instead of those of passing bats. Due to this hard-wired rejection system, the lowest frequency that can be recorded is 16kHz. Therefore, if you intend to work with Molossids, the batcorder is NOT your detector of choice (in this case a conventional microphone and 96kHz recorder should do the job!). Theoretically, one could use the Batcorder in the hand as a batrecorder with a heterodyne detector in the other hand to detect the bats. The pre-trigger time is fixed at 50ms. This is not very long, but if you see the bat coming one simply has to press the button in anticipation of its arrival near the microphone.
The Batcorder system costs 2400 euros for the detector and 700 euros for the software. It is possible to download the recorded sound files from the detector without the Ecoobs software, but the files are not in .wav format so they need to be transformed to analyse with Batsound. Clearly, the intended mode of operation is by using the Ecoobs software. To run the software you will need a Macintosh computer. We don't have any detailed user's reports on the software so far. The software seems to be very easy to use. It can be used to automatically identify the species (nearly all european species at present) and it gives you the certainty of each automatic identification. The software will be improved over the years to come so it will be able to tell you with more certainty the species you recorded. I understand that many of you will be sceptical about the automated ID, knowing how hard this is. Uli and Volker will write a publication about the batcorder ID method. On their website you can find performance tables of how well the software can identify different bat species. Their database for each species is enormous and it really contains the most open-environment calls down to the most clutter adapted calls each species can produce. However, all calls follow certain quality criteria. The highest frequency of the call isn't used in their algorithms. Calls are divided into sections whose slope is measured. The slope in combination with the frequency in the lowest part of the call are very important criteria in the species ID algorithms used by Ecoobs. This also applies to the slope of the final FM-tail that some bat species have. Ecoobs is therefore using quite similar methods as used in the Anabat-community and probably close to the way Michel Barataud judges the sound quality of calls. We hope to approach this methodology in the future by presenting similar slope values in the batecho tables.

User's report: Using the Batcorder for its intended purpose: monitoring

Christian Dietz reported back to us that he has been using the Batcorder for monitoring purposes in some forests in southern Germany. He also uses the Ecoobs software. He says the Batcorder saves you a lot of time. You come back to pick it up and let the computer automatically process hundreds of files. It was having some trouble with Barbastella barbastella sometimes, but generally seemed to be doing a great job. Leaving the detector for weeks at one spot offered the possibility of monitoring migration of Pipistrellus nathusii. This is a good example of using the Batcorder with a particular plan or idea, which is essential. Christian ended up getting thousands of soundfiles, a complete overkill. He decided only to look at Pipistrellus nathusii and to plot the number of recordings per night over a period of many weeks. This way he did something with the Batcorder that would have been extremely demanding to do with a normal bat-detector.
Even if you have some serious doubts about IDs made by the Batcorder, it may point you at species (or call aberrations) you normally wouldn't notice. One can always go back to a particular location to put it full of mistnets. If you only get pipistrelles on the Batcorder you can spare yourself the effort.

User's report: Using the Batcorder in the hand in rainforests.

I do not possess a MacIntosh computer and never used any Ecoobs software but I have now used the Batcorder only as a simple recording device for nearly a year in Java, Sulawesi, Bali, Flores and Papua.
The detector has 3 modes: 1) A programmable timer that wakes it up to start and stop being active whenever you want. 2) A mode in which it has started already (active) and then scans and starts recording whenever a bat triggers the system. 3) A manual mode in which you can simply push a button to start and stop a recording.
The design of the batcorder is so simple that you don't even need a manual to figure out how the modes work. Since I haven't got the slightest idea of what I am recording in this part of the world, I mostly use setting 3 and wait until I see the bat coming. In twilight, cities and villages this works very well, much less so in forests. A raw file is created immediately on an SDHC card. Using a simple card reader I can later transfer the files to my computer. I use Matlab to import each file, normalise its amplitude, and then read it out with 50kHz, so the recording gets slowed down 10x, but this can also be done manually with freeware from the web, for example Wavosaur. I have tested setting 2 as well. From 17:50 every day here, bats can be seen virtually everywhere, even on a highway in the fumes of buses and motorcycles, flying past and over you at different altitudes. At this point, the Batcorder will store files at a pace of 10 per minute. This is a real consideration to make when using the batcorder: what will be the purpose of your study (also see user's report above)? Do you want to analyse 500 files after each night? Of course, in temperate regions of the world, this problem won't be as grave, but still, if a pipistrelle decides to start hunting for 20 minutes around the batcorder... The filedata do show the time the bat was recorded, so this is not a technical limitation but it is something you as a researcher have to think about in advance!
In the active mode (user going to cinema/restaurant or in bed) the Batcorder scans for bats and puts their calls on the memory card for you. However, it doesn't simply record ALL bat sounds. It only accepts "good" calls and rejects everything that is not "good" enough. This is in fact what everybody does (or should do!) manually when analysing recordings. Below, I will discuss in more detail the "trigger happiness" of the batcorder for different species of bat. When it is triggering, it sometimes records only a couple of pulses from an inferno of flying bats. I get the impression that this often happens when it has just finished storing a long sequence.
When it comes to recording quality, the batcorder is good. The antenna-microphone is meant to eliminate any secondary reflections from the casing, which is a problem in the D980/D1000. There are still some reflections, but unless the pulse is very long, they appear after the recorded pulse. The secondary reflections can be nearly eliminated by using a triangular reflector or a piece of sound absorbing foam. The signal is captured quite well with the 500kHz sampling rate. Often, when a bat passes though, the recorded pulses are still -10dB below saturation level. According to the specifications, the amplification of the batcorder is such that saturation is reached if a 40kHz tone hits the microphone with 96dB SPL. I can see why Ecoobs did it this way because the loudest bats do nearly reach saturation level when passing. Loud bats would get distorted if the amplification had been higher. Still, if a very quiet bat would pass, we may be as far as -20dB below saturation level and I am not sure if the S/N ratio won't deteriorate at some point. Maybe Ecoobs can comment on this. In any case, the 16 bits should be sufficient to capture the amplitude modulation of signals well, even if they are well below saturation level. My experience so far is that only the Molossid Chaerephon plicata, still at >5m distance managed to distort (slightly) by throwing more than 96dB SPL at the microphone, pipistrelles are always a fraction under the limit and all other bats are well below it. The batcorder picks up horseshoe bats and hipposiderids quite well, but this is also because they often fly low. Hipposideros ater however, which emits around 160kHz and is renowned for being very quiet, failed to trigger the Batcorder on more than 10 occasions, even when flying as close as 20cm to the microphone. Part of the problem here may be the relative insensitivity of the Batcorder to very high frequencies (see below). Very low frequency bats (below 22kHz) are also easily missed by the batcorder. Such bats really have to be close and loud. Strange pulse types, such as used by Emballonurids are no problem. Of all the insects encountered so far, one bushcricket on Flores and 2 species in Sulawesi managed to trigger the Batcorder. It has to be said that one of them in Sulawesi really looks like a bat call. West of the Wallace-line I have had no problems so far. Car breaks also trigger the Batcorder, so be careful using the unit near roads. I have also tried to use the detector from a bus (packed with people, pigs and chickens). I got no false triggers, but none at all, maybe because the bat needs to be close long enough to trigger the Batcorder. From a motorcycle (not too old), the bike itself kept on triggering the Batcorder, so test your means of transport before you decide on buying the Batcorder.
One group of bats I keep on missing are: Kerivoula / Murina / Phoniscus which use quiet very high frequency FM sweeps. Even when very close, they won't exceed the threshold to activate the trigger. The same may very well apply to phyllostomids in South-America, so I really believe some improvements are required on this front. To record a Murina, one really has to manually trigger the detector (and then you do see the signals). But how often is it that you see a bat coming in complete darkness? If at least the pre-trigger time would be 2 seconds or so, this would have helped me enormously to get the close bats you see passing in the light of the head torch. Now, I am often too late pressing the button as the pre-trigger time is negligable. Still, despite not working well for very high frequencies and weak sounds, the present trigger function is a great help to find bats also when using the Batcorder in the hand. When using a heterodyne detector to find bats, one tends to miss many rhinolophids and hipposiderids. At least half of them I would never have recorded without the trigger function! A broadband detector to hear bats coming in this case wouldn't help that much either since the choir of insects is pretty overwhelming. My conclusion is therefore that the automatic trigger mode works quite well in principle, but some modifications could make it work perfectly in the tropics.
The Batcorder is supposedly weatherproof. Even inside non-leaking houses in Indonesia, CDs get mouldy because of high humidity. The telephone cable corrodes inside its mantle and cameras and lenses get condensation inside even when packed in plastic with silica gel. Now imagine what conditions are like in rainforests! So far, no problems whatsoever. The Batcorder kept on recording fine when I left it outside during light rain. It can withstand conditions a D1000 microphone would never survive. The microphone even survived several attacks by Myotis muricola, a bat emitting pipistrelle-like calls and which, somehow often mistakes the glint from the microphone for a prey.
I like the Batcorder so much that it has now become my main recording device, with my D240 just serving as a scanner to see what's around. Of course, I still have to note down that the funny Eptesicus-like bat was recorded on file 109 and some high flying bat at 110. One could use a dictaphone for this. I can imagine using the Batcorder together with a cheap heterodyne detector to find the bats, with the Batcorder to record them well, once found. In this case, it would be best to look for some kind of carrying bag to carry it around your neck / waist. To this purpose, the pre-trigger time really should be increased in future versions, though.
The Batcorder is powered by an external rechargeable battery that is part of the kit. The battery will run for 100 hours, typically data can be recorded over 5-20 nights. The battery is small and doesn't weigh that much.
To make the detector as energy efficient as possible, the hardware uses a single passband filter (Butterworth 10th order = 60dB/octave) to serve both as an insect-noise rejector and at the same time as an anti-aliasing filter to reject very high frequencies. The cut-off points (-3dB) of this filter are at 16 and 150kHz. As a consequence, the recorded sounds at 200kHz are suppressed by 28dB and by 47dB at 250kHz. Since some bats on this planet start as high as 185kHz (some maybe even at 220kHz!), this is a bit of a shame. Suppression at this frequency is 21dB, so these bats have to be extremely loud to come through OK and are nearly always missed in scanning mode. As explained above, the choice to use only 1 bandpass-filter (instead of a separate anti-aliasing filter) was to save on power consumption.
I think that the choice of 150kHz as cut-off frequency in combination with the low gain (see above) of the detector makes the Batcorder less suitable for use in tropical regions of the world where some bats species tend to use weak high frequencies. I would urge Ecoobs to choose 16-200kHz as cut-off frequencies for the bandpass filter in future models. Suppression at 250kHz will then amount log2(250/200)*60+3=22dB. 22 deciBels, combined with the decreased microphone sensitivity at 250kHz and the little energy in this range arriving at the microphone anyway, should be sufficient to minimise aliasing.
For demonstration purposes the Batcorder is obviously completely useless as you first need a computer to make anything audible: it doesn't have a loudspeaker. I am usually not bothered by this as I use a D240 in the other hand. This other detector (a simple heterodyne is fine) is really important if you want to use the batcorder as a recording machine for unknown bats. Check at the spot what the batcorder is recording (be it in scanning mode). I once didn't do so only to discover at home that 2 species had been present instead of 1, too late!


Here are the main conclusions and recommendations for future models:
Selling points: easy to operate, good design, weatherproof, long battery life, very good recording quality, matching the most expensive detectors, can be used in hand, but also automatic (camping, flying routes, private gardens/parks).
Points to improve in future models: Widening the upper frequency range by choosing 200kHz as cut-off frequency instead of 150kHz. Lengthening the pre-trigger time to at least 1 second, but preferably 2 seconds. Incorporating a gain control so quiet bats can be detected too. This would be at the expense of no longer being able to analyse loud bats (pipistrelles), but this would then be up to the user.
Added note: Ecoobs are developing a playback system and are thinking about making the Batcorder more sensitive to higher frequencies.

p.s. Mind you that the software only covers european bats (nearly all of them at present).



Last updated 5.1.2010