A collapsible and easy to build dish microphone
I wanted to have a directional microphone for bird and insect sound recordings in the field. The local electronics shop offers a small plastic dish reflector (diameter of about 30cm) which I purchased. But the first test was more than disappointing. Both gain (signal to noise ratio) and directionality effect were miserable with this small dish so I decided to build something more capable. It should still be easily transportable, though. The larger dish therefore had to be made collapsible in order to fit into a tramper backpack. I came up with the following design.
The small dish is used as a base to carry a set of two extension rings that approximate the parabolic form well enough. The whole thing is held together by screws. This picture shows the first build which I cut from polyacrylic (Plexiglass®) sheet which turned out to be an expensive mistake. Polyacryl is too brittle, it will break and splitter while cutting, and it can not be bent well enough for the purpose without the risk of breaking it. Polyethylene (aka polycarbonate) will probably work better, and I found polystyrene most comfortable to cut and handle (2mm PS sheet can actually be cut using scissors). In my country, there is plenty of this material used (and dumped) as advertisment panels on construction sites. Naturally, with the extensions the dish becomes quite deep. I did not find this to impose negative effects on the frequency response.
If you don't have a base dish available to build upon you can also simply use a flat plate on the back side. This will lower the performance by only a small fraction.
The extension rings have been cut into pieces of two (inner ring) and three (outer ring) in order to make them transportable (the assembled outer ring is more than 1m wide). Assembling it in the wild takes about 5 minutes.
Calculation and Construction
I include a thurough explanation of how to calculate the dimensions of the extensions as a PDF file. The calculations involve some fairly simple trigonometry. There's also a spreadsheet available that does the math for you (even in a more exact way than how it's done in the PDF). You only type in the basic dimensions and how big you want to have your extension(s). The spreadsheet should work in LibreOffice or in OpenOffice.
Explanation of the calculations, PDF
Explanation of the calculations, OpenDocument source file
Spreadsheet for the calculations, OpenDocument file
Adding to this, I custom built a stereo microphone from arrays of 5 mic capsules in parallel at each channel. Such an array assembly helps to dramatically raise the S/N ratio: the audio signals from the capsules add up because of their synchronicity while the noise from the individual capsules (which is stochastic) tends to cancel out. The mics I used had a sensitivity of -58dB (IIRC), the best I could get at this size and at a reasonable cost at that time. I estimated that the sensitivity of the whole assembly (microphone array and dish) is comparable to that of a human ear. Not bad!
The spacing between the two arrays has been made so that a wall of cardboard or plastic can be put in between. This wall would act as the key element in separating the channels, by practically cutting the dish in halves. It should fit tightly around the mic area and be the size of the whole cross-section of the dish. I never got around to making a more permanent solution than crudely cutting in a piece of cardboard (but it works).
Size matters. Because the mic cartridges are small and because an electret microphone responds to air pressure waves rather than to motion of the air, they should practically be no obstacle for sound waves down to a wavelength of double their size. The ones used are 4mm and therefore respond in an omnidirectional (spherical) manner at up to 42kHz. Field recordings (with Crickets) show that this particular type works well above 30kHz (I think the datasheet showed linear response until 28kHz, sorry I forgot the exact type).
There are also cartridges available that have cardioid caracteristics. Due to the parabolic approximation of the dish, I expect a "focal spot" rather than a focal point. My feeling told me to put the capsules closely together, so to have them all in this area. Peter from South Dakota brought me to the idea that arranging them in a spherical shape (instead of flat) would make them space a little more and block each other even less. Using cardioids pointing outward from the focal point would also help in separating the stereo channels.
(to be continued...)