We hooked up the 40 kHz ultrasonic transducer to the frequency generator, and listened to it using the Secret Sounds toy. Direct sound is very narrow, with only a small range (10 deg cone, 5 foot limit). BUT, bouncing the signal off our rough-textured wall made the signal much more robust - cone nearly 180 deg, and audible as far away as we can get inside our room.
So, we were inspired to build a portable version of the frequency generator so that we could test things outside.
555 Timers Rule
We searched for a 555 timer 40 kHz circuit and then built it, with 9V battery.
Taking it outside, we discovered that distance is no problem - the signal can be clearly heard at 25 feet (and probably much further). BUT, the angle problem is back - both the receiver and the emitter have to be pointed very accurately, otherwise the emitted sound blends into the background noise. (the background noise seems to be mostly other bounces / bleed from the emitter - if you turn the emitter off, the ambient ultrasound is very quiet)
To try to solve this, we desoldered the mike from the secret sounds and replacing it with a 40 kHz transducer put onto the breadboard really close to the emitter's speaker. We built a small sound baffle by cutting up some rigid foam we found on the hack shelf.
The new receiver works - we are able to hear echos very well, as before. Unfortunately, there is considerable bleed from the emitter to the receiver, such that there is almost always a large amount of background noise. But, we were able to hear people walk by with the device, and able to hear walls and other objects. The most interesting sounds were when the board was quickly moved towards a wall or other solid object.
Due to the bleed, we weren't actually able to tell what the maximum audible range between board and the object that it's reflecting on is. We built a baffle 2.0 from some anti-static foam twice as high as previous baffle (with solid wire loops to hold it in place), but it didn't make any noticeable improvement.
We need to do some proper 'control/blind' experiments to verify our detection of walls/etc.
We need to build a pulsing circuit, so that instead of emitting a continuous sound, we can emit very brief pulses, and listen for the echo. This would enable a proper 'time of flight' echolocation rig, rather than the present amplitude/harmonics mechanism.
We need to investigate baffle technology.