From Noisebridge
(Difference between revisions)
Jump to: navigation, search
(Hack Notes)
Line 73: Line 73:
[[UltraEcho_2009Aug10|August 10, 2009]]
[[UltraEcho_2009Aug10|August 10, 2009]]
[[UltraEcho_2009Aug24|August 24, 2009]]

Revision as of 20:12, 24 August 2009


Ultrasonic Hearing & Echo Location

Inspired by this Wired Article on learning to do echo location by clicking your palate, wouldn't it be great to be able to use echolocation?

Expanding out the idea, most species which do echolocation use higher frequencies, because the spacial resolution goes up with decreasing wave size (=increasing frequency). From my (Eric Boyd's) email to the cyborg list:

Having thought about it more, I am keen on an "echo location" kit. Imagine some electronics which generates an ultra-sonic noise (40kHz? 60kHz?), then receives it (via normal microphone?), and then frequency shifts the sound back into human hearing range, and uses ear-buds to display it to you. If the frequency shift is done correctly, you could even pre-process the sound data to help "amplify" the difference that close object bounces creates in the sound. This leaves the real data processing to the brain, of course - it's still going to just be a bunch of sounds, not a map of what's around you. But I actually think this could be way superior - who knows what kind of patterns your brain could pull out of the sound if you just wore the electronics for a week? I have no idea how complicated the frequency shift math might be, but I think the electronics for this should be doable using Arduino-class hardware?

I also have some thoughts about the armature. According to Song of the Mouse, mice make many noises in the ultrasonic region. I think it's only natural that a device which allows you to hear mice should be made from Disney Mouse Ears! This also means it's located conveniently near your ears, where it must display it's data anyway...

As I see it, the ultrasonic noise generating device is actually a separate thing entirely, expanding the original "ultrasonic hearing" device into an echo-location device. So in terms of working towards a prototype, first you build the ultrasonic hearing rig, then you build the emitter and tie the two together to get the ranging information.

Interesting Links

Development of a Bone-conducted Ultrasonic Hearing Aid for the Profoundly Sensorineural Deaf

Ultrasonic listening devices: listen to bats, insects, and other high-frequency noises

Wikipedia: Animal echolocation

Ultrasonic Ranging Circuit

Uncle Milton Secret Sounds, see also the manual (I am buying one now)

Build A Simple Bat Detector - tons of great info and schematics, see especially the link Bertrik link at the bottom of the page for yet more circuits and groups interested in listening to bats...

Wikipedia Human Echolocation Page - interesting! See also this cool story about Ben Underwood.

Neuroethology - bat brains - information about how the brain of the bat manages to do what it does as regards echolocation. The complexity of the stuff that the bat brain does is awesome - but this level of detail also makes it possible for us to engineer something. In fact, some germans have already done this, pretty awesome! There stuff isn't portable or even real-time though...

Research Notes

I did a whole bunch of reading about "bat detectors". Basically they come in two varieties, the Heterodyne, and the Frequency Division. The former takes e.g. a tunable 8kHz zone in ultrasonic, and moves it into an 8kHz zone in the audible region. You can think of it something like a radio for ultrasonic sounds. It preserves amplitude information, but since it only accesses a tiny fraction of the ultrasonic region at once, you can easily miss lots of stuff. Frequency Division is taking e.g. an 80kHz band of ultrasonic sound, and dividing it down to an 8kHz audible zone. The division is done digitally, so the typical result is a loss of amplitude information; the resulting sounds are comparable to a "Geiger counter" experience of ultrasonic sound. There is also an expensive technique of time expansion - record 1 second of data at e.g. 50kHz, and then play it back at 5kHz for 10 seconds.

The Bertrik page lists many sources for bat detector kits, but all but one appear to be dead, and that's not so much a kit as an offer of a custom PCB and BOM, see Tony Messina Bat Shop. You must order the entire BOM yourself. The list of products fairs much better, with most still being available, but the prices are quite high. Market entry is about $80, and the sky is the limit - there are some thousand dollar+ products for "professionals".

Bertrik himself in 2006 was working towards an amplitude preserving version of the Frequency Division idea, but he appears not to have finished the effort. This is a promising approach to a more convenient device for listening to all ultrasonic sound. He also has some better info up at Digital Bat Detector Wiki

Another approach is to use more modern digital technology. With a modern DSP you could acquire the entire ultrasonic region, and then use FFT and some fancy algorithms to "compress" the data into the audible region. There would be some tradeoffs, of course, but I think it would be better than the amplitude preserving frequency division idea, if a lot more work.

I would like to encourage one of the people to finish the kit and then sensebridge would help them to sell it (i.e. put it on our website and help them with China manufacturing and distribution through various hobbiest sites, like our plans for North Paw). I would then take one of these kits and add extra stuff to support the echolocation functionality.

Uncle Milton's Secret Sounds

That's the name of the ultrasonic toy that I ordered from Amazon. I received it on Wednesday, it works once you get a working 9V battery into it. It appears to be an untunable hererodyne circuit, but there is no information on what frequency it operates at. I was able to hear their "bat sound" emitter (which is actually just a nice wave pattern), plus the sound of CFL bulbs and the sound of bouncing coins. Everything else (computers, the North Paw, etc) was quiet.

We hooked the speaker in the Secret Sounds toy up to frequency generator. Using the Secret Sounds toy, you can hear sound up to 16kHz-22kHz (depending on your age, mostly), and the device seems to work from 10kHz-80kHz. It's much louder at lower frequencies though - not sure if this is because of the frequency response of the speaker or the microphone. I believe this confirms the device as some kind of frequency division device, but you can hear some amplitude information as well, as verified by pressing the attenuation buttons on the frequency generator.

Looking at the parts on the circuit inside the listening device, we have:

  • HC193 - "SYNCHRONOUS UP/DOWN BINARY/DECADE COUNTER". There are actually two of these chips, they are clearly the heart of the thing, dividing the signal into the audible range.
  • HC4066 - "Quad Analog Switch/Multiplexer/Demultiplexer". Not sure what this is used for?!?
  • 464C EZ5G723 - Quad "Output rail-to-rail operational amplifiers". This maybe handles the volume control. And probably the amplitude preservation stuff (we don't know how that works)
  • HC4046AM - "High-Speed CMOS Logic Phase-Locked Loop with VCO", this is probably the signal conditioner between the mic and the binary counter.
  • 6A77 - a little IC, marked BATT, maybe it's an LDO or similar?
  • two NPN transistors on the back side
  • 4 big caps - 2x10uF, 1x47uF, 1x100uF
  • A diode
  • Dozens of little passives
  • POT, the volume control knob thingy. Also contains the main switch (4 wires, one of which goes to battery)
  • Battery connector for 9V battery
  • Two wires (red and a black shield) to the mic, but can't see the mic without destroying it
  • Wires to LED for power status
  • Two wires to the headphone jack

Hack Notes

August 10, 2009 August 24, 2009

Personal tools