Ultrasonic leak detector?

dougo

New member
I have been reading about Ultrasonic Leak detection that apparently generates noise at around 40Khz.
My aim is to (Blower door) pressure test a Passive House I wish to build. Apparently the 50Kpa (about 50mm water
offset in Bowden tube) is enough to cause Ultrasonic noise at leaks. To achieve PassivHaus certification, the leaks
need to cause less than 0.6 Ach (Air changes per hour), & the smaller the house the harder to get it leak free.
My project is 105 sq M, so fairly small. What I want is a detector I can walk around with looking for leaks.
My first check will be smoke to find the big leaks, but one needs something more sensitive for the tiny leaks.

Has anyone tried the TeensyBat for such an application? As a retired Electronics tech, the build will be straightforward.

btw, in Australia, the

Teensy 4.1 Development Board​

is available from K&A Electronics for $Aus 55 atm, a saving of $9. (Cheaper than AlieXp)
 
50mm head of water is about 500Pa, not 50kPa, you are out by a factor of a hundred.
50kPa (note the capitalization) is 1/2 an atmosphere. If you pressurize a house to that level it will literally explode. This is how tornado's destroy buildings. Yes a tornado will generate ultrasound!
 
Sounds like a cool project.

I haven’t played with the bat detector, but I can suggest a couple of things.

First, you need a microphone that is sensitive in the ultrasound frequency range. Many mics aren’t rated for the ultrasound range, but I would expect smaller diaphragm condenser mics (as in MEMs) to be more sensitive in that range. See if the Bat detector thread has any suggestions.

Instead of a visual interface to show you what’s going on (Bat detector plays around with analysers I believe), why not get a Teensy to slow down the ultrasound to a range for the human ear, so you can "hear the ultrasound"? So I might suggest that you record the sound at 192KHz sampling rate, then slow it down by a factor of 4 (you would have to remove 3/4 of the samples). That would make your 40kHz sound convert to 10kHz, which should be audible.

Your setup would be a pair of headphones, an audio adapter board a Teensy 4.X, and an I2S microphone. I honestly believe the human ear and brain would be a better "computer" to analyse such a thing as leaks.

You might have trouble finding an i2s mic that operates at 192kHz frequency, but reading some of the datasheets suggest that you may be able to achieve it by reducing the bitdepth of the input (I'd be interested to know what others think). FYI - a sampling frequency of 192kHz theoretically means you can sample sounds up to 96kHz frequency. So you could still record your 40kHz sound at a lower 96kHz sampling frequency, (sounds in the top 10-15% of the frequency ceiling don't tend to be good quality, so be warned).

For the processing chain, I guess you would need to make your own code. What you are looking to achieve is a pitch shifter / octave shifter, which is an interesting project in its own right. I can think of a few approaches to this, but will leave this to your imagination and research.

Good luck
 
Another thing to consider is CO2 build-up in a small sealed building... You definitely don't want leak-free, or atmospheric pressure changes would pop the windows, and the occupants would gradually suffocate! Listening for ultrasound to detect leaks is completely infeasible as I've pointed out, perhaps another approach is needed, like the normal thermal camera method?
 
Being an architect myself, I can tell you these pressure tests for air tightness of buildings are quite standard… they cover over the controlled ventilation openings, pressurise the house, and look for leaks around windows.

It’s all about making sure that air comes in through the controlled centilation system where heat from the exhaust vent can be recovered.
 
Thermal camera seems ideal for finding the leaks. 50kPa you are not going to get, but warm air seeping and warming the site of the leak seems very plausible and much easier to track down.
 
Back
Top