Portable Bark Killer - how to complicate uncomplicated thing

[blankfield]

Hi all,
in this topic I'll describe design & construction progress of my version Portable Bark Killer. In my neighbourhood I have a noisy dog that long makes me wake up early. Therefore, this is the perfect project for me. I love dogs so I decided to make intelligent version of this unit to reduce emission of unnecessary ultrasound. Really portable unit needs good power management so design will be optimized for low power use in standby mode. I'm a fan of Texas Instruments MCU's so my first design assumption is to use msp430 device with cc430 to make wireless link. I would like to use remote control from my TI EZ430-Chronos watch and some digital signal processing for self-trigger mode.

Here is my ultrasonic consultant  :







Probably final functional diagram:


More updates as soon as possible.

Cheers
blankfield

[blankfield]

First update.

If someone looking how to simple generate ultrasonic waveform, there is nothing simpler than famous IC NE555. Just one IC, 2 resistors, 2 caps that's all. I was really surprised how loud can it be, even with small dynamic speaker, all from 12V power supply. LTspice results R2=R3=1k, C1=22nF, C2=100nF but in real live correct frequency achieved by change value of C1 to 10nF.


A basic circuit:


On breadboard:





Output waveform 22kHz


Zoomed spike shape, NE555 overshoot or "kickback" from dynamic speaker induction.


Output waveform 15kHz, something for human ears. 



I've made some tests with step up transformer to increase voltage level from 4Vpp to about 30Vpp. This is the best and simple way to drive piezo's from one Li-ion cell (10 cells in parallel). No need for additional step up controller and power consumption in idle mode is a 100% dependent and controlled by uC.

My development tools ready to go with coding:


I decided to use 2/4 pieces of these piezo speakers:

[MJLorton]

Blankfield...thanks very much for the informative post and keeping us up to date with your progress.

Cheers, Martin.

[blankfield]

Thanks Martin, I need to tell You this project is great I have a lot of fun making this. Here are my recent progress from yesterday.

Battery pack.
I taken Li-ion cells from my faulty Dell Vostro 1500 battery, one cell died but all other are to good to throw them away so I decided to make my own battery pack.
Yes I know, soldering Li-ion cells is not good idea but I did it quick so they survived :-). Now I have battery pack with 8 cell in parallel 4.1V fully charged, capacity is about 14Ah maybe I should measure it with some constant load???
To charge this package easily, I have modified my TrustFire charger (from dx.com).


Step-up transformer - "the black box".
Primary side : 20 turns  R=0,5R
Secondary side: 200 turns R=5R
Ratio 10:1 (secondary:primary).
Core: EI Ferroxcube E22/15/6  http://www.ferroxcube.com/prod/assets/ei22156.pdf

Unloaded secondary side can be dangerous (voltage peaks above 100V Vpp @ 1kHz)!

For switching mosfet I chosen MTP75N03HDL it is great logic level mosfet works nice with 4V Vcc (above 40A @ 3Vgs).
Anyway mosfet should have fast acting reverse polarity diode connected between drain and source.


Enclosure.
To better performance I decided to use 4 piezo's to cover more space around unit, piezo's are angled down because "Portable Bark Killer" will be placed about 3m above earth level.
In enclosure I will place:
1x on/off switch
1x fuse holder
1x light sensor
1x led
1x DC jack as charge port
2x uSwitch


MCU and power.
To meet requirements for low power I'm using one 1N4002 diode in series with 4V rail to drop 0.6V, MSP430 Vcc maximum is 3.6V so it's ideal solution.
Deep sleep modes in MSP reduces power consumption below 1mA, but of course this value will increase when I add RF and some simple "DSP" processing.

Planned features:
* automatic bark detection (it will be hard to implement)
* wireless control
* day/night detection
* battery voltage measurement
* statistics (no. of trigging's per day etc.)
* different frequencies and power (PWM)
* outside temperature measurement (msp430g2553 have integrated temp sensor)

One important notice this unit is extremely loud, during test at audible frequencies  put on hearing protection, this isn't a joke.

P.S. I'm sorry for the mess in my garage, I need to change table mat for something better.

[MJLorton]

Thanks Martin, I need to tell You this project is great I have a lot of fun making this. Here are my recent progress from yesterday.

Brilliant to see your progress...but very pleased to hear it is fun!

[blankfield]

Hi all,
There is a time to write about final results of my project, this week will be very busy for me so I've decided
to write some description now. First of all I want to tell you about my challenges with this project,
now I know - output power is the most important factor to make a device meeting the job.
My unit produces almost 10W RMS output power (measured with audio output power meter with bandwidth
up to 22kHz) (0,95A @ 12V @ 20kHz) it's enough to intrigue dogs in range about 5m but not more.
Noisy dogs in my neighbour  stops barking only for some while. My next step will be rewinding transformer
with thicker wires to use a maximum power of battery pack. Now I drain battery at 4A current level (@4,1V)
so it's low value like for 8 Li-ion cells, some space for improvements is available . Another fail like for me
is bark detection now is working like analog comparator realised via ADC interrupt, my attempts to realise some
DSP takes to much energy and MCU power so I decided to drop my plans. I found really nice program to design
digital filters - WinFilter easy and works! In present configuration standby power don't exceeds 900uA which
is great result. My observations led me to set frequency below 20kHz I think that cheap piezo have huge drop
of power above 20kHz, my piezo speakers are affected by this effect.
but

Block diagram.


Final schematics.


Some internal mess:-)


Assembled portable bark killer.



Output waveform loaded with 4 piezo speakers.


Output waveform shape unloaded.   


A new position in my watch menu.


At last I did some cell phone video, sorry about poor quality.
http://youtu.be/fhTo13PFoFI

Some source code.
Portable Bark Killer - firmware
Watch - firmware

Regards,
blankfield.

[blankfield]

Hi All,
At first thanks for voting my project.

Finally my portable bark killer works since today! And I'm very happy about this.

I spent today's evening to disassemble step up transformer that I've made some time ago.
It was a very productive time, I found some mistakes in alignment of 2 part ferrite core
that caused a high power loss. Next I have tried a brand transformer exactly
Colilcraft FA2443-AL (6:1 ratio) and results just shocked me! Huge different in power.
This inspired me to make another attempt to rewind my new transformer, I used
thicker 0.5mm enamelled wire 5 turns in primary side and 70 turns on secondary side.
PBK became so loud so I had to increase frequency from 16kHz to 20kHz. Output power
become even more stable than before. I thought for some while that I damaged
the MSP430 MCU but only 5A fuse just blown. For now power consumption in mode of 100%
output power rises above 33W (7,9A @ 4,2V) (power from battery pack).
Mosfet become quite hot (about 60 'C after 30s) but there is no need for heat sink.

My future steps will be:
* find 'of the shelf' transformer that will be have decent windings and 20:1 ratio and capable to transfer 50W
* design proper PCB
* develop code that can be easily modifiable
* release all files as OS

Neighbour’s dogs are not so happy like I

Take care,
blankfield.


Some additional photos:

Colilcraft FA2443-AL:



'Naked' core:





And some ugly windings (it's hard to rewind transformer without core casing):




Test jig:





And a evening mystery - blown 5A fuse:

[Geekinator]

Hi blankfield,

Thanks for posting such detailed pics and firmware!  Did you ever get this device working to your satisfaction?  I think you said it stops dogs barking for a while but then they start again.  I was reading on this site that the key is volume.  He links to some 120dB piezo speakers you might try.

My biggest concern is bark detection.  I tried an off-the-shelf ultrasonic bark deterrent with an audible mode and it was going off all the time due to any "loud" noise - even wind!  How often does your device go off for things other than barks?  If you're happy with your bark detection, does the latest firmware files you posted have your latest bark detection algorithm?