Wireless Sip and Puff Switch for Computer Access (University of Arkansas)


This project, Wireless Sip & Puff switch for computer access, came from an interest in improving the way that a person with a severe disability clicks the mouse when using alternative access methods. The current method of clicking generally used is dwell (holding the cursor still for an amount of time) and takes more time than is necessary. The final design of this project is a small wireless sip & puff switch that could be used attached to a wheelchair like George is using it in our pictures or attached to the head using a headband.  George and others were able to issue left and right clicks while using a head tracking system to control the movement of the cursor. Also, the interface allowed for ideas of more work that can be done to find alternative user interfaces once in production.

Switch mounted on George's wheelchair

Switch mounted on George's wheelchair


The standard click mechanism used today for most types of alternative mouse movement, besides blink clicking for eye-tracking, is a dwell. Dwell refers to when the user must hold the cursor steady within a certain area for a defined time to cause the click to occur. The user must also select which type of click will occur when they dwell before the time of the click; therefore, the user must dwell twice if not using the default click type. The objective of this project is to create another mechanism to click by using a sip & puff interface.


The main goal at the beginning of this project was to create a sip & puff interface that could be worn on the person’s body and be able to get the current pressure being applied to the device at any moment. This set the first two specifications that the device must be both wireless and small. Then, since this project was done with help from and will eventually be commercialized by InvoTek Inc., the packaging that is currently being used on their laser pointer for their Safe Laser System (InvoTek Inc.) was planned to be reused to package the result of this project. Therefore, the reuse of packaging set the exact dimensions of the circuit board that was to be designed for this project. The picture below shows the finished switch in a modified version of the laser pointer packaging, to make room for the large pressure sensor, beside a standard computer mouse to show size.

Sip & Puff in laser package

Wireless Sip & Puff in proposed laser package

The next step in the project was to determine the type of pressure sensor to read the range of pressure that can be created by a person as well as read both positive and negative pressure values. This gives the ability to tell when both the user sips and puffs. A sensor (Freescale Semiconductor, 2009) was chosen which can read -3.6 to 3.6 psi and has an analog output ranging from .2 to 4.7. A microcontroller (Microchip, 2011) was chosen that had a built in analog to digital converter. The wireless components were chosen based on prior use by InvoTek in multiple products, and the power circuitry was then built around these parts which required both 3.3 volt and 5 volt power supplies.

Once the hardware was chosen, the user interface to display the data from the sensor was developed. This interface is a window (pictured below) which contains a debugging window, a bar to show the current pressure and two sliders to determine the threshold of when the program is to issue a switch closure. This switch closure was chosen to be a left click for a sip and a right click for a puff. For example, the program issues a left down command when the pressure drops below the lower threshold and then issues a left up command when the pressure raises back above the lower threshold. This allows the user to issue single and double clicks as well as click and drag using their breath without having to select the type of click that is to be issued as with the dwell mechanism.

User Interface

Graphical User Interface on a PC


The system delivers the ability to send the value that is being read from the sensor as well as issue clicks based on user-selectable thresholds. The ability to send both the value and issue switch closures on the computer leads to a number of different user interfaces that can be developed to take advantage of whatever control and abilities the person with a disability has. These interfaces could be as complex as directly selecting choices based on the incoming pressure to as simple as issuing clicks without a constantly running graphical user interface.  In addition to clicks, the system could act as pressing different keyboard buttons or joystick buttons so common programs like The Grid 2 (Sensory Software International Ltd) will be able to use its output as switch inputs.

The interface chosen for the purposes of this project shows its promise as a clicking mechanism while being used with InvoTek’s AccuPoint head tracker (InvoTek Inc.), with dwell detection disabled. A number of non-AAC users have tried and succeeded in using this set-up and have succeeded after a very short period of learning the system. While demoing the system, different people were able to walk up to the system and used the switch easily to play a game of Minesweeper. Minesweeper was chosen to demo the system because users could use AccuPoint to move the mouse cursor and issue left and right clicks (both of which are needed) using the wireless switch.

The interface was also used by a man named George whom Jakobs works with regularly through his work with InvoTek. George suffered a C4 spinal cord injury during a car accident in 1998 and has been using a sip & puff switch as the mechanism to control his wheelchair for years. He has also been using InvoTek’s AccuPoint system for head tracking as his method of using the computer since 2008. George had no problem using the switch and commented that the amount of pressure he had to use was very similar to the switch on his wheelchair after some quick manual calibration using the sliders on the user interface. He also commented that the ability to use something other than dwell as his selection method would be very helpful while he is on the internet. Websites are getting more jumbled with links and pop up advertisements so there is no telling where he may end up if he accidentally clicks. Below are two pictures of George using the device. The first is of his face and the device while using it and the second is a view over his shoulder while using the system.

George using Sip & Puff

George using Sip & Puff to control cursor click

George using Sip & Puff 3

George using the wireless sip & puff switch to issue mouse clicks on the laptop.


In conclusion, the project was able to show its usability as an access method by collecting the sensor data, transmitting it to a device connected to the computer, displaying the sensor data on the screen and doing an action based on the sensor data. This is very promising because, by making the device generic, there are a lot of different things that can be done with the information once it gets to the computer; clicking the mouse being an obvious use of the data.

This project proved to be more difficult than was originally thought when the pressure sensor was chosen. The sensor component is much larger than any other part on the board and not available in a smaller package. Because of this, the size and dimensions of the board turned into the most difficult part of the layout of the board. The students attempted to get the parts as close together as possible and decided the best chance was to add parts to the back of the board, but Barret Ewing at InvoTek was able to move parts around in a better way that allowed for all the parts to be on one side of the board. Another major problem was found when writing the microcontroller software. It proved that one must always completely set up a chip the way it is wanted since one part of the configuration was not set and the program set default ended up being the opposite of what had been expected.

Another major issue that ended up being found during the process is that not everyone will be able to apply the same pressure to the sensor. Therefore, the thresholds must be settable for each user. These values also change as the user gets more accustomed to using this type of switch. In order to address this issue, sliders were added to the user interface, under the pressure bar, to set the levels that cause switch closures. The user interface also allows the user to turn on and off clicks while still showing when the closure would occur by changing the background of the threshold red when crossed. This allows a user to find the appropriate threshold values for themselves without clicks being issued.

For this project, two switches were built to give the students a back-up. The total cost for these two systems was about $450 with about $210 being spent on the boards and $240 on the parts to place on these boards. The boards were purchased with a very short lead time and in small quantity meaning that the cost was $105 on boards per switch. In production, at least 50 would be made and the short lead time would not be necessary. This is quoted to be $700 for 50 switches bringing the board cost down to $14 per switch. The parts were ordered for 4 switches to give the students the ability to have extras for fixing problems and debugging. Therefore, the approximate cost per board is $60, but it will go down as the number of switches being produced goes up due to discounts for larger numbers of parts. An additional cost is the wireless to USB dongle. This will be about $12 for the board and $20 for the parts and packaging per switch when producing 50 at a time for a total of about $32 each. This brings the cost of materials to about $105 per switch.


We would like to thank George for trying our device as an alternative to his current clicking mechanism. We’d also like to thank Tom Jakobs and Barret Ewing from InvoTek for their help with the design. Also, Jerry Russell at InvoTek for his help placing some of the smaller parts on the board which were beyond our abilities as students. In addition, we are grateful to Arkansas Power Electronics International (APEI) for use of their equipment to place most of the parts on our boards. Lastly, thank you to our senior design advisor Dr. Magda El-Shenawee and TA Andrew Dodson for keeping us on track and making sure that we documented the process we took to get to our final design.


Freescale Semiconductor. (2009, January). MPXV7025 Data Sheet. Retrieved May 2, 2012, from Freescale Semiconductor: http://www.freescale.com/files/sensors/doc/data_sheet/MPXV7025.pdf

InvoTek Inc. (n.d.). AccuPoint. Retrieved May 2, 2012, from InvoTek Inc.: http://www.invotek.org/products/accupoint/

InvoTek Inc. (n.d.). Safe-Laser System. Retrieved May 2, 2012, from InvoTek Inc.: http://www.invotek.org/products/safe-laser-system/

Microchip. (2011, February 8). PIC16(L)F1826/27 Data Sheet. Retrieved May 2, 2012, from Microchip: http://ww1.microchip.com/downloads/en/DeviceDoc/41391D.pdf

Sensory Software International Ltd. (n.d.). The Grid 2. Retrieved May 2, 2012, from Sensory Software International Ltd: http://www.sensorysoftware.com/thegrid2.html


Powered by WordPress. Designed by WooThemes