ART TEC - Guy Marsden
Product Design My Artwork Living Sustainably
Art Engineering Levitation Kit Solar Power Chevy Volt
About Me Blog Solar Heating Thermal Windows
Contact Twitter Solar Hot Water Solar Mower
home > Engineering Prototypes > Sample Design Process
A short version of the story below was reproduced in an article in the January/February 2003 
issue of Inventor's Digest Magazine
Click here to download a printable version of the article.


After signing a non-disclosure agreement (now moot, see bottom of this page) and faxing it back to me, a private inventor asked me to build a pill timer prototype that would be incorporated into the cap of a prescription bottle.  It would beep to remind people to take their medication and re-set when the cap was replaced on the bottle.   I provided him with an estimated price to design a working model.  This was not a fixed bid due to the R&D nature of this design (I normally try to provide a fixed bid when the project falls within my design experience).  After some discussion I suggested that I make 2 prototypes, one that would work in actual hours, and a "demo" unit that would simulate hours by using minutes instead.  This would make it easier to demonstrate the alarm for a sales pitch. 

My first step was to visit a Pharmacy to obtain some bottles and caps.  I acquired the largest size available to ensure that I would have room for the electronics.  I determined that the smallest off-the-shelf LCD was a 2 digit unit with 1/2" characters.  The ideal display would have a regular clock display with 4 digits and a colon (12:25) but I could not locate a standard component that would fit inside the cap.  I asked my client if he thought that 2 digits representing hours and not minutes would be OK.  He agreed that it would.  I then researched batteries, battery holders, alarm speakers and the electronic components needed to drive the LCD and beeper.

After designing the electronic schematic, I designed a PC board on a CAD program.  The circuit board holds all the electronic parts and creates all the electrical connections between them.  I sent this design out to be fabricated, and e-mailed a copy of the image above to my client to show my progress.


I used 2 boards, one stacked under the other.  The lower one supports the battery holder.  I cut out the 2 boards and soldered all the components to them and I emailed a picture of the board to my client.


Then I started on the firmware.  This is the code that goes into the microcontroller chip.  A microcontroller is a tiny computer chip that is designed exclusively to control things (like the LCD and beeper).  The one I used measures less than 1/4" square and runs at 4 MHz with 2K of program memory.  The code includes a clock timer, a way to respond to the switch that senses when the cap is on or off the bottle, and a set-time routine that responds to a button inside the cap.

Here is the image that I e-mailed to my client showing the working LCD.  The ruler is there for visual scale reference.  The clip leads on the right connect to an external microcontroller that I can easily re-program without having to remove and re-solder the chip on the circuit board.

I bolted the display circuit board to the top of a cap with the battery board under and inside the cap.

The large square chip at the left is the LCD driver, the smallest black chip under the purple wire is the microcontroller.  To the right of that is the sounder or beeper device.  The LCD sits above the board and plugs into the 2 rows of sockets along the top and bottom of the circuit board.

At this point I had to revise my estimate, due to the extra time that the cap fabrication and code development took, and my client agreed to the increase before I proceeded with final cap fabrication and assembly of the second unit.  I then took a second cap and cut a hole in it for the LCD and glued it over the top circuit board to cover the electronics:

Note the label that indicates that this is the "demo mode" prototype that runs in minutes rather than hours.


I ran into a small problem when the switch that senses the cap on the bottle failed to work reliably when the cap was turned past the safety click.  The cap would spring back up off the switch when it was turned past the safety click.  I solved this by gluing small pieces of plastic to the bottle to prevent the cap from clicking into a locked position.  My client approved this change for the prototype, allowing that the production unit would have the safety click feature, and this prototype did not really need to demonstrate that feature.

My next step was to run a battery life test.  My prototype circuit draws more power than a fully optimized production design and I was concerned that the battery might not run for more than a day or so.  Below is a data log graph that I sent to my client showing the battery voltage over time :

I was surprised to find that the battery only lasted 4 hours!  I re-wrote the code to save power by not flashing the period on the left side of the display -- allowing the chips to "sleep" 99% of the time thus saving power.   I had put the flashing period in to have some indication that timing was in progress, but my client and I agreed that it was not needed.  The battery lasted over a week with the revised code, but I decided to try using 2 "N" type batteries (short versions of the AA battery) to give my client a longer lived prototype for the "real time" model.



The next graph shows a much better power curve for these batteries, they ran for over 3 weeks:

Note that the graph does not show dips from the alarm as before.  The alarm circuit draws a fair amount of power, however these larger batteries can handle lit.  In a production design the power consumption could be optimized to run for over a year.

I e-mailed this image to my client to show both working models, and he sent me the final payment, I then sent them to him.



Several weeks after I had shipped the prototypes to my satisfied customer, my wife was in the Pharmacy and discovered that a Pill Timer was already on the market!  I bought one (for $9.95) and immediately e-mailed an image of it to my client!  I was quite surprised that he had not done a patent search!  This unit was patented in 1993 (patent number 5,233,571) and was being manufactured by The Pill Timer company.  I looked up the patent and mailed a copy of the front page and the product to my client to ensure that he was aware of it. 

Here is the manufactured item:

On the one hand I was rather pleased that my prototype closely resembles a manufactured item, however this serves as a caution to all inventors -
DO YOUR RESEARCH!   A patent search could have saved my client several thousand dollars.
It is not my job to ensure that your idea is original and not patented, I am an engineer not a patent attorney!

Another interesting note:
I spoke with the owner of the Pill Timer company about the design and development of his product and he said that one of the most significant challenges of the design was battery life.  In fact the product almost went to market with faulty code that would have drained the battery in a few days!

A short version of the story above was reproduced in an article in the January/February 2003 issue of Inventor's Digest Magazine Click here to download a printable version of the article.

back to top of page