My initial thought on this was “Hey, you could do this with a 555 timer!” Seriously–tie the reset/enable pin to a pressure sensor or photodiode, tie the output to one LED or the other, and have a capacitor and very-high-resistance resistor on the threshold/discharge pin.

@Mohonri Stop trolling please.

Very nice hack btw!

Mohontri’s comment may appear as a subtle troll since anybody who would recommend a 555 for a three day period (259,200 seconds) clearly lacks experience using the 555.

When designing your RC network, the minimum value of R should be about 1k to prevent too much current flowing into the 555. The maximum value of R should be about 1M so that enough current can flow into the input of the 555 while there is also current to allow for the electrolytic capacitor’s leakage current.

The minimum value of C should be around 100pF to avoid the timing equation being too inaccurate. The maximum value of C should be about 1000uF as any larger value of capacitor will discharge too much current through the 555.

These maximum and minimum values give a minimum period of around 0.1 us and a maximum period of 1000s.

Saying throw high values of R or C at it will not work. Not for a three day timer. And it won’t have any accuracy. I can see the advantages of using an RTC; the timer will be accurate and it can even fire at a predetermined time.

Personally, if I were using a microcontroller, I’d choose something a lot cheaper (like an ATtiny13A, which should be up to the task of bit-banging a couple of i2c commands every so often. Alternatively, use a 555 feeding into a couple of counter chips. A denary example: a 555 with a period of 1000s feeds into a decade counter. The 10th output of the counter feeds into the input of another decade counter. Which feeds into another decade counter. You’re now counting hundreds, tens and units. Throw in a three input and gate and you can count to 259 and you have a three day counter without using a microcontroller. Make an astable multivibrator from a pair of transistors and use a switch with a thyristor in series to switch a transistor powering the multivibrator. Simple!
That microcontroller based solution is looking pretty attractive right now, isn’t it…

I love 555s as much as the next guy (and I love working out how to make stuff with discrete logic, too!) but just give in sometimes. If someone can’t program an ATtiny chip, and Arduino will work wonders for them.

Bonus solution: have a read of Maxim’s DS1307 RTC datasheet. Once programmed (through i2c – a piece of cake with an Arduino), you can enable a square wave output. There’s plenty of things you could do with that (and it’s temperature compensated – it will keep better time than a 555).

For someone who takes just a few pils once or twice a day, I give you kudos…. For the person who takes a figuritive pharmacy upwards of five times a day, a more elaborate solution will be required.

msc: Seconded — it’s especially not trolling when, per Pedro’s comment, a 555 wouldn’t work in this application anyway.

This is an awesome hack! The only thing I might change about it would be to make the section with the pill-bottle holes a separate insert that could be replaced. That way, if you need to remind yourself about a different medication, you can not only reprogram the AVR to a different timing, you could also cut out a new insert with holes precisely sized to fit the bottles you’re using.

i just wanna hug you guys.

Why was the RTC needed at all? I’m working on building a clock, and I am just planning on using an arduino… really, just the microcontroller from the arduino. I thought I would need an RTC at the beginning, but I didn’t find anything that couldn’t be done with just the arduino. Is there a reason to have the RTC?

“these particular pills had to be taken every three days, and he wanted a solution that required “zero effort.” Although one might question whether his solution actually took this amount of effort,”

I think by zero effort, he meant remembering to take the pills requires zero effort.

@LeeRay
The regular Arduino clock (16 MHz ceramic resonator) is not precise enough: it’s rated for 0.5% or about 7 minutes of drift a day. You would have to go with some sort of calibration, and even then, as conditions change, it would not hold time very precisely.

There seems to be a way to hook up a 32KHz quartz to an Atmega, but I never tried. It’s much easier to use a RTC to obtain reasonable precision. For instance, the DS1307 drifts by about 1 minute a month.

If you are (crazy) serious about building a clock, the precise-to-the-microsecond way to go is, IMHO, to use a GPS receiver over a serial connection along with the heartbeat line of the receiver.

(and maybe I’ll post something when I’m done, now that HAD seems to have cleaned their act)

So cool !!