Illustration by Dan Matutina

Personal Project: Powering a Small Wearable

In my free time, I’m working on a small battery powered project – which is essentially a small broach covered in RGB LEDs along with a microcontroller and infrared LEDs for communication. 

There are a variety of engineering challenges (like creating a reliable IR wireless protocol!) but right now I’m working on the power source. 

As the device is designed to be wearable, it needs to be light enough to be easily worn on a shirt or jacket, but last for an entire day of continuous use. 

The nucleus boils down to the following:

  1. Look awesome (IE bright and colorful)
  2. Be wearable (IE small and light)
  3. Last for 8 or more hours on a single charge/cell/set of cells
  4. Be easy to assemble
  5. Be cheap to make
  6. The power source needs to be replenishable or replaceable 
  7. … etc

This is actually revision two of this project. For now, the most important fact that I learned from revision one is that if I ignore the LED datasheet and drive them at 3.3v, I can reduce power consumption and meet my brightness needs. 

Based on those modifications and testing with the first version, I know the whole system typically draws an average of 40mA. This is what I will use for my estimates. 

In version one I had unclear product goals and little time, so without measuring power usage, I made a conservative bet for a power source – I used a 9v battery and an LDO. 

This was… unwieldy to say the least, but it worked!

It required a total of 3 components and the broach ran for 16 hours with one battery. Certainly sufficient for a rushed first version.

But sufficient isn’t good enough!

The 9v battery was convenient because it ran at a much higher voltage than I needed, so all that was required was to waste energy in the form of heat (the purpose of an LDO) until I got to the desired 5v. 

I could replace the single large 9v with a series of smaller batteries, but my power needs are very modest and that would still be volumetrically large. 

After discussing my needs with my coworkers a new plan was formulated; I would try a boost converter. 

Instead of using a single (or group) of cells at a higher voltage than I needed, I would use a single small cell and boost it up to 3.3v to run my system. 

As an added bonus I can choose a part which is designed to run with a variable input voltage that reaches far below the cells nominal voltage and drain nearly every coulomb out of the cell.

So, now to choose a battery cell. 

I immediately discarded rechargeable cells as they are generally less energy dense than their primary cell counterparts. 

Less energy density means larger or more cells, which violates nucleus item 1. 

Consulting Wikipedia’s unsurprisingly generous list of battery geometries provides many primary cell options – and many that can be immediately discarded. The following was my thought process:

  • CR2032: The classic coin cell. Browsing datasheets they looked promising! Depending on chemistry, they could be had with nearly 250 mAh of energy. But then I noticed the currents cited in their performance charts. The charts for the Energizer cell linked above are many orders of magnitude below my current draw needs. So these are out.
  • AA: Fairly large and much more capacious than I need. Could definitely meet the current draw needs though.
  • AAAA: Exotically named, it turns out these tiny cells are fairly easy to purchase as they are used for laser pointers and tablet stylus’. According to another Energizer data sheet for my expected current draw, I would see about 400mAh of capacity. Not terrible given the size but not stellar either. 
Battery size comparison, modified from Wikipedia

Battery size comparison, modified from Wikipedia

Other cell choices I deemed too large, too expensive or too hard to casually purchase. 

Freddie and I calculated that, with an estimated average cell voltage of about 1.15v under load and over the full discharge cycle, the actual current draw on the Vin side of a particular (a Skyworks AAT1217) boost converter would be about 140mA. 

Based on all the above it looks a single AAA (up to 1250mAh for 8.9 hours of runtime) best fits my size, cost, and availability needs. 

Expect this post to be updated with real power numbers from an Othermill’ed test board soon.