Power Supply Design
 
You might think power supplies are the most boring thing in the world, and you might be right. But, still, it’s got to be done and it has a major effect on the size of the watch.
 
The basic design problem is that the voltage of the battery is not the same as the requirements of the things it has to power. We need to change the voltage, and we need to do it efficiently.
 
Here are the basic requirements of the parts I’ve selected:
 
  1. Battery: provides nominally 3.6v, but could be as high as 4.24v when charging.
  2. Microprocessor: Requires 1.8 - 3.6v, at about 1 mA or less
  3. Display requirement #1: 2.4-3.5v at 0.2 mA
  4. Display requirement #2: 12-13v at about 0.75 mA (playing pong) or up to 40 mA (with a fully lit display - if I decide to make the watch do something other than pong)
 
If the battery was taken out of the watch to charge, I could probably power the microprocessor and the display by connecting it directly. But, the battery won’t be easily removable, so I want the watch to run while it charges.  We’ll use a simple regulator that basically wastes some of the power to slightly drop the voltage down to a nice round 3.0v.
 
The display is interesting because it is a merging of two types of technology:
  1. •The controller chip is basically a fancy memory chip with a circuit that automatically reads it out and drives the rows and columns of the display. Like most memory chips, this part was designed to run off of a low voltage  -- any more than 3.6 volts would just waste power and require a physically bigger (and therefore more expensive) chip.
  2. •The display is made up of a totally different (and relatively new) technology - a light-emitting OLED film.  Traditional LEDs required only 3 volts to light up, but it’s very difficult to build these into a tiny dot-matrix display. OLEDs are easy to build into tiny displays, but the chemists who developed have not been able to get them to light at battery-friendly low voltages. We’ve got to live with that limitation for now, and maybe it’ll become easier as the technology matures.
 
So, how do we generate 12 volts with only a 3.7 volt battery?
 
The obvious solution is more batteries. Add 3 in series and we’ll have 11 volts -- but, that isn’t quite the 12-13 volt range we need.  We could use 4 batteries to generate 15 volts, and then use a regulator (like on the 3.0v supply) to drop it down to 12. That would work, but 4 batteries will take quite a lot of space. The smaller batteries get, the less energy-per-volume they have, so this pack wouldn’t last nearly as long as one big 3.7v battery of the same size. There are a bunch of other issues, too, like “what happens if one battery goes dead before another” and “how do we charge them evenly”, so let’s try another tact: some circuitry magic.
 
Generally speaking, there are 4 basic kinds of components in electronics:
 
  1. •Resistors - waste power (waste isn’t always a bad thing!)
  2. •Capacitors - temporarily hold energy in an electric field
  3. •Inductors - temporarily hold energy in a magnetic field
  4. •Semiconductors - are basically resistors that can be controlled. They generally control control the flow of energy by being low-resistance at some times (letting current flow) and being high-resistance at other times (stopping flow). There are a lot of devices in this big category.
 
Resistors and semiconductors by themselves can’t really generate more voltage by themselves.  But, when used with capacitors and inductors, you can build a circuit to do this.
 
Power supplies can get pretty complicated, but we’ll only consider two types.  One is based on capacitors, and the other is based on inductors. This makes sense, because capacitors and inductors are like yin and yang..
 
--  --  --
 
A voltage-doubling circuit using capacitors switches between two states. It’s very similar to using more batteries -- so much so that I’ve used batteries in the illustrations below.
 
First, the two smaller capacitors are charged in parallel. They will reach the input voltage of 3 volts.
 
Then, these capacitors are disconnected from the input, and then connected in series to another output capacitor. They charge this capacitor to 6 volts (minus some losses).
 
The third state is really the same as the first, except you can see that the big capacitor supplies the power to the output for the time which the smaller capacitors are charging.
 
Hopefully that’s easy enough to understand - you could even recreate this experiment with rechargeable batteries (I wouldn’t recommend it, but you’d have to make sure you use the right voltage batteries).
 
So, capacitors are nice. But, there are some reasons why this circuit would be hard to build in to a watch.
 
First off, this circuit only doubles the voltage -- we need much more voltage. Assume that 3.7v gets “doubled” (after losses) to 6v. Six volts isn’t enough, so we’d need to double it again, but losses would probably make it only 10.6 volts. So, we’d need a third stage or do something funky (maybe charge 3 capacitors in parallel) to get to 12 volts. Three stages is 9 capacitors, and we’ve got to wire them all up with some switching circuits that I haven’t shown. Physically, this would take a lot of space. Plus, those losses that I mentioned -- those affect battery life!
--  --  --
 
So, the second type of voltage-increaser is called a Boost Converter and it uses an inductor instead of the two small capacitors.  The neat aspect of this circuit is that it can provide 2x, 3x, or 4 times or more voltage without changing the layout of the circuit or adding stages.
 
The wikipedia article is good, and it’s harder to explain this type of circuit, so I’ll just give a rough description.
 
Inductors conduct current, and they don’t like when the current through them changes. In response, they generate a voltage that tries to counteract the current change. It’s not that they really have a personality with intentions and all that, but... Anyway, like an electric eel, we can agitate an inductor (by changing the amount of current through them often) to generate lots of voltage. The voltage will only last a split second, but, like above, we can store it in a capacitor temporarily and agitate often.
 
(circuit from tabtronics)
 
Like the capacitor voltage doubler, this circuit uses two cycles:
 
During the first half of the cycle, the transistor switch (“Q”) is closed so that it pulls current through the inductor (“L”). This energy doesn’t actually get wasted -- it is stored in the inductor’s magnetic field.
 
The second stage of the cycle is where the action is. The transistor turns off, so current no longer flows through it. The voltage on the inductor becomes as high as necessary to conduct current - if the transistor switch is abrupt enough, this could be enough voltage to generate a spark! (that’s how a basic spark plug works in a car).  In this circuit, the diode (“D”) turns on and the high voltage from the coil charges the capacitor (“C”). The load then draws from the capacitor, just like in the voltage-doubler circuit.
 
--  --  --
 
So, I’m going to use an inductive circuit to boost the battery voltage to 12 volts need by the display.  There are lots of IC’s that automatically control the timing of the switch, and the one I’m planning on using even includes the transistor switch and the diode built-in.
 
There are lots of factors that go in to designing this circuit, but important is the physical size (inductors are hard to miniaturize) and the efficiency of the circuit (so that the battery will last long).
 
So, I’m off to order parts and build a prototype of the power supply.  Here’s a teaser for what I hope to start in two-three weeks.
 
 
Saturday, September 23, 2006