Read our ISSCC 2008 coverage of the San Francisco conference
Belgian research organisation IMEC has described a thermo-electric power conditioner consuming only 1.4µA at ISSCC, aimed at tiny radio sensor nodes requiring between 50 and 100µW.
The boost DC-DC converter in the maximum power point tracker (see box) is a switched capacitor design, required because the Q of integrated inductors, according to IMEC, is too poor for this application.
The eight-stage converter uses 175pF metal-insulator-metal capacitors to reduce loss into the substrate - three per cent compared with up to 20 per cent for poly-poly capacitors, said IMEC. The active area of the chip is 3.01mm2.
DC-DC efficiencies are 82, 77, 74 and 73 per cent for one, two, thee and four stages respectively. A control circuit (see diagram) selects the number of stages needed to deliver peak power.
Clocked at 60kHz, the control circuit measures current delivered to the output buffer capacitor - an analogue of output power - every half a second. During the measurement, output power is detected with: the existing number of stages, one less stage, then one more stage - with a 260µs delay between each for settling.
If no change in operating point is required, another 0.5s delay is initiated. If a change is needed, it is done immediately and another measurement initiated. With a small 100kO impedance thermoelectric generator, the whole tracker offers up to 58 per cent efficiency.
Losses include 109nA tail current in the linear regulator, 491nA in the analogue current sensor and comparators, and 800nA for the digital state-machine and clock.
Like solar cells, thermoelectric generators have an output characteristic that depends on environmental energy.
For any particular input energy there is only one load impedance that extracts maximum power from this energy, and no amount of conventional feedback can keep a load at this point.
Instead, 'maximum power point tracking' is used, controlling a variable operating point DC-DC converter that makes the real load look like the ideal load.
The controller finds the best converter operating point by regularly moving the existing operating point slightly up and down. It takes note which direction of dither produces most power at the load, then resets the operating point in this direction.
By looping around through this procedure, the operating point is nudged to track the ideal operating point.