Tuesday 22 June 2010

Low Power

Seiko Instruments Inc. S-882Z Ultra-Low Voltage Charge Pump

Working at the extremes of very low power provides a greater understanding and appreciation of all aspects of this type of pulse motor design. Our definitive benchmark in low power consumption is based on the circuit on the right.

These circuits place no reliance on semiconductors; other than the solar panel. The trigger is a simple reed switch which requires no power to operate other than an infinitesimally small force to close.

Solar cell specifications look good on paper – 0.65V at 4mA. But indoors, under ‘normal’ lighting it’s more like 0.2V at 100µA ! Pointing a 60W light bulb at it produces some improvement – only in full sunlight is the promised power delivered !

Using a magnetically levitated rotor, the first circuit worked but it required the close presence of a 60W light source; maybe only to be expected with barely 0.3V at 1.5mA available. The rotor ‘chugged’ along, hesitantly, at a sluggish 100 RPM.

Adding the 100µF capacitor changed the performance dramatically. The rotor sprang into life accelerating up to a healthy 2,000 RPM. Even with the 60W light source removed, it continued to produce around 200 RPM !

Why did the capacitor produce such a performance improvement ? Nope – not ‘radiant energy’ ! The answer is that it created an energy reservoir for the solar cell. For 80% of the time the solar cell’s power was not being used. Only when the reed switch closed was there a demand for power – and there was hardly any to be had from the tiny solar cell. But, by adding the capacitor, it allowed a charge to accumulate until the next trigger.

(This is technique called ‘burst’ supply – power is delivered in pulses rather than continuous; we use this approach in our designs to cater for situations when power availability becomes very low.)

With a lot of effort to optimize the reed switch location, we successfully got the motor to run from the lowest power levels yet achieved – the motor operated at 200 RPM using 0.15V at 500µA (75µW).

Silicon based semiconductors do not work at such voltage levels. They need a minimum of 0.65V to operate. No work has been done using germanium devices. They are considered to be archaic and are becoming obsolete; but they work down to 0.25V.

The reed switch is hard to beat. They’re cheap and you can replicate similar motors for the cost of a reed switch, a bobbin of copper wire, a few magnets and a ‘borrowed’ solar panel using a garden solar lamp – or just use an old 1.5V battery.

In our implementation, using a 1.5V battery, we reached speeds where magnets were being thrown off the rotor !

To date, we’ve not bettered the reed switch performance – we did toy with the idea of forgetting all the electronic design; to go with a product that used something that was so simple. But there are a few problems with reed switches. To get the extremes of performance, the switch position and rotor geometry was critical. The other point is that a reed switch ultimately will fail through metal fatigue. The best estimates are 109 operations before failure – this sounds a lot but for a rotor using 2 magnets, spinning at 10,000 RPM 24/7 means that the reed switch might fail in less than a year.

Reed switches have a much reduced lifespan when high currents are involved. For those, making ‘bigger’ motors, this can be remedied using a power transistor that is driven by the reed switch.



Our test results from the above circuits are detailed below. We used three multimeters, an old analog AVO 8, Fluke 85 and a very cheap digital meter whose origin is unknown - model no. N72CG.

These results show the problems associated with measurement of pulse driven systems. Using solar cells exaggerates the problem - unless drawing a relatively high current, battery voltage will remain constant - a solar cell does not !

Solar Cell O/C Voltage S/C Current
AVO 8 0.35 1.35mA
Fluke 85 0.35 1.38mA
N72CG 0.35 1.41mA

  Motor (no capacitor) Motor (100µF capacitor)
Volts Current Volts Current
AVO 8 0.15 1.10mA 0.15 1.00mA
Fluke 85 0.58 1.85mA 0.15 1.00mA
N72CG No reading No reading 0.24 3.35mA

From examination of the above test results, it is clear that the AVO 8 provides the most consistant figures. We presume this to be due to the averaging effect of the meter ballistics. Both digital meters appear to display the peak transient levels rather than an average value. (In fact the cheap meter could not cope with the pulse nature of the first motor and displayed random numbers !)

If we had extrapolated these measurements, it is clear that we could be led to believe and substantiate evidence to show that `extra´ energy was being created.

Again we highlight the danger of interpretation of measurements from test instruments. In our experience, digital meters tend to measure peak transients ... and that then leads to conclusions that make no sense ... and permits us all to enter the strange world of `free energy´ !



(We had intended to discuss ‘Energy Harvesting’ and DC-DC Charge Pumps but that subject will be covered later.)

Visitors