Computing stuff tied to the physical world

It’s all about the startup “hump”

One way to test the latest MPS design, is to feed it a minimal amount of energy, i.e. to lower the AC current to a point where the CT barely delivers enough energy to get the µC going:


As you can see, it takes a lot more time for the energy harvesting to reach an acceptable level (almost 0.5 second), but the switch-on behaviour remains well-defined and snappy.

If you look closely, you can see the brief voltage dip in the purple Vdd trace, just after it has risen to about 2V. That’s the µC coming out of reset, doing its thing, and entering deep power-down mode right away. As before, the proper operation of the µC can be seen (just barely) through the brief dimming of the red LED once every 3 seconds.

But starting up is only half the story. We also need to verify that the circuit behaves in a well-defined manner when energy runs out.

The following capture shows what happens when AC mains power is suddenly turned off:


After about a second, the Vres and Vdd levels have dropped to the point where there is no hope to get going again. The NPN transistor wants to turn off the circuit, as can be seen by its collector voltage rising and even nearly reaching Vres (in yellow, CH1), but by now there is simply no voltage left to keep that NPN transistor going, let alone the P-MOSFET.

At this point the whole circuit switches off, Vdd collapses (purple, CH3), and Vres gradually drops further, due to leakage from the various components in the circuit.

All the voltage levels are now well below the point where startup happens. Once power comes back, the sequence will start from scratch – with the same voltage levels as before.

Again, looking closely at the trace, you can see the last µC wakeup “blip” on the far left.

So there you have it: life and death of an MPS, exactly as we need it. Success at last!

[Back to article index]