4 MIN READ

Graceful shutdown

Battery-powered and battery-backed systems are now so commonplace that even to remark on them feels incredibly old fashioned. We are all so used to charging our mobile phones, electric toothbrushes, game controllers and Bluetooth headsets that it has become as automatic as hanging up our car keys. And it comes as a huge surprise if we can’t find the keys, or our phone has no charge. So much effort has been put into battery technology, both in terms of capacity and consumption, that battery life is now so good that it is rare indeed to find our phone flat.

And here we find a largely unconsidered corner of the battery life design equation – the graceful failure. What is actually supposed to happen when the power source can no longer support its host device?

For the electronic engineer this is a classic conundrum. Just at the time when energy becomes limited, we would like to warn the user of this – but we don’t want to consume more energy doing so! An LED doesn’t take much power, and is actually more likely to be noticed if it’s flashing (thus consuming no power for some of the time); but it’s no use if the device is in your pocket, or another room. Sound is good, but the user is going to get annoyed if woken at 3am with an alert that the toothbrush could do with a charge.

Even in the absence of a good warning strategy, there remains the problem of optimal shutdown. For your phone there’s no major problem. Provided you don’t lose all your contact numbers you’re unlikely to worry too much and after all, in the end it’s your fault the battery went flat. But for a complex medical system other choices could be made; by shutting down a power-hungry heater, for example, it may be possible to keep a vital oxygen-level control system running for a few precious extra minutes. At the very least, one needs enough warning to be able to save the system state so that operation can resume cleanly once power is restored. This requires input from the whole design team. Amongst the myriad of other considerations battery shutdown can easily get short shrift, along with adequate system cooling and labelling. Yet ignoring it can have enormous consequences, and leaving it until later will often limit the scope of what can be done, making it difficult or even impossible to then change a sub-system to a less power-hungry one, or even just to one that can be shut down on demand.

Even a sophisticated charge-counting system, which monitors current into and out of the battery, can come unstuck.

None of this is made any easier by the difficulties involved in monitoring battery charge state. Simply monitoring battery voltage is rarely sufficient, as it will vary with temperature and load. Even a sophisticated charge-counting system, which monitors current into and out of the battery, can come unstuck. At Team we have direct experience of this when using an ‘intelligent’ battery module. This came complete with charge counting, status communication and automated charge termination, but could suddenly go from 40% charge to 0%, shutting the system down without warning. The reason? It took several weeks to determine, but it turned out that the system would fail if put on charge straight after being brought into a building. The charge termination would sense the fast temperature rise, assume that it was due to the battery reaching full charge, and not only stop charging but set the status to indicate a fully-charged state – even if moments before it had indicated 30%!
 
So when designing any device or system reliant on battery power remember these key points:

  • As part of the design process, include a thorough review of which sub-systems might be shut down early as power levels fall
  • Involve the whole design team
  • Make sure you fully understand the charging and discharging regimes – and how these will work in the real world
  • Don’t leave it too late in the design process!

This article was taken from issue 4 of Insight magazine. Get your free copy of the latest issue here.

 

Like it