Wow, is it middle of February already? Time flies.

Quite frequently, I've been asked "So, how much lower power is Bluetooth low energy compared to Bluetooth?". Although it may sound like a cop-out answer, I'll usually answer "It depends".

The reason I give this answer is whenever you look at power consumption, you need to have a use scenario in mind. How much data is being sent? How often is data sent?

Bluetooth low energy was designed for infrequent communication of small amounts of data. This means that it has low packet overhead, it is designed to keep the radio on for the minimum time possible, and the some of the more power-hungry PHY parameters of Bluetooth have been relaxed. Under those circumstances, it provides substantially lower power consumption than Bluetooth, as low as 1/10th or even better. That does not mean BLE will be lower power for all applications. For instance, if you were trying to stream high-speed data (audio), BLE could be higher power than Bluetooth, because Bluetooth could send much longer packets and therefore have lower packet overhead. Also, with modern Bluetooth chips, you could use EDR, boosting the data rate to 3 Mbps and in this way have your radio on for a much shorter time than you would if you could only do 1 Mbps.

It works the other way as well. I once heard a competitor bragging about their very low energy/bit number. This means that they spend very little energy to send one bit. That sounds good, right? Except it doesn't say anything about packet overhead, start-up times or how much power you spend keeping your link alive. WiFi 802.11n has an extremely low energy/bit figure. This means it is very efficient at moving large amounts of data, which coincidentially is what it was designed to do (wireless Ethernet). However, it is very inefficient at moving small amounts of data as it has a high packet overhead, it takes a long time to open a link, has a high peak current which limits what kinds of batteries can be used and it does not have a good power-off mode. So it would not be the right choice a system that only wants to transmit a little bit of data every now and then.

Another fallacy is over-simplification. I've seen people compare a single parameter in two datasheets and then think that this tells the whole story about power. It doesn't. To make a relevant comparison, you need to put together a model that takes the usage pattern into account. This doesn't need to be very complex, but you need to consider how much time is spend in various states and how much current is drawn in every state. Don't just look at the radio; you need to look at the MCU as well as any sensors or other circuitry included as well (LEDs can consume quite a bit of current, for example). Based on this, you can calculate an average current consumption, which together with the rated capacity of your chosen battery will give you an idea of what kind of battery life you can expect. Spreadsheets are ideal for making these types of calculations. We are currently making a spreadsheet model of this type for BLE, which we will be making available as we get closer to launch of the CC2540. It is very instructive; by playing around with the parameters, you can easily see that the slow clock accuracy has a huge impact on the average current consumed by a BLE connection, for example.

Some people think 1W makes a processor low-power. Others measure their power consumption in uW. "Low power" means nothing if it is not put into context.

Best regards,

Karl