Mind the Raspberry Pi power usage

The RaspberryPi is a powerfull little gadget that can add a lot of good features to your robot. For under $60, including cables, dongle and SD-card, you get a lot of processing power, data storage, wifi and audio. So I ordered a RaspberryPi to be used as the brain for my R2D2 project.

The Raspberry needs a regulated 5V to run and it has a 3.3V linear regulator for the Arm processor and some other parts. I guess the 5V is mainly for the USB. As I plan to power R2D2 with 8 x 1.2V NiMH batteries, I will need to bring 9.6V down to 5V. When the batteries are fully charged, they will most likely provide allmost 11V.

raspberries.png

Raspberry model B (left) vs. model A (right). note the missing ethernet/USB chip on model A

Using my newly acquired adjustable lab power supply I set up a few tests to help me figure out how to power R2D2.

First I used a regular LM2940CT-05 volt regulator. A rather efficient linear regulator much like the 7805. The power supply is set to about 10V and the multimeter (or measuring thingy) is showing the current flowing to the Raspberry. The multimeter is set to 10Amps so 0.49=>490mA The RaspberryPi model B was fitted with a wifi-dongle and an SD-card with the OS and booted up.  

Raspi_B_linear.jpg

As you can see; just under 500mA constantly being drawn from the 10V power supply and about 480mA going at 5V to the raspberry. The regulator quickly became very hot.

The second test was with a Recom R-78B5.0-1.0 switching DC-DC regulator using the same Raspberry setup. The meter on the power supply shows only half the load it had before. Because the voltage is cut in half from 10V to 5V, the 250mA is about as good as can be expected.

Raspi_B_switched.jpg

Other articles about the raspberry show that the Ethernet/USB chip is a major power consumer and gets hot. So lets see if removing that part will make a difference.

Running the same test, but this time with the Raspberry model A, which has no ethernet port and USB hub, shows another drop in power consumption.

Raspberry_A_switched.jpg

This time only 120mA is being drawn from the 10V supply at 220mA going to the raspberry. The 380mA difference between the first and last test is huge as it represents possibly hours of battery life. 

The money I spent on the switching regulator is just about the same as the price difference between the Raspberry model A and model B. So if you are considering using a RaspberryPi in your battery powered robot project, you know where to spend those $10.