Raspberry Pi 3 Model A+ Unboxing

Sizing such a supercap would be very difficult: So you measure for a working day, find that it uses less than 1.5W all day and decide to use a 1.5W powersupply. So you add you supercap and set it running.

Then... the next day you find it crashed during the night because it rotated the logfiles and compressed them. So while compressing it used over 2W for ten second.

So, when you have that figured out, you add a bigger capacitor and set it running again. Now it runs nicely for a while until say some monthly job runs that uses the "2W" for much longer than just the compress-the-normal-logfile. So again it crashes.

It is really difficult to find an appropriate value that will always work. But the 1.25W "normal" and just over 2.5W peak numbers are so low that a mains power supply will probably be an 2.5 or 5W USB charger. That will give you some headroom.

Hi Shabaz,

That is really interesting and I'm sure you are the first to produce such a plot. Following on from some recent discussions on E14 community, perhaps a super capacitor could be utilised to enable the Pi to operate on a much lower VA power supply; supplying the extra energy during WAN activity and intensive processing. The risk would be that, without any form of closed loop energy monitoring/throttling back the ARM activity, the capacitor becomes exhausted, the light-weight PSU cannot provide the demanded current, Vout drops effectively turning the Pi off without a proper shut down (possibly damaging the OS image).

On reflection, probably best to just use the recommended PSU

Rod

Here's the current consumption at power-up.. the scenario was that the Pi 3 A+ was connected via HDMI, and a Logitech Unifying receiver (used for wireless keyboard and mouse) was plugged into the single USB - hopefully that should be representative of how the Pi may be connected for other users too.

To do the test, power (5V) was applied to the pins on the 40-way connector instead of the USB connector.

The Logitech receiver consumes 25mA, so if it is not plugged in, then 25mA can be subtracted from the values in the chart. If I do not plug in the HDMI connector, then a further 16mA can be removed.

Note that as soon as there is WLAN activity, or any significant processing of course, then the current will spike up as expected. The spikes can be very brief for the wireless, so the connected power supply needs to be able to support that. Here's some typical behaviour over about 10 seconds, when the system is doing nothing.. there is the occasional spike due to brief processing by the OS, network and so on.

I didn't run any load tests, but it is normal under load for the current consumption to significantly rise (perhaps 1.5A in extreme cases - just a guess based on old Pi 3 model B behaviour) depending on exactly what is running.

Hi Allen,

I'll take some measurements tonight hopefully, I'm guessing it could be maybe close to 3B levels (but slightly less since there is no USB hub/Ethernet chip). I need to locate a microSD card.. always forget to order those! : )

Hi @shabaz. I agree this model is a nice addition to the rPi line.  The quad processor makes for good performance.  But it’s still a cost vs. performance decision.  In the US, Microcenter sells the Pi Zero for as low as $5.  Great for IoT projects, with the lower power requirements, with slow processing speed.  And when you want to hide the IoT project in a small space the Pi Zero is impressive.  The microcenter.com web site says they won’t have the model 3 A+ until November 28, at a price of $25.  I didn’t hear a power usage mentioned in the video above.  The text says 2.5 amp power supply, but what's the actual amps used.  I’ll look around for that information, the model 3 B+ is power hungry.

Decisions, decisions . . .

Same performance but by loosing some memory and a few USB connectors they have brought the price down considerably - what a great addition to their range.

Rod

Nice!

I had not realized that the new 3 Model B+ and now the 3 Model A+ have dual-band WLAN... so I'm glad you mentioned it in the video!!

Also it supports 802.11ac which is highly attractive, I saw some test online that shows the new models can have at least twice the throughput compared to the older 3B.

However it could be possible to go higher (I don't know what speed the SPI interface to the Cypress/Broadcom chip gets to), a lot of the tests were not clear on what the access point was capable of.

I think for all-round value for money, and looking at price-performance, based on the specs, this renders the Pi Zero W almost obsolete.