MOX idle consumption

Hi everybody,
I was interested in MOX consumption, especially the idle consumption because that is what my router is doing ~95% of the time. I found no numbers, so I measured it myself and I am sharing the results for the benefit of Turris community:

A: 2.6W - microcomputer, the version with 1GB RAM
B: 0.0W - mPCIe slot, without mPCIe card
C: 0.5W - 4xLAN, each connected LAN appends roughly 0.1W
E: 1.2W - 8xLAN, each connected LAN appends roughly 0.1W
F: 1.1W - 4xUSB, no connected devices
G: 1.1W - mPCIe slot, without mPCIe card

For various combinations of modules, just sum their consumptions. A+E+C was the only combination I found, deviating a little bit, resulting in 4.5W instead of predicted 4.3W. Additional 0.2W is probably taken by E module that needs to establish high-speed data link with module C. All measurements was made with fully booted TurrisOS with reset to factory defaults and disabled updates, to make sure it is idle.

Consumption of add-ons:

WiFi SDIO

  • disabled - 0.1W
  • 5GHz network - 0.8W
  • 2.4GHz network - 0.5-0.6W

WiFi mPCIe - WLE900VX

  • disabled - 0.4W
  • 5GHz network - 1.6W
  • 2.4GHz network - 1.2-1.3W

SATA mPCIe card - Delock MiniPCIe 2xSATA 6Gb/s

  • no disk connected - 0.9W
  • 1 disk connected - 2.1W (0.9W card + 1.2W spinning disk)

Disk Seagate BarraCuda 5TB (ST5000LM000) + ICY BOX USB/SATA adapter (IB-AC6031-U3)

  • spinning disk - 1.6W
  • stopped disk - 0.7W
  • spinning disk itself (without adapter consumption) - 1.2W
  • various USB/SATA adapters might have different consumptions

Ethernet measurements:
I already wrote that connected LAN port (no traffic) takes about 0.1W (I do not have means to measure it more precisely). I was using short LAN cable. However, I found huge variations of measured values. It seems that 0.1W is the value for high quality devices, like Dell laptops, connected to Turris. Other devices, surprisingly including old HP Elitebook made it 0.6W when connected to the port. The same for ethernet port on my Gigabyte motherboard in my desktop computer. A Dell docking station when connected to Turris LAN port, was consuming 0.6W until I switched on my laptop. I am suspicious that there is a kind of power-save mode for ethernet link and not all devices use it properly. So, each LAN cable plugged into your Turris might cost 0.1 or even 0.6W.

Finally, one warning: No guarantee that the consumption will not vary between hardware revisions and software updates. The numbers are provided ā€œas isā€, just as I measured them, mostly January 2022.

In conclusion, if you fight for low consumption to save your money and the nature, A+B + WiFi mPCIe will give you 4.2W with 5GHz WiFi and all the power of Turris. Additional 0.7W for sleeping disk drive will turn your Turris into NAS provided that you will not make it spinning most of the time. Ethernet is good, but with C module it might push you up by 0.5+(4x0.6), e.g. 0.9 to 2.9W. SDIO WiFi seems more power-wise, but you can find info about some difficulties with it on this forum.

Edit: Typos.

13 Likes

thank you for this great job!

1GB :wink:

1 Like

Fantasticā€¦ thank you for that. Just out of curiosity, how did you manage to do measurements with such level of details?
I have just bought a raspberry Pi hat to do some energy measurements around the houseā€¦ my MOX is on the list so I will publish results once I have completed the tests. But it will only be the overall consumption at the mains, inclusing the losses at power supply (I will disable wifi to check the difference, at least).

Also very interesting that ethernet controller does use power even with no devices connected.
And I would expected mush higher consumption for mechanical hard driveā€¦

You are welcomed! Thanks for this 1MB correction :slight_smile: . Text updated.

I bought a consumption meter. Just pay attention to the precision of the device. Many meters stops measuring bellow 5W. I bought the one that can measure 0.1W. Then, you need much of patience and experiment with your hardware to understand it and to measure it properly. During the process, you might need to re-measure everything again as a new unexpected knowledge about the hardware (and software) suddenly appears :-).

Yes. It is like with a processor. It consumes power even if it does nothing. You have to keep clocks, bus, timers, interrupt logic, etc. Well designed devices consume much less power when idle.

Me too contrary to the results :-). These 2.5" disks are designed even for laptops, so just spinning them is quite power efficient. But peak consumption is written on them: 5V and 0.85A, e.g. 4.25W.

Edit: Added more details.

1 Like