Recommended budget upgrade Mi 3G V1?

I basically agree with everything you said so far, the ASUS GT-AX6000 would not have made sense for the OP (especially considering the budget constraints and given that there is no OpenWrt support for that device either), but the 65 watt PSU should be rather misleading. Yes, fast 802.11ax runs hot and power hungry, yes 10GBASE-T (WAN) drives up the power consumption (even in idle states) even more (and the two 2.5GBASE-T ports don't really alleviate this situation either), but unless ASUS has made really wrong design decisions, I would expect an idle power usage around 20-25 watts (which is obviously a lot for a 24/7 device) and not much more than 30 watts under (normal-high) load, the rest is probably reserved for the USB ports (USB 3.2 gen1, USB 2.0).

Good options (especially with OpenWrt in mind) would have been the RPi4+USB3-ethernet, NanoPi r4s or an x86_64 mini-pc, either reusing the existing mir3g as AP/ switch (well, with 2+1 ports it's not really much of a switch) or adding something like rt3200, xiaomi ax3200, ax3600 (~7 watts idle; pending support). This combination would have combined comfortable performance (especially x86_64, VPN at 1 GBit/s might be a bit much for RPi4 or r4s) with low wattage (well selected x86_64 ~6-15 watts), requiring a budget between 150-300 EUR, depending on the details (e.g. second hand hardware would have been an option for x86_64).


For absolute power usage, agreed. Clearly 24W measured on an AX6000 is only ~third of its 65W PSU. And 2.2W idle and 5W with load measured on a NanoPi R4S is only ~quarter of its 20W PSU. The absolute rating of a PSU is a terrible indication of absolute device power usage. Clearly some margin is added, among other things

and also margin to keep the PSU in its most efficient load range; luck of the draw margin - the ratings of available PSU's on the market closest to the manufacturer requirement might be a larger difference on device A than device B; and other variability sources.

Yes - the AX6000 is quite the power hungry device, isn't it? While its PSU rating does not predict its actual usage, comparing its power supply rating to the sum of those provided for the NanoPi R4S and RT3200 still seems to get us to the right place on helping decide whether one option is materially more power hungry than another option. That could be a matter of luck in this case, but I still think there is some merit in relying on manufacturers not spending more on PSU's than they have to for a relative comparison - with large uncertainty bands on the technique to be sure.

Speaking of uncertainty bands, good point. Every USB port on a device, whether it is used or not, is one to which the manufacturer needs to allocate some PSU watts, just in case they are all used. Differences here could throw a simple relative comparison metric of adding up PSU ratings to the four winds without a watt/USB port adjustment.

1 Like