Reducing power draw of router, minumum draw for basic functions?


#1

I have a GL-AR300M-Ext, I’ve turned off the wifi as I have no need for it and am just using both of the wired ports, I powered it via the USB port on the side of my Roku 3 and has been working well with an OpenVPN connection for hours.

I was considering buying a GL-AR750 as well since the extra LAN port would be helpful on occasion, looking at the power specs though was surprised to see it listed as 5V/2A and <6W power consumption which was much higher than I expected. That prompted me to check the GL-AR300M-Ext which lists 5V/1A and <2W power consumption.

As far as I remember a basic USB2 port gives a max of 0.5A and 2.5W I think, by turning off the wifi it looks to have dropped the GL-AR300M-Ext power draw sufficiently for Roku 3 USB2 port to power it.

My real question is by disabling both wifi settings on GL-AR750 is that likely to get power draw low enough to also run off a standard USB2 port eg Roku 3?

Power draw numbers in listed specs are much higher for GL-AR750 but I imagine the 5ghz wifi etc accounts for majority of that since it has same CPU/RAM etc as my GL-AR300M-Ext. Has anyone tried it and managed OK?

Will try and find my basic power plug meter then measure Roku 3 on own vs roku 3 with GL-AR300M-Ext but I don’t think it’s precise or accurate enough to give me any useful insights sadly.


#2

Wow, you are hardcare. I live out of a suitcase but this is next level.

The CPU in the AR750 is going to use more power than the 300M as it has more squirrels inside.


#3

I power a 750S off 2 x USB ports with the aid of a Dual (2 > 1) USB cable - runs perfect with both 2Ghz and 5Ghz WiFi enabled and 1 LAN cable connected.


#4

For calculating power consumtion, It can you should look for data sheet of:

  • used CPU
  • max power of deactivated 2.4 and 5 GHz WIFI
  • USB port
  • ethernet port

Or, messure the power consumtion…


#5

OK thanks, sometimes it’s awkward to get more than one power socket especially if travelling plus the less things I have to carry around the better! having a short 15cm LAN cable and 15cm USB cable to go between the Roku and 300M router is a great way to provide the power/LAN connection between them without the need for extra plugs :slight_smile:
Needing to carry a 2nd power adapter and maybe a power bar etc would really harm the portability element of the micro router, if no-one can check and confirm it works I’m better off with what I have than all the hassle for an occasionally useful 2nd LAN port…


#6

true but I don’t have anything accurate enough to measure the consumption of the one I already have to make that determination.

For the rest of it:
they both use the same cpu
not bothered about wifi, will be turning it all off
both have 1 usb port
One has an extra LAN port but I’d expect additional power requirements for this to be minimal

Sadly without buying it and hoping I don’t have to return it I’m not sure I’ll get a confirmed answer unless someone has one to try with? Ideally with a roku to power from (not that I think the roku usb is out of spec vs usual usb ports on other devices)


#7

Glitch, have you ever tried with just one USB? (I guess you are using 2x USB for a reason but then you also use both wifi which would dramatically increase the power draw…)


#8

Yes I have tried one USB but my tests were not extensive or conclusive.
If I remember correctly, one port powered the AR300M OK but the AR750S would keep rebooting (although I also had other teething problems with the latter which might have been reason and not lack of power).

Glitch


#9

The GL docs list both devices as using the same CPU… so where are the extra squirrels again?

Cannot be sure without testing but I expect with wireless disabled your power consumption on the AR300M would be close to the 750. From a basic design perspective they’re very similar boards, just with different wireless configurations and an extra hardware port exposed.

Note that as Glitch mentioned the AR750S is a different beast as it has different CPU with upgraded features but also higher base power draw.

The only way to know the actual consumption for sure though would be go have somebody who owns one configure it exactly as youre suggesting and then meter it. Unfortunately I’ve only got a 750s so won’t be much help.


#10

Hi, yes would be really helpful if someone with a 750 could turn off the wifi (single click for each wireless) and do a quick test over just the wired ports, the hardware is very similar so I’d expect to be OK but if the GL-AR300M-Ext I already have is right on the limit then those few changes may still be too much.
Hadn’t noticed Glitch had a 750S but yes that has different hardware so possibly uses more power (even though power specs are the same), I would have looked at the 750S as the speedier CPU would certainly be helpful but those non removable antennas have killed that idea, it’s annoying enough with my GL-AR300M-Ext and the antennas unscrewed, definitely not buying something where they are fixed on.
Glitch - if you did get a chance to test with all the wifi off that would still be helpful to know, if the 750S works ok then the 750 certainly should, the updated QCA9563 possibly has same power use as the 750 CPU anyway due to improved newer CPU design offsetting the extra Mhz?

That said my GL-AR300M-Ext whilst great in every way from a form factor perspective is a bit slower than I’d like when using OpenVPN, I don’t really need decent/any encryption so was hoping to get up and running with an L2TP connection. Can’t find much info around setting this up and my question on that here has had no replies at all, if I can’t find a way of getting that setup may be forced to look at alternative products which would be very annoying since this hardware is the perfect combination of small form factor + quick (over L2TP would easily manage 60mbps down which is about my max connection anyway).
Can anyone here help with that as well? Took less than a minute on an old tp-link router I had to get this working, no idea why it needs to be so awkward on these…


#11

OK two things:

  1. Just did a load test on my AR750S with wifi turned off. Typical draw is about 300-380 mA, however during boot up (presumably when the wireless drivers load or something CPU intensive happens) it peaked up to 510mA which is technically above what you would get from an in-spec USB port.

I confirmed once booted that if you max out the CPU it jumps back to that 500-520mA level so that pretty much says the AR750S isn’t going to work for you. It’s close though, so there’s hope the original AR750 might.

  1. For improved VPN performance look at using Wireguard instead of OpenVPN.

#12

Great thanks, very helpful to know (if slightly annoying that it’s a tiny fraction above what I’d hoped), would probably get away with it though I’d expect if only briefly above 500ma, especially if using a less CPU intensive workload eg basic unencrypted L2TP rather than OpenVPN/Wireguard which works the CPU a bit…

What I could really do with is a decent guide for using L2TP on one of these, there’s a guide on OpenWRT pages it seems but definitely not written in an easy to follow way and quite a few steps to go through. DD-WRT looks a lot easier to do through the GUI but would seem the only product supported is GL-AR150 which is rather slow and just two wired ports, might be OK with basic unencrypted L2TP. Might end up forced in this direction which is crazy, this should be simple…

As I mentioned on the other L2TP thread sadly my current VPN providers don’t have any test implementations of Wireguard running and not looking to sign up with yet another VPN provider, need to get this working with the ones I already have, L2TP works perfectly over that very cheap tp-link router I had but that’s not very portable and needs another plug, fine at home but not for travel

EDIT: Actually, looking under the DD-WRT release notes for 18.06.0 there’s mention of fixes relating to three GL.iNet products: GL-AR750, GL-AR300M and GL-AR150
https://openwrt.org/releases/18.06/changelog-18.06.0
Is DD-WRT actually working ok on GL-AR750 / GL-AR300M and just the documentation not updated to show now works with them?

Is anyone using or has tried DD-WRT with GL-AR750 or GL-AR300M? Would be great if I could load on my GL-AR300M for testing then maybe risk buying a GL-AR750 or play safe with another GL-AR300M if all goes well


#13

Testing an AR750S (“Slate”) here, with the v3 -testing build from January 28th, I see a peak current draw during boot of 520 mA with no wired connection, but wireless enabled (with no clients). It idles around 410 mA.

This is significantly higher than my AR300M-Lite. There is a CPU difference between the two units, with the AR300M-Lite using a QCA9531 at 650 MHz and the AR750S using a QCA9563 at 775 MHz. The AR750 (no “S”) uses the QCA9531 at 650 MHz and probably has lower power consumption than the Slate due to the slower clock rate.

Adding an Ethernet client on my AR300M-Lite added about another 40 mA of idle draw.


Do you have control over both endpoints, or are you “stuck” with L2TP at the other endpoint?

If you have control over both, I’ve had good luck with GRE tunnels, both L2 and L3.

For unencrypted L2TP, https://support.aa.net.uk/L2TP_Client:_OpenWRT looks to be a good reference, though I have not tested any of it.

Note that

        option _orig_ifname <anything>
        option _orig_bridge <anything>

are remnants of previous LuCI-based configuration and can be safely deleted/omitted.