"Turn off router for 10 seconds" - Quantifiable?

Question

Is there any quantifiable evidence to support power cycling routers for 10 seconds or any arbitrary amount of time, instead of just unplugging and plugging back in?

This is related to troubleshooting a misbehaving router. The theory is based on 'things' needing to be cleared from memory, and that this could take a few seconds.

This is also a theory related to electronics from over a decade ago, and I'm sure it was equally as anecdotal then.

As a person allergic to anecdotes, I became curious when I recognized that I never investigated this issue.

Is there any quantifiable reason to support power cycling routers for 10 seconds or any arbitrary amount of time, instead of just unplugging and plugging back in? Sources welcome

Solution 1

Yes there is.

Any electronics device will have capacitors that will store energy even after you unplug it. You may have noticed when you unplug, say a monitor or TV, the little diode will take another second or two to discharge remaining energy from the capacitors in form of electricity and stop glowing.

This residue energy may not allow memory chips to wipe and you may have problems once your router starts again.

As for the sources - well it's really common sense to someone with basic electronics knowledge, like the sky is blue, water is wet, so I recommend reading about capacitors to see what they do and you will understand it.

The point is that electronics components are far from perfect and any interference may produce unpredictable results.

Solution 2

Ten seconds is an arbitrarily long amount of time, but yes, it does take time for electronic devices to discharge themselves completely because of the capacitance of the circuits within. Some of this capacitance is intentional; some of it is not.

It's impossible to say exactly how much time is needed, as the bleed-off of that capacitance varies with environmental factors like temperature, humidity, and background EMI generated by nearby electronics. The RAM in your computer, for example, can take minutes to discharge fully.

But there is a shortcut. If the router has a button of any kind on it (WPS button, or a reset button), this will usually discharge any residual electric charge immediately. This is because the button places a load on the circuit(s) holding the charge and there is no power going into the device.

In fact, in the old days of parallel ports, this used to be a guaranteed way to correct a stubborn printer. Unplug the printer, unplug the computer, and unplug the parallel cable. Then hit the power button on both devices. Then plug everything back in. Worked every time. Parallel SCSI busses had this problem sometimes too.

Solution 3

I think its worth considering what you're actually trying to do. Turning off a router for 10 seconds is probably longer than the time it takes for any residual power to discharge (likewise, the old 30/30/30 technique could be a 10/10/10 technique). Ten seconds is a simple and arbitrarily large enough time for this to work.

I would however consider any troubleshooting techniques involving singing, or animal sacrifice somewhat suspect, but you're free to unplug and wait longer than 10 seconds.

Solution 4

Working as tech support for 3+ years, I can tell you that 10 seconds is surely arbitrary, but easy to communicate, and is intended to be a little longer than necessary (probably 5 or 6 would work well) but when you power-cycle, you only need to do it once. Unplug your modem and router, wait 10 seconds (my counting is likely different than the customer's), plug in the modem first, wait till the connect light comes on (or wait 10 seconds more) then plug in the router. If the issue is a frozen modem/router, this works like a charm. Every time. Guaranteed.

PS- When I have network issues with my ADSL/router/2x Internet TV boxes/NAS + Webserver setup, I still count to 10. :)

Solution 5

Another reason for cable modems specifically, which I heard from a technician:

Modems will "phone home" to the ISP every 30(ish)-seconds to let the ISP know they're still connected. Some types of failures or settings-updates can only be resolved after the ISP's systems know the modem has been disconnected. They can't know that until the 30-seconds have passed, which is why they ask you to wait at least 1 minute.

View more solutions

Comments

[Ƭᴇcʜιᴇ007]

The power-cycle time suggested is usually general so that it can be applied to any device, not just routers.

[Tyson]

Is there a differentiation in methods? You refer to "power cycling" and "plug cycling", and there is a difference. "power cycling" refers to using a switch to interrupt the power to the device, leaving ground intact which helps drain the electronics. Plug cycling causes ground to also be removed, meaning discharge has to occur thru the environment.

[cqm]

@Tyson in practice I am referring to plug cycling, this morning I unplugged my router, waited ten seconds, and plugged it back in. before questioning my life and coming here (this did not make my router work, but it is likely an ISP issue)

[Tetsujin]

It really has nothing to do with the grounding - most PSUs for equipment like that wouldn't have the ground actually connected anyway - it's more to to with power-soak from the capacitors. Switching off from the device itself is a total power interrupt; from the mains allows power to gradually soak out of the capacitors, keeping the device partly powered for several seconds.

[Jason]

@cqm Have you ever taken note of what happens when you unplug (from the wall) a laptop or desktop power supply that has an LED on it? The LED stays on for a few seconds. Depending on the model, it might be just one second, or several seconds. Visibly quantifiable.

[Keltari]

I know this is a dupe, I cant find the original question on the site

[Brian Knoblauch]

On the ASUS board, that's why those boards usually have a jumper you can short (once battery is safely removed) to drain the stored power. It's basically shorting the system to force an immediate dissipation.

[David Richerby]

Obviously, you have good intuition about the behaviour of electronic components. However, somebody with intuition good enough to understand your answer wouldn't need to ask the question in the first place.

[kasperd]

I agree no animal sacrifice should be needed, since it is highly unlikely there are any SCSI devices connected to the router.

[Michael Hampton]

If you tell someone 30 seconds, and they don't have a stopwatch, they usually do 10-15 seconds...

[Piper McCorkle]

I once cycled my power strip for my desktop and there might have been RAM corruption, but I was in the BIOS and nothing happened.

[Dewi Morgan]

Further to K7AAY's answer: capacitors typically discharge exponentially, the voltage out from them dropping sharply and then trailing off as it approaches 0V, so they theoretically take forever to fully discharge: not "8 seconds"! Practically, their discharge rate depends on the current drawn from them. For a smoothing capacitor, you want it to stay almost completely charged between voltage cycles, since you want its output voltage to stay as close to the input voltage as possible: so, hitting them with voltage before full discharge won't damage them, as it happens 100x/sec in normal use!

[Dan Is Fiddling By Firelight]

Anecdotally, several 30+ minute outages I've had over the last decade did end within a minute of my giving my cable modem a 2+ minute off period. OTOH the other ~80% of the time I tried doing so it didn't help.

[tvt173]

10 seconds is also good as it gives equipment upstream and downstream a decent chance to notice that the router or modem is no longer available.

[TOOGAM]

Reminds me of a friend's findings. He logged into a (DSL) modem and read the script that runs when the device starts up. The script ran a command that caused the light to blink for 20 seconds and then to turn on solid. Then the script did some other stuff that took 8 seconds to do. The device wasn't doing anything noteworthy during the remaining 12 seconds of blinking, and after it was done with its 8 second bunch of tasks, the modem worked just as well as if you waited for the blinking to stop.

[StygianAgenda]

Do you recall when that feature came into use? I ask because I'm not certain my board has that jumper, although that is possible, and I may have simply missed it during the stress of the moment.