hub 5 - simple straight swap from Hub 3?

[allmat]

A recent call from the promotions team offered me an upgrade from my current M350 and Hub 3 to 1Gb and Hub 5.  I run the Hub 3 in modem mode with a Deco Mesh and an 8 port switch, plus a couple of Homeplugs (one of which feeds my VM TV box).  Am I going to be able to simply swap the Hub 3 for the Hub 5 and put that in modem mode, or will things get more complicated?  Detailed advice would be welcome.

Separately, I see from another post that the Hub 5 needs a different PSU to the Hub 3 - do I assume one comes with it?

[MOD EDIT: Subject title changed for clarity]

[carl_pearce · Accepted solution]

Yes, it should just be a simple case of:

• Powering off the Deco.
• Power off and replace the HUB 3 with the HUB 5.
• Power on the HUB 5 with nothing connected other than a laptop or other device capable of accessing the HUB 5 web interface.
• Put HUB 5 into modem mode, wait for it to reboot (Power LED will be green indicating modem mode) then power off.
• Plug Deco into HUB 5 via ethernet (Any port should be ok).
• Power on HUB 5 and wait to boot.
• Power on Deco.

Yes, the HUB 5 will come with a new PSU (Well, it should!).

One thing to note is you'll be limted to 900-940Mbps due to the 1Gbps ethernet ports.

[allmat]

Thanks for that advice Carl.  Followed above steps and system up and running – but:

I’ve two Windows 10 PCs connected via a Gigabit switch to the main Deco unit, which is directly connected to the Hub5.  VM’s speedtest shows one with a download speed of 438 and upload of 105, the other (a more powerful machine) with a download speed of 191 and upload 103.  My previous hub 3 and 350 mBit package achieved close to optimum speeds on both machines.  Nothing’s changed – same cables etc. So why am I not getting anywhere close to the one gig download speed, and why is one now much faster (relatively!) when both were similar before?

I’ve also got a Microsoft Surface Pro which connects using the Deco mesh wifi. Close to the main Deco unit it returns 400 / 105, but in a different location via a slave Deco this drops to 161 / 73. Again, any idea why?

[carl_pearce]

Could you please try this site, as I find it more consistent than others:

https://proof.ovh.net/

Could you also try a speed test in safe mode with networking:

https://www.bleepingcomputer.com/tutorials/how-to-start-windows-10-in-safe-mode-with-networking/

If the Deco has an 'QOS' settings please disable.

[allmat]

Thanks for the very prompt further advice.  As far as I can see QoS is off on the Deco as I'm offered an option to set it up. Tried speed test via ovh having restarted in safe / networking mode (Interesting aside - I normally use a Bluetooth mouse and keyboard, which aren't recognised! Fortunately I have others). Chrome result: 492 / 109; Firefox result: 493 / 106.  I'm awaiting a new (longer) Cat 7 cable to check out a direct connection between the hub and the PC. Meanwhile, have you any other suggestions?  

[allmat]

Further point - The Deco itself reports 764 down and 105 up.

[carl_pearce]

If you get chance put the HUB 5 back into router mode and run this test:

https://samknows.com/realspeed/

It tests the speed to the HUB over the VM network.

Note it only works in router mode as the test runs from the HUB itself.

[allmat]

Hi Carl

Samknows/realspeed shows the Hub 5 is working properly – download speed is 1128. My setup links it direct to a Deco M5 (one of three in a wifi mesh) thence to a TP-Link gigabit 8 port switch. Connected to this are two PCs (both with Realtek 1GB NICs), a Raspberry pi server, a Brother colour laser printer, and a Solwise 1200AV2 powerline adapter which sub-network supports another Raspberry pi server, a Samsung TV, and a Virgin Tivo box.

The Deco itself reports an average download speed around 630, the two PCs around 300.  While that’s more than sufficient for anything they’re used for (so why am I concerned?) I’m just intrigued about what’s happening. Any ideas?

[legacy1]

test speed with PC direct to hub in modem mode

[Kain_W]

Hi allmat,

Thanks again for your prompt responses on the matter.

Are you able to as stated by @legacy1 to provide the results from that test if possible?

Let us know,