What's new

Mesh setup for a 3 floor concrete house and 1gig internet

  • SNBForums Code of Conduct

    SNBForums is a community for everyone, no matter what their level of experience.

    Please be tolerant and patient of others, especially newcomers. We are all here to share and learn!

    The rules are simple: Be patient, be nice, be helpful or be gone!

Sheridans

New Around Here
Good afternoon from sunny Portugal :)

This is my first post, however I’m a long time reader of this forum, from which I have learned a lot (combined with Google... of course).

This time around, after reading many threads about the topic I mentioned in the title, I feel I’m walking in circles and therefore I would really appreciate a more one-to-one advise.

So, the topic is: I have a 3~4 floor house (mezzanine built), totally built in concrete in which I have installed a Velop Tri-band AC system with 3 nodes, ie the main node plus 2 satellites, with the main node on the first floor, the second also on the first floor to go around an obstacle and the third node on the floor since, since there’s no obstacle between the first and last floor (as mentioned the house is built in a mezzanine configuration).

However, and although I get around 950Mbps from the ISP, my WiFi speed when connected to the primary node varies a lot between 100 and 500ish Mbps, although always sitting on 5Ghz and about 10ft away.

When connected to the first satellite, sitting on 5Ghz still and at 5ft away, I get a drop to 200~300Mbps. This satellite is around 30ft away from the primary node with only a brick wall in between.

Then, when connected to the second satellite, which is placed around 15ft away from the previous satellite, in a almost strait vertical position above, in open space (just imagine two nodes on top of each other, 15ft away without anything in between), the maximum speed I take out of it is 100Mbps, still sitting on 5Ghz.

I have to say that I never have more than 12~15 devices connected and when i tested the speeds and latency (always between 13~17ms with 0-10% jitter) no other devices were making any intensive use of the service.

Given this performance, and the fluctuation/inconsistency which is very annoying, I’m willing to start a new mesh from scratch, exchanging this one for something which gives me better performance, and willing of course to spend the necessary buck.

Any advice please? Going for a wired backhaul is not possible/feasible.


Thank you!
 
Last edited:
If you're intending on continuing to use pure-wireless mesh, you probably won't get any higher throughput for your wireless clients, with the possible exception being a switch to a system that uses 4x4 or 8x8 backhaul radios, such as Orbi tri-band, AmpliFi Alien or similar. That said, neither system will likely decrease (and may actually increase) your latency and/or jitter. The only product worth switching to in that regard would be Eero Pro, which runs SQM QoS between nodes and out to the internet, to likely reduce both ping and jitter to almost wire-like levels. Eero's ability to re-purpose radios between fronthaul and backhaul also may produce more consistently higher average throughput, but that's far from a guarantee.

At the end of the day, if you want to significantly increase endpoint throughput and lower latency/jitter, you'll need to figure out a way to establish at least some amount of a wired backbone to hard-wire your nodes. That, and converting to conventional wire-first APs (TP-Link Omada, Ubiquiti UniFi, etc.) instead of consumer mesh, as mesh makes all nodes use the same channel for backhaul and fronthaul, thus not allowing more available bandwidth than what's usable on a single channel. Wire-first APs don't have that limitation, and can provide much more available fronthaul capacity, due to their ability to use non-identical fronthaul channels on a per-AP basis.

Hope that helps paint a picture of the practical limitations you're looking at.
 
Another thing that is driving me crazy with my Velop setup is that I run a test now and get for example 200Mbps reading and making a second run test just 2 or 3 minutes after may produce results such as 90Mbps (less than half!!!) with the same conditions (same laptop, location, etc. )

Any explanation for this? Or is Velop just a crappy solution?
 
I’ve used Velop 3 node system, Google Nest Wifi, Ubiquity, Netgear Orbi (both AC and AX systems), Asus Aimesh systems, you name it.

Pretty much the same situation as you have. No way to wire backhaul, about 2400 sq. ft. 2 stories.

I’m now using a Netgear Orbi RBK752 with one node. Fast, full ISP speeds everywhere, no dropouts, fast backhaul channel, easy setup. After the initial setup I tweaked a couple of settings, moved the node about 35 ft. away and haven’t touched it since. Netgear finally made a decent Orbi that works as intended. Might look into this or the Asus Zenwifi system. Both the same price.
 
I’ve used Velop 3 node system, Google Nest Wifi, Ubiquity, Netgear Orbi (both AC and AX systems), Asus Aimesh systems, you name it.

Pretty much the same situation as you have. No way to wire backhaul, about 2400 sq. ft. 2 stories.

I’m now using a Netgear Orbi RBK752 with one node. Fast, full ISP speeds everywhere, no dropouts, fast backhaul channel, easy setup. After the initial setup I tweaked a couple of settings, moved the node about 35 ft. away and haven’t touched it since. Netgear finally made a decent Orbi that works as intended. Might look into this or the Asus Zenwifi system. Both the same price.

The Orbi appears to have a 1-year warranty. If the router dies after that, can you replace it or do you need to buy a new system?

OE
 
If you're intending on continuing to use pure-wireless mesh, you probably won't get any higher throughput for your wireless clients, with the possible exception being a switch to a system that uses 4x4 or 8x8 backhaul radios, such as Orbi tri-band, AmpliFi Alien or similar. That said, neither system will likely decrease (and may actually increase) your latency and/or jitter. The only product worth switching to in that regard would be Eero Pro, which runs SQM QoS between nodes and out to the internet, to likely reduce both ping and jitter to almost wire-like levels. Eero's ability to re-purpose radios between fronthaul and backhaul also may produce more consistently higher average throughput, but that's far from a guarantee.

At the end of the day, if you want to significantly increase endpoint throughput and lower latency/jitter, you'll need to figure out a way to establish at least some amount of a wired backbone to hard-wire your nodes. That, and converting to conventional wire-first APs (TP-Link Omada, Ubiquiti UniFi, etc.) instead of consumer mesh, as mesh makes all nodes use the same channel for backhaul and fronthaul, thus not allowing more available bandwidth than what's usable on a single channel. Wire-first APs don't have that limitation, and can provide much more available fronthaul capacity, due to their ability to use non-identical fronthaul channels on a per-AP basis.

Hope that helps paint a picture of the practical limitations you're looking at.


Thank you for the feedback Trip.
Would you consider an AX system, for example the Asus Zenwifi XT8 instead of Eero Pro. If not, why?

:)
 
Hi @Sheridans - I personally wouldn't consider an AX mesh system right now on the merits of it being AX alone. A single all-in-one router? Maybe. But a mesh system? Likely not.

The reason is that with mesh, especially multi-hop/multi-layer mesh, there are many more factors at play determining the overall quality of the endpoint experience than just the wireless standard. Several of them relate to things like real-time RF adjustment, fronthaul/backhaul radio re-purposing and/or proper QoS on backhaul/WAN links. As you might guess, hardly any consumer products have enough of them baked in and working, and none of them happen to be AX systems. At the end of it all, I would take a 200Mb/s transfer rate over 600Mb/s, if the former was solid and consistently low-latency, and the latter was not.

That's just my stance, though, as I personally don't need that much extra bandwidth (beyond 200-300Mb/s). For you, though, if you'd really like to try and max things out and you don't mind the expense, maybe buy two or three AmpliFi Alien units, the top Orbi AX system, or two or three Asus AiMesh AX routers, and see what they can do. Maybe it will be worth it for you. You can't really can't be 100% sure until you try them in your environment.

Hope that helps.
 
Last edited:
Ethernet not possible? Ever heard of MoCa- Ethernet over coax? Cable TV and Internet are likely compatible with it. If you've got coax in your home, it's possible you can have Ethernet on the same wall port as the coax, even extend to multiple devices with an Ethernet switch. And even a mesh node like Netgear Orbi or Asus Aimesh or ZenWifi wired backhaul; I've tested both of these with my MoCa adapters. Cable internet and coax lines at central point at house's utility box are connected to MoCa switch. The other ends of coax in each room connected has a MoCa adapter. These adapters have two coaxial ports; one feeds modem and the others are lan extensions of the MoCa network. MoCa 2.0 is rated at 1Gbps so there's virtually no loss like typical 1Gbps lan ports on an Ethernet switch. Apologies if you're aware but not usable for whatever reason.

Asus ZenWifi XT8 with Ethernet backhaul here. The extension node is essentially as fast as the router with the modem connected to it. The 5G-2 can be used on devices with 160 MHz bandwidth if the SSID is hidden. Extra bandwidth penetrates walls and gives me strong WiFi 5G outside despite my exterior walls being stucco.

I've got a small two story but WiFi is tricky here as evidenced by my former Orbi RBK50. When I installed the satellite or router in living room, the nearby staircase and duct overhead would limit 5G and slower 2.4G wasn't ideal. I moved the Orbi router to above the first wrung of stairs and dealt with weaker in the living room until the 2.4G radio got too weak to hold my Logitech Harmony hub 2.4G connection. I think all the radios were weakening over time. I can recall when it was new, getting maximum speed tests throughout my house. But the time I added cameras outside, I was moving my router and satellite closer to the cameras and dealing with weaker WiFi.

I'm using the XT8 without the AX enabled. So it's essentially being used as an AC router. But it's still better than the RBK50 Orbi.

Sent from my Pixel 4 XL using Tapatalk
 
Thank you again for the reply.

My pings with the present Velop configuration vary between 15 and 20 ms with jitter between 0 and 5 ms, being the worse results measured on the 3rd node and the best on the primary node.

The ping measured on the ISP router, via Ethernet cable is 8ms.

Obviously all the above measured on the same server for sake of comparison of results.



Given the above, what would be in your opinion the expectable results with the Eero Pro?
I’m just trying to collect further information before deciding to spend another 500 bucks on a new mesh ;)


Thank you!
 
Honestly, that's relatively good as far as mesh goes. I'm not sure you're going to see much better with Eero Pro, if it's better at all. I'd probably pocket the $500 for now, unless you're really burning to see if you can max out your client speeds with something like AmpliFi Alien or Orbi AX6000, which would set you back $700-750 for a 2-node setup of each, $1000-1100 for a 3-node setup. And at that point, if it's just for the performance alone, I would rather you move heaven and earth to find a way to hard-wire Cat6 or 6a, even if it has to be surfaced-mounted inside some discrete raceway, and run a pair of Asus AX88U's hardwired to each other in AiMesh.
 
Honestly, that's relatively good as far as mesh goes. I'm not sure you're going to see much better with Eero Pro, if it's better at all. I'd probably pocket the $500 for now, unless you're really burning to see if you can max out your client speeds with something like AmpliFi Alien or Orbi AX6000, which would set you back $700-750 for a 2-node setup of each, $1000-1100 for a 3-node setup. And at that point, if it's just for the performance alone, I would rather you move heaven and earth to find a way to hard-wire Cat6 or 6a, even if it has to be surfaced-mounted inside some discrete raceway, and run a pair of Asus AX88U's hardwired to each other in AiMesh.

I’m really not willing to burn money for nothing (this brought to my mind Dire Straits :) ). If that’s the case then I’m just gonna stick with these Velops until something would come out.

However what’s drives me crazy are not the pings but rather the speed fluctuations.
One moment I’m measuring 300Mbps and on another, just seconds apart and under the same circumstances, measure 100Mbps. If this is something Eero Pro could solve it would worth the money already....
 
It seems to me you trying to spend a lot money on something you will never achieve until you run wire. It can't cost near as much as you are trying to spend to install wire for a back haul.
 
However what’s drives me crazy are not the pings but rather the speed fluctuations.
One moment I’m measuring 300Mbps and on another, just seconds apart and under the same circumstances, measure 100Mbps. If this is something Eero Pro could solve it would worth the money already....
I'd be inclined to think Eero Pro would be able to give you more consistent throughput, but there's really no guarantee until you try them. If you can float the $500, I'd say pull the trigger off Amazon and find out. That way you get a quick refund option if it doesn't pan out. If nothing else, it will be an excellent data point for yourself (and us!).
 
I didn't see anyone pick up on the MoCA 2.0 suggestion...if you have cable TV coax in your walls, this is a good way to use them for ethernet, rather than putting in ethernet cables. Our house is on a slab, without a full attic, so we can't really put in ethernet cables without really tearing things up, or surface mounting them. On the other hand, the MoCA 2.0 adapters give us pretty much full speed at the coax outlets where we have eero remote nodes. Gives us real speedy wireless, and the wired speed at the remote eero nodes is just about full speed. Works well.

Anyways, good way to go if you've had (or have) cable TV. There are a few things to think about, you can read about them in the MoCA forum here *smile*.
 
I didn't see anyone pick up on the MoCA 2.0 suggestion...if you have cable TV coax in your walls, this is a good way to use them for ethernet, rather than putting in ethernet cables. Our house is on a slab, without a full attic, so we can't really put in ethernet cables without really tearing things up, or surface mounting them. On the other hand, the MoCA 2.0 adapters give us pretty much full speed at the coax outlets where we have eero remote nodes. Gives us real speedy wireless, and the wired speed at the remote eero nodes is just about full speed. Works well.

Anyways, good way to go if you've had (or have) cable TV. There are a few things to think about, you can read about them in the MoCA forum here *smile*.
Yep. I've fished Ethernet through walls, attic and along walls in previous single story dwellings. It's no picnic. But my current dwelling is 2 story with low attic overhang. No one wants to get in the attic to run wires and exterior wires are unsightly and may not last unless ran through pipe.

MoCa 2.0, even at 1Gbps for Ethernet backhaul, is better than wireless backhaul. Too many obstacles in a house to expect WiFi backhaul to deliver speed and consistency of wire. The coax is pre-installed with short runs, protected and minimal loss compared to wireless or even powerline. Powerline was a hassle and unreliable for me. Secure line needed constant resetting and eventually just broke. Never was fast as Ethernet anyway.

I was disappointed putting router and node equipment where I wanted and running speed tests- wireless backhaul vs MoCa 2.0 No comparison; wireless loses way too much speed and may introduce packet loss.

Sent from my Pixel 4 XL using Tapatalk
 
Yes, with MoCA 2.0 I get about 400Mbps wi-fi at the remote eero nodes from internet speed tests. Without MoCA (wireless backhaul), I get about 250Mbps on internet speed tests. And with MoCA, much less variable than wi-fi backhaul.
 
Hi there again fellows

Well, I’ve managed to connect the primary node on the first floor to the slave node on the third floor with CAT cable :).
The slave node on the second floor is connecting via Wi-Fi to either node on the first or third floor.

Given this setup now, what would be your recommendation? Should I keep the Velop as it is? I can measure pings of 11ms and WiFi speed of 600Mbps on both wired nodes on first and second floor and I get around 300Mbps on the node of second floor.
These measurements are taken using an ipad 12 Pro, always connected to the same server.

Just as reminder, my ISP speed is 1Gbps.


Thank you!
 
Okay your is peculiar and since no wiring possible and when you mention Wifi 5ghz its struggling can you confirm what clients capability as it can also sometime adds to delay and latency and freezing .

As currently not much clients in AC can hit full 1gbs . Max i seems is around 870Mbps on A7000 usb 3.0 client when doing a local transfer from ssd to m.2 as the speed also getting throttled on storage iops limitation.


Mesh system will be hard as it will have limitation on walls and metal objects. Repeater with fast lane can be option. But unless a floor diagram exist and current devices coverage will be hard to recommend any.


Sent from my iPad using Tapatalk
 

Similar threads

Support SNBForums w/ Amazon

If you'd like to support SNBForums, just use this link and buy anything on Amazon. Thanks!

Sign Up For SNBForums Daily Digest

Get an update of what's new every day delivered to your mailbox. Sign up here!
Top