MoCA latency issue

[alcaas99]

I've added 5 MoCA adapters *behind* my router using a a PPC-9M-U/U splitter, to leverage the existing coax wiring in the house since WiFi can be spotty in the rooms further away.
The adapters are Actiontec ECB6200 and ECB7250. I reckon that the network will be limited to MoCA 2.0 but that's fine for now.
I have only internet connection coming into the house (Spectrum in NC, 200Mpbs); no cable TV nor phone.

However, I'm experiencing very annoying latency issues. For example, establishing a Remote Desktop session to my remote office computer takes 7-20 seconds, where if I instead used WiFi it will connect within 2-3 seconds consistently. Speed tests show the download bandwidth to be pretty good once the download test starts (200+ Mpbs). However, it can take 30 secs easily to get going (but is much quicker with WiFi).
This latency is also obvious when loading websites that perform several operations to fetch their content (most, nowadays).

Devices connected directly to the router's ethernet ports or to its Wifi do not experience this problem. It's only when the connection crosses the MoCA network.
Strangely, this issue sometimes is much less severe, but is seldom and doesn't stay like that for long. Those times seem be at very off-peak hours, like midnight or later. But not even that is consistent.

I'm wondering if there is an issue in my network setup, or perhaps coexisting the 6200 vs 7250 requires special settings.
When looking at the Phys rates in the 7250, they seem normal from what I've seen in other posts.
I did notice that the three 6200s report a "MoCA password" that is different from the one reported by the two 7250s as "enhanced privacy pwd" but unclear if this is a concern or not.

Attached is a diagram of my network with all the devices included and some screenshots of the devices.

 

[krkaufman]

Quick thoughts…
* coax structure seems correct; modem isolated from “MoCA” coax is a good thing (for security and DOCSIS 3.1+ compatibility);
* TX Power numbers don’t look great for a supposed “designed for MoCA” amp.; the spec’d range is -30(great) to +3(maxed-out);
* you’re not amplifying any incoming signal, so the amplifier is sub-optimal; you’d be best served by using a passive MoCA 2.x splitter sized to just what you need, possibly with a MoCA filter for performance improvement. (Though you might try a 4-way splitter sans MoCA filter, with the main node connected via the input port … *if* the Guest Room connection isn’t required.)

edit: p.s. you might try using a MoCA adapter at the central junction to check the stats for a direct connection to each of the remote MoCA adapters, to see if one of the lines is more out of spec than the others.

 

[krkaufman]

Of course, you’d also want to confirm that all the adapters are running the latest firmware available.

 

[alcaas99]

Quick thoughts…
* coax structure seems correct; modem isolated from “MoCA” coax is a good thing (for security and DOCSIS 3.1+ compatibility);
* TX Power numbers don’t look great for a supposed “designed for MoCA” amp.; the spec’d range is -30(great) to +3(maxed-out);
* you’re not amplifying any incoming signal, so the amplifier is sub-optimal; you’d be best served by using a passive MoCA 2.x splitter sized to just what you need, possibly with a MoCA filter for performance improvement. (Though you might try a 4-way splitter sans MoCA filter, with the main node connected via the input port … *if* the Guest Room connection isn’t required.)

edit: p.s. you might try using a MoCA adapter at the central junction to check the stats for a direct connection to each of the remote MoCA adapters, to see if one of the lines is more out of spec than the others.

Thank you for taking the time to look into this.
Yes, all adapters running latest fw.

Your suggestions prompt me to start testing end to end in each segment... and in the process I isolated problems with the Netgear hub in the Porch (hard time getting DHCP to work when directly attached was the first clue, then seeing faster rates when skipping it). Since all MoCA traffic flows thru this to get to the router, it makes sense that it'll affect things. So I replaced that hub and is working much better.

I'm still seeing poor Tx Power numbers though in most of the connectoins reported by the ECS6250s, so I'll test the other hubs next (any tools to stress test a hub? not the first time that I see one misbehave but is very hard to tell because is inconsistent), and more segment checking.

Questions about your suggestions:
- How much of a difference would using a passive splitter make here? Isn't the amplifier without power and with no input signal also acting as a splitter, or do you mean that it has more signal loss? Pondering how worth is to mess with that next.
- Why add a MoCA filter, if no cable/phone signal are incoming to the MoCA network? Future protection?

 

[degrub]

The moca filter can improve the moca network signal quality.
i would use a pair of moca 2.5 modems and test and document the sync rates and other values for each cable segment until you find the issues. Then, as you reconnect cables, test and document.

 

[krkaufman]

I'm still seeing poor Tx Power numbers though in most of the connectoins reported by the ECS6250s, so I'll test the other hubs next (any tools to stress test a hub? not the first time that I see one misbehave but is very hard to tell because is inconsistent), and more segment checking.

MoCA TX Power values won’t be affected by the gear attached via the MoCA adapter’s Ethernet port. It’s strictly related to the coax path to other MoCA nodes.

 

[krkaufman]

Why add a MoCA filter, if no cable/phone signal are incoming to the MoCA network? Future protection?

In your case, with the MoCA coax segment isolated from the outside world, a “PoE” MoCA filter would only provide a performance benefit (no security benefit), by efficiently reflecting MoCA signals back onto your coax. However, you wouldn’t be able to use a MoCA filter in a passive splitter setup, replacing the amp, if you end up connecting one of your MoCA nodes via the splitter’s input port.

As for the amp, it *should* work, however wasteful, but replacing it with a passive MoCA splitter is something worth testing given the MoCA stats. And removing the amp would also allow removal of the amp’s power inserter, in case that is the component throwing off the numbers.

Also, I don’t know how the amplifier would behave sans any power.

 

[krkaufman]

Separate from the MoCA network quality vein of troubleshooting, are the remote wireless devices (Airport Xpress, Linksys E2000) configured as access points (rather than extenders)?

 

[alcaas99]

In your case, with the MoCA coax segment isolated from the outside world, a “PoE” MoCA filter would only provide a performance benefit (no security benefit), by efficiently reflecting MoCA signals back onto your coax. However, you wouldn’t be able to use a MoCA filter in a passive splitter setup, replacing the amp, if you end up connecting one of your MoCA nodes via the splitter’s input port.

As for the amp, it *should* work, however wasteful, but replacing it with a passive MoCA splitter is something worth testing given the MoCA stats. And removing the amp would also allow removal of the amp’s power inserter, in case that is the component throwing off the numbers.

Also, I don’t know how the amplifier would behave sans any power.

- In this case seems to behave the same without the power.
- Yes, the other two are set up as APs and broadcast their own SSID.

However, here's an update after replacing the hub: The issue came back after a couple days of progressive worsening. I've tried rebooting the hub and doesn't help. I've tried unplugging the other devices from the hub, and leaving just the connection from the router and the connection to the MoCA adapter, no change. The only thing that fixes it 100% is if I bypass the hub and connect the ethernet cable from the router straight to the MoCA adapter. Then I get excellent results on the laptop at the office end for example, zero latency, consistent all day.

 

[degrub]

What is this netgear hub that you keep mentioning ?
Model # ?
Is this another GS108 ? That is a switch, not a hub.

how did you physically bypass the porch switch ?
Were you using the same cables ?

Maybe there is a power delivery issue to / at the switch ?

based the testing you have done, that or maybe a cable/termination issue appears to be the issue rather than a moca problem.

 

[alcaas99]

What is this netgear hub that you keep mentioning ?
Model # ?
Is this another GS108 ? That is a switch, not a hub.

how did you physically bypass the porch switch ?
Were you using the same cables ?

Maybe there is a power delivery issue to / at the switch ?

based the testing you have done, that or maybe a cable/termination issue appears to be the issue rather than a moca problem.

Yes, it’s another GS108. (Unmanaged) switch as you pointed out, - not a hub, sorry. I have a TPLink that I’ll try next.

The bypass wa stone by connecting the cable that comes from the router directly into the MoCA adapter (instead of into the switch).

I also tested replacing the UTP cable between the MoCA adapter and the hub, no difference.

Porch is covered, outside, equipment is inside a cabinet. Both the switch and the MoCa adapter are plugged to the same power strip along with the other equipment shown in the diagram.

 

[degrub]

when you bypassed the switch, did you let that run that way for at least as long as it would take for the problem to show up when the switch was installed ?

i have used various versions of those switches for many years. As long as moisture cannot condense and they don’t get too hot ( mine seemed to run toasty ) , the only problems i had were with the wall wart power block failing. I suspect that is your issue - a degrading wall wart power supply.

 

[alcaas99]

Ok. So far I’ve run it for 12h and will check again tomorrow. But the consistency of the performance is very telling so far.

How did you diagnose the power supply? Just checking the voltage output with a multimeter?

 

[degrub]

i replaced with one from a spare same model switch
testing would have to be under load condition somehow.

 

[Reverse]

I'm just a noob, but looking at the diagram and the photo of the PPC-9, I don't understand how you are powering the splitter?

 

[degrub]

on the diagram, top right blue shaded box - "power inserter". It is a wall wart power supply connected to a coax three connection block. one for coax in, one for coax out, one for DC current in. RF (sine wave) signal (MOCA, TV, etc) on the coax does not see interference from a DC signal ( not a sine wave). DC current flows from their across the coax to the powered splitter.

 

[Reverse]

I see the power inserter, but the photo of the splitter shows a termination on the power connection on the splitter? I must be missing something?

 

[degrub]

power comes in on the coax. That cable has to be plugged into a specific port. See the cable labeled "porch" in the photo ? look real close at the label on the splitter for that port - it is an alternate DC power in.