Inconsistent speeds with moca 2.0

[krkaufman]

The -15dBm TX Power value is a good sign, since the gear has backed-off from max amplification and is now actually reducing the power level (I believe max reducation is -30 dBm); however, the RX PHY rate indicates the adapter is connecting at standard MoCA 2.0 (sub-700 Mbps) rather than bonded MoCA 2.0 (near 1400 Mbps).

Given these stats, you should see effective throughput somewhere between 300-400 Mbps (standard MoCA 2.0 rate), but the goal would be getting as close to 800 Mbps as possible.

This is one reason I suggested the direct connection test between the MoCA adapters. You'd want to verify how to configure the adapters to maximize their connection rate before putting them into the thick of it, where you're left to wonder if the throughput is falling short because of a configuration issue or something amiss with the coax plant.

very simple TEST setup ... keep the main bridging MoCA adapter connected to your router via Ethernet, but bring the other adapter to the same room and connect them directly to each other via a known good coax cable ... "IN" port to "IN" port. Then power them up, and use your PC or a laptop connected to the second adapter to test your Internet download speed, if you wish, or use a tool such as LAN Speed Test, iPerf or jPerf with two computers to test the actual effective throughput over the coax/MoCA segment.

 

[Bacala muntoni]

The -15dBm TX Power value is a good sign, since the gear has backed-off from max amplification and is now actually reducing the power level (I believe max reducation is -30 dBm); however, the RX PHY rate indicates the adapter is connecting at standard MoCA 2.0 (sub-700 Mbps) rather than bonded MoCA 2.0 (near 1400 Mbps).

Given these stats, you should see effective throughput somewhere between 300-400 Mbps (standard MoCA 2.0 rate), but the goal would be getting as close to 800 Mbps as possible.

This is one reason I suggested the direct connection test between the MoCA adapters. You'd want to verify how to configure the adapters to maximize their connection rate before putting them into the thick of it, where you're left to wonder if the throughput is falling short because of a configuration issue or something amiss with the coax plant.

​

Ok ill test it in the morning i thought u needed two computers

 

[degrub]

it depends on the test you want to run. IF you want to verify what the modems are capable of , then you need something like IPERF and two computers. IF you just want to run the variable internet speed test, then just the one. They give you two different results. The first is good for diagnosing setup and local issues.

 

[krkaufman]

Ok ill test it in the morning i thought u needed two computers

it depends on the test you want to run. IF you want to verify what the modems are capable of , then you need something like IPERF and two computers. IF you just want to run the variable internet speed test, then just the one. They give you two different results. The first is good for diagnosing setup and local issues.

Right.

There are 4 levels of testing/verification you could do with the MoCA adapters direct-connected via a coax cable:

1. Check the link lights to confirm connectivity (obvious)
2. Check the MoCA stats (ideally seeing near -30 dBm TX power and approaching bonded MoCA 2.0 PHY rate, 1400 Mbps)
3. Perform an Internet speed test (understanding that the Internet connection would likely be the bottleneck in the test, so it wouldn't reveal the true performance of the MoCA connection)
4. Perform a LAN segment speed test, requiring two computers, one at either end of the segment to be tested, to measure the maximum effective throughput of the network segment.

If you can't get near bonded MoCA 2.0 rates with the MoCA adapters direct-connected, there must be some configuration or firmware issue with the adapters. The adapters would surely need to be capable of these rates direct-connected before one would expect to reach the rates with the adapters placed into their permanent locations on the coax plant.

 

[Bacala muntoni]

it wont let me run iperf3

 

[sfx2000]

iperf3 is built into MobaXterm - free for personal use...

https://mobaxterm.mobatek.net/

Also, much more capable than Putty.

 

[Bacala muntoni]

iperf3 is built into MobaXterm - free for personal use...

https://mobaxterm.mobatek.net/

Also, much more capable than Putty.

My laptop doesnt have a gigabit port dam.... I saw one guy on another fourm with my same problem but he had -27 tx power dbm but hes still only getting 600 mbps

 

[Bacala muntoni]

Can i still do a iperf3 test without a gigabit port? My laptop only has a 100 megabit port

 

[sfx2000]

Can i still do a iperf3 test without a gigabit port? My laptop only has a 100 megabit port

Yes

 

[krkaufman]

Can i still do a iperf3 test without a gigabit port? My laptop only has a 100 megabit port

Sure, but it will be of minimal value if the goal was to measure the throughput of the MoCA segment, expecting test results between 300-800 Mbps. The laptop's NIC will bottleneck the throughput at 100 Mbps or lower.

I saw one guy on another fourm with my same problem but he had -27 tx power dbm but hes still only getting 600 mbps

Can't beat -27 dBm for TX power, but what is the 600 Mbps referring to, PHY rate or effective throughput?

I'm not entirely certain what the reported PHY rates would be for bonded MoCA 2.0. It's possible the PHY rate would still be reported in the 600-700 Mbps range, but since the link is using 2 MoCA channels (i.e. bonded) the effective throughput is doubled.


edit: p.s. Short of being able to run an iPerf test on the MoCA segment, have you checked bullet #s 2 & 3 from the above post, what your MoCA stats report with the adapters direct connected, and your Internet speed test results?

 

[Bacala muntoni]

Those are his stats but his rx phy rate is still 600 mbps?

 

[krkaufman]

View attachment 12178
Those are his stats but his rx phy rate is still 600 mbps?

It's a bit blurry but doesn't that say 670 Mbps for the RX PHY Rate, near the theoretical max of 700?

I need to research, but my earlier thinking, as previously commented...

The -15dBm TX Power value is a good sign, since the gear has backed-off from max amplification and is now actually reducing the power level (I believe max reducation is -30 dBm); however, the RX PHY rate indicates the adapter is connecting at standard MoCA 2.0 (sub-700 Mbps) rather than bonded MoCA 2.0 (near 1400 Mbps).

... has evolved to:

I'm not entirely certain what the reported PHY rates would be for bonded MoCA 2.0. It's possible the PHY rate would still be reported in the 600-700 Mbps range, but since the link is using 2 MoCA channels (i.e. bonded) the effective throughput is doubled.

Lacking bonded MoCA 2.0 gear, I can't say for sure w/o add'l research. I was hoping your iPerf test would confirm it.

 

[Bacala muntoni]

Ok i downloaded mobapt how do i do a iperf test on mobapt

 

[Bacala muntoni]

Ok i now have the two moca adapters directly connected with a 3 foot coax but the rx phy rate is still 670 mbps? This is soo cofonfusing

 

[degrub]

you need a longer cable. Try 20-30 ft.

 

[krkaufman]

View attachment 12181
Ok i now have the two moca adapters directly connected with a 3 foot coax but the rx phy rate is still 670 mbps? This is soo cofonfusing

As I've recently posted, I now believe that may be the best TX PHY Rate that you'll see for MoCA 2.0 (700 being the theoretical max; just like -30 dBm being the max theoretical power reduction). But the "1400 Mbps" rate would come from bonded MoCA 2.0 using 2 MoCA channels (the "bonded" bit!), rather than the single channel of standard MoCA 2.0. (2x 700 Mbps = 1400 Mbps; so your 670 Mbps PHY rate should be, effectively, 1340 Mbps if using two bonded channels)

The only way to confirm the theory would be if you had 2 PCs with Gigabit NICs, allowing you to test the MoCA segment's throughput using iPerf or another LAN testing tool. Unfortunately, even a solid Internet speed test tool wouldn't be able to differentiate between a standard or bonded MoCA 2.0 connection for your adapters, since your spec'd Internet rate is below the expected standard MoCA 2.0 effective throughput of 300-400 Mbps.

I'm annoyed enough by this that I've ordered a couple of the new Motorola MM1000 adapters from Amazon, both to review the PHY Rates for standard and bonded MoCA 2.0, and to see if the adapters are actually capable of reaching the "TURBO" MoCA 2.0 effective throughput of 1000 Mbps. We'll see. edit: Order on hold.

 

[krkaufman]

you need a longer cable. Try 20-30 ft.

The short cable was just for performing the direct-connect test for the adapters, to determine their best possible performance.

The MoCA stats look good, but "Bacala" should run an Internet speed test, to compare to what they were seeing in the OP...

My problem is i am getting fluctuating speeds, i pay for 250 down 20 up but when i do speed tests from my room i am getting 250 down then it will go down to 100ish constantly fluctuating.

... and ideally should find a friend with a laptop having a Gigabit Ethernet port, to allow them to run an iPerf/LAN test, *if* they want to confirm the direct-connect "best possible" performance.

Once they've checked-out what the MoCA adapters can do direct-connected, it's back to figuring out how they perform in the actual setup and why they're falling short.

 

[degrub]

sometimes the 1 meter cable is too short for the driver power auto adjust.

 

[krkaufman]

sometimes the 1 meter cable is too short for the driver power auto adjust.

Interesting.

But should they expect better than -27 dBm and 670 Mbps (-30 and 700 being the theoretical max for each)?

 

[krkaufman]

Once they've checked-out what the MoCA adapters can do direct-connected, it's back to figuring out how they perform in the actual setup and why they're falling short.

Oops, I forgot that Bacala had already made some improvements to their setup, dramatically improving the MoCA stats.

View attachment 12145

So i changed the coax cables feeding the moca adapter and modem and the stats changed on the node?