Signficant loss of networking speed!

darrenxh

Standard Member
Hello all - I am a networking novice, so go easy on me....

I have a router in the house - getting 30mbps download / 10 mbps upload.
I have an outbuilding (seperate from the house) with lots of requirement for wifi/internet.
I have a 50m cable buried from the router to the outbuilding - it is cat6.

When it enters the outbuilding the speed has dropped to 10 mbps download / 9 mbps upload (so losing 20 mbps on download and 1 mbps on upload).
As it comes into the outbuilding it goes into a netgear 8 port switch. If I connect a laptop directly via a cable to the router, the speed again loses a little - to 8bps on download, but maintains the 9 mbps on upload). I have 3 wifi hotpsots coming off the router in the outbuilding - when I connect via wifi to one of these and I get a large drop to 2.5mbps download but still 9 mbps upload.

My question ... is it usual to see such a drop on download speed over a) the 50m cable run [as I thought this was the most efficient way to do this] b) why the degregation only on download c) is it usual to also see the 66% drop from the router speed to the wifi hotspot speed?

How can I fix this - its rather urgent I do so!! All help will be appreciated!

Darren
 

neilball

Well-known Member
How are you measuring speeds? If it’s internet based speed tests then this may be variable do would not be the greatest way to check the local wiring. Are you also sure that no one else is using bandwidth on the network while you are attempting your tests?

How have you wired up the cable - directly crimped RJ45 plugs at each end or punch down modules with patch leads? What wiring configuration did you use (hopefully TIA-568B or A). Have you tested the cable? Is it full copper Cat6 and not copper clad aluminium (in which case it should not be labelled as Cat6)?

A correctly wired ethernet cable will give you gigiabit speeds (assuming the ports at each end are gigabit) regardless of length upto 100m, it simply does not slow down or degrade. If you have mis-wired then it may run slower, either because it’s only linking at 100mb (100mb only uses 2 of the 4 pairs in the cable), or because it’s barely working at all which makes it seem slow.

Why do you have 3 APs in the outbuilding? If they are all close to each other you could be creating lots of interference between them leading to much lower rates on each. Are they 2.4 GHz, 5GHz, or dual band? Have you set their channels and radio output to avoid interference?

If there are lots of clients then does the router have the power/memory to keep up? Some ISP-provided routers do not cope well in large networks which can impact network performance.

So your best approach is to prove your cable is 100%, and once this is fertain you can concentrate on wifi AP settings and check out the router.
 

mickevh

Distinguished Member
As Neil suggests, Internet "speed test" sites are not the best way to assess your internal network links. If you can get you hands on a couple of PC's (laptops are idea) tools such as NetIO or iPerf are better for testing your local links - both are free. It's kind of like having your own speed test server on site instead of relying on one out on the Internet and will do a much better job of stress testing the local network links than any Internet "SpeedTest" as you are removing the Internet from the test path.

Ideally you'd run up NetIO/Iperf in "server" mode on a PC cabled to your router, then run the the same programme in "client" mode on a laptop and roam around your environment testing the wired/Wi-Fi links as you go. Run the tests a few times and at different times of day and look for the trend.

Don't stress over the absolute values of the results - these tools are rather imprecise (and like SpeedTest, don't actually measure the "speed" of anything, they send measured amounts of data and compute statistical averages) - what you are really looking for is order of magnitude and trend.
 

darrenxh

Standard Member
Yes I just used an internet based speed test - on the laptop - not a local client tool. I can try this.
This is brilliant insight by the way - in terms of the cable, it was just CAT6 cable from the building merchants - unsure how to tell beyond this. It had the plastic cross shaped separation to the twisted pairs, and a foil cover before the plastic sheaf. Have attached 3 pictures of the cable detailed stamped on the outside. In terms of how it is wired - I used a crimp tool and changed the blue / green pairs as per instructions on the internet ... same both ends obviously on the 50m run. As you rightly say above, this loss of 30mbps to 10mbps across this 50m cable is the main issue I need to solve initially (then I can look at how the hotspots are configured / placed and if too many).
Is there an alternative way I should be crimping the 50m cable?
726836E1-5463-476A-AB97-4D6EEA1B11C7.jpeg
0C1BC04C-8E79-42DB-A1A7-1F93E1E2E8FF.jpeg
F962DC1D-0004-40C5-B7E6-D2A059F8CE7B.jpeg
 

neilball

Well-known Member
Your RJ-45s should be wired the same at both ends, and ideally use the following sequence...

1596880263056.jpeg

If your cable is shielded then you should have used shielded connectors too, and shielded network ports etc, otherwise the shield is useless. In fact in certain circumstances an improperly terminated cable shield can become an antenna, actually picking up interference on the cable instead of keeping it clean! Most installers stick with UTP (unshielded) cabling in non-industrial installations as dealing with cable shielding correctly takes a lot of time and effort for little or no benefit.

In terms of other things to be aware of is that ethernet cable should be kept at least 30cm away from power conductors when running parallel and it should not be tightly clipped as crushing the cable with tight cable ties can also cause issues with signal integrity.
 

mickevh

Distinguished Member
Cable says it's cat6. I can't see a brand name, (or at least, if there is I don't recognize it) so I guess we can only assume it is what is says it is. Cat6 UTP should be good for ethernet up to at least 100m.

Strictly speaking the foil shield should all be earthed, but I think Neil is more competent to talk about that than me.

Ethernet works at fixed speeds and thusly either works full speed, whatever the cable length, or not at all. It's extremely rare for it to "work but slowly" as a result of a cable problem. (Technically what happens is you get a lot of corrupt packets that get silently discarded, and when you watch if using a speed test etc. the performance gets erratic, but generally if the cable is "that bad" it tends to just not work at all.)

I think the next step it to try and prove your cable link either with one of the NetIO/iPerf tools, or if you haven't got the kit to do that, try a locally copying a big file that you know the size of - a DVD rip for example - time how long it takes on a watch and do the math to compute the average (which is pretty much all that speedtest/NetIO/iPerf does.) Again, we're not really looking for the minutia of the numbers, we're just looking for orders of magnitude and trend (is it the same or similar each time or does it vary wildly.)

BTW - is the pin outs on the cable the same both ends...?
 
Last edited:

sraper

Active Member
A very simple local speed tester is Totusoft | LAN Speed Test. Very simple to use. Need 2 pc's/laptops and run softw on one of them. Specify a "shared folder" on the other eg \\computer2\c$\windows\temp specify the a file size eg 100Mb and job done
 

ChuckMountain

Distinguished Member
Agree with everything Mick and Neil say what I would add though is that it sounds like there is a wiring issue.

I recently had a device which only connected at 10Mbits due to an issue with pins 1 and 8 effectively being disconnected.

It sounds like you could have a similar issues as all your speeds are below 10Mbps in the shed.

If you connect your laptop directly to the cable, what does Windows report the connection speed as?
 

darrenxh

Standard Member
Sounds like a good first action it to re-crimp rj45 on each side and see if that fixes? If not, then run the totusoft (I have two laptops I can do this with). Should I do this with shielded connectors (with take me a while to get hold of them) or standard rj45 ends?
 

darrenxh

Standard Member
Agree with everything Mick and Neil say what I would add though is that it sounds like there is a wiring issue.

I recently had a device which only connected at 10Mbits due to an issue with pins 1 and 8 effectively being disconnected.

It sounds like you could have a similar issues as all your speeds are below 10Mbps in the shed.

If you connect your laptop directly to the cable, what does Windows report the connection speed as?
how do I check the windows connection speed?
 

ChuckMountain

Distinguished Member
They keep changing it in Windows but basically under network connections it will tell you the connection speed of a wired (and wireless) network.
 

ChuckMountain

Distinguished Member
Are you using the right type of connectors. Obviously they should be cat6 but they should also be for the right type of cable normally stranded but they should be limited to shorter runs. Normally permanent runs use solid core and have a different connector. Some do both but but the wrong one will terminate badly and often not get a valid connecting
 

mushii

Distinguished Member
A couple of things. How tight are those cable ties? What is running in that grey flexible conduit?
 

darrenxh

Standard Member
Pretty tight cable ties and it is a power line in the conduit. What are the right type of connectors? I am using basic rj45 I had left over from cat5e a few years ago
 

ChuckMountain

Distinguished Member
What are the right type of connectors? I am using basic rj45 I had left over from cat5e a few years ago
Well that is one issue. CAT 5e cable has a smaller individual core diameter than CAT 6 and so you have to use the right size for the cable otherwise you have termination issues.

However is your cable stranded or solid, it should be solid (and not CCA). Ideally this should be into a pinning down socket either as part of a wall mount or patch panel. Then connect a short patch lead between devices.
 

sraper

Active Member
Cat6 and cat5e cable comes in two main "flavours"
Solid
Stranded

Solid is normally used for long runs between patch panels (or wall sockets) whereas stranded is used for the shorter and more flexible patch cables between sockets and devices. The rj45 crimp connector often come in two styles to match the cable ie the bit that pierces the cable has a different shape to make a better electrical connection with either the stranded or solid cable.

Si if you use a stranded crimp connector on a solid cable or vice versa you may get issues which would show up as slower or unreliable connections

Saying that this claim to work on both types!
 

darrenxh

Standard Member
Cat6 and cat5e cable comes in two main "flavours"
Solid
Stranded

Solid is normally used for long runs between patch panels (or wall sockets) whereas stranded is used for the shorter and more flexible patch cables between sockets and devices. The rj45 crimp connector often come in two styles to match the cable ie the bit that pierces the cable has a different shape to make a better electrical connection with either the stranded or solid cable.

Si if you use a stranded crimp connector on a solid cable or vice versa you may get issues which would show up as slower or unreliable connections

Saying that this claim to work on both types!
Will try new connectors
 

mushii

Distinguished Member
Pretty tight cable ties and it is a power line in the conduit. What are the right type of connectors? I am using basic rj45 I had left over from cat5e a few years ago
If they are tight enough to deform the cable sheath then they are too tight. Deforming the cable can affect the performance of the cable as you crush the twisted pairs and the internal wave guide. You need to replace all of your cable ties with either looser cable ties or better use velcro strips.

Running CatX cables parallel to power cables for any significant length is not recommended as you can induce current in both the twisted pair but even worse in the shielding (which should be grounded at one end).

1596973727484.png




Both of the above can have detrimental affect on the the performance of your CatX cabling
 

ChuckMountain

Distinguished Member
@mushii while I don't disagree with anything you have put, there are plenty of installs that will work but will never be certified with running together and cable ties.

However there are far fewer that work correctly with the wrong size and type of rj45 connectors.

Replace these first before doing anything else.
 

darrenxh

Standard Member
@mushii while I don't disagree with anything you have put, there are plenty of installs that will work but will never be certified with running together and cable ties.

However there are far fewer that work correctly with the wrong size and type of rj45 connectors.

Replace these first before doing anything else.
That is great advice - want to start with the highest probability fix ... as I am losing 2/3 of speed over 50m!
 

mushii

Distinguished Member
I agree, I have crimped up literally thousands of RJ45 connections. Normally though it is very difficult to feed Cat6 into Cat5 RJ45s because of the difference in conductor size.

Before I would re-crimp anything I would do an iperf test on the cable as it will give you a better idea if the cable is the problem.

@darrenxh what is the router in the house and what is the switch that you are using?
Just to confirm you are only running a single Cat6 cable between the house and garage.
 

Andy98765

Distinguished Member
I do not have the high tech tools as many on here may have but my very simple Tester shows up many connector failures over the past God knows how many years.
 

darrenxh

Standard Member
I agree, I have crimped up literally thousands of RJ45 connections. Normally though it is very difficult to feed Cat6 into Cat5 RJ45s because of the difference in conductor size.

Before I would re-crimp anything I would do an iperf test on the cable as it will give you a better idea if the cable is the problem.

@darrenxh what is the router in the house and what is the switch that you are using?
Just to confirm you are only running a single Cat6 cable between the house and garage.
Router is a livebox sagecom. Switch is a netgear 8 port switch. And yes single cable from the router to the switch in outbuilding
 

Similar threads

The latest video from AVForums

Podcast: Panasonic HX800 TV + Sony HT-G700 Soundbar reviews, movie and TV show news and reviews

Trending threads

Latest News

What's new on Sky and NOW TV UK for October 2020
  • By Andy Bassett
  • Published
Linn upgrades Majik LP12 turntable with new tonearm
  • By Andy Bassett
  • Published
VPI announces new Prime 21 turntables
  • By Andy Bassett
  • Published
Skyworth launches 8K Q71 TV
  • By Andy Bassett
  • Published
AVForums Podcast: 23rd September 2020
  • By Phil Hinton
  • Published
Top Bottom