Ethernet CAT7 disaster

There are a few possibilities

a) The cable you have is not 8 wire (4 pair), although unlikely

b) The two extra send and receive pairs required for 1000BaseT are not wired correctly, but the two pairs for 100BaseT are correct wired

c) There is a fault in the cable for the extra two pairs.

100BaseT requires 2 pairs duplex, 1000BaseT requires 4 pairs (1Gbps is always duplex)

Simon

It's an 8 wire cable, I terminated it myself using very fool-proof snap-in plugs that don't require a crimping tool. I did it with a friend and we checked, double-checked and triple-checked that all pairs are correctly connected. Then the electrician tested it with a simple ethernet cable tester and it was all green.

What really worries me is that on some fora I read of similar problems, e.g. here: 

http://www.tomshardware.co.uk/...onnect-1gbps-100mbps

I might have made a huge mistake by buying CAT7 cable, should have gone for CAT6a...

The weird thing is that the cable works fine at 100mbps when connected between the living room router and the gigabit switch in the study, 30 meters away. But if I connect my laptop in the study directly to that cable (instead of via the switch), the laptop cannot see the cable at all!

Gajdzin, your new cable is not up to spec. as Simon said.

That's possible, but how can it be measured? A simple cable tester shows that all leads are connected correctly. Is there some way to check if I have to rip the whole cable out and buy and install a new one (oh, my Dog...) or just crimp on different plugs, or the cable as such is fine but broken in a specific spot?

My guess would be the termination or the termination sockets. Keep the twisted pairs twisted right up to the connection. Maybe try crimping on some RJ45s on the ends as a test? Paul

You are right, Paul, that's the first thing I'll do, because throwing away four high-end snap-on connectors (same as on AudioQuest Vodka) at 10 Euro a piece is still much cheaper than replacing the whole cable... Some fora suggest that it could be the shielding, some cheaper switches or routers don't like shielded cables... So should I crimp on regular, plastic RJ45s and leave the shield unconnected just as a test? Or try the shielded crimp-on variety?

I think the fact that my switch recognizes the cable (albeit allows only 100mbps) while my laptop does not recognize at all the same cable plugged directly into it is in some way significant and could tell me something, except I've no idea what

How have you wired the shields on the twisted pairs? or the overall shield? I trust you have left floating as I suggested elsewhere?

I would try cheep plastic RJ45s first. Try one end first and it it works well you know that end is the problem.

Also you can hire expensive test equipment that will tell you the colour of the wire with the problem.  

Good luck tomorrow. Paul

Errr... No If you remember that thread, I wrote there I'm still going to connect the shield at both ends. I really wanted the maximum benefit from it, and you wrote that if I only ground it from one end it will be like a Faraday cage and only effective at some frequencies... Strangely enough that's what some fora suggested, that it's grounding the shield that's causing the speed drop...

So now I have 2 options: crimp on simple RJ45s leaving the shields floating or crimp on CAT6 connectors with metal sides and ground the shield on one side only...?

I think the speed problem is severe in that not only the switch and the router recognize the cable as 100baseT not 1000baseT, but any computer connected behind the switch only gets 79mpbs from the internet, not even close to 100, not to mention the 250mbps that I get if I plug a computer directly into the router...

No, I don't think it's the shield - I have justed appended the end of the long CAT7 cable with a plastic (no shield) female-to-female RJ45 coupler followed by a short run of UTP CAT5 cable. So I effectively disconnected the shield on one end. No change - I only get 100mpbs.

Yes, I made an appointment for Tuesday with my internet provider, they will bring some solid test equipment and replace the whole cable if needed. Expensive but in addition to music streaming (for which 100mbps would actually be plenty) this link is my lifeline - I project manage IT projects in the USA out of my home office in Poland, so I spend 8 hrs a day on Microsoft Lync voice and slide sharing, without a solid internet connection I can't work! The old one had some issues so I tried to improve it (also for audio streaming quality) and looks like I only made things worse...

Hi, Ethernet patch leads don't quite work like that with respect to speed. They will only work at 10/100/1000mbps

10 and 100 are wired the same and 1000 have two extra pairs wired.

If the leads check cut correctly on a wire test - and i assume you have wired correctly as cross over and not pass thru for all 4 pairs, then I can only assume you have a ground loop with the shield which is causing a relatively high circulating current which is upsetting the line drivers.

So as a test float the shield - or at least have only connected at one end, and make sure it is not connected to any of the cable wires.

That is why I said unfortunately grounded shields can be a pain on Ethernet leads.

Simon

In the connectors what have you connected the shields to?

Try using different plastic connectors?

Then i would check the cable checker - how did you check and what did it confirm. Did you wire as cross over? Did you check it once installed? If not, are there any tight bends on the CAT7 cable.

I am intrigued when you say one of your devices didn't see the network at all..

If the link is half duplex, what colour are the LEDs then on your equipment.

If you set the equipment to manual speed/duplex as opposed to Auto is there any difference to behaviour?

Ultimately I can't see there are not many variables here. It does sound like there is a either fault with the cable or connectors.

Good luck!

Simon

See if you can borrow a Network Qualification tool from a structured wiring installer. SignalTek, Fluke and several others make them. It will test your cable using for full  duplex 1000BaseT data transfer. If there is a fault in the cable with bad cross talk, damaged twists, poor design, or bad connector interfacing technique, it will not show up with a simple DC pinout check, but will show up when you do a data cable check when the cable is running at frequency.

Great advice, Simon. Rather than running around like a chicken with its head cut off I will get here somebody with a Fluke CableIQ or similar tool.

Without being an expert, it seems some wires contain a high %iron instead of copper. Using a magnet, when a wire sticks you could found the problem.. 

Earlier I found these experiences, hopefully usefull for you (Google pagetranslate wil do the trick)

http://gathering.tweakers.net/...t_messages/1491393/1

good luck, Fred

Thanks, Fred. I'll do the magnet test tonight On Tuesday morning a crew from my ISP comes down with a Fluke CableIQ and we shall know the truth... I hope it's just the fancy plugs I put on - they are the same ones as on AudioQuest Vodka, so I thought they'd be a Mercedes Benz of RJ45, but now I read on several fora of problems with such plugs... Hopefully changing them to ordinary CAT6 shielded crimp-on RJ45s will solve the problems, I would hate to have to re-cable the house again.

http://www.dailymail.co.uk/new...blackened-pants.html

Re-cabling could be the least of your problems!

Keith

I thought it might work well in a Polish context 

I had a conversation with Phil at Naim last year and he highlighted the perils of earthing CAT7.

Audioquest upgraded the Vodka cables after I bought them. They very kindly replaced my old ones free of charge. Again, I believe the change in design was related to earthing (according to the dealer).

All this, plus you experiences, would suggest that making home brew Vodka is a tricky business.

I've had no problems that I know of, but I did check the colour of the LEDs on my switch (they're green)! I also looked at Cisco switches. They're not cheap, are they? Also, they don't seem to be available from the likes of Amazon. Also, they're managed so you have to configure them which is a pain.

Keith

Gajdzin - I would let the shields float both ends and take it from there.  

Remember the cables are twisted pairs anyway so leakage is quite low, and you can let the shields capacitively couple to ground on their own....

PS Cisco switches and routers are still typically accessed via Telnet / SSH for IOS scripting via a loopback address or USB/RS232  console

Simon

Sir, YES SIR! Since you REALLY know your stuff, I'm going to let the shield float solely based on your advice. I must be crazy

PS. The big boxes are, but the little wi-fi cable routers made by Cisco for home use like mine are consumer products with a browser console with menus, help, etc. Which is good, because I had my Cisco certification in Japan in 1995 and I don't remember a single command I used to design networks in those days, but soon after that switched to project management and I left the networking world. Now I simply ask experts. Like you.

Gajdzin, I am no expert I just earn a living from this stuff

Simon

The very fact that you earn living from it AND you can quickly respond to somebody's problems or doubts and just nail the solution, and you know how to SELL your solution so that somebody will understand it and will actually want to use it, makes you one HECK of an expert in my book. And trust me, just like you live from IT, I live from identifying and employing experts - it's almost a definition of my job as a project manager. So yes, you are an expert. Sorry if I changed your self-image