CAT3 vs. CAT5

[rbertram]

I just setup a wireless Internet connection. My antenna plugs into a Linksys router w/f port switch. If I connect into the router/switch with my PC, down/up speeds are correct. When I uplink the router to another 16-port switch, my download speeds are cut by a third, but the upload is still ok.

So I started playing around with different CAT5 and CAT5e cables and found no difference. When I tried a CAT3 cable, the download speed went back to normal. To confirm that my CAT5 cable was ok, I cut one end off and converted it to CAT3, and it worked perfectly.

Why would I get higher download speeds with CAT3? It is not a big deal since the broadband connection is only 2.5Mbps, but it prevents me from get higher network speeds in my LAN if I want to use the other 3 ports on my router.

 

[vaoldschl]

Sounds like the end on your original cable was bad. I don't fully understand what you mean by converting a CAT5 cable to CAT3, you can't convert a cable from CAT5 to CAT3, you can put a CAT3 end on a CAT5 cable, though so I assume that's what you did. Anyway, if you cut the end off a cable that was giving you trouble, reterminated it and it stopped giving you trouble (CAT5 v. CAT3 aside) I'd say the original end was defective.

 

[rbertram]

Yes, I just made one end CAT3.

It is hard for me to believe that I had 3 bad CAT5 cables. The chance that I randomly cut one end off of one of the CAT5 cables and converted it to CAT3 would make a difference is very slim.

 

[jimbopalmer]

try to explain what 'converting to Cat 3' means to you, as I am confused by the term, are you just using an old end on it? untwisting the wire to defeat the cat5 noise resistance? are you making a crossover cable? I tried to remain child-like, all I acheived was childish.

 

[rbertram]

What I did exactly is on the one end I only installed the Tx and Rx wires. Completely removed the unused blue and brown pairs (pins 4,5,7 & 8) from that end.

No, I did not make or use a crossover cable. My 16-port switch is auto sensing. However, I did try both the sraight and crossover positions on the uplink port on the router, and results were identical.

 

[jeffmoss26]

That would be 10BaseT (2 pairs )wiring vs. 100BaseT wiring (which I believe uses 4 pairs, correct me if I am wrong) jeff moss

 

[vaoldschl]

Putting a CAT3 end on a CAT5 cable doesn't make the whole unit CAT3, it just makes it not CAT5. Also, removing pairs 4,5,7&8 doesn't constitute CAT3 either, CAT3 cabling also consists of 4 pair. I'm just offering this as information since I believe it to be moot in terms of the problem.

Pertinent information is that when you reterminated the end of that cable it worked. Where did your three cables come from? Did you make them or buy them? Did you run a scanner over them? I say the cables are the weak link here, if you made them consider buying some standard patch cables, if you did buy them I'd throw a scanner on, if available and see what kind of performance you're getting.

 

[jeffbouldin]

jeffmoss,

10baseT and 100baseT use the same two pairs, on both you can leave off the blue and brown and it will work.

 

[jeffmoss26]

thanks for the clarification jeff moss

 

[daronwilson]

Interesting. Just to clarify, the Cat3 cable has to do with the bandwidth of the cable, a function of the number of twists per inch on each pair (among other things) and Cat5 had many more twists per inch and a higher bandwidth. Bandwidth on either cable is more than enough for what you are doing. How many wires you terminate out of the 8 have no corelation with Category 3,4,5,6, etc. However I will tell you that generally the circuit is designed to have all four pairs terminated on the jack/plug.

I think we are dealing with two different issues here.

First - ethernet connectivity - that is the two pair you need to make 10 mbs or 100 mbs ethernet work on your switch. Assuming you have regular Cat5 patch cables, the test SHOULD be downloading a file from one PC on the network to another PC on the network via the ethernet connection. Now...assuming that transfer is good, I would use it to measure the quality of your cables and connections. Using your internet is a very undependable method as you have no control or verification on what is happening on the WAN side of the router, any number of things could adversely affect your throughput.

Second - decreased throughput when the 16 port switch is connected. Again this may be a bit tough to troubleshoot as the WAN side of your router may not provide you a reliable, but if I understand your post, you had good speeds when plugged into the router, but when you go through the switch you get 1/3 the throughput. Another good test would be to see if this configuration slows down your PC to PC transfer or if it only affects your internet downloads. Perhaps swapping the switch with another one could narrow it down.

Personally I vote for bad/slow switch or bad patch cables.

One problem for us trying to help is we need the details. Since there is already some confusion about what cat3 and cat5 are, perhaps if you can give us the model of the switch, what else you have connected, speed of the switch, where you went to test this download and upload speed, etc. I know it may seem like a lot to explain, but there are enough people here who want to help and can be very helpfull if armed with the information.

Good Luck, let us know what you find out!

It is only my opinion, based on my experience and education...I am always willing to learn, educate me!
Daron J. Wilson, RCDD
daron.wilson@lhmorris.com

 

[jimbopalmer]

A fair number of the Wireless Access Points allow the use of POE (Power over Ethernet) using the 'unused' pairs, if the other end also 'paid attention' to the 'unused' pairs (100baseT4, 1000baseT, etc) then cutting the unused pairs out may have stopped a conflicting protocol. Just a thought. I tried to remain child-like, all I acheived was childish.

 

[rbertram]

OK, I have run some more tests. Unfortunately, I do not have all of the necessary equipment to test the termination of my cables.

Test #1 – Throughput between 2 PCs connected directly to 16-port switch.
94.118 Mbps

Test #2 – Throughput between 2 PCs. 1 PC connected to the router and the other connected to 16-port switch. CAT5e (with all pins) uplinking the router to the switch.
91.954 Mbps

Test #3 – Throughput between 2 PCs. 1 PC connected to the router and the other connected to 16-port switch. CAT5e (with only Tx & Rx pins) uplinking the router to the switch.
93.023 Mbps

The above throughput figures are the average of 5 tests each.

The router is a Linsys (Network Everywhere) NR041.
The 16-port switch is a Dell PowerConnect 2016.

To test the LAN troughput, I am using NetIQ Qcheck.
To test Internet speed, I am using

http://chi.speakeasy.net.

My Internet antennae is a Motorola Canopy Subscriber Module.

 

[rbertram]

I did one more test.

Test #4 – Throughput between 2 PCs. 1 PC connected to the router and the other connected to 16-port switch. CAT3 (with only Tx & Rx pins) uplinking the router to the switch.
80.001 Mbps

I als get the higher Internet speed with the CAT3 cable.

 

[Bobg1]

If you are thinking that you want low numbers on these tests
you are making, you are mistaken. The higher the number the better. And from what I see the test you have labeled as cat5 are better than the one labeled cat 3.

 

[rbertram]

I am not sure what your point is Bobg1.

I know my throughput test for the CAT5e are better than the CAT3, and it should be.

I posted the test because they show that there is no apparent problem with my cables, and that I should not be getting a slower Internet connection using the CAT5e cable versus the modified CAT5e or CAT3 cable.

 

[daronwilson]

SO...how are you testing the slow internet connection??

I see your test data, I understand that no matter which cables you use that your throughput through the switch is very good. So...I would say we could rule out the cables and the switch.

Now if we examine it further it looks like the connection between the router and the switch is as fast as just from switch port to switch port (roughly). So...I would say we can rule out the connection between the switch and the router.

So what is left? Your test results are showing me that data flow between PC's and router/switch are all acceptable no matter which cables you use. If that is the case, there can't be a slowdown in the switch/cable/router end of things.

The only thing left is the WAN (internet) side of the router. So what were the test results on the Internet download speed tests?
It is only my opinion, based on my experience and education...I am always willing to learn, educate me!
Daron J. Wilson, RCDD
daron.wilson@lhmorris.com

 

[rbertram]

My wireless Internet connection speed is 512/512 with a 2.5 Mbps burst rate with a 12 MB cache.

These are speed tests done when connected to the 16-port switch with a standard CAT5e cable uplink to the router.

749/496
436/501
644/495
469/471
519/497

These are the speed tests using the CAT3 or modified CAT5e uplink cable.

1512/492
1368/496
1708/461
1702/496
1691/496
1502/497

Clearly shown is a very constant upload speed, but there is about a 3x difference on the download rate.

 

[franklin97355]

What model is the second switch(the 16 port one)?

 

[rbertram]

Look 7 posts up.

The 16-port switch is a Dell PowerConnect 2016.

 

[daronwilson]

Well we can't seem to get you past the Cat5e Cat3 thing, which confuses me when you post your results. I have never heard of a speed decrease of any type with a 4 pair Cat5e patch cable (provided it is good). I can see no reason for an increase when using two pairs of wires on exactly the same circuit UNLESS the other two pairs were being used by one of yoru devices for PPOE or something similiar. Odd, but I suppose that could have some effect.

In your post of internet download tests, the slow speed one says you are through the 16 port switch, but I'm unclear what the configuration is on the second (faster) tests. Is the second list of faster tests STILL going through the 16 port switch just with your 2 pair cable?

Sorry, but if we can stick with accurate terminology and overly descriptive posts on your tests and setup, hopefully we can help you narrow it down.

Rereading I see that perhaps we are only dealing with the uplink cable as being the problem? The cable between the router and the switch? If THAT is the case, I too vote for something different being done on that router port (such as PPOE) that you are eliminating when you drop those two extra pairs. Although it seems that the throughput you measured from PC to PC via the same setup was ok. That would make me think of some filtering/firewall issue, also I'd want to know that you only used IP for your testing.

Hmm, lots to consider, interesting problem, I hope you can nail down the specific of what it is.

Good Luck!

Have we looked at the router configuration and/or changed the ports out of the router that we are using? Are you using any firewall software in the router? It is only my opinion, based on my experience and education...I am always willing to learn, educate me!
Daron J. Wilson, RCDD
daron.wilson@lhmorris.com