The Ethernet speed is expressed in GB/sec second based on Shannon’s law…and the test data saying you meet it with connectivity attached as a permanent link or channel (patch added).
Each upgrade is 1.2 gb/sec on up to 4.8 gb/sec and 10G is 10 gb/sec. The usable guaranteed analog (not digital) BW is listed to the right. This, over 328 foot / 100 meter lengths,
1200 cat5 - 160 MHz
2400 cat5e - 250 MHz
3600 cat6 - 280MHz
4800 cat6+ - 300 MHz
10G Ethernet - 625 MHz
100baseT Ethernet is 100MB/sec or 800 bits/sec.
400 mbps is only 400/8 (eight bits in a byte) = 50 MB/sec.
What all that says is that something is really broken in the channels. Connectors or something, 5e can easily handle your system.
The NIC card should auto negotiate (LED’s on the card show what it connects as) to 10/100 baseT older cards or 100/1000 newer cards baseT Ethernet. What are those NIC card LED’s showing? THAT is the BW the “channel” is meeting per the NIC cards data test.
Flashing means it is sending data. The color means the achieved data rate.
Blue Jeans will be happy to test a few of your cords. The connectors assembly quality and design will massively impact data tests. The data doesn’t lie so let’s get the data. I’m not sure what you are referencing as the “data rate” but the NIC card negotiates to the fastest speed the cable allows. As short as your cables are, I doubt you’ll negotiate slower than 1000baseT based on the NIC cards LED status.
Trust me on this, we sell millions of miles of Ethernet cable and if it wasn’t working to 10/100/1000/2.5G/5G/10G we’d hear about it. FAST.
Galen, I was an electrical engineer in my former life, with some small experience in both fiber and copper network design, so I am familiar with all this. I tried to set up tests with only one variable changing - the Ethernet cable being used. I bought 3 (maybe 4 - I did this last year) CAT6 cables from Blue Jeans in 1.5M lengths. With as controlled a test as I can make it - using the same computer with reboots after a cable swap, the same port on my Ethernet switch, the same regional servers selected from Ookla’s Speedtest, all tests done consecutively at the same time of day, the Blue Jeans cables (all of them) showed Speedtest results of 260-270mbps, consistently. As I said, my provider is supplying me with a connection capable of up to 400mbps. With any random CAT6 cable I had bought from Amazon I was (in these one after another test conditions) consistently hitting the expected 400mbps, if not often better. All I changed was the cables. Didn’t matter in what order the cable swaps were done. I can redo these tests to get the data, if you want, but for the moment you’ll have to trust an engineer’s reporting on what was seen.
Like I said, I’ve used plenty of Belden cable otherwise both personally and professionally, and have had no reason to question performance (although in truth that work was all with RF cables).
Well I just purchased many feet of the Belden Cat 6a cable from Blue Jeans for in the walls of my new house, so I surely hope that these cables are not constricting throughput compared to random cat6 cables from Amazon… And to be honest, I do not think they will. There must be some other cause to the results above. I just cannot fathom that the Belden cables themselves are the cause, and perform so far below their design capacity (for which BlueJeans also provides individual test reports per order). Good luck with the analysis and keep us in the loop. It surely is intriguing.
Frankly I can’t believe it either. We’d used so may Belden products during my professional life simply because they met or exceeded any requirements we had.
No, I don’t trust an engineer but the real data provided…that’s or job like it or not, to get the right numbers repeatable by anyone. So no offense to you or any engineer. But we make mistakes and data has to be reviewed when it makes no sense. Send one of the “working” cables in for test. The results aren’t even logical per Shannon’s law data supplied with your cables. Something is wrong and it isn’t the cable. I trust the data on the component first. Then we test ancillary stuff to make sure it is verified to, or well in excess of spec.
Once we test BOTH cables at the same time and with the same technician…let’s see what the data says.
There are too many variables in your test that you don’t control, like the Internet. Setup iperf on two endpoints on your LOCAL network and rerun your tests. If you’re on gigabit Ethernet, theoretical max is 940mbps, but you’ve got the switch and two endpoints to optimize.
Once you verify it’s going excess of 400bps, then look at your router/gateway to see if it can operate at that speed. Hook up a computer direct to it and turn off any packet inspection systems that tax your router’s CPU and rerun your speedtest to an endpoint run by your ISP. Even the best consumer routers can’t do packet classification or intrusion detection (IDS) at 400mbps, so these features have to be off. This is where you will find your problem.
Well @rower30, I didn’t like the ambiguity I had seen last year, so yesterday I reran the same tests I ran last year. Same equipment, same switch port, same cables. How, in just the space of literally less than two minutes for Ethernet cable swaps, I could get such consistently differing results last year, and when repeated just yesterday get no meaningful difference whatsoever between the Belden and the other cables, I can’t explain - nor do I really want to. All that matters is the cables are not the oddity in whatever happened last year. I withdraw my crude observations with apologies.
Now, as to what to run through the walls for the embedded infrastructure , this is something I don’t want to have to do more than once, if at all possible. I need to run copper for the backbone, but will also run fiber to at least the listening room (along with CAT). I don’t want to get into a Ken Olsen (paraphrased) ‘no one needs a computer in their home’ disagreement, but I think CAT6 or 6a might be not as future proof as I’d like. The question I have is whether running anything beyond 6a - at this point in time - will be problematic for existing hardware? (switches, etc, even PoE cameras).
That’s why we are all here…to help us get correct answers even if it takes a stab or two or three. We ALL make errors! Engineering can’t be done with too many errors and get meaningful results. Mother nature is mean like that. No one’s mad at you, or should be. Why? You put the problem out there for exactly what you got, another view and some extra thinking. I do the same for you making my cables. I allow my data to be criticized for accuracy. This is good for all of us.
OK, the future. It all boils down to how much data a home network will really need. Your home wire is your INTRAnet). Once you go outside your walls, to the ETHERnet that’s the real bottleneck. A home network can FAR exceed what’s coming into your house with typical 10G Ethernet cable wired INTRAnet. If you run 50 micron fiber, the current data standard size, make sure you get fiber VCSEL hardware approved and 10 G bandwidth rated. Fiber SHOULD have ample headroom. You can later use 850 or 1300 nm or even WDM, wavelength division multiplexing as BW needs arise. When they do arise I want you to tell me WHAT you are DOING!
I’m currently using Ethernet Power-Line Wall wart and cat6 patch! These power-line modems are like 30MEG (240 mbps) and stream fine. So we have the “worst case” INTRAnet! My INTERnet tests fine @ 235 mbps to the outside world. I have 200 mpbs service. So I don’t see the top-end bottleneck yet.
Yeah, we’re nowhere near pushing the limits right now of what 6 or 6a can do. But, to your knowledge are there any backward compatibility issues if one were to go with CAT7 or CAT8?
No, none I’m aware of up to 10G. The NIC cards are the limit on the top end so use a 10G NIC cared to get 10G. They did split up the 10G into 2.5 G (common on PC’s now) and 5 G. But those are more NIC card changes and not the cable. 10 G NIC hardware was too expensive for everyone jumping from 1G.
Thanks Galen. I searched back through this forum and reread your posts on CAT7. What I got from it, I think, is that if improperly used it could cause more problems than it’s worth.
Yes, shielded systems are for experts only to get the best out of them. The wrong ground system and or the removal of cable grounds ANYWHERE and combined with the damage to cables ALL impact performance significantly.
UTP is far, far more tolerant of system irregularities and cable and connector applications. STP and ISTP is “better” but ONLY if it is used 100% correctly and if not, by DESIGN, it gets far worse than UTP.
You can use it, yes, but make sure you do everything right and really be careful with the cable itself as damage to the shield alters the internal capacitance along the cable and thus the impedance uniformity and return loss reflections. Solid and even with bonded pair insulated UTP cable is far more tolerant of jacket wall pertubations.
Just adding a FTP (overall) shield to a “perfect” cable core reduces frequency swept internal NEXT 3-6 dB on average. This is from internal EM coupling through the shield. There is zip you can do to stop this. To prove it, remove the shield and retest NEXT! BINGO, it gets better.
If you use an ISTP, EM inductive coupling is largely removed because the pairs can’t see each other (the individual shield!). It is still evident as the frequency drops to 1 MHz, though so be aware of that. We shift from pure EM up high to magnetic as we go lower in frequency.
With ISTP we still have the impedance stability that is impacted by the close proximity to a tight ground plane which is the inside surface of the shield. The closer it is, the more ANY geometric shift alters the impedance more and more the closer the shield is. This hurts RL performance as the cable is not perfect, and we can damage the cable in use. Those FOAM dielectric of high air content used for attenuation and cable physical size are FRAGILE and DO NOT recover from being crimped or crushed.
Not to scare everyone but this is just the way it is as we increase performance on stuff. Knowledge is the key to getting what you pay for same as any high tech device and of any sort. As they say, it gets complicated.
Thanks Galen. To my understanding (again), the Belden 4800 is a CAT6 rated cable - is that correct? I ask about this one as you recommended that as the best, plus the one to use if one were to have PoE utilization. I have PoE devices throughout my home, but would like to lay in a 6A infrastructure. Is there a 6A equivalent to the 4800?
For 1000baseT Ethernet applications (no ALIEN NEXT between cables) the 4800 internal electrical exceeds 6A. The 4800 is the best POE and Ethernet, both design but it is $$$ as it uses a larger copper. DT600e’s ~0.0235" copper uses a special spline to maximize the Shannon’s law BW and still has good POE, but not as good as the 4800 that has that in mind but BOTH meet POE+. 4800 has a higher (lower DCR) MARGIN built-in with ~0.0243" copper.
To meet 6A you trade-off some internals for ALIEN and SIZE. We have two solutions;
10GXS series is a reduced size 6A design. The floating ISO-shield isolates the ANEXT between cables so size can be reduced.
And 10G series is a UTP type 10G, larger but no shield issues to deal with. To meet alien ANEXT the UTP has to be larger to attenuate the coupling between cables.
But if I want to swing for the cost, it sounds like 4800 is the way to go. How does one / can one go about ordering this cable at desired lengths and terminated in say a keystone jack configuration?
, this is something I don’t want to have to do more than once, if at all possible. I need to run copper for the backbone, but will also run fiber to at least the listening room (along with CAT). I don’t want to get into a Ken Olsen (paraphrased) ‘no one needs a computer in their home’ disagreement, but I think CAT6 or 6a might be not as future proof as I’d like. The question I have is whether running anything beyond 6a - at this point in time - will be problematic for existing hardware? (switches, etc, even PoE cameras).
