Hello Ronny! Indeed, if you get a kink that occurs during installation, you can typically remove it by massaging it out. The cable will likely Certify just fine. The trick is spotting it before there is not enough excess slack to remove the kink.
what really matters in my mind with my experince is the insulation o nthe actual pairs, it is soft and mash till the copper shows through. as long as they aren't shorted or leaking
@@cdoublejj Hello and thanks for your comment. You should be concerned about much more than just shorts. Ethernet is not electrical cable, and cannot be treated as such. Ethernet cable construction is precision engineered for balance in order to provide Category performance. Although relatively durable, it is possible to pull too hard and distort the conductor pairs that will unbalance the cable and cause transmission issues. It is also possible to apply too sharp of a bend (especially to solid copper Ethernet) and ruin performance that way, too.
This test was with a patch cable, which are designed to be flexible enough to soak up funny handling. A solid core isn't going to behave as well with an actual kink. Maybe it will still pass that cert test but it's not designed for this kind of abuse. Either way as you said this just a theoretical test at best conditions possible, real environments don't care about whether the cable is certifiable they care about how many packets get corrupted, esp as the electromagnetic fields along the length of the cable fluctuate.
Hello pubcollize! Getting empirical data on performance loss would be nice. As you point out, a Fluke DSX-8000 won't reveal on-the-ground performance loss data since it takes a brief snapshot test for Certification purposes. Agreed, solid copper Ethernet will be far more sensitive and provide a truly interesting test. We are working on a mad scientist lab for this kind of testing and more, with a BERT tester that will show packet loss in real time. We might even take crazy test requests and do content on them and credit the requestors. How cool is that?
That is a great idea and one we are thinking of doing. This would require a BERT tester, which measures packet drops and can be dialed up to 10G as well. Separation distances at various speeds, for example, would be something we would focus in on. The Fluke DSX-8000 we use provides a snapshot test that does not demonstrate EMI issues (unless the EMI issue is severe enough to cause a failure during the snapshot test). We are looking into various options, but that is something we wish to do, and I don't believe anyone else has done it to date. Who knows, we might change our recommendations based on empirical data as opposed to NEC/BICSI/TIA guidelines!
I don't think that will affect ethernet signals... it is an easily rejected "common mode" low frequency signal... the transformers on either end of the line will just not pass it. of course I know there are rules of thumb to keep your low voltage away from the AC wiring but it's not as real a problem as people think for all signal types.
@@Sean_y4k2l5 I read a comment where they said VOIP only started working reliably when they switched to shielded ethernet cable that had to run alongside mains cable. Surely VOIP is just data packets so could some interference run down the ethernet cable to affect analogue telephone adapters?
@@BALLOOROOM interesting. I don't know... I mean, if the link is bad you will see packet loss regardless of what protocol you run over it... that is one thing to measure if trying to get an objective sense of what's going on. it is possible the interference was having some secondary effect on the electronics, besides, or in addition to the ethernet link.
@@Sean_y4k2l5 Hello Sean! Yeah, the bends will actually affect packetized data if severe enough. The issue shows up more with solid copper Ethernet, in real installations. Totally agree on AC branch circuit (and breaker box) separation. We have something in the works to test all of this and more. I can see from the number of folks requesting that kind of data that those videos will be very popular and satisfy curiosity. I have to admit, I want some hard data and my own curiosity need to be satiated as well.
It would be nice to see how this (crosstalk and return loss) translates to real world performance. Like will it limit the throughput to 800Mbit rather than 1Gbit or similar. Or are bends/kinks purely about earlier than expected failures?
Hello Goon! Agreed. The damage and performance loss OVER TIME is the larger consideration (plus early failures), but getting empirical data on performance loss would be nice. A Fluke DSX-8000 won't really reveal on-the-ground performance loss data since it takes a brief snapshot test for Certification purposes. The kind of testing you (and I) both would like to see involves a BERT or Bit Error Rate Tester that allows for a certain speed to be established (10G, for example) and then what-if scenarios to be experimented upon. The same thing would also allow for exhaustive AC power circuit and Ethernet separation testing! We are working out a testing lab and video studio combination that will allow that and much more. Don't worry, we are going to go all "mad scientist" at some point!
Several years ago, I had to replace some cables in an office. The Bozo that installed them put them hard into and around corners and also used bare wire staples to hold them down. That sort of thing might work with phone cables, but not Ethernet.
Hello James! Indeed. Using metal staples for Ethernet is a serious no-no. Cable jacket compression caused by the staple, plus the metal of the staple itself cutting into the jacket over time, will cause problems. Hard kinks with solid copper Ethernet are also not good.
I'd be really curious to see how these numbers compare to, say, if you seriously kink the cable and then straighten it out again (and if it changes if you repeatedly bend/unbend the cable multiple times). This is a common scenario, and it seems to me that the process of bending _and then unbending_ the cable (especially repeatedly) might cause the internal twisted pairs to loosen or deform, which could affect signal quality even if the final result looks "straight enough". Also, it would be really nice if at the end of tests like this you actually put up a table with the different values under the different conditions, so the results can be easily compared for those of us watching.
Hey, again Foogod! Yup, and I will be sure to add that to the what-if testing scenario. It should be interesting to see how solid copper deals with it vs. stranded copper, that is designed to handle that. We are likely to use a Softing BERT device (Bit Error Rate Tester) that is going to cost some serious coin, but will be worth it. It can inject 1G/2.5G/5G/10G application packets across a cable and then do error checking. You can also let it run for days if need to!
Thanks! I wonder if older cables, such as Cat3, were more susceptible to tight bends. Also, bending an unbending say 10 times or more... would that degrade the cable?
Hello John! I don't think bend issues are Category-related, or at least not per the ANSI/TIA standard. They are related to the overall cable diameter and typically stated as 4X OD bend radius for common U/UTP or 8X OD bend radius for F/UTP (overall foil shield), unless otherwise specified by the manufacturer. The issue is not just performance related, but cable jacket damage over time is another issue. As for bending a cable repeatedly, YES, if the conductors are solid copper you can break the conductors that way. I have not tested the number of times it takes to do it, however. This is why solid copper Ethernet is for permanent installation and patch cords are made from stranded copper conductors (flexible & bendable without damage).
What is a inexpensive and/or DIY solution for analyzing cable run performance in my house? The fluke tools are too expensive for me. Ideally, I would like an open source pcb design and software for a testing tool with a couple of ESP32's or other SBC boards like the raspberry pi-based solutions I could build and operate for testing my cable runs with a good deal of granular insight into the quality of my network wiring and terminations. EMF issues would be good too.
Hey Karl. My first response: Yikes. Thorny subject. Basically, if you want to DIY it, you would first use a simple wire map tester (which tells you nothing about performance) and then use some sort of software tool to check for packet loss while you shovel large files (1TB size, for example) across the cable. That is pretty much the least expensive way. I used to use Linux boxes and iperf to do that. One machine is set up as the "server" and the other as the "client," with a switch in the middle. As far as a discreet device that can check cables by itself for any degree of performance and accuracy (like a bandwidth speed test at various speed settings), you just signed up for $2,000 minimum. The Fluke LinkIQ is the least expensive cable "qualifier" out there that actually works and gives you anything useful. It can bandwidth qualify to 10G. Rumor has it that it is eerily accurate, like Fluke DSX-8000 accurate, for spotting bad runs. It provides accurate length and diagnostic data, too. EMI/RFI issues are best checked with a dedicated BERT tester that you can set and let run at certain speeds (big money). Interestingly, the Linux box + iperf method is good for that kind of test!
Yeah, for personal use I generally just get two computers (e.g. laptops) with gigabit adapters and do some sustained high-speed transfers, and measure the transfer time and packet loss reported by the OS (make sure you check them first with a known-good connection, though (i.e. a short, known good patch cable), because sometimes some network adapters or software can't actually do the full performance they claim to, even when the cable is just fine, for various other reasons). Ultimately, dB numbers and cable certification tests aren't what really matters in the end anyway; what matters is whether you actually get significant packet loss or not when you actually try to use it, which any PC should be able to tell you, really (with the right software).
@@foogod4237 Hey Foogod! I cannot help but chime in: "because sometimes some network adapters or software can't actually do the full performance they claim to, even when the cable is just fine, for various other reasons)". I think I will start a controversy and put the "various other reasons" as any variation of Windows OS. I have had a lot of trouble getting a good bead on NIC > NIC performance using it. Windows is too chatty and has tons of overhead. That is why I use Linux and iperf. It seems able to provide a better picture. Ultimately your stated method is going to be good enough short of someone who installs structured cabling systems for a living. Note on patch cords: WATCH OUT. Industry estimates are that 75% of imported patch cords don't pass patch cord Certification. I have tested hundreds of patch cords and can say for a fact that is the case. I often cannot get so-called "Cat6A" patch cords to even Certify on Cat6 patch cord testing. Many won't even Certify to Cat5e. How does that show up? It shows up when the patch cord is combined with a solid copper permanent link and used to complete the overall channel. In early May, trueCABLE will be selling 100% Certified Cat6 UTP 28AWG patch cords that you won't have to wonder about. They are guaranteed to pass on a Fluke DSX-8000 with Cat6 PCA adapters and the correct test limits.
Your video is very good, just would like to add one thing. Seems that the cable you are testing could be damaged, a good cable Cat 5 cable should be over 27 dB on Next value and over 10 dB in return loss. Maybe you can change the cable and show a bigger difference on how bending have a risk on the cable?
Hello Omar! The values you provided are likely for Channel testing, where you would expect to see such excessive and unrealistic dB numbers when testing a patch cord alone. Remember that Channel testing does NOT measure the very end-point terminations, which means when you Channel test a patch cord that only has the two end-point terminations, you are effectively only testing the cable and not the cable + terminations. Channel testing has high limits as it expects three cables and multiple terminations in the middle (keystones, etc.). I was using a far tighter Patch Cord (PCA) test, which holds the test limits to those of Permanent Links and also tests NEXT and RL at the connector. Patch cord testing assumes only one cable and two terminations and significantly tightens up the test limits accordingly. I think that is the difference you spotted and the reason for the disparity.
Hello Elijah! Correct. Don't kink cables, don't scuff the cable jacket when pulling, and don't pull too hard. Pretty basic stuff, but you would be shocked at how many issues in the field are related to poor installation practices!
While pulling cables, a cable might get kinked. When we notice that, we "massage" the kink out of the cable and that usually does the trick.
Does it? Does re-straightening a kinked cable restore the original performance?
Hello Ronny! Indeed, if you get a kink that occurs during installation, you can typically remove it by massaging it out. The cable will likely Certify just fine. The trick is spotting it before there is not enough excess slack to remove the kink.
what really matters in my mind with my experince is the insulation o nthe actual pairs, it is soft and mash till the copper shows through. as long as they aren't shorted or leaking
@@cdoublejj Hello and thanks for your comment. You should be concerned about much more than just shorts. Ethernet is not electrical cable, and cannot be treated as such. Ethernet cable construction is precision engineered for balance in order to provide Category performance. Although relatively durable, it is possible to pull too hard and distort the conductor pairs that will unbalance the cable and cause transmission issues. It is also possible to apply too sharp of a bend (especially to solid copper Ethernet) and ruin performance that way, too.
This test was with a patch cable, which are designed to be flexible enough to soak up funny handling. A solid core isn't going to behave as well with an actual kink. Maybe it will still pass that cert test but it's not designed for this kind of abuse.
Either way as you said this just a theoretical test at best conditions possible, real environments don't care about whether the cable is certifiable they care about how many packets get corrupted, esp as the electromagnetic fields along the length of the cable fluctuate.
Hello pubcollize! Getting empirical data on performance loss would be nice. As you point out, a Fluke DSX-8000 won't reveal on-the-ground performance loss data since it takes a brief snapshot test for Certification purposes. Agreed, solid copper Ethernet will be far more sensitive and provide a truly interesting test. We are working on a mad scientist lab for this kind of testing and more, with a BERT tester that will show packet loss in real time. We might even take crazy test requests and do content on them and credit the requestors. How cool is that?
Maybe run some cables next to some mains lines and see what that does to performance?
That is a great idea and one we are thinking of doing. This would require a BERT tester, which measures packet drops and can be dialed up to 10G as well. Separation distances at various speeds, for example, would be something we would focus in on. The Fluke DSX-8000 we use provides a snapshot test that does not demonstrate EMI issues (unless the EMI issue is severe enough to cause a failure during the snapshot test). We are looking into various options, but that is something we wish to do, and I don't believe anyone else has done it to date. Who knows, we might change our recommendations based on empirical data as opposed to NEC/BICSI/TIA guidelines!
I don't think that will affect ethernet signals... it is an easily rejected "common mode" low frequency signal... the transformers on either end of the line will just not pass it.
of course I know there are rules of thumb to keep your low voltage away from the AC wiring but it's not as real a problem as people think for all signal types.
@@Sean_y4k2l5 I read a comment where they said VOIP only started working reliably when they switched to shielded ethernet cable that had to run alongside mains cable. Surely VOIP is just data packets so could some interference run down the ethernet cable to affect analogue telephone adapters?
@@BALLOOROOM interesting. I don't know... I mean, if the link is bad you will see packet loss regardless of what protocol you run over it... that is one thing to measure if trying to get an objective sense of what's going on.
it is possible the interference was having some secondary effect on the electronics, besides, or in addition to the ethernet link.
@@Sean_y4k2l5 Hello Sean! Yeah, the bends will actually affect packetized data if severe enough. The issue shows up more with solid copper Ethernet, in real installations. Totally agree on AC branch circuit (and breaker box) separation. We have something in the works to test all of this and more. I can see from the number of folks requesting that kind of data that those videos will be very popular and satisfy curiosity. I have to admit, I want some hard data and my own curiosity need to be satiated as well.
It would be nice to see how this (crosstalk and return loss) translates to real world performance.
Like will it limit the throughput to 800Mbit rather than 1Gbit or similar.
Or are bends/kinks purely about earlier than expected failures?
Hello Goon! Agreed. The damage and performance loss OVER TIME is the larger consideration (plus early failures), but getting empirical data on performance loss would be nice. A Fluke DSX-8000 won't really reveal on-the-ground performance loss data since it takes a brief snapshot test for Certification purposes. The kind of testing you (and I) both would like to see involves a BERT or Bit Error Rate Tester that allows for a certain speed to be established (10G, for example) and then what-if scenarios to be experimented upon. The same thing would also allow for exhaustive AC power circuit and Ethernet separation testing! We are working out a testing lab and video studio combination that will allow that and much more. Don't worry, we are going to go all "mad scientist" at some point!
Several years ago, I had to replace some cables in an office. The Bozo that installed them put them hard into and around corners and also used bare wire staples to hold them down. That sort of thing might work with phone cables, but not Ethernet.
Hello James! Indeed. Using metal staples for Ethernet is a serious no-no. Cable jacket compression caused by the staple, plus the metal of the staple itself cutting into the jacket over time, will cause problems. Hard kinks with solid copper Ethernet are also not good.
@@trueCABLE That too, but the metal that close to the pairs can cause an impedance bump.
@@James_Knott Hello James. Yup, combine all that and you have problems!
Ahahaha bozo made me laugh
I'd be really curious to see how these numbers compare to, say, if you seriously kink the cable and then straighten it out again (and if it changes if you repeatedly bend/unbend the cable multiple times). This is a common scenario, and it seems to me that the process of bending _and then unbending_ the cable (especially repeatedly) might cause the internal twisted pairs to loosen or deform, which could affect signal quality even if the final result looks "straight enough".
Also, it would be really nice if at the end of tests like this you actually put up a table with the different values under the different conditions, so the results can be easily compared for those of us watching.
Hey, again Foogod! Yup, and I will be sure to add that to the what-if testing scenario. It should be interesting to see how solid copper deals with it vs. stranded copper, that is designed to handle that. We are likely to use a Softing BERT device (Bit Error Rate Tester) that is going to cost some serious coin, but will be worth it. It can inject 1G/2.5G/5G/10G application packets across a cable and then do error checking. You can also let it run for days if need to!
Thanks! I wonder if older cables, such as Cat3, were more susceptible to tight bends. Also, bending an unbending say 10 times or more... would that degrade the cable?
Hello John! I don't think bend issues are Category-related, or at least not per the ANSI/TIA standard. They are related to the overall cable diameter and typically stated as 4X OD bend radius for common U/UTP or 8X OD bend radius for F/UTP (overall foil shield), unless otherwise specified by the manufacturer. The issue is not just performance related, but cable jacket damage over time is another issue. As for bending a cable repeatedly, YES, if the conductors are solid copper you can break the conductors that way. I have not tested the number of times it takes to do it, however. This is why solid copper Ethernet is for permanent installation and patch cords are made from stranded copper conductors (flexible & bendable without damage).
What is a inexpensive and/or DIY solution for analyzing cable run performance in my house?
The fluke tools are too expensive for me. Ideally, I would like an open source pcb design and software for a testing tool with a couple of ESP32's or other SBC boards like the raspberry pi-based solutions I could build and operate for testing my cable runs with a good deal of granular insight into the quality of my network wiring and terminations. EMF issues would be good too.
Hey Karl. My first response: Yikes. Thorny subject. Basically, if you want to DIY it, you would first use a simple wire map tester (which tells you nothing about performance) and then use some sort of software tool to check for packet loss while you shovel large files (1TB size, for example) across the cable. That is pretty much the least expensive way. I used to use Linux boxes and iperf to do that. One machine is set up as the "server" and the other as the "client," with a switch in the middle. As far as a discreet device that can check cables by itself for any degree of performance and accuracy (like a bandwidth speed test at various speed settings), you just signed up for $2,000 minimum. The Fluke LinkIQ is the least expensive cable "qualifier" out there that actually works and gives you anything useful. It can bandwidth qualify to 10G. Rumor has it that it is eerily accurate, like Fluke DSX-8000 accurate, for spotting bad runs. It provides accurate length and diagnostic data, too. EMI/RFI issues are best checked with a dedicated BERT tester that you can set and let run at certain speeds (big money). Interestingly, the Linux box + iperf method is good for that kind of test!
Yeah, for personal use I generally just get two computers (e.g. laptops) with gigabit adapters and do some sustained high-speed transfers, and measure the transfer time and packet loss reported by the OS (make sure you check them first with a known-good connection, though (i.e. a short, known good patch cable), because sometimes some network adapters or software can't actually do the full performance they claim to, even when the cable is just fine, for various other reasons).
Ultimately, dB numbers and cable certification tests aren't what really matters in the end anyway; what matters is whether you actually get significant packet loss or not when you actually try to use it, which any PC should be able to tell you, really (with the right software).
@@foogod4237 Hey Foogod! I cannot help but chime in: "because sometimes some network adapters or software can't actually do the full performance they claim to, even when the cable is just fine, for various other reasons)". I think I will start a controversy and put the "various other reasons" as any variation of Windows OS. I have had a lot of trouble getting a good bead on NIC > NIC performance using it. Windows is too chatty and has tons of overhead. That is why I use Linux and iperf. It seems able to provide a better picture. Ultimately your stated method is going to be good enough short of someone who installs structured cabling systems for a living. Note on patch cords: WATCH OUT. Industry estimates are that 75% of imported patch cords don't pass patch cord Certification. I have tested hundreds of patch cords and can say for a fact that is the case. I often cannot get so-called "Cat6A" patch cords to even Certify on Cat6 patch cord testing. Many won't even Certify to Cat5e. How does that show up? It shows up when the patch cord is combined with a solid copper permanent link and used to complete the overall channel. In early May, trueCABLE will be selling 100% Certified Cat6 UTP 28AWG patch cords that you won't have to wonder about. They are guaranteed to pass on a Fluke DSX-8000 with Cat6 PCA adapters and the correct test limits.
Your video is very good, just would like to add one thing. Seems that the cable you are testing could be damaged, a good cable Cat 5 cable should be over 27 dB on Next value and over 10 dB in return loss. Maybe you can change the cable and show a bigger difference on how bending have a risk on the cable?
Hello Omar! The values you provided are likely for Channel testing, where you would expect to see such excessive and unrealistic dB numbers when testing a patch cord alone. Remember that Channel testing does NOT measure the very end-point terminations, which means when you Channel test a patch cord that only has the two end-point terminations, you are effectively only testing the cable and not the cable + terminations. Channel testing has high limits as it expects three cables and multiple terminations in the middle (keystones, etc.). I was using a far tighter Patch Cord (PCA) test, which holds the test limits to those of Permanent Links and also tests NEXT and RL at the connector. Patch cord testing assumes only one cable and two terminations and significantly tightens up the test limits accordingly. I think that is the difference you spotted and the reason for the disparity.
So basically, so long as you use a little common sense when installing ethernet cables you'll probably be perfectly fine.
Hello Elijah! Correct. Don't kink cables, don't scuff the cable jacket when pulling, and don't pull too hard. Pretty basic stuff, but you would be shocked at how many issues in the field are related to poor installation practices!