Cable validator

[Raj]

So I saw this video from Linus tech tips-
https://youtu.be/u6lx1ntNoxE?t=88
I thought, cable testing just needed a continuity tester (be it modded to beep when continuity is broken) and sockets (to actually get access electrically, of the pins of the connector) as long as you have the cable, well under standard maximum supported length.
Sure, you can have fixtures for a factory environment but expensive oscilloscopes? for digital signal? simple lcr meter ain't enough?

[Kleinstein]

For really demanding cables one may actually need quite some effort to really test a cable all the way to the details. The box does not look really dodgy, though may be already at the limits with HDMI.
The video however looks a bit odd - for me the way the video is made rings some alarm bells and makes me a bit suspecious.

[TimFox]

I haven't viewed the video, but there are cable faults that will not be measurable with continuity or LCR meter.
The usual test equipment required for some types of faults is "TDR" for time-domain reflectometry.
For example, back in WW II, it was noticed that the manufacturing process for coaxial cable could produce a periodic deviation in the dielectric diameter along the cable length, which would result in interesting interference effects at certain frequencies where a small reflection at each cycle of the deviation could add up to serious problems for a long cable.

[Kleinstein]

The cable tester seems to measure the eye diagrams, which is standard for testing transmission systems. However a large part of the error is usually from the send and receive side and more like an amplifier if present. The cable is only one part. The more tricky part could be the connectors and here half of them are in the tester, with additional one for the front ends.

[Raj]

I haven't viewed the video, but there are cable faults that will not be measurable with continuity or LCR meter.
The usual test equipment required for some types of faults is "TDR" for time-domain reflectometry.
For example, back in WW II, it was noticed that the manufacturing process for coaxial cable could produce a periodic deviation in the dielectric diameter along the cable length, which would result in interesting interference effects at certain frequencies where a small reflection at each cycle of the deviation could add up to serious problems for a long cable.

aren't those for RF? I do know the phenomena of line ringing and importance of termination of coax so they are indeed more complex than twisted pairs and straight lines that are generally used in HDMI, ethernets and USB.

[Kleinstein]

One can get the same periodic destortions with twisted pairs - just a periodic patern in how close the twists are can cause this. Worst case this could act as a kind of bragg refletor, adding a sharp peak to the frequency response. This may need a rather long cable to really become a problem.
In that respect twisted pair is not better than coax, it may be even more tricky to get a really homogenous cable. This could especially be a problem with many thin twisted pairs bundled.

Twisted pair termination being differential can be also complex - one can get refections there as well. The consumer systems have quite some SNR reserve and can tolerate some reflections, more than more analog RF communication systems.

[Ranayna]

In a twisted pair ethernet cable, it is important to not split the pairs. One pair each to pin 1 and 2, to pin 3 and 6, to pin 4 and 5, and the last pair to 7 and 8.
If you split one of these pairs, for example you just do 1/2, 3/4, 5/6, 7/8, a cheap continuity tester will not find the error if you do it like that on both ends. But network performance will be severly limited, if you even get a connection at all.
Also, as far as i know, the twist ratio for each pair is different, which can cause issues if you swap pairs.

I have personal experience with dicky connections, that could not be located with a cheap tester. The cheap crap claimed the cable is ok, but i could not even get a 100 Mbit/s link. Getting the big guns out, i.e. the cable certifier, immediately found the issue.

[AVGresponding]

The twist ratio is indeed different for each pair, which helps reduce crosstalk: https://en.wikipedia.org/wiki/Category_5_cable

I have a couple of sets of Wavetek LT8600s that I use; other brands are available. These will analyse the properties of the cables you just pulled in, to make sure the apprentice didn't bugger them up...
They can tell you the length of the cable, dB losses over a frequency sweep etc etc.

[ConKbot]

aren't those for RF? I do know the phenomena of line ringing and importance of termination of coax so they are indeed more complex than twisted pairs and straight lines that are generally used in HDMI, ethernets and USB.

What part of a cable that passes signals in the GHz range isn't RF? HDMI 2.1 demonstrated in the LTT video passes 48gbps over 4 pairs(HDMI finally adopting using all 4 pairs for data and a packet-based data transfer like display port has used for over a decade...), 12gbps each, so at least 6GHz of bandwith for each pair.  There is equalization and pre-emphasis that can help correct for some amount of loss, but it can only help so much.

The signals are running so fast, that a 3m long HDMI 2.1 cable at the full data rate will have 28-36 bits of data in the cable at any given moment depending on the velocity factor of the transmission lines.
Even A lowly USB 3.0 cable that is 2m long will have 2-3 bits in it (each direction) at the same time.

[newbrain]

A small personal experience with twisted pair:
I have an Ethernet connection to my ISP, coming from some optical network equipment in the building basement.

Some years ago, my ISP switched for free from 100 Mb/s to 250 Mb/s, after some days I was sure I was still getting only 100 Mb/s.
Moreover, I was never able to get even that unless I used some specific HW (e.g. an Asus router would work, a Linksys one would not).

The ISP technician came, confirmed that the ports were working correctly, and proceeded to test the cabling:
Electrical connection was there with low resistance, but the max speed his tester achieved was 10 Mb/s - so it seems my Asus router was really overachieving on a marginal cable!

Luckily, a parallel, identical, cable was already installed from the basement to my apartment which tested perfectly up to 1 Gb/s.

So, no, connection is not enough, even for very tolerant standards such Ethernet (I once wired a magjack totally wrong to the PHY, still got 100 Mb/s on a couple of meters).

[ANTALIFE]

Total Phase are known to make quality gear, so I would not have a reason to think that the cable tester is dodgy

Like others have said, signal integrity at higher frequencies is tricky to test and I am fairly sure they would be using the correct eye diagram limits. What I will add is that cable SHIELD plays a key role in signal integrity, and what you will find with chapter cables is that the SHILED does not have a good RF bond to the connector body. For example with cheap HDMI cables the bond would usually be via a single strand of wire to the connector SHIELD (like a pig tail), which is not ideal as it becomes a high inductance path effectively reducing EMI resilience

[sandalcandal]

Cross referencing the other thread on the same product and LTT video here: https://www.eevblog.com/forum/chat/is-this-product-a-bs-didnt-it/