Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

A nearly perfect USB cable tester

blog.literarily-starved.com

Post reply on HN
Print view View original

151–160 of 181 posts

Re: A nearly perfect USB cable tester

#151

Post by SAI_Peregrinus »

atoav wrote: post #9

One thing to realize is that especially for high resolution video cables these cheap testers can't really deliver. The way to test them is a eye diagram (see: https://incompliancemag.com/eye-diagram-part2/ ) and testers with that capsbility cost upwards of 10.000 Eurodollars.

Keysight are nice enough to provide prices on their web site for all their cheaper equipment. I priced out a setup for USB 2.0 eye diagram compliance testing on their site in this comment¹ and it's more like $40,000 than $10,000.

¹https://news.ycombinator.com/item?id=47568399

Re: A nearly perfect USB cable tester

#152

Post by SAI_Peregrinus »

tom_alexander wrote: post #39
jmalicki wrote: post #24

Earlier quoted context omitted.

His link made very clear the issues of jitter and flickering.

These two statements aren't mutually exclusive. The link is looking at the analog signal through an oscilloscope. The person you replied to is pointing out that after decoding and applying error correction, you can still end up with the same digital signal output. So the eye diagram charts are useful for detecting the quality of the cable, but as long as the quality is past a certain threshold, it does not matter.

And that threshold is "baked in" to the eye mask pattern you load into the tester. If the eye stays out of the masked areas, it passes, if it goes into the masked areas it fails. Oscilloscopes capable of eye diagram testing can trigger on failure, so if it passes an eye test it'll reconstruct correctly with proper timing.

Re: A nearly perfect USB cable tester

#153

Post by SAI_Peregrinus »

fmajid wrote: post #23
Onavo wrote: post #2

I just want one that tells me the maximum voltage and current supported by a USB C cable.

The Treedix will tell you that, as it is a feature of the eMarker chip (no chip means 60W).

Assuming the chip isn't fraudulently added. Like in the article, some manufacturers are shady & will sell cables with e-marker chips for capabilities the cable can't actually support.

Re: A nearly perfect USB cable tester

#154

Post by atoav »

jmalicki wrote: post #36
atoav wrote: post #30

Earlier quoted context omitted.

Well the thing is better doesn't mean better quality here. Better means you can use a longer cable or abuse the cable for longer till it dies. This is a big part of what makes any pro gear expensive: reliability. If you just connect your home hifi to your speakers in an acoustically untreated space, you could also just use a bunch of steel wire coathangers and get an indistinguishable result. Even a el-cheapo store b…

I have seen shielding and gauge make quite a difference for cables carrying analog signals!

Sure shielding is important, especially for high impedance signsls or very small signals (e.g. microphone cables). But the difference in shielding between a 10€ cable and a 100€ cable is not going to be audible in anything but edge cases.

Gauge is only important if the driving or receiving sides suck or you're using ridiculously small gauges or driving high currents (e.g. speaker signals) for long distances. In my lab tests I drove line signals over a spool of hair thin wire without audible difference. A famous experiment that gave me a good chuckle came to similar results: https://www.diyaudio.com/community/threads/copper-wire-vs-ba...

You could literally use CAT 6 ethernet cables and call it a day. There have been tests running balanced audio signals over a kilometer of ethernet cable without audible loss. And since you got 4 twisted pairs you could run 4 channels over one cable (commercially available here for example: https://store.monkeywrenchpro.com/thomann-5fc-cat-snake-spli...). I use this in combination with an ethernet patchbay to route signals around our event spaces and it works well even for unamplified mic level signals.

Re: A nearly perfect USB cable tester

#155

Post by bsder »

Telaneo wrote: post #113
graemep wrote: post #110

Earlier quoted context omitted.

Most users tend to ignore diagnostic information. "What did the error message say" "I don't know."

This is because error messages have historically been bad, unintelligible, un-actionable, and hard to separate from soft errors that don't actually matter. 'Segmentation fault. Core dumped.' 'Non-fatal error detected. Contact support.' 'An error occurred.' 'An illegal operation was performed.' 'Error 92: Insufficient marmalade.' 'Saving this image as a JPG will not preserve the transparency used in the image. Save an…

Don't blame this one on programming techies. This one is ALL the fault of shitty UI designers abusing modal dialog boxes.

A modal dialog is supposed to be for something damn near irreversible--like about to blow away your application because of error. You are supposed to STOP and go get the guru or you are about to lose, badly.

Unfortunately, UI designers throw them up for everything and people get used to simply clicking "OK" to make them go away so that they can get back to doing their task. So, when the user gets an actual error, they've already blown away the dialog box with information.

Your 'Saving this image as a JPG will not preserve the transparency used in the image. Save anyway?' line is a horrifically excellent example. That is a standard "Save As..." response, and it should NEVER have been. That should have always been under "Export..." as saving should never throw away information and it would be perfectly fine to regenerate a JPG as long as you have the full information still available in the original file.

This is the stuff that infuriates me about the UI designers. Your job is about interactions, first, and pixels, second.

Re: A nearly perfect USB cable tester

#156

Post by Dylan16807 »

AceJohnny2 wrote: post #116
seanalltogether wrote: post #20

I wasn't surprised to learn that when Linus Tech Tips released those new usb-c cables, that they all sold out almost instantly. They put their entire reputation on the line to claim (and label) the exact capabilities of their usb cables. Isn't that all we really want?

They lost me at "our conductors are coax!". USB is designed around differential signaling, which is what twisted pair excels at.

Twisted pair is good but it only gets your losses so low at these speeds. Keep in mind that USB cables have a very small budget for signal loss, and at 40Gbps they're carrying frequencies 25x higher than 10gig ethernet.

Re: A nearly perfect USB cable tester

#157

Post by atoav »

SAI_Peregrinus wrote: post #151
atoav wrote: post #9

One thing to realize is that especially for high resolution video cables these cheap testers can't really deliver. The way to test them is a eye diagram (see: https://incompliancemag.com/eye-diagram-part2/ ) and testers with that capsbility cost upwards of 10.000 Eurodollars.

Keysight are nice enough to provide prices on their web site for all their cheaper equipment. I priced out a setup for USB 2.0 eye diagram compliance testing on their site in this comment¹ and it's more like $40,000 than $10,000. ¹ https://news.ycombinator.com/item?id=47568399

I said 10k because I think I saw a specialized product just for HDMI starting at roundabout that price bracket once mentioned in a youtube video, csn't really remember the brand, but it was relatively unknown in terms of test equipment. More generic (and flexible) equipment like the one mentioned will indeed tear a deeper hole into your pocket, but comes with its own perks.

Re: A nearly perfect USB cable tester

#158

Post by jmalicki »

atoav wrote: post #154
jmalicki wrote: post #36

Earlier quoted context omitted.

I have seen shielding and gauge make quite a difference for cables carrying analog signals!

Sure shielding is important, especially for high impedance signsls or very small signals (e.g. microphone cables). But the difference in shielding between a 10€ cable and a 100€ cable is not going to be audible in anything but edge cases. Gauge is only important if the driving or receiving sides suck or you're using ridiculously small gauges or driving high currents (e.g. speaker signals) for long distances. In my la…

I have seen cables for powered computer speakers make cat5, or coat hangers, look amazing

Re: A nearly perfect USB cable tester

#159

Post by Dylan16807 »

HDBaseT wrote: post #149
the_biot wrote: post #52

Earlier quoted context omitted.

Yup, that's the sort of thing that's typically missing from cable testers. I have a USB cable that normally works fine, but introduces errors when doing full blast USB 2.0 bulk transfers. I keep it around just in case I ever come across a tester that can show me this in hard numbers.

Products that do exactly that do exist, although are prohibitively expensive. [0] It is difficult to send 10Gbps, 20Gbps, 40Gbps or even 80Gbps, then receive it, then validate it. [0] - https://www.totalphase.com/products/advanced-cable-tester-v2...

I'm sure it's hard to do a detailed eye analysis at those speeds. But come on, I can get a hub that does 10Gbps per lane (so 20Gbps equivalent) for $13, 40Gbps SSD enclosures for $40, and 80Gbps SSD enclosures for $150. Making that signal do a loop, maybe adding some artificial attenuation, and checking how many bits get corrupted shouldn't need fancier hardware than that.

Re: A nearly perfect USB cable tester

#160

Post by Barbing »

dijit wrote: post #122
joshvm wrote: post #120

Earlier quoted context omitted.

I started buying Belkin TB5 cables which are around $50 a pop. They can easily power a laptop at full load and can stream video at any reasonable resolution/framerate I might need. I've yet to find a need for an NVMe faster than 20 GBps nevermind finding USB4 enclosures, or that the cable supports up to 80. They're also not nearly as chunky as the Dell cables, which are good, but seem to have very rigid shielding. I…

I do something largely similar. It alleviates the anxiety of knowing what cable does what. I use Apples Thunderbolt 4 or USB-C cables exclusively: if its white its for charging and low data, if its black its for high data. I’ve been doing this for a few years, but its really costly as those Apple Thunderbolt cables are crazy expensive.

I wish the USB spec had mandated labeling. There must be a lovely label printer they make for cables or something that shouldn’t be too expensive these days. Label a few cables a day, finish the whole house in a jiffy.

LTT or another big YouTuber made a cable and made sure to get it labeled. Also complained how difficult it was to find a supplier willing to make a better cable than usual.

Post reply on HN