USB extension cables

micky wrote on 4/27/2026 5:07 PM:

If it's working for you, why worry about it?

The only suggestion I have is to keep spares for the cables and stuff that you know work well.

What is ASI? Artificial Super Intelligence?

Yes, depending on the USB-C mutation and what you define as a signal. For example:

USB 2.0 (via USB-C):

USB 3.2 Gen 1 (5Gbps):

USB 3.2 Gen 2 (10Gbps):

USB 3.2 Gen 2x2 (20Gbps):

USB4/Thunderbolt 3 & 4:

USB4 Gen 4: Up to 80 Gbps (emerging standard).

For the various speeds (data rates), there are also cable length limitations. Too long a cable means the higher speeds don't work because of distorted waveforms, cable losses, timing issues, etc. For DC power, increased cable resistance might introduce losses.

If the errors are caused by too long (or too wrong) a cable, chances are good that repeating the same data on the same wrong cable is going to repeat the errors. Using various forms of error correction will largely eliminate that problem at the cost of additional delays and the associated throughput reductions.

Sorry, no suggestions because you didn't provide any numbers for what manner of hardware and data protocols you're working with. Of course, you could just plug in an extension cable and see what happens by measuring the error rate (assuming your hardware is managed and provides SNMP info, including data rate). As long as the extension is wired correctly, you're unlikely to do any damage.

USB hub or USB switch? Since it's plugged into a laptop, it's probably a hub or docking station. I don't know about "most people" but I use USB mostly for flash drives and mice. I'm not an Apple computer user and therefore don't have any Thunderbolt hardware.

signal goes in both directions in the same wire, so not that simple to regenerate the signal. The cable needs intelligence, like a hub, to do regeneration. Maybe a hub with a long cable exists.

Artificial Semi-Intelligence!

Okay.

It might provide SNMP info. I'll look, or I'll try to add it.

USB hub, although the one currently in use, made by Sabrent with 4 output ports, has an on-off switch for each port. I dont' think that makes it a USB switch, though.

Yeah, I use it for the mouse, an external keyboard, and flashdrives, but lately for the bare drive I use for imaging. Amazon doesn't sell the connection device anymore, but it has its own power supply. And I'm using USB-A so I'll give it a try with a 6" extension.

Higher spec cables often contain chips at each end, to tell what it's connected to what speed/power it can handle, if you whack an extension on the end, that information will be wrong and you risk sending over 100 watts of power down a cable not rated for it ...

micky snipped-for-privacy@fmguy.com wrote: [...]

Just connect the fast stuff, i.e. the disk drive, directly to the laptop and connect the slow stuff, i.e. keyboard and mouse and if need be the USB memory sticks, to the hub. You can use extension cables for the keyboard and mouse, but I would use no or only a short extension cable for the USB memory sticks.

FWIW, I also have a laptop which has only 3 USB ports, but I do not use a hub. If I'm out of ports for a certain task, I just disconnect the mouse and use the built-in touchpad.

I have two current laptops (and a pile of old ones) they both have two USB-C and one USB-A.

The "home" laptop has one port used for power, one used to connect a FIDO2 fingerprint dongle, one is spare for occasional devices.

The "work" laptop has a thunderbolt dock with power, 2.5Gb ethernet, dual monitors, keyboard, mouse, backup drive, CF/SD reader, audio DAC, webcam, spare USB ports.

You can do data-transfer tests, to assure yourself that an interface is in reasonable shape. You can run SHA256 on the destination file, and see if the value is the same as the SHA256 on the source file.

dd.exe --list # Get the name of the # random number generator

dd.exe if=/dev/random of=D:\source.bin bs=1M count=10240 # Make a 10GB file

copy source.bin dest.bin # Copy it somewhere

sha256sum source.bin # Compare the two files sha256sum dest.bin #

And you can make the test file, as large as your storage devices will allow.

That "dd" program, can make zeros at 3.5GB/sec and random numbers at 1.2GB/sec. The disk drive, not so much. (Even whizzy TLC NVMe, can slow down to 1.2GB/sec if you write them for long enough periods of time, as they run out of SLC cache after a while.)

A three foot cable should be suitable for USB3 500MB/sec work. I have an enclosure, and that is the cable that came with it.

Anything faster than that, is a "mystery meat" :-) There is very little user test of any of the faster stuff, available to read about. Anandtech is gone, and Toms kinda sucks for this sort of thing (they're in some sort of groove now, churning out chucky cheese articles).

I can see me now, doing space cadet testing on USB4 with a "31 inch cable". You know, taking the scissors, cutting it to 31 inches, taping the connector back on the end, then test that the files aren't getting corrupted. I think that's what USB.org wants us to do :-) And for USB4 Rev2.0, they'll want us to pick up the cables, with tweezers. I can't wait.

Paul

It contains a full set of wires and is electronically marked (E-marked): It contains an E-marker chip that responds to the USB Power Delivery Discover Identity command, a kind of vendor-defined message (VDM) sent over the configuration data channel (CC). Using this command, the cable reports its current capacity, maximum speed, and other parameters.[41]:  §4.9  Full-Featured USB Type-C devices are a mechanic prerequisite for multi-lane operation (USB 3.2 Gen 1×2, USB 3.2 Gen 2×2, USB4 2×2, USB4 3×2, USB Gen 4 Asymmetric).[41]

USB-C devices support power currents of 1.5 A and 3.0 A over the 5 V power bus in addition to baseline 900 mA. These higher currents can be negotiated through the configuration line. Devices can also use the full Power Delivery specification using both BMC-coded configuration line and the legacy BFSK-coded VBUS line.[41]: §4.6.2.1 "

Now, does USBPD run over a USB-A connector ? I'm guessing we can expect at least USB-C to be marked that way.

"There is nothing like the smell of 48V @ 5A flowing through a USB-C connector, in the morning."

Paul

If you know much about electricity, you know that watts/amps don't flow unless there's some kind of load. Also, the damaged chip is likely to be non-functional, and the power is limited to 60W.

Sure, current is pulled rather than pushed, but the comms between source and sink device can set the supply to 48V which most devices won't be ready for.

Latest USB PD3.1 spec caters for up to 240W

Why would they set something they can't use? IIRC, 48V can't be set unless the cable has a working chip in it.

IIRC, 60W is the limit without a functioning chip in the cable. It doesn't matter if the source and/or load are capable of more.

BTW, I haven't yet seen one of these devices, although C to C cables are often marked for 240W.

Okay

I'll think about that.

Okay, that sounds like something I want to avoid.

Off-topic, turned on the upstairs laptop yesterday, and couldn't do anything. Kept getting menus. Restarted 3 times

Today I looked at it in a fresh light and figured out that my mouse keys were reversed. I DID NOT TOUCH them. I wasn't even in Settings. Some weird force must have changed it.

A polterguest

If you think of an Open Circuit as a load of Infinite Resistance, there IS an infinitesimally small amount to Current flowing .... but, in any case, Watts don't flow, they are just dissipated. ;-P

When reading, I had thought "Ghost" but sure, "poltergeist" but even "polterguest" works, too. ;-P

I wonder what else it's doing.

# Theme from GhostBusters