Twisted USB

[Ogitek]

According to baseline USB 2.0 specifications, the USB cable must have signal (green-white) wires as a twisted pair with 90 Ω differential impedance.

What if you use one of the twisted pair as V+ or ground and the data in one of the original V+, ground wire (for sake of understanding it). Would the USB still work? If it would still work, how many speed decrease would it encounter? 50% 80%

[wraper]

According to baseline USB 2.0 specifications, the USB cable must have signal (green-white) wires as a twisted pair with 90 Ω differential impedance.

What if you use one of the twisted pair as V+ or ground and the data in one of the original V+, ground wire (for sake of understanding it).

Low speed (1.5 mbps) devices like low end mice or keyboard most likely will work. Full speed (12mbps) YMMW, anything faster, no.

Would the USB still work? If it would still work, how many speed decrease would it encounter? 50% 80%

There would be not speed decrease but communication failure.

[Ogitek]

According to baseline USB 2.0 specifications, the USB cable must have signal (green-white) wires as a twisted pair with 90 Ω differential impedance.

What if you use one of the twisted pair as V+ or ground and the data in one of the original V+, ground wire (for sake of understanding it).

Low speed (1.5 mbps) devices like low end mice or keyboard most likely will work. Full speed (12mbps) YMMW, anything faster, no.

That means any serial port devices that has bps like 115,200 bps can still work? 115,200 bps being 0.1152 mbps only?

Does this mean for serial port devices. Even a plain 4 wire stranded wire (without twisting) can make it run?

Would the USB still work? If it would still work, how many speed decrease would it encounter? 50% 80%

There would be not speed decrease but communication failure.

What? you just said 1.5 mbps would still work, why communication failure?

[wraper]

USB devices work at predetermined speed (USB LS, FS, HS mode), they don't adjust their speed depending on how well communication works.

[Ogitek]

USB devices work at predetermined speed (USB LS, FS, HS mode), they don't switch their speed depending on how well communication works.

What do you mean? If the serial port is only 115,200 bps. then the USB to serial port converter would only be 115,200 bps, isn't it.

Can you give some figures.

[magic]

I once connected a high speed card reader to motherboard's USB pin header with some random 10-pin IDC cable, which I think was originally meant for audio panels. It worked.

Like all dodgy stuff, you won't know for sure until you try.

What do you mean? If the serial port is only 115,200 bps. then the USB to serial port converter would only be 115,200 bps, isn't it.

It can be transmitting nothing and yet USB link bitrate is what it is
https://duckduckgo.com/?q=USB+LS%2C+FS%2C+HS

[Ogitek]

Supposed you have an ADC module connected to USB 2.0.  The ADC speed is much less than 480Mbps (speed of USB 2.0). And you use simple 4 stranded wire as USB cable, and the USB speed gets down to 1.5 mbps. Let's say the ADC only need 115,200 bps. It won't work?

[dobsonr741]

Do I understand OP’s proposal right? Swapping the white-green with black-red pair.
Given in the usual cables, even for high speed, they are exactly the same material/size, for cost saving, so it does not matter. Only the color changes.

Where you start to see different wire size, tight twisting vs no twisting on power is on the USB-C and above. I did not yet see differentiation on the 4 wire cables. Or maybe I did not cut into enough cables?

Bit rate: b/c the baud of a virtual com port is 115200, it does not mean the USB packets are using a lower bit rate. The bit rate is whatever the peripheral descriptor tells, low/full/high, during enumeration, not what data you will be sending over. The serial port data is wrapped into USB packets ant sent as bursts. You will not see 115200 clocked pulses when you look at the waveform on an oscilloscope.

[Whales]

USB devices work at predetermined speed (USB LS, FS, HS mode), they don't switch their speed depending on how well communication works.

What do you mean? If the serial port is only 115,200 bps. then the USB to serial port converter would only be 115,200 bps, isn't it.

No.

The USB bus will still be sending data at a higher rate, just with gaps in between.  A bit like how you can either walk to the shops OR sprint then rest then sprint then rest.

[wraper]

I once connected a high speed card reader to motherboard's USB pin header with some random 10-pin IDC cable, which I think was originally meant for audio panels. It worked.

Like all dodgy stuff, you won't know for sure until you try.

Depends on length. Short wires will work.

[Ogitek]

Do I understand OP’s proposal right? Swapping the white-green with black-red pair.
Given in the usual cables, even for high speed, they are exactly the same material/size, for cost saving, so it does not matter. Only the color changes.

Where you start to see different wire size, tight twisting vs no twisting on power is on the USB-C and above. I did not yet see differentiation on the 4 wire cables. Or maybe I did not cut into enough cables?

What I mean is the V+ gets into one of the twisted pair due to rewiring of the USB cable with a LEMO plug at one end with wrong pins. For example. You purchase this USB to LEMO cable:

https://www.amazon.com/gp/product/B06XKV73C1/ref=ewc_pr_img_1?smid=A1PUUT3OIOMJ0X&psc=1

But in your equipment LEMO has other pinout than it. So what you do is cut the USB cable into half, and rewire them to match the pins. What would happen is either the V+ or V- can become go inside one of the twisted pair.

If USB 2.0 always run at 480Mbps even if the data is only 4800 bps. And you rewire the cable above. then it won't run anymore even if the USB source and your equipment has the right pins already??

[Andy Chee]

USB devices work at predetermined speed (USB LS, FS, HS mode), they don't switch their speed depending on how well communication works.

What do you mean? If the serial port is only 115,200 bps. then the USB to serial port converter would only be 115,200 bps, isn't it.

Can you give some figures.

The speed between your legacy serial device and the converter would be 115,200bps.  But the speed between the converter and USB host/PC would 1.5MB/s.  Of course, the speed mismatch means the USB converter has to "translate" between the two.

[Ogitek]

I don't get your logic in this. You can change wires at LEMO connector end any way you want, it does not change how wires are connected at USB connector side.

I don't know how to solder the micro LEMO pins.. so instead I plan to rewire it at the middle of the cables by cutting it and rewiring the individual wires. and not changing the pin ends. So it won't work at all?

[wraper]

But in your equipment LEMO has other pinout than it. So what you do is cut the USB cable into half, and rewire them to match the pins. What would happen is either the V+ or V- can become go inside one of the twisted pair.

If you do that, you rewire either within LEMO connector of as close as possible to it. Watch some videos about how to properly solder FFS.

[Andy Chee]

I don't get your logic in this. You can change wires at LEMO connector end any way you want, it does not change how wires are connected at USB connector side.

I don't know how to solder the micro LEMO pins.. so instead I plan to rewire it at the middle of the cables by cutting it and rewiring the individual wires. and not changing the pin ends. So it won't work at all?

If the converter is located within the molded USB plug body like so, then the rewiring should be ok.  Strictly speaking, anything after the converter, no longer requires to be compliant with the USB protocol.

[Ogitek]

Today I got the above USB cable. I plan to solder the end to the LEMO plug clone when it arrives next week. Let's say I would be successful in soldering them to the plug. And it will be used in USB 2.0 equipments. Are you sure it won't run at 480Mbps (with a length of it not twisted)?

Or did you mean even if you use straight stranded wires. Sometimes the USB 2.0 can run. Sometimes it can't.  Or 100% can't all the time?

[ejeffrey]

Supposed you have an ADC module connected to USB 2.0.  The ADC speed is much less than 480Mbps (speed of USB 2.0). And you use simple 4 stranded wire as USB cable, and the USB speed gets down to 1.5 mbps. Let's say the ADC only need 115,200 bps. It won't work?

Yes.  What matters is the USB operating mode not the amount of data transmitted.

Keep in mind USBs "interesting" naming.  The USB 2.0 standard superseded 1.1 and continues to include lower speeds. So while normal people make the totally reasonable association that "USB 1.1 is 1.5 or 12 megabit, USB 2.0 is 480 megabit" that's not actually how USBIF wants you to use those terms.  USB 2.0 devices can be low speed (1.5) full speed (12) or high speed (480). 

A high speed device will transmit data at 480 megabit / s always and requires appropriate signal integrity.

[Ogitek]

I don't get your logic in this. You can change wires at LEMO connector end any way you want, it does not change how wires are connected at USB connector side.

I don't know how to solder the micro LEMO pins.. so instead I plan to rewire it at the middle of the cables by cutting it and rewiring the individual wires. and not changing the pin ends. So it won't work at all?

If the converter is located within the molded USB plug body like so, then the rewiring should be ok.

(Attachment Link)
(Attachment Link)

I was talking about rewiring in middle of wire with a USB and LEMO male at its end. I wasn't referring to the converter in the rewiring. But then even in the conventer. If you rewire the middle (let's say the USB male pin is incorrect), and the V+ gets into the twisted pair, and you are running it at USB 2.0 at 480 Mbps. Why would it run at all? I thought 480 Mbps require the middle to be twisted in the data lines.

[Ogitek]

Supposed you have an ADC module connected to USB 2.0.  The ADC speed is much less than 480Mbps (speed of USB 2.0). And you use simple 4 stranded wire as USB cable, and the USB speed gets down to 1.5 mbps. Let's say the ADC only need 115,200 bps. It won't work?

Yes.  What matters is the USB operating mode not the amount of data transmitted.

Keep in mind USBs "interesting" naming.  The USB 2.0 standard superseded 1.1 and continues to include lower speeds. So while normal people make the totally reasonable association that "USB 1.1 is 1.5 or 12 megabit, USB 2.0 is 480 megabit" that's not actually how USBIF wants you to use those terms.  USB 2.0 devices can be low speed (1.5) full speed (12) or high speed (480). 

A high speed device will transmit data at 480 megabit / s always and requires appropriate signal integrity.

So if the equipment is high speed. It won't decrease to low speed (1.5) mode just because it couldn't negotiate at the higher speed?

[Andy Chee]

The minimum USB speed is 1.5Mbits/s

https://en.wikipedia.org/wiki/USB_communications

If a USB 2.0 is plugged into a USB 1.0 port (or vice versa), then the speed will be negotiated at 1.5Mbits.

[Ogitek]

The minimum USB speed is 1.5Mbits/s

https://en.wikipedia.org/wiki/USB_communications

So USB 2.0 can readjust from 480Mbps to 1.5 Mbps if the wiring is not good. Why did you guys say the speed is fixed?

[Andy Chee]

The minimum USB speed is 1.5Mbits/s

https://en.wikipedia.org/wiki/USB_communications

So USB 2.0 can readjust from 480Mbps to 1.5 Mbps if the wiring is not good. Why did you guys say the speed is fixed?

Incorrect.

Bad (or good) wiring has nothing to do with negotiation.  It's the USB device and USB host that does the negotiation.  The USB device/host has no idea of the condition of the wiring.

[Ogitek]

I first became aware of all this a while ago when googling whether USB cable should be shielded or not. Then I came across this:

https://electronics.stackexchange.com/questions/430914/a-few-of-my-cables-dont-work-is-it-because-they-werent-shielded-properly

"According to baseline USB 2.0 specifications, the USB cable must have signal (green-white) wires as a twisted pair with 90 Ω differential impedance.

And no, the shield should connect only metal shells, and shouldn't be connected to GND wire at all, see section 6.5.2 of USB 2.0 specifications.

Some of your cables do not work reliably likely because you have too wide cut-offs near soldering area, which creates impedance discontinuity, which causes reflections and intersymbol interference. USB cables must be tested on TDR (Time-Domain-Reflectometer) to ensure impedance uniformity across soldering joints."

That page is saying that even in normal fully twisted USB cable, the soldering at the end can even impact it and you need to use Time Domain Reflectometer to check??

A while ago. I was about to transact with a China cable fabricator at Ebay who will assemble USB cable with LEMO clone. So without proper soldering or using the right cable. It won't even run even if all pins rightly connected!

I think I'd better buy the original $400 USB-LEMO cable for my sound system. I want to save on it by trying to recreate it, but encountered so many issues.

[Andy Chee]

I first became aware of all this a while ago when googling whether USB cable should be shielded or not. Then I came across this:

I think I'd better buy the original $400 USB-LEMO cable for my sound system. I want to save on it by trying to recreate it, but encountered so many issues.

It completely depends on where in the transmission chain you want to rewire.  If it's a low speed non-USB section (for example RS232), then it shouldn't be a problem.

But yes, your LEMO device looks like it relies on high speed signalling, so better go with an original cable.

[magic]

USB 1.x defines two speeds: low (1.5Mbps) and full (12Mbps).
USB 2.0 defines a new speed: high (480Mbps) and still permits the lower speeds to be used.

HS hosts and devices can fall back to FS if the other end doesn't support HS.
If the cable is very dodgy, it may look like HS is not supported and the link will work at FS.
If the cable is slightly dodgy, HS may be enabled and then fail to actually move data.

There is no fallback from FS to LS. An FS device simply fails if connection is too bad for FS.

UART/RS232 dongles usually are FS only. Some fancy ones may be HS.