Length of USB cable w/o twisted data pair to reach Hi Speed 480Mbps

[Ephant]

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 to reach 480Mbps (Hi Speed). Without it, supposedly the speed is only Full Speed (12Mbps).

But if the cable is only very short. like 1 foot. Can the wire without twisted data paths reach Hi Speed of 480Mbps? If not, what is the technical reasons it just isn't possible? (perhaps with exception of just 3 inches of untwisted wire length which can work?).

 

[Baluncore]

If the wires are twisted, they are close together, and are balanced. Another pair in the same bundle will be twisted at a different pitch, so the coupling and cross-talk between pairs is also balanced.

But if the cable is only very short. like 1 foot.

It seems you are seeking the point of maximum unreliability.
That is an anathema, to an engineer.

The signal travels about 1 foot in one nanosecond, 1 ns, it is then reflected by the impedance mismatch at the far end of the sloppy untwisted line. The reflection gets back 1 ns later, at 2 ns, just as the next bit is being transmitted. The line is long enough to contain one bit. You have built a resonator, not a transmission line.

For a transmission line to carry 480 Mbit/s, a minimum bandwidth of about 2 GHz is required. A 3 inch mismatched line, would resonate at about that frequency.

To radiate a 2 GHz waveform efficiently, a dipole antenna would need to be about 3" long. The 5'th harmonic of 480 MHz falls at 2.4 GHz, in the ISM band. Do you want the signal energy to remain in the transmission line, or to radiate, and interfere with wireless Bluetooth, in the 2.45 GHz ISM band?

 

[Ephant]

If the wires are twisted, they are close together, and are balanced. Another pair in the same bundle will be twisted at a different pitch, so the coupling and cross-talk between pairs is also balanced.It seems you are seeking the point of maximum unreliability.
That is an anathema, to an engineer.

The signal travels about 1 foot in one nanosecond, 1 ns, it is then reflected by the impedance mismatch at the far end of the sloppy untwisted line. The reflection gets back 1 ns later, at 2 ns, just as the next bit is being transmitted. The line is long enough to contain one bit. You have built a resonator, not a transmission line.

For a transmission line to carry 480 Mbit/s, a minimum bandwidth of about 2 GHz is required. A 3 inch mismatched line, would resonate at about that frequency.

To radiate a 2 GHz waveform efficiently, a dipole antenna would need to be about 3" long. The 5'th harmonic of 480 MHz falls at 2.4 GHz, in the ISM band. Do you want the signal energy to remain in the transmission line, or to radiate, and interfere with wireless Bluetooth, in the 2.45 GHz ISM band?

Can termination like USB male head or even LEMO plug can cause impedance mismatch at the ends supposed the wire is twisted correctly?

Also when creating USB cable. Does the manufacturer simply get their roll of wires and say cut every 3 feet of the wires and put plugs at its ends or do they have to compute the twisting of each wire?

 

[Baluncore]

Can termination like USB male head or even LEMO plug can cause impedance mismatch at the ends supposed the wire is twisted correctly?

The termination must be designed to have the same characteristic impedance as the twisted pair.

Also when creating USB cable. Does the manufacturer simply get their roll of wires and say cut every 3 feet of the wires and put plugs at its ends or do they have to compute the twisting of each wire?

The cable is made by twisting separate pairs that are many hundreds of metres long, then those pairs with different twist rates are laid together, then often wrapped, or screened with foil, before being encased in the extruded outer jacket.

By making hundreds of metres of cable, the entire length can be tested with TDR while it is on the roll. Any faulty section can be identified and removed, then diagnosed to identify the cause of the fault in the cable manufacturing process.
https://en.wikipedia.org/wiki/Time-domain_reflectometer

To make a short cable, the cable is cut to length, the insulation is stripped, and the connector terminals bonded to the conductors, before the connector outer housings are moulded in place.

 

[Ephant]

The termination must be designed to have the same characteristic impedance as the twisted pair.

The cable is made by twisting separate pairs that are many hundreds of metres long, then those pairs with different twist rates are laid together, then often wrapped, or screened with foil, before being encased in the extruded outer jacket.

By making hundreds of metres of cable, the entire length can be tested with TDR while it is on the roll. Any faulty section can be identified and removed, then diagnosed to identify the cause of the fault in the cable manufacturing process.
https://en.wikipedia.org/wiki/Time-domain_reflectometer

To make a short cable, the cable is cut to length, the insulation is stripped, and the connector terminals bonded to the conductors, before the connector outer housings are moulded in place.

Does a LEMO plug have the same characteristic impedance as the twisted pair in a USB 2.0? If not, what will happen?

 

[Baluncore]

Does a LEMO plug have the same characteristic impedance as the twisted pair in a USB 2.0?

Probably not, but it will be close.
You will need to look up the impedance data on those coupled plugs and sockets. Multipole plugs tend not to work well with twisted-pair transmission lines at such high frequencies.

The impedance of a parallel conductor transmission line, is determined by the ratio of conductor diameter to separation, and the dielectric constant of the insulation.
When a twisted pair is connected to two adjacent pins on the plug, the pin separation is fixed, so the diameter and insulation must be adjusted to maintain the impedance. If you follow through the connection, you will find it difficult to avoid steps in impedance, that will look like small series inductors, or parallel capacitors, which will reflect some signal energy.

 

[Ephant]

Probably not, but it will be close.
You will need to look up the impedance data on those coupled plugs and sockets. Multipole plugs tend not to work well with twisted-pair transmission lines at such high frequencies.

The impedance of a parallel conductor transmission line, is determined by the ratio of conductor diameter to separation, and the dielectric constant of the insulation.
When a twisted pair is connected to two adjacent pins on the plug, the pin separation is fixed, so the diameter and insulation must be adjusted to maintain the impedance. If you follow through the connection, you will find it difficult to avoid steps in impedance, that will look like small series inductors, or parallel capacitors, which will reflect some signal energy.

Have you (or anyone) ever built a USB 2.0 cable with LEMO 4-pin (multipole) plug at one end? Did it run at Hi speed of 480Mbps?

 

[Baluncore]

Have you (or anyone) ever built a USB 2.0 cable with LEMO 4-pin (multipole) plug at one end? Did it run at Hi speed of 480Mbps?

Certainly not I.

 

[Ephant]

Certainly not I.

Are the normal USB male plugs molded in a USB 2.0 cable impedance matched with the twisted wiring?

You said that for the Lemo plug, the diameter and insulation of the wiring must be adjusted to maintain the impedance. So if one can design the wiring to become bigger as it nears the solders to the pin with thicker insulation, it can be impedance matched. But how do you make the wiring bigger as it gets near the pins?

 

[Baluncore]

But how do you make the wiring bigger as it gets near the pins?

You don't. You remove some dielectric insulation, so the capacitance stays the same with the slightly larger pins. Larger pins also have lower series inductance, which makes it more difficult since the impedance, Z, is proportional to √(L/C) ;

The problem also occurs when a pin is plugged into the socket. The OD of the pin, becomes the OD of the socket, so pins also need to be smaller.
Notice how the pins on high speed data cables have been getting smaller. That raises the connector impedance to work with twisted pairs having thinner wires.

The type-N coaxial connector was invented by Paul Neill, hence the type-N name.
https://en.wikipedia.org/wiki/N_connector#Design
Neill was then joined by an applied mathematician, Carl Concelman. Together they designed a smaller coaxial connector, the -NC. It came in bayonet, BNC, and in screw, SNC, forms. The centre connector was dimensioned for either 50 or 75 ohms impedance. The -50 and -75 were not physically compatible and could be damaged if mixed.
https://en.wikipedia.org/wiki/BNC_connector#Origin

 

[Averagesupernova]

The -50 and -75 were not physically compatible and could be damaged if mixed.

I learned that the hard way many years ago.

 

[Vanadium 50]

This question is "if I push a cable beyond its specs will it work?" and the answer is "you can't count on it".

This seems like a very odd way to save $1 on cables.

The next question is "Will this worked with mismatched impedance?" and the answer is "you can't count on it". FWIW, USB is 90 ohms and LEMO has 50 and 75 ohm series.

Now it sounds like you are building some kind of Franken-cable. These exist commercially (heaven knows why) but they are not so cheap, and if I had to guess, I would guess that these "cables" have some active components to ensure they work correctly.

 

[Ephant]

You don't. You remove some dielectric insulation, so the capacitance stays the same with the slightly larger pins. Larger pins also have lower series inductance, which makes it more difficult since the impedance, Z, is proportional to √(L/C) ;

The problem also occurs when a pin is plugged into the socket. The OD of the pin, becomes the OD of the socket, so pins also need to be smaller.
Notice how the pins on high speed data cables have been getting smaller. That raises the connector impedance to work with twisted pairs having thinner wires.

The type-N coaxial connector was invented by Paul Neill, hence the type-N name.
https://en.wikipedia.org/wiki/N_connector#Design
Neill was then joined by an applied mathematician, Carl Concelman. Together they designed a smaller coaxial connector, the -NC. It came in bayonet, BNC, and in screw, SNC, forms. The centre connector was dimensioned for either 50 or 75 ohms impedance. The -50 and -75 were not physically compatible and could be damaged if mixed.
https://en.wikipedia.org/wiki/BNC_connector#Origin

Someone has the original USB-Lemo cable. I tried zooming in the cable picture and seemed to make out the words "CMG LL84201 CSA CABLE COPARTNER FT4". I googled it and the following page comes up.

https://www.ebay.com/itm/374199291173

UTP Patch Cable? Can you use a UTP patch cable on a USB 2.0 cable? What happens if you use it? There is more impedance match or mismatch?

 

[Ephant]

By the way, by UTP patch cable (see last message) is meant ethernet cable with 4 pairs of twisted wires. Did the original USB Lemo cable uses this to match impedance? I read ethernet wire has higher impedance like 100 Ohm). Is this to match the Lemo plug which is 100 Ohm for impedance match. Or is it irrelevant and one can use USB cable or ethernet cable for USB 2.0?

 

[Ephant]

My bad. I thought UTP cable automatically meant ethernet cable. So the label MG LL84201 CSA CABLE COPARTNER FT4 still refers to USB cable.

My last questions about this. First thanks for all the insights.

AC can cause transformers to have impedance. Why does a USB cable have impedance when the source is DC? Remember the impedance is formed from the current like in transformers. Can you give examples of other circuits where DC can pull off the stunts of AC and cause impedance to form?

Also about impedance mismatch. Can you give analogy. Is it like voltage divider in a resistors with different values?

 

[Baluncore]

The impedance of a cable is the relationship between the amplitude of the voltage wave and the current wave that travel along the transmission line. Where two transmission lines meet, with different impedances, some energy is reflected from the junction.
https://en.wikipedia.org/wiki/Impedance_matching#Transmission_lines

 

[Ephant]

The impedance of a cable is the relationship between the amplitude of the voltage wave and the current wave that travel along the transmission line. Where two transmission lines meet, with different impedances, some energy is reflected from the junction.
https://en.wikipedia.org/wiki/Impedance_matching#Transmission_lines

It means the DC can turn to AC when the USB is transmitting at 480Mbps? Can pure DC also do that?

 

[Tom.G]

Another way of stating what @Baluncore said above:

Please note that the RF impedance of a transmission line is an inherent characteristic describing the Voltage/Current ratio when driven with a given power.

For twin-lead or coaxial cable, the Characteristic Impedance is a function of conductor diameter, conductor spacing, and the dielectric constant of the material between the conductors.

The reason that high frequencies are sensitive to this is that the successive cycles of the waveform are so close together,

For a not-very-good analogy consider this:

1) You run full speed into a soccer goal net.​

Of course you wiil suddenly stop and probably bounce back a little.​

You recover and walk away.​

​

2) You and several other people in a line behind you run full speed into a soccer goal net.​

2a) If the people behind you are several feet apart, result is as in 1) above.​

2b) If the people behind you are close to you and each other, maybe 1 foot apart, when you bounce off the goal net you will interfere with the person behind you... and they will interfere with the person behind them...​

​

Of course if you are a signal, some part of you will get thru that soccer goal net, and only the part that bounces off will cause interference to those behind.

Well, as I said above "a not-very-good analogy", but I hope it helps some!

Cheers,
Tom

 

[Ephant]

Another way of stating what @Baluncore said above:

Please note that the RF impedance of a transmission line is an inherent characteristic describing the Voltage/Current ratio when driven with a given power.

For twin-lead or coaxial cable, the Characteristic Impedance is a function of conductor diameter, conductor spacing, and the dielectric constant of the material between the conductors.

The reason that high frequencies are sensitive to this is that the successive cycles of the waveform are so close together,

For a not-very-good analogy consider this:

1) You run full speed into a soccer goal net.​

Of course you wiil suddenly stop and probably bounce back a little.​

You recover and walk away.​

​

2) You and several other people in a line behind you run full speed into a soccer goal net.​

2a) If the people behind you are several feet apart, result is as in 1) above.​

2b) If the people behind you are close to you and each other, maybe 1 foot apart, when you bounce off the goal net you will interfere with the person behind you... and they will interfere with the person behind them...​

​

Of course if you are a signal, some part of you will get thru that soccer goal net, and only the part that bounces off will cause interference to those behind.

Well, as I said above "a not-very-good analogy", but I hope it helps some!

Cheers,
Tom

Ok. But the power souce is DC, let's say a laptop on battery or already fully charged. So when it uses the USB. It changes DC power to AC signal? Can you put it that way? I thought DC can only turn to AC using an inverter.

Edit: For an ordinary USB male plug. Is it impedanced matched to the wire? What is the impedance of a USB male plug?

 

[Tom.G]

I thought DC can only turn to AC using an inverter.

If you have DC and turn it On-Off-On-Off quickly, you end up with what is effectively AC. Like this: __---__---__--__--__---__--

Technically it is "pulsed DC", but because it acts much like AC it is often called that, especially when it is changing fast and we have to worry about running into that soccer goal net.

It is the transitions between levels that causes problems at the higher frequencies we have been talking about.

The AC at your wall outlet really is AC (Alternating Current) in that it changes polarity from + to - to + every cycle. The signals on the USB cable just change between (zero) 0 to +5 to 0 every cycle... but because of the rapid changes, it acts more like AC.

Cheers,
Tom

 

[renormalize]

@Ephant, I'm confused by what you're trying to accomplish. In your original post you asked:

Can the wire without twisted data paths reach Hi Speed of 480Mbps?

but in your last post you stated:

But the power souce is DC, let's say a laptop on battery or already fully charged.

Can you clearly state the purpose of your custom cable?
Do you want the cable to:

1. Send DC power from a source to a device?
2. Or transmit data signals at 480 Mbps between two digital devices?
3. Or both?

For 1) any simple wiring will work. But for 2) or 3) you need to carefully consider the impedance of your cable and the impact of mismatch and noise.

 

[Ephant]

@Ephant, I'm confused by what you're trying to accomplish. In your original post you asked:

but in your last post you stated:

Can you clearly state the purpose of your custom cable?
Do you want the cable to:

1. Send DC power from a source to a device?
2. Or transmit data signals at 480 Mbps between two digital devices?
3. Or both?

For 1) any simple wiring will work. But for 2) or 3) you need to carefully consider the impedance of your cable and the impact of mismatch and noise.

No. I just wondered how an equipment that was completely powered by DC can produce AC signal at the USB port.. I guess it's the USB chips that can do that. I was only familiar with impedance from the working of transformers. So I thought impedance can only came from AC. So this time it is chip based.

 

[Guineafowl]

No. I just wondered how an equipment that was completely powered by DC can produce AC signal at the USB port.. I guess it's the USB chips that can do that. I was only familiar with impedance from the working of transformers. So I thought impedance can only came from AC. So this time it is chip based.

Just to re-iterate: it’s not AC. The voltage pulses from 0 to 5V, so the current doesn’t reverse.

Impedance comes into play whenever the current is changing.

 

[f95toli]

"Generic" Lemo connectors were originally NOT meant for high frequencies. As I mentioned before we do use they quite a lot, but only for DC or low-frequency measurements where you typically don't have to worry about the impedance but you need a rugged, reliable connector. Note that Lemo was selling the exact same type of connector well before USB was even a thing.

That said, Lemo has in recent years designed some inserts to be used specifically with USB and ethernet.
https://web.lemo.com/img/resources/catalog/ROW/UK_English/High_speed_data_transfer.pdf

Hence, as long as you are using that specific model with the right type of cable you should be OK.

 

[Averagesupernova]

No. I just wondered how an equipment that was completely powered by DC can produce AC signal at the USB port.. I guess it's the USB chips that can do that. I was only familiar with impedance from the working of transformers. So I thought impedance can only came from AC. So this time it is chip based.

No offense intended but before you can get much out of the advice here besides accepting yes or no answers to your questions you need to hit the books.
-
A book on signals and systems is a good place to start. A transmission line with pulsating DC on it is generally thought of as having an AC signal with a DC offset. The transmission line doesn't care that the DC offset is there.

 

[Ephant]

Probably not, but it will be close.
You will need to look up the impedance data on those coupled plugs and sockets. Multipole plugs tend not to work well with twisted-pair transmission lines at such high frequencies.

The impedance of a parallel conductor transmission line, is determined by the ratio of conductor diameter to separation, and the dielectric constant of the insulation.
When a twisted pair is connected to two adjacent pins on the plug, the pin separation is fixed, so the diameter and insulation must be adjusted to maintain the impedance. If you follow through the connection, you will find it difficult to avoid steps in impedance, that will look like small series inductors, or parallel capacitors, which will reflect some signal energy.

Without special inserts (which ordinarily people won't add). Would the following pin configuration make the connection somehow be able to cause impedance matching for 480Mbps transfer? Notice the D- and D+ are both at the top of the 4-pinouts. This is for sake of discussion. I'm not making a cable.

 

[Vanadium 50]

@Ephant , will you just tell us what the heck you are trying to do? If you won't tell us, and you won't take the (excellent) advice to learn about Systems and Signals, you are leaving us all to guess. This takes a lot of work on our side for a very low probability of success on your side.

Whatever you are trying to do, there is likely a better way than what you propose.

 

[Ephant]

I just want to know if the china Lemo clones especially the 4 pin can also work in 480Mbps. f95toli. I looked at your hi-speed catalog. Does it mean all 4 pin LEMO series are automatically hi-speed? because I can't see the difference in the model between ordinary and hi-speed.

Im deciding whether to have a China based custom wire maker build me the cable for my Audiophile equipment that needs the lemo plug using their lemo clone that is way cheaper than genuine Lemo which happens to have the FGC.0B.304 model not available anywhere in the United States.

And oh, I saw the folloiwng at the internet. Left side is genuine Lemo, right side is china clone.

 

[Ephant]

clearer copy version

https://a4.pbase.com/o12/74/7061074/1/174281425.j39S8PMu.LEMOS2.jpg

 

[Vanadium 50]

I just want to know if the china Lemo clones especially the 4 pin can also work in 480Mbps.

Just out of random curiosity? Pick a cable at random, a connector at random and a protocol at random? I don't believe it.

You don't have to provide context, but if you make us guess, you are going to get worse advice. Or, as Barbara Billingsley once said...

 

[berkeman]

Just out of random curiosity? Pick a cable at random, a connector at random and a protocol at random? I don't believe it.

You don't have to provide context, but if you make us guess, you are going to get worse advice. Or, as Barbara Billingsley once said...

I just want to know if the china Lemo clones especially the 4 pin can also work in 480Mbps.

Lemo connectors are for audio, not half-GHz RF. This thread of yours has gone on long enough, and is now closed.

 

[berkeman]

UPDATE -- I missed this helpful post by @f95toli and I did not realize that there were special versions of LEMO connectors that were designed for RF:

That said, Lemo has in recent years designed some inserts to be used specifically with USB and ethernet.
https://web.lemo.com/img/resources/catalog/ROW/UK_English/High_speed_data_transfer.pdf

Hence, as long as you are using that specific model with the right type of cable you should be OK.

Thread is reopened provisionally.

I just want to know if the china Lemo clones especially the 4 pin can also work in 480Mbps. f95toli. I looked at your hi-speed catalog. Does it mean all 4 pin LEMO series are automatically hi-speed? because I can't see the difference in the model between ordinary and hi-speed.

Unless you use a connector that is specifically designed to maintain an RF $Z_0$ (characteristic impedance) all along the signal propagation path from cable through the connector(s), there will be reflections that will cause jitter in the receive waveform and "close the eye diagram", which can result in data errors. This includes how the wiring/cabling is handled once you get through the RF LEMO connectors in into the enclosure that you have shown in another picture. You would need to control the $Z_0$ of the interconnect inside the box as well, all the way to the USB IC that is receiving the signals.

And your comments about a DC device actually using AC signalling on USB are a bit baffling to me. You do understand how the 480Mbps signalling works on USB 2.0, right? (lower speed example shown below from Wikipedia)

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

 

[Averagesupernova]

And your comments about a DC device actually using AC signalling on USB are a bit baffling to me.

The reason for my post. In way over his head at that point. Have not had a reply on that. I meant it in a constructive manner.

 

[berkeman]

The reason for my post. In way over his head at that point. Have not had a reply on that. I meant it in a constructive manner.

Yeah, it makes me wonder why we are trying to give detailed $Z_0$ transmission line advice when they ask this:

No. I just wondered how an equipment that was completely powered by DC can produce AC signal at the USB port.. I guess it's the USB chips that can do that.

...but I guess we all have to start learning somewhere...

 

[Ephant]

Ok last request, let's compute it because in another electronics group, people are convinced all LEMO connectors can do USB 2.0 or 480Mbps. Theys said for example:

"You would have to make a catastrophically bad connector to fail to work for usb 2.0. A bad cable maybe but the connector is just too small and absent deliberate inductors in the signal path or failing to connect a wire altogether its going to work."

I showed f95toli document and one commented:

"I saw that document long ago. It’s just their recommendations. As far as I can tell, they aren’t special insert configurations, at least not for USB 2, they’re just their standard 4-pin inserts in the recommended sizes, with a recommended pinout. They probably calculated which of their existing inserts works best for USB 2 and recommend that connector size. Because if you look at their overall catalog, “304” is one of their standard insert sizes, with no mention of USB 2. I think they list that on some SKUs simply as a convenience."

"Note that in that datasheet, when they say “proprietary interface” as the insert type, they mean “LEMO design, not an open standard”, NOT that it was designed specifically for USB!".

Ok. Let's compute it. Why can't ordinary LEMO connectors survive 480Mbps?? In original USB male plug, is it impedance matched to the twisted wires? Please show computations as one last request. Thank you..