USB cable question

  • Thread starter Bill Bonde { ''Direct And On Point'')
  • Start date
B

Bill Bonde { ''Direct And On Point'')

I'm reading the following from usb.org and they seem determined to
prevent anyone from making cables longer than 5 metres. But if you
look at their protestations, isn't it true that the timing spec
issues aren't based on 5 metres but instead on a worst case
scenario with 5 hubs and 5 cables? That would be more like 25
metres.

So this leads to the question, at 5 metres, is the problem with the
USB data lines or with the five volt power lines? If it's the power
lines, could one provide the five volts to the device directly and
create a longer cable for the data? If not, why not?


http://www.usb.org/developers/usbfaq/#sig7
#begin quote
Cables and Long-Haul Solutions

1. Why are there cable length limits, and what are they?

A: The cable length was limited by a cable delay spec of 26ns to
allow for reflections to settle at the transmitter before the next
bit was sent. Since USB uses source termination and voltage-mode
drivers, this has to be the case, otherwise reflections can pile up
and blow the driver. This does not mean the line voltage has fully
settled by the end of the bit; with worst-case undertermination.
However, there's been enough damping by the end of the bit that the
reflection amplitude has been reduced to manageable levels. The low
speed cable length was limited to 18ns to keep transmission line
effects from impacting low speed signals.

top top
gray line

2. I want to build a cable longer than 5 meters, why won't this
work?

A: Even if you violated the spec, it literally wouldn't get you
very far. Assuming worst-case delay times, a full speed device at
the bottom of 5 hubs and cables has a timeout margin of 280ps.
Reducing this margin to 0ps would only give you an extra 5cm, which
is hardly worth the trouble.

top top
gray line

3. What about using USB signal repeaters to make a cable longer
than 5 meters?

A: Don't bother. The best solution is self-powered hub with a
fixed 10m cable that had a one-port bus powered hub in the middle.
The maximum range will still have to deal with the timeout, so any
out of spec tweaking of the terminations between the two hubs and
the timing budget still won't yield more than 5cm of extra
distance. A better solution is described in the following question.

top top
gray line

4. I really need to put a USB device more than 30 meters away
from my PC. What should I do?

A: Build a USB bridge that acts as a USB device on one side and
has a USB host controller at the other end. Use a long-haul
signaling protocol like Ethernet or RS-485 in the middle. Using
cables or short-haul fiber, you can get ranges upwards of a
kilometer, though there's no reason why the long-haul link in the
middle of the bridge couldn't be a pair of radio transceivers or
satellite modems.
Embedded host solutions capable of doing this already exist. Also,
two PCs connected via USB Ethernet adapters are essentially a
slave/slave version of this master/slave bridge.

top top
#end quote
 
P

Paul

Bill said:
I'm reading the following from usb.org and they seem determined to
prevent anyone from making cables longer than 5 metres. But if you
look at their protestations, isn't it true that the timing spec
issues aren't based on 5 metres but instead on a worst case
scenario with 5 hubs and 5 cables? That would be more like 25
metres.

So this leads to the question, at 5 metres, is the problem with the
USB data lines or with the five volt power lines? If it's the power
lines, could one provide the five volts to the device directly and
create a longer cable for the data? If not, why not?


http://www.usb.org/developers/usbfaq/#sig7
#begin quote
Cables and Long-Haul Solutions

1. Why are there cable length limits, and what are they?

A: The cable length was limited by a cable delay spec of 26ns to
allow for reflections to settle at the transmitter before the next
bit was sent. Since USB uses source termination and voltage-mode
drivers, this has to be the case, otherwise reflections can pile up
and blow the driver. This does not mean the line voltage has fully
settled by the end of the bit; with worst-case undertermination.
However, there's been enough damping by the end of the bit that the
reflection amplitude has been reduced to manageable levels. The low
speed cable length was limited to 18ns to keep transmission line
effects from impacting low speed signals.

top top
gray line

2. I want to build a cable longer than 5 meters, why won't this
work?

A: Even if you violated the spec, it literally wouldn't get you
very far. Assuming worst-case delay times, a full speed device at
the bottom of 5 hubs and cables has a timeout margin of 280ps.
Reducing this margin to 0ps would only give you an extra 5cm, which
is hardly worth the trouble.

top top
gray line

3. What about using USB signal repeaters to make a cable longer
than 5 meters?

A: Don't bother. The best solution is self-powered hub with a
fixed 10m cable that had a one-port bus powered hub in the middle.
The maximum range will still have to deal with the timeout, so any
out of spec tweaking of the terminations between the two hubs and
the timing budget still won't yield more than 5cm of extra
distance. A better solution is described in the following question.

top top
gray line

4. I really need to put a USB device more than 30 meters away
from my PC. What should I do?

A: Build a USB bridge that acts as a USB device on one side and
has a USB host controller at the other end. Use a long-haul
signaling protocol like Ethernet or RS-485 in the middle. Using
cables or short-haul fiber, you can get ranges upwards of a
kilometer, though there's no reason why the long-haul link in the
middle of the bridge couldn't be a pair of radio transceivers or
satellite modems.
Embedded host solutions capable of doing this already exist. Also,
two PCs connected via USB Ethernet adapters are essentially a
slave/slave version of this master/slave bridge.

top top
#end quote

Maybe you could ask over on the USB.org forum ? I don't know
if registration is restricted to card carrying members or not.
I tried their search engine, but didn't see any good discussions
on cable length.

https://www.usb.org/phpbb/

Looking at the 5.8MB 650 page spec myself, I'd say it was a
conspiracy to support interoperability :)

On the one hand, if I look at 7.2.2 Voltage Drop budget, the 95mV
cable component of the drop is neatly met by using 20 gauge wire
in the power/ground pair, and limiting the length to 5 meters,
at a current flow of 500mA. The 500mA current flow would only
occur if the cable was between a host and a high power function,
while the diagram shows a hub in the picture. But I expect the
same voltage drop budget exists in either case. So that is one
constraint on the cable. If you tried to make your own longer custom
cable, and have it carry 500mA with less than 95mV drop, then
the power/ground gauge would have to be thicker than 20 gauge.
So the cable design seems predicated on satisfying host to
function direct connect, at 5 meters of 20 gauge, while the
string-of-hubs configuration probably doesn't stress that budget.
(Limited to a 100mA load at the end function ? Five times longer,
five times less current.) If you made your cable to only suit
your particular requirements (such as only self powered device
at the end), then power/ground is not such a problem.

7.1.19.2 shows where the 1500 nanoseconds comes from, which is
a number used for timeout timers. Part of the delay is
"thru-delay" in each hub. 26ns per 5 meter cable is the
cable budget. They work out the round trip delay as
721 bit times, times 2.08ns/bit, gives 1502ns or so.
The timers are set a bit higher than that number.

Now, if you made a hypothetical custom setup, where
your equipment did not interoperate with every possible
legal configuration of gear allowed by the spec, I expect
you could change things. The ingredient I'm missing at the
moment, is a discussion of what the signals look like on
the cable. They talk about reflections and settling time,
as if the cable is not terminated properly.

Table 7-1 mentions the 480Mbit/sec mode uses current
mode drive of 17.78 mA and doubly terminated lines
(45 ohm resistor at either end of each wire, 90 ohm
differential impedance). So doubly terminated, should
give good characteristics.

In section 7.1.17, there is a graph of attenuation versus
frequency, for the cable. Loss is 5.8dB max for the 5 meter
cable, at 400MHz. The signal is carried differentially, and
presumably has a decently tight skew between D+ and D- (I
see 100pS mentioned), so cheating and using two low loss
coax might not work out too well. Maybe something like a
twinax might work, but it would be sheer chance if the impedance
spec happened to match up with something like that. (Picking
cables for stuff, is a PITA.) What I'm thinking about here,
is where I'd find a low loss substitute for a cable.

Perhaps the reflections and settling issue, has to do with
using lower speed protocols first ? I understand that a
USB2 device, first operates at the lower speed, before a
bus reset, and it switches to USB2 480Mbit/sec mode. So
perhaps your longer cable, would fail to make USB 1.1
transmission work properly.

Note - I don't work with USB, so this is just a casual glance
at the spec. And there is a lot of content in the spec, that
needs checking.

I expect making your own, good quality cables, is not going
to be that easy.

Paul
 
B

Bill Bonde { ''Direct And On Point'')

kony said:
No, that is still 5 meters. Distance is from one active
device to the next, which does send/receive of signals not
just passive connectios.
But the worst case *timing* they were whining about is for the
entire link.

data lines and ground

Better question is what are you trying to do? There are
lots of other ways to connect gear, usually, instead of
trying to violate specs on a bus that isn't suited for a
purpose.
I want to violate the specs because that's the cheapest way to hook
up the USB devices.


However, if you had a low enough gauge cable and were
pulling enough current, 5V (and ground again) would also be
a problem, and to some extent it is supposedly possible to
keep a cable working at a little longer distance by using
higher gauge wires.
That sounds like the issue is really the 5v and not the data line
distance.


Because the data lines matter too.
That's the sort of answer I was getting at the USB website. They
don't want you to do it because it means they are competing with
more expensive technologies.


The "don't bother" is a false bit of nonsense. What is
meant is it will work but they *decided* you might as well
use a more fully functional device instead.
Will their other solution cost money?


In other words, don't use USB for it, use the right tech and
convert that to USB as required, if possible.
A converter costs how much?



They're overcomplicating what can be said more simply. USB
can't do it. Must convert to a communication standard that
can then convert back again if using USB at all is really
necessary.

None of this really gets at the heart of why you want a
longer USB cable. There are ethernet or wifi networking
devices that can act as print servers, NAS, etc, and other
models of some peripherals that don't need USB at all for
the distance the data travels, for example a bluetooth USB
dongle plugged into the host system for a wireless link to a
mouse or phone.
I'm not arguing that these things don't exist.
 

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments. After that, you can post your question and our members will help you out.

Ask a Question

Top