Quality USB hub?

[ultraport]

They use logic to buffer the signal?
So the properly buffered cable can be any length?
Is 16' too long for one cable segment?

Thanks.

 

[Pen]

They use logic to buffer the signal?
So the properly buffered cable can be any length?
Is 16' too long for one cable segment?

Thanks.

It's right at the cusp of go/nogo. The question isn't
buffering, rather it's delay. Too long a delay and the
sender thinks there's no one there. The spec is 5 meters.
So you'll have to try it and see.

 

[ultraport]

On 9/22/2011 3:48 PM, (e-mail address removed) wrote:> They use logic to bufferthe signal?

It's right at the cusp of go/nogo. The question isn't
buffering, rather it's delay. Too long a delay and the
sender thinks there's no one there. The spec is 5 meters.
So you'll have to try it and see.

Total cable length is usually limmited to 5 meters?
I see your point, but I suspect the logic used effects the lag.
Thanks.

 

[Loren Pechtel]

They use logic to buffer the signal?
So the properly buffered cable can be any length?
Is 16' too long for one cable segment?

Thanks.

You can't get around the length limit other than by stringing hubs
together.

 

[ultraport]

You can't get around the length limit other than by stringing hubs
together.

And some hubs are faster than others?

 

[Loren Pechtel]

Total cable length is usually limmited to 5 meters?
I see your point, but I suspect the logic used effects the lag.
Thanks.

The lag is set by Einstein.

 

[Paul]

Total cable length is usually limmited to 5 meters?
I see your point, but I suspect the logic used effects the lag.
Thanks.

USB travels in "hops", for all practical purposes.

You can get a passive cable, which is 5 meters long. But if
you want to extend it, you plug in a hub, then another 5 meters of
cable and so on. (The USB spec is in nanometers, and if someone
were to change the "time of flight", perhaps a small alteration
in reach could be made. Really long reaches, are done with bridging.
The USB spec assumes a cable propagation delay of 5.2ns/meter.)

Since cable+hub is a common configuration, they also make "active" 5 meter
cables, which have a "one port hub" on the end. That blob, terminates
and effectively buffers the signal.

Since bus powering is affected by all that length of cable, best
practice is to put a powered hub on the end of the chain. So perhaps
your "last" extension cable is passive, then a wall-wart powered hub,
then your USB peripheral plugs in there. That config covers cases
where the peripheral needs 500mA, and the wall wart powered hub
can provide that current. Active buffer cables (running off bus power)
would be limited to 100mA loads, as far as I know, so only things like
a keyboard or mouse, would go on the end of a bunch of active cables.

All of the info is here, if you need it.

http://www.usb.org/developers/usbfaq/

One other quote for you:

According to the USB Specification, the maximum limit of hubs cascaded in
series cannot exceed five.

(Figure 7-41 of USB20 spec, has the maths nearby for this.)

http://www.usb.org/developers/docs/usb_20_071411b.zip (extract USB20.pdf from the zip)

So now the picture looks like this. This is an example of a 30 meter
reach, with full power available at the end (500mA). The wall powered
hub down near the end, ensures you've got good quality power.

MB
SB 5M
USB --active_cable--(hub) 5M
\--active_cable--(hub) 5M
\--active_cable--(hub)
\
(hub)--active_cable--/
5M / 5M
Wall_powered_hub --passive_cable--
/
\--passive_cable--500mA_USB_peripheral
5M

Note: Certain Intel chipsets of very recent vintage, have a "hub"
inside. Intel decided, after having the same USB architecture
for eight or nine years, to just arbitrarily change it. The
USB1.1 and USB2.0 logic blocks traditionally provided in the
Southbridge (SB) have been changed to one unified "hub" block.
As far as I know, that cuts down the external reach to 25 meters.
If you buy a USB2 add-in card for $10-$15 or so and slap that
into the motherboard, you can go back to 30 meters again :-(

Much longer reaches are possible with various bridging approaches. And
you can spend the whole night researching those. I see Newegg lists
some cables, of dubious length, with no technical jargon to back up
the product.

This is an example of a 5 meter cable, using a USB hub chip in the "blob"
on the end. You could buy four of these, as part of the above diagram
(the rest of the reach being done with passive cables). Note that not
all these active cables are built equally. Years ago, there were
some which could not be cascaded and a chain of two would fail. And at
the same time, other brands worked. Now, I see at least one brand
has managed to make cables where they're all bad out of the box. Bravo!
Always read the reviews, before you buy, and make sure that *somebody*
has tested that the cable works when cascaded. If there aren't reviews
of your liking on Newegg, check out Amazon reviews instead.

http://www.newegg.com/Product/Product.aspx?Item=N82E16812339009

So 5 meters is the limit per hop, unless something non-standard is used
for an implementation. And then it's up to the manufacturer to guarantee
it will work.

HTH,
Paul

 

[Paul]

And some hubs are faster than others?

There are always going to be standard solutions
and non-standard solutions. For the standard solutions,
we don't have to worry about how they work - they just
work. A standard solution will list a 5 meter or 16 foot reach.

If I look at this 150 foot solution, it turns out to run at USB 1.1 speed.

http://www.amazon.com/Cables-Go-293...YK/ref=sr_1_18?ie=UTF8&qid=1316751563&sr=8-18

OK, try another one. Different price bracket.

http://www.amazon.com/Gefen-EXT-USB...JY/ref=sr_1_40?ie=UTF8&qid=1316751996&sr=8-40

"USB2 over 330ft"
http://www.gefen.com/kvm/dproduct.jsp?prod_id=5529

But we'd need some thorough reviews to know whether it's the real deal.
And I don't think I've seen a real review of a non-standard extender.

There are no prices on these units, but one of them claims a 10km reach
over fiber optic cable. Implying, with bridging, the delay doesn't
affect the basic protocol. But it might affect what you're doing
(since normally, extra long delays over USB aren't expected).

http://www.zantech.com.au/extender/usb/r442-r444.html

If you've got the money to try products like that, well,
tell us how they worked out The first item, that
ran at USB 1.1, at least one user waited more than 30 days
to test his product, so got stuck with it and couldn't
return it.

Paul

 

[ultraport]

USB travels in "hops", for all practical purposes.

You can get a passive cable, which is 5 meters long. But if
you want to extend it, you plug in a hub, then another 5 meters of
cable and so on. (The USB spec is in nanometers, and if someone
were to change the "time of flight", perhaps a small alteration
in reach could be made. Really long reaches, are done with bridging.
The USB spec assumes a cable propagation delay of 5.2ns/meter.)

Since cable+hub is a common configuration, they also make "active" 5 meter
cables, which have a "one port hub" on the end. That blob, terminates
and effectively buffers the signal.

Since bus powering is affected by all that length of cable, best
practice is to put a powered hub on the end of the chain. So perhaps
your "last" extension cable is passive, then a wall-wart powered hub,
then your USB peripheral plugs in there. That config covers cases
where the peripheral needs 500mA, and the wall wart powered hub
can provide that current. Active buffer cables (running off bus power)
would be limited to 100mA loads, as far as I know, so only things like
a keyboard or mouse, would go on the end of a bunch of active cables.

All of the info is here, if you need it.

http://www.usb.org/developers/usbfaq/

One other quote for you:

    According to the USB Specification, the maximum limit of hubs cascaded in
    series cannot exceed five.

(Figure 7-41 of USB20 spec, has the maths nearby for this.)

http://www.usb.org/developers/docs/usb_20_071411b.zip  (extract USB20.pdf from the zip)

So now the picture looks like this. This is an example of a 30 meter
reach, with full power available at the end (500mA). The wall powered
hub down near the end, ensures you've got good quality power.

    MB
    SB         5M
    USB --active_cable--(hub)      5M
                           \--active_cable--(hub)      5M
                                               \--active_cable--(hub)
                                                                   \
                                              (hub)--active_cable--/
                                       5M       /         5M
              Wall_powered_hub --passive_cable--
             /
             \--passive_cable--500mA_USB_peripheral
                      5M

Note: Certain Intel chipsets of very recent vintage, have a "hub"
       inside. Intel decided, after having the same USB architecture
       for eight or nine years, to just arbitrarily change it. The
       USB1.1 and USB2.0 logic blocks traditionally provided in the
       Southbridge (SB) have been changed to one unified "hub" block.
       As far as I know, that cuts down the external reach to 25 meters.
       If you buy a USB2 add-in card for $10-$15 or so and slap that
       into the motherboard, you can go back to 30 meters again :-(

Had not thought of that, thanks.





..

 

[Loren Pechtel]

But we'd need some thorough reviews to know whether it's the real deal.
And I don't think I've seen a real review of a non-standard extender.

There are no prices on these units, but one of them claims a 10km reach
over fiber optic cable. Implying, with bridging, the delay doesn't
affect the basic protocol. But it might affect what you're doing
(since normally, extra long delays over USB aren't expected).

How I've figured they build such things is that the USB receiver is in
the very start of the cable and the data gets sent over it in some
different format and then converted back on the other end.