Post by thread:[EE:] USB cable length

At 03:18 PM 1/9/2004 -0500, Bob Blick wrote: >Before I try, has anyone experimented with extending USB cable length on a >cheap webcam? > >If so, how far could you go? 10 feet? 20 feet? 30 feet? > >I had an old parallel port color quickcam, it seemed that 15 feet was OK, >20 feet started to get flaky. I'd expect USB to work a little farther, but >not much. You can use hubs or powered repeaters to extend the distance. Also, IONetworks makes a USBAnywhere that lets you plug USB devices into your computer over a twisted pair lan. I have one at home, it's pretty spooky to plug in a mouse across the room, and watch it running around on the main system screen just like you were sitting there. Great prank potential, you could mess with someone's mouse from a completely remote place. It only works when the lan is up, but unplugging the lan would not be my first guess when having mouse trouble. -- http://www.piclist.com hint: The PICList is archived three different ways. See http://www.piclist.com/#archives for details.

Bob Blick <spam_OUTPICLISTTakeThisOuTmitvma.mit.edu> wrote: > Before I try, has anyone experimented with extending USB cable length > on a cheap webcam? > > If so, how far could you go? 10 feet? 20 feet? 30 feet? > > I had an old parallel port color quickcam, it seemed that 15 feet was > OK, 20 feet started to get flaky. I'd expect USB to work a little > farther, but not much. Bob: You can go five meters with just a cable, or up to 30 meters with a series of cables and hubs (or a series of active USB extension cables). USB cables are limited to 5 meters because that length is short enough to allow reflections from the far end of the cable to settle between bits; if a longer cable is used, reflections won't be damped as well, and the line voltage could rise high enough to cause physical damage to the drivers. You can connect a series of 5 USB hubs and cables to get a maximum host-to-device length of 30 meters (the "active USB extension cables" are really just single-port hubs between two cables), but you can't use more than 5 hubs. The 5-hub limit on serially-connected hubs is a consequence of the USB spec for maximum turnaround delay between an outgoing packet and the incoming response; that spec allows for 70 ns of propagation delay through each cable/hub combination, and 30 ns through each cable alone. The sum of the propagation delays through 5 hubs and 6 cables (and back), plus the delay allowed in the device itself, beats the full-speed timeout spec by less than half a nanosecond. Adding even a few inches of cable to a full-length serial string of USB cables and hubs would violate that spec. There are more-exotic USB extenders that can give you miles of range, but they're expensive... They work by putting what looks like a USB device right next to the PC host, and what looks like a USB host right next to the device, then communicating betweeen the two using something other than USB. -Andy === Andrew Warren -- .....aiwKILLspam@spam@cypress.com === Principal Design Engineer === Cypress Semiconductor Corporation === === Opinions expressed above do not === necessarily represent those of === Cypress Semiconductor Corporation -- http://www.piclist.com hint: The PICList is archived three different ways. See http://www.piclist.com/#archives for details.

Andrew Warren wrote: > Bob Blick <PICLISTKILLspammitvma.mit.edu> wrote: > > > Before I try, has anyone experimented with extending USB cable length > > on a cheap webcam? > > > > If so, how far could you go? 10 feet? 20 feet? 30 feet? > > > > I had an old parallel port color quickcam, it seemed that 15 feet was > > OK, 20 feet started to get flaky. I'd expect USB to work a little > > farther, but not much. > > Bob: > > You can go five meters with just a cable, or up to 30 meters with a > series of cables and hubs (or a series of active USB extension > cables). And the absurd thing is that the LOW SPEED USB devices have a much SHORTER cable limit than the high speed ones? Any rational explanation for that would be appreciated. > USB cables are limited to 5 meters because that length is short > enough to allow reflections from the far end of the cable to settle > between bits; if a longer cable is used, reflections won't be damped > as well, and the line voltage could rise high enough to cause > physical damage to the drivers. It take it that the USB designers never heard of 'terminations' to prevent reflections? Or 'transient clamp diodes'? {Quote hidden}> You can connect a series of 5 USB hubs and cables to get a maximum > host-to-device length of 30 meters (the "active USB extension cables" > are really just single-port hubs between two cables), but you can't > use more than 5 hubs. The 5-hub limit on serially-connected hubs is > a consequence of the USB spec for maximum turnaround delay between an > outgoing packet and the incoming response; that spec allows for 70 ns > of propagation delay through each cable/hub combination, and 30 ns > through each cable alone. The sum of the propagation delays through > 5 hubs and 6 cables (and back), plus the delay allowed in the device > itself, beats the full-speed timeout spec by less than half a > nanosecond. Adding even a few inches of cable to a full-length serial > string of USB cables and hubs would violate that spec. Would it not help to use a 'low capacitance' cable, which has a higher propagation velocity? I've read of a videomaker in Montreal who runs firewire links to his camera that are WAY over spec (150' he claims) by using Cat5 wire. I've been meaning to do this with my firewire camera, but have not been able to find a Male-Female firewire extension cable I could hack up. > There are more-exotic USB extenders that can give you miles of range, > but they're expensive... They work by putting what looks like a USB > device right next to the PC host, and what looks like a USB host > right next to the device, then communicating betweeen the two using > something other than USB. So what's the point of using USB then? How long before we see ethernet mice/keyboards and cameras? Looks to me like a market opportunity, particularly for cameras. If I can get a 100BT card for under $10, putting the chip IN the camera can't be THAT expensive. Robert -- http://www.piclist.com hint: The PICList is archived three different ways. See http://www.piclist.com/#archives for details.

I guess this is mainly a question for Andy, as he's likely pretty up to date on the USB standard. There is a local store here that I've seen USB A-B cables in 25 foot lengths. Now, the packages could be mislabeled, I'm not sure. These are the no name, made in China imported jobbies. These cables would seem to be definately out of spec. They are not powered extensions, they don't have a big enough "hump" on the cable. If I read your post correctly, having a cable that long could damage the drivers, correct? So why would anyone make one that long? Is it possible that it might work? Josh -- A common mistake that people make when trying to design something completely foolproof is to underestimate the ingenuity of complete fools. -Douglas Adams Andrew Warren wrote: > USB cables are limited to 5 meters because that length is short > enough to allow reflections from the far end of the cable to settle > between bits; if a longer cable is used, reflections won't be damped > as well, and the line voltage could rise high enough to cause > physical damage to the drivers. -- http://www.piclist.com hint: The PICList is archived three different ways. See http://www.piclist.com/#archives for details.

Josh Koffman <.....PICLISTKILLspam.....mitvma.mit.edu> wrote: > There is a local store here that I've seen USB A-B cables in 25 > foot lengths. Now, the packages could be mislabeled, I'm not sure. > These are the no name, made in China imported jobbies. These cables > would seem to be definately out of spec. They are not powered > extensions, they don't have a big enough "hump" on the cable. If I > read your post correctly, having a cable that long could damage the > drivers, correct? It could, yes. In the real world, it probably won't, but it's certainly out of spec. The trademarked USB logo is legitimately carried only by products (including cables) that have passed USB compliance testing. Lack of a logo is a good clue that you're looking at a non-compliant (or at least non-tested) product. > So why would anyone make one that long? Is it possible that it > might work? Sure, it could work. Using a 25-foot cable is loosely analogous to overclocking a PC video card (or a PIC): In any specific system, it may work well enough for a particular application, but it may not meet all of its specifications... And under those conditions, it's certainly not guaranteed to do so by its manufacturer. -Andy === Andrew Warren -- EraseMEaiwspam_OUTTakeThisOuTcypress.com === Principal Design Engineer === Cypress Semiconductor Corporation === === Opinions expressed above do not === necessarily represent those of === Cypress Semiconductor Corporation -- http://www.piclist.com hint: The PICList is archived three different ways. See http://www.piclist.com/#archives for details.

It is actually outside the USB spec to make any 'extension cord' or overlong cable. For example a A-male to A-female or B-male to B-female cable are "illegal" because they could be used to make a connection that is too long. The design decisions made in the USB spec were intended so that no matter what you did, if you could hook something up, it would work. (The only exception is that you could hook up a chain of more than five hubs, which is relatively unlikely to happen). Another design decision involved having low speed devices requiring captive cables (or cables with a custom (non-USB connector) end at the device) that are not longer than a certain limit. I forget the details on this, but I do remember that it was a requirement to ensure reliable operation (otherwise the user could use a cable designed for full speed operation and get problems). And yes, low speed devices are limited to shorter cables than full speed devices because of the simpler signaling scheme used. Unfortunately idiotic companies don't follow the spec properly, leaving support folks to deal with the grief. Bob Ammerman RAm Systems {Original Message removed}

> > You can go five meters with just a cable, or up to 30 meters with a > > series of cables and hubs (or a series of active USB extension > > cables). > > And the absurd thing is that the LOW SPEED USB devices have > a much SHORTER cable limit than the high speed ones? > Any rational explanation for that would be appreciated. My guess is low speed devices are expected to be very low power, i.e. keyboards, mice, etc. Low power means weaker drivers. TTYL > So what's the point of using USB then? Originally? To replace the parallel, keyboard, mouse, serial and other legacy ports. In most cases these devices are all low power, low speed, and close to the PC. > How long before we see ethernet mice/keyboards and cameras? > Looks to me like a market opportunity, particularly for cameras. > If I can get a 100BT card for under $10, putting the chip IN > the camera can't be THAT expensive. Not necessary. Necessity is the mother of all invention. Lack of need means something won't catch on. It seems you don't understand what USB was originally meant for and expect more from it then you should. TTYL ---------------------------------- Herbert's PIC Stuff: http://repatch.dyndns.org:8383/pic_stuff/ -- http://www.piclist.com hint: The PICList is archived three different ways. See http://www.piclist.com/#archives for details.

Robert Rolf <PICLISTspam_OUTmitvma.mit.edu> wrote: > the absurd thing is that the LOW SPEED USB devices have a much > SHORTER cable limit than the high speed ones? Any rational > explanation for that would be appreciated. Reading the USB spec (available for free at http://www.usb.org) would explain it. See Chapter 7. Paraphrased from the spec: Low-speed USB signalling has slow rise and fall times (no less than 75 ns and no more than 300 ns under specific test conditions). The 3-meter maximum cable length is to avoid transmission-line effects: Limiting the cable length to 3 meters (18 ns propogation delay) ensures that reflections will appear during the first half of the rise or fall, which allows the cable to be approximated by a lumped capacitance. > It take it that the USB designers never heard of 'terminations' to > prevent reflections? Or 'transient clamp diodes'? Yeah, I'm sure they never heard of any of that. Sounds too complicated for those dumb USB guys to understand. > > The sum of the propagation delays through 5 hubs and 6 cables > > (and back), plus the delay allowed in the device itself, beats > > the full-speed timeout spec by less than half a nanosecond. > > Adding even a few inches of cable to a full-length serial string > > of USB cables and hubs would violate that spec. > > Would it not help to use a 'low capacitance' cable, which has a > higher propagation velocity? The USB spec is already written and allows 30 ns for the cable propagation delay. If you have a cable that meets all the other requirements and has a propagation velocity closer to the speed of light than what the USB spec assumes, you can make a cable longer than 5 meters. If you can exceed the speed of light, that'd be even better. > I've read of a videomaker in Montreal who runs firewire links to > his camera that are WAY over spec (150' he claims) by using Cat5 > wire. I've been meaning to do this with my firewire camera, but > have not been able to find a Male-Female firewire extension cable I > could hack up. Firewire's significantly different from USB; there's no necessary correlation between what works for one and what'll work for the other. > > There are more-exotic USB extenders that can give you miles of > > range, but they're expensive... [they communicate] using > > something other than USB. > > So what's the point of using USB then? The point is mostly contained in the phrase "but they're expensive". There's a LOT more, but cost is a big-enough deal that further explanation is usually unnecessary. > How long before we see ethernet mice/keyboards and cameras? Ethernet cameras? There are already some; they have an absolutely insignificant share of the webcam market. You'll never see ethernet mice and keyboards. > Looks to me like a market opportunity, particularly for cameras. If > I can get a 100BT card for under $10, putting the chip IN the > camera can't be THAT expensive. Keep studying the problem; you'll see why USB is better. -Andy === Andrew Warren -- @spam@aiwKILLspamcypress.com === Principal Design Engineer === Cypress Semiconductor Corporation === === Opinions expressed above do not === necessarily represent those of === Cypress Semiconductor Corporation -- http://www.piclist.com hint: The PICList is archived three different ways. See http://www.piclist.com/#archives for details.

Bob Ammerman wrote: > > It is actually outside the USB spec to make any 'extension cord' or overlong > cable. For example a A-male to A-female or B-male to B-female cable are > "illegal" because they could be used to make a connection that is too long. > > The design decisions made in the USB spec were intended so that no matter > what you did, if you could hook something up, it would work. (The only Yeah, right. Some crap driver s/w won't work even with a short cable! (WaveX MP3 player for one). > exception is that you could hook up a chain of more than five hubs, which is > relatively unlikely to happen). Happened at a Comdex demo of early USB. "Look how many USB devices we can hook up". They supposedly made it out to 250 something before the software broke. With the typical 4 port hubs of that day, that would mean 3**5 =243 maximum devices. > Another design decision involved having low speed devices requiring captive > cables (or cables with a custom (non-USB connector) end at the device) that > are not longer than a certain limit. I forget the details on this, but I do > remember that it was a requirement to ensure reliable operation (otherwise > the user could use a cable designed for full speed operation and get > problems). And yes, low speed devices are limited to shorter cables than > full speed devices because of the simpler signaling scheme used. And yet they use type A connectors, for which extension cords are readily available. Part of the distance issue is that manufactures use the cheapest cables they can get away with. Higher grade (low capacitance) cables (as used on IEEE 1288 printer cables) extend the range markedly. > Unfortunately idiotic companies don't follow the spec properly, leaving > support folks to deal with the grief. You're talking mostly about MicroSloth I presume... Q: How many MicroSoft programmers does it take to change a light bulb? A: NONE. They just define "Darkness" to be the new standard. R -- http://www.piclist.com hint: The PICList is archived three different ways. See http://www.piclist.com/#archives for details.

Herbert Graf wrote: > > > Andrew Warren wrote: > > > > Looks to me like a market opportunity, particularly for cameras. > > > > If I can get a 100BT card for under $10, putting the chip IN the > > > > camera can't be THAT expensive. > > > > > > Keep studying the problem; you'll see why USB is better. > > > > Standalone webcams with built-in Linux based web-server > > connected to intranet. How USB is supposed to be better than > > ethernet to this very much real case? > > > > Mike. > > Very simple: price. Only if the camera is near a PC. If you need a 150' cable run, it gets awfully expensive, Or you're not -really- running USB on the long path. Robert -- http://www.piclist.com#nomail Going offline? Don't AutoReply us! email spamBeGonelistservspamBeGonemitvma.mit.edu with SET PICList DIGEST in the body

On Friday, Jan 9, 2004, at 14:23 US/Pacific, Robert Rolf wrote: > How long before we see ethernet mice/keyboards and cameras? > Looks to me like a market opportunity, particularly for cameras. > If I can get a 100BT card for under $10, putting the chip IN > the camera can't be THAT expensive. > > However, the software stack to support something like video over ip over ethernet is pretty expensive, once you add the baggage "necessary" to ensure reasonable Quality Of Service over a protocol that was not designed to provide it. USB took the desire for video into account during its design, and requires fewer fixes... (and we won't talk about the relative costs of administering an IP ethernet network vs a group of USB devices, either.) BillW -- http://www.piclist.com#nomail Going offline? Don't AutoReply us! email TakeThisOuTlistservEraseMEspam_OUTmitvma.mit.edu with SET PICList DIGEST in the body

William Chops Westfield wrote: > > How long before we see ethernet mice/keyboards and cameras? > > Looks to me like a market opportunity, particularly for cameras. > > If I can get a 100BT card for under $10, putting the chip IN > > the camera can't be THAT expensive. > > > However, the software stack to support something like video > over ip over ethernet is pretty expensive, once you add the > baggage "necessary" to ensure reasonable Quality Of Service > over a protocol that was not designed to provide it. USB > took the desire for video into account during its design, > and requires fewer fixes... How much is "software stack to support something like video over ip over ethernet is pretty expensive"? From http://www.aegi.com/store/netipcam.html +++++++++++++++++++++++++++++++++++++++++++++ Intellinet Network IP Camera $365.00 . No special software or additional hardware required. Simply type the camera's assigned IP address into your browser and view live, full motion video . Access the camera from any PC on your LAN . Assign a "live" IP address for anytime, anywhere viewing over the Internet . Ideal for monitoring buildings, offices, daycare, schools, homes, etc. - anywhere with a network connection . Embedded operating system streams live video directly to your website . Records streaming video or snapshots on your PC . Automatically uploads images to FTP server or sends images out via email, fully configurable . Built-in Motion Detection Function . Lifetime warranty Specifications: . JPEG compression, 10 levels . Image Rate: Up to 30 frames/second, European version 25 frames/second . Resolution: 640 x 480, 320 x 240, 160x120 . Supported Protocols: TCP / IP, UDP, ARP, FTP, TFTP, HTTP, DHCP, SMTP, SNTP, SNMP, etc. . Output of 12V to signal external devices, max 50mA . Embedded operating system . CMOS lens +++++++++++++++++++++++++++++++++++++++++++++ Mike. -- http://www.piclist.com#nomail Going offline? Don't AutoReply us! email RemoveMElistservTakeThisOuTmitvma.mit.edu with SET PICList DIGEST in the body

{Quote hidden}> Herbert Graf wrote: > > > > > Andrew Warren wrote: > > > > > Looks to me like a market opportunity, particularly for cameras. > > > > > If I can get a 100BT card for under $10, putting the chip IN the > > > > > camera can't be THAT expensive. > > > > > > > > Keep studying the problem; you'll see why USB is better. > > > > > > Standalone webcams with built-in Linux based web-server > > > connected to intranet. How USB is supposed to be better than > > > ethernet to this very much real case? > > > > > > Mike. > > > > Very simple: price. > > Only if the camera is near a PC. > If you need a 150' cable run, it gets awfully expensive, > Or you're not -really- running USB on the long path. > > Robert Which is what USB was NOT meant for, I don't know why you STILL don't get that. For most users a webcam WILL be near to PC, for the others there is ethernet, which is more expensive, or other means (such as WiFi which is even more expensive). ---------------------------------- Herbert's PIC Stuff: http://repatch.dyndns.org:8383/pic_stuff/ -- http://www.piclist.com#nomail Going offline? Don't AutoReply us! email listservEraseME.....mitvma.mit.edu with SET PICList DIGEST in the body

Herbert Graf wrote: > > > > > > Looks to me like a market opportunity, particularly for cameras. > > > > > > If I can get a 100BT card for under $10, putting the chip IN the {Quote hidden}> > > > > > camera can't be THAT expensive. > > > > > > > > > > Keep studying the problem; you'll see why USB is better. > > > > > > > > Standalone webcams with built-in Linux based web-server > > > > connected to intranet. How USB is supposed to be better than > > > > ethernet to this very much real case? > > > > > > > > Mike. > > > > > > Very simple: price. > > > > Only if the camera is near a PC. > > If you need a 150' cable run, it gets awfully expensive, > > Or you're not -really- running USB on the long path. > > > > Robert > Which is what USB was NOT meant for, I don't know why you > STILL don't get that. Because he is talking about "market opportunity". Look at the top of this post. Is he allowed to talk a bit about that? > For most users a webcam WILL be near to PC, for the others > there is ethernet, which is more expensive, or other means (such > as WiFi which is even more expensive). Who are you to forecast future? There is an essential difference between potential webcam usage and usage of keyboard, mouse, monitor and similar I/O devices. These devices are needed to operate a computer, But webcam is/will be used to gather video data under or not computer control. Perhaps it's against the law in Canada to use webcams at some distance from a computer? Linux and Linux-based web servers are open source - Boa (http://www.boa.org/) and thttpd for example. SOCs are getting cheaper and cheaper. It's a thankless business to forecast what "most users" will prefer in the future. Best Wishes, Mike. -- http://www.piclist.com#nomail Going offline? Don't AutoReply us! email EraseMElistservmitvma.mit.edu with SET PICList DIGEST in the body

> > > Only if the camera is near a PC. > > > If you need a 150' cable run, it gets awfully expensive, > > > Or you're not -really- running USB on the long path. > > > > > > Robert > > > Which is what USB was NOT meant for, I don't know why you > > STILL don't get that. > > Because he is talking about "market opportunity". Look at the top > of this post. Is he allowed to talk a bit about that? What does that have to do with complaining about a technology that doesn't work because it's being used out of spec? > > For most users a webcam WILL be near to PC, for the others > > there is ethernet, which is more expensive, or other means (such > > as WiFi which is even more expensive). > > Who are you to forecast future? Not the future. Look at today: what percentage of users of webcams want them more then say 10 meters from their PC? >There is an essential difference > between potential webcam usage and usage of keyboard, mouse, > monitor and similar I/O devices. These devices are needed to > operate a computer, But webcam is/will be used to gather video > data under or not computer control. Yes, which is why there are other options. What point are you trying to make? > Linux and Linux-based web servers are open source - > Boa (http://www.boa.org/) and thttpd for example. SOCs are > getting cheaper and cheaper. It's a thankless business to > forecast what "most users" will prefer in the future. > > Best Wishes, > Mike. What point are you making? Or are you back to your old habits? ---------------------------------- Herbert's PIC Stuff: http://repatch.dyndns.org:8383/pic_stuff/ -- http://www.piclist.com#nomail Going offline? Don't AutoReply us! email RemoveMElistservEraseMEEraseMEmitvma.mit.edu with SET PICList DIGEST in the body

On Saturday, Jan 10, 2004, at 06:55 US/Pacific, Mike Singer wrote: > How much is "software stack to support something like video > over ip over ethernet is pretty expensive"? > > From http://www.aegi.com/store/netipcam.html > Intellinet Network IP Camera $365.00 > That makes it out to be about the same price as a typical USB camera, plus a low-end PC clone for the SW. Yeah, that's about right... And I didn't see anything in the protocol list about RSVP, IPRTP, or any of those 'fancy' protocols. Running ONE ethernet camera across a 100Mbps ethernet without much other traffic is comparitivly easy. Try sharing your 10M ethernet with a bunch of other peripherals and the video camera (which is analogous to what USB tries to do out of the box) and you may start to miss the extra capabilities. BillW -- http://www.piclist.com#nomail Going offline? Don't AutoReply us! email RemoveMElistservspam_OUTKILLspammitvma.mit.edu with SET PICList DIGEST in the body

On Saturday, Jan 10, 2004, at 08:47 US/Pacific, Herbert Graf wrote: > Which is what USB was NOT meant for, I don't know why you STILL > don't get that. For most users a webcam WILL be near to PC, A forum like PICList is likely to rate the merits of a comm port based somewhat on how easily the defined "protocol" can be "bent" to other purposes. Use the printer port for input? Cool. Put a disk drive on it? Excellent! And still have the printer work? Uh... The bidirectionally-communicating printer? ummm... And the SERIAL port, well. Hardware flow control. 115200bps. Deeper and deeper FIFOs, but don't get incompatible with that venerable 8250 - too much legacy SW! Generating Vpp from the rs232 voltages, and using the modem signals for IO. Back to cool! I don't know who said that the legacy ports "just worked" - that certainly doesn't match MY memories of struggling with (especially) serial port cables and adaptors. Things have settled down a bit now, thanks to ... extinctions of entire species, essentially, but I think the most common serial comm problem is STILL a wrong cable somewhere. And I don't want to talk about lovely external interfaces like HPIB and SCSI, where the cables often out-weighed the peripherals, and had their own problems with chaining and termination... Now, USB was designed (as I see it) to: 1) address the requirements of the sorts of devices people were mis-using serial and parallel ports to connect. 2) cut down on the scope and magnitude of the user errors that could occur in connecting devices to the computer. 3) permit a greater number of devices to be connected without interfering with each other. They did pretty good, IMO. USB2.0, OTOH, is a rather embarassing afterthought. A sort of "oops, here are the things we forgot about (or specifically excluded), cause after all we don't want firewire to own those pieces..." There's no doubt in my mind that USB can be abused. Longer cables are clearly possible, especially if you're willing to give up things like hot-swap out at the end of that cable. (yeah. termination. looks to me like USB is designed to keep working even if one or more ends of the max-specified cable suddenly become un-terminated. What a good idea!) It would probably be good for the computing world in general if intentional abuses of the USB spec do not become rampant the way they did with serial and parallel. It's bad enough that "evil" extension cables are widely available. No selling USB cameras with 50 foot of cable on them, please... BillW -- http://www.piclist.com#nomail Going offline? Don't AutoReply us! email RemoveMElistservTakeThisOuTspammitvma.mit.edu with SET PICList DIGEST in the body