On Mon, 2008-01-21 at 23:51 +0800, Ed Greshko wrote: > Craig White wrote: > > On Mon, 2008-01-21 at 16:16 +0800, Ed Greshko wrote: > > > >> Craig White wrote: > >> > >>> On Mon, 2008-01-21 at 16:02 +1030, Tim wrote: > >>> > >>> > >>>> Do you realise how stupid it is to even ask that? How many hundreds of > >>>> brands may be available for purchase? How many different brands in > >>>> different places that aren't available at the other places? How many > >>>> that are fine, but aren't mentioned by someone, so people won't buy > >>>> them? How many of the same product are sold by different labels? How > >>>> many of different products are sold under the same label? > >>>> > >>>> > >>> ---- > >>> I was hoping that you could name just one commercially available USB > >>> cable that was not usable for USB 2.0 > >>> > >>> Craig > >>> > >>> > >> Somehow I can't imagine how unimportant that question is.... But, if > >> you want me to pick you up a few of them from the cheap computer swap > >> shops here in Taipei I'll get a few and send them to you. While I'm at > >> it, I'll even get you some cheap CDs or DVDs that produce coasters 5-10% > >> of the time. > >> > > ---- > > I've made my share of coasters but that is an entirely separate topic. > > > No, it isn't. > > There are substandard products everywhere one looks. And there are > products that were built to satisfy a given standard but don't pass > muster when the standard is updated. You just seem bent on trying to > have Tim name a given product that will work on USB 1.0 but won't work > on USB 2.0. Frankly, it is boring. > > I don't doubt that there are cables out there that work perfectly well > with USB 1.0 and fail with USB 2.0. Just like I don't doubt that there > are 10BaseT cables that work just fine at 10Mbps and not so good at > 100Mbps....because I have experienced that end of the spectrum. But, I > don't have to see them and touch them to know they exist. > > They exist. And, frankly, having been on the hardware end of > manufacturing super computers, I could tell you some wonderful stories > about coaxial cables used in the Control Data Cyber 205. But, you > probably won't believe me either. > > Oh, that that "supervisor" that wasn't concerned or wouldn't even > consider a cable problem is just plan...well how can I put > it...ahhh....wrong. > > But, tell you what....you can go on believing what you will. > > But, it is truly unimportant why you are trying to "prove". So, > whatever..... > I'm no engineer, but I do have over 40 years experience with RF, including precision RF measurement (I was a calibration technician in the US Navy.) I can assure you that Tim and Ed are both correct. Cables do indeed affect signals, especially when you go above 50MHz, and with digital signals such as USB, the total bandwidth necessary for good signal fidelity is 10x the base rate. In otherwords for you cable to work correctly for USB 2.0 its bandwidth would have to be nearly 5Ghz. If it were for example only 4Ghz, you would have occasional dropouts of single bit signals. Thus 110011 or 11100111 would transfer correctly, while 10111 or 000100 would not. This leads to intermittant operation. It is further compounded if the standard has signal encoding used to reduce the total bandwidth consumption, because then the occasions for single bit issues would be far rarer, but not eliminated. In otherwords if the signal is known to be ascii text, the text could be encoded such that vowel consonent pairs would produce lower frequencies, effectively increasing the cables reliability without affecting bandwidth. There are a number of papers detailing the issues involved, and if you are really interested, I am sure all of us could point to some LaPlace documentation or Fourier documentation, or even Karl could point you to some papers and manuals for Amateur Radio, which would be lighter on math, and heavier on text. Regards, Les H
