I probably meant controlled impedance, but controlling the impedance means controlling the inductance ad capacitance in general. A quick example demonstrating a change in impedance can be seen in the common rules for wiring Cat5/6 cable; if you were to tape a section directly to a metal pipe, the impedance would be affected for the length of the contact due to capacitive and inductive coupling. You actually get a (partial) reflection at the start of the tape-job and again at the end and you may have a length affecting the signal in general for the length of impedance mismatch. . In theory if you were to wrap the UTP really tight (spec'd in number of twists per inch or other length" ) an external ground proximity wouldn't affect the differential mode signals as much, however the common -mode coupling to ground or "free air" goes up. In real life, assuming that the cable is the same (and not fanned out or packed or rolled or in proximity to metal down its length,) the impedance changes mostly from the cable to the connector, and then again from the connector to the PCB (on both ends). Using a sensitive Time Domain Reflectometer (TDR) you can see the disturbances along the length If you look at http://www.altera.com/literature/an/an315.pdf ; on page 5 they show the change in inductance going through an SMA connector and page 7 shows a signal start at 50 ohms and hit a peak of 56 ohms at one point in the length. At the frequencies of the old CBM stuff, changing impedance by 10-20 ohms probably isn't that critical most of the time. Bil -----Original Message----- From: firstname.lastname@example.org [mailto:email@example.com] On Behalf Of Nate Lawson Sent: Wednesday, December 18, 2013 1:43 PM To: firstname.lastname@example.org Subject: Re: Ribbon Cable I thought the main reason for multiple ground wires is current return. If you have a balanced flow (spatially), the wires are more resistant to cross-talk. Taken to an extreme, you get differential signaling. Can you explain what you mean by "controlled inductance throughout the length"? Thanks, Nate On Dec 15, 2013, at 3:36 PM, Bil Herd wrote: > I missed the beginning of the conversation but saw a question about > ribbon cable propagation. Depending on the quality of the ribbon you > can get a couple of hundred megahertz (what we call a Teflon ribbon > sometimes seen on ISCSI3) and even a cheap ribbon cable can be somewhat tamed. > > The two things to control is impedance and crosstalk and they have a > common remedy which is t include a lot of ground wires in with the > signal wires. The ideal situation is every other wire is a ground > wire which drastically reduces the ability of signals to capacitivly > or inductively couple. Every other wire being being a ground also > gives a controlled inductance throughput the length and then the > designer's job is to make the transition back to the PCB not have huge > changes is impedance that will cause a reflection. A ribbon cable in > this mode will usually have an impedance of 100-120 ohms so you need > to have a receiver that absorbs the signal by matching the > impedance... if the signal gets to the end and finds a 10K load after > traveling down a 120 Ohm pipe the signal will reject and ring. > > If anyone has a question about a configuration and concerned about > propagation give me a shout, there are also a lot of (java based) > tools on the net that help calculate impedance for things like PCB > traces and cables. > > Bil > > -----Original Message----- > From: email@example.com > [mailto:firstname.lastname@example.org] On Behalf Of Jim Brain > Sent: Sunday, December 15, 2013 11:24 AM > To: email@example.com > Subject: Re: Thanks for the Verilog help > > On 12/15/2013 6:32 AM, firstname.lastname@example.org wrote: >> On 2013-12-14, at 21:49, Jim Brain <email@example.com> wrote: >> >> Really? I didn't think us to be in such frequency range here so that >> it > would matter that much. > CMD cautioned many times about SuperCPU usage on the SX. Mine worked, > but that was to across the board. I'm sure some of it is no doubt > hurt by the fact that the expansion port drives through 3 circuit > boards, 1 ribbon cable, 3 connectors, and adds 8 inches of travel path > to the unbuffered expansion port signals. >>> and EasyFlash and a lot of the newer cart options (Chameleon, 1541U, > etc.) won't work when doing things that require tight timing. If you > could turn the EF3 KERNAL replacement function off, I bet the EF > portion would work, and 1541U no doubt works as long as the function > you are requesting does not require tight timing. >> I don't have the Chameleon (tried to put my hands on one for some >> time) > but 1541U-II works well in the very same machine. I mean the KERNAL > replacement function, which is the most timing critical AFAIU. > Without looking at the differences, I cannot explain that. >> >>> It's the cable, pretty much. I bet if you pulled the case, and >>> ordered > a shielded ribbon cable (or better, created a non ribbon cable > option), the problem would disappear. >> Pardon the ignorance but what's so wrong with the ribbon at the > frequencies we deal here with? Or - generally - with the ribbon. How > much it differs from e. g. traces running parallel across the big C64 > PCB? What with the (even longer) ribbon used for the USER PORT in SX-64? > Well, I think it's probably no worse than an 8" cart expander with > parallel lines, and most such carts won't work if I attach them to 2 > X-Panders in series and use the last slot on the second XPander. > That's not a completely fair test per se, but I think the cross talk > of those lines would wreak havoc no matter what tech was used. > > Jim > > Message was sent through the cbm-hackers mailing list > > Message was sent through the cbm-hackers mailing list Message was sent through the cbm-hackers mailing list Message was sent through the cbm-hackers mailing listReceived on 2013-12-19 23:00:04
Archive generated by hypermail 2.2.0.