On Sunday 22 April 2018 13:30:18 Chris Albertson wrote: > Back to the title subject. I'm going to need to buy a Mesa card > soon. Is there any reason to prefer the 7i90 over the 7i92 or other > ethernet cards. It seems to me that even if I still owned a PC with > a parallel port the computer would also have an Ethernet port which is > 10X faster and uses cheaper cables. Possibly there is something > about the 7i90 I'm over looking?
Yes, the latency in the ethernet path. tcp is not exactly instant. And lag between the issuance if the command, and the message back from the hardware will be seen as a following error. The 5i25, is a pci card, talking to a couple parports, each of which I've heard can run at the equivalent of around 5 megabaud. The std parport's speed can be all over the place but is generally under a megabaud in effective bandwidth. Ethernet will be slower yet. Think in terms of the ping time reported, and thats an icmp transmission, faster than tcp. SPI can be much faster to the 7i90HD. On my rpi-3b the spi is a 32 bit packet sent at 41 megabaud, and the reply is sent back at 25 megabaud, usually a millsecond later. However that bus is only about an inch long in my install. Due to the puny gpio line drivers, I doubt those speeds could be done over a 5 foot cable. Maybe not even a foot. But IMO we need a bit more development to optimize the SPI buss. Its really intended to be an interchip bus for an inch or so, and figuratively speaking at those distances, proper termination of a transmission line is not quite as important. But all of this stuff really needs to be treated as a transmission line, because at the speeds they talk, improperly terminated lines can lead to data trashing echo's. The characteristic of the common ribbon cable where every other conductor is a ground, is about 122 ohms. They can be load terminated, which takes power and generates heat, or in the case of the spi, is often source terminated with a 120 ohm resistor (some of that can be in the source driver stage) at the source of the signal. This takes less power because it controls the dv/dt of the signal waveform on the cable. > > On Sun, Apr 22, 2018 at 10:19 AM, John Dammeyer > <[email protected]> > > wrote: > > Well said Chris. > > > > I worked on a system once where the 24V system bus shared a common > > ground with a 12V instrument bus. The system bus could have as much > > as 20A flowing > > through it as a Servo motor moved one section. There was also other > > equipment that sporadically drew more and less current. They used > > one common ground from the battery/charger systm to all the devices > > tapping onto > > this ground wire where needed. > > That would spoil many a nights sleep for me. > > On a small panel that approach had worked for them. When the > > batteries/chargers and 12V power supply along with one instrument > > were 30m away from the rest of the instrumentation they had all > > sorts of induced noise issues. They argued that they had measured > > the noise on the 24V system in the control box with a scope and it > > was minimal. > > > > I demonstrated that if they ran the scope ground 30m back to the > > battery and > > measured the ground at the control box they would see lots of noise. > > Splitting the 12V and 24V grounds made the problem go away. The 30 > > meter distance was still an issue since it wasn't physically > > possible to run a star ground for ever device back 30m to the > > battery. Add that some of the 24V devices had a non-isolated CAN > > bus and shared their CAN bus ground with the 24V system the > > potential for the 24V return path to take the 12V return path. > > > > But the major 24V ground separation from the 12V was a big step in > > the right > > direction. > > > > John > > > > > Gene is right. Many people don't understand grounds. The star > > > ground is not magic. It is based on two basic laws of physics. > > > Just two things > > > > to > > > > > remember > > > > > > 1) It is a fundamental law of electricity the current ONLY flows > > > if there is a loop. Take a straight piece of wire and connect > > > one end to a > > > > battery > > > > > and leave the other end free and absolutely zero current flows and > > > the wires has EXACTLY the same voltage all along it's length. > > > THIS is why you use a star ground. No current will flow in the > > > ground wire if ONLY > > > > one > > > > > end of each wire is connected to your ground reference. Make a > > > loop and you have current. > > > > > > 2) Mr. Ohm has a law that says if there is current you have > > > voltage. In fact the volts are proportional to the current. So > > > if you make a loop with your ground wires there will be current > > > and hence different parts of the loop will have different voltages > > > on them. It will not be zero everywhere as you might think. > > > > > > Basically the combination of Kichoff's and Ohm's laws. People > > > try can get around this by using real heavy and thick ground > > > wires. But they forget (or never knew) that impedance has an > > > imaginary component and the thick wire only effects the real > > > component. This noise that is transmitted is high frequency AC > > > not DC. you have to make the path "low impedance" if you want to > > > get around Ohm's law. This noise is basically in the radio > > > spectrum, big wires don't help much there. > > > > > > Just make sure ALL you grounds are "dead ends" There are places > > > where where loops sneak in. One of then is shielded cables where > > > the ground shield is connected to each end of the cable. > > > > > > There are more sophisticated systems where they shunt the AC to > > > ground using capacitive coupling but don't even try unless you > > > are a specialist engineer. You do see this in some equipment. > > > They use a capacitor in parallel with a high value resister and > > > they connect signal and > > > > protective > > > > > grounds this way. What they are doing is making a low impedance > > > path > > > > for > > > > > AC to ground. > > > > > > The other thing you can and should do is try to stop this > > > electrical > > > > noise > > > > > at the course. The best way to is again remember Kichoff. > > > Current > > > > flows > > > > > in a loop. So any time to have a wire with current, say going to > > > a > > > > motor, > > > > > some place there is another wire with that same current flowing in > > > the opposite direction. Always route those to wires as close as > > > possible Best way to do that is to twist them together. The > > > opposing directs > > > > will > > > > > cancel any electric field. At about 8 wire diameters, from a > > > twisted > > > > pair > > > > > the field will be nearly zero. So if you d this the noise will > > > only radiate for less then about an inch. Wire routing and > > > layout matters. > > > > > > > But this brings up a second recommendation, which is to > > > > establish a single bolt as a common ground point, with the > > > > commons of everything else connected ONLY to this bolt. This is > > > > known as a star ground > > > > system. > > > > > > Stuff connected to the nearest ground often results in having > > > > more than one ground and that constitutes a ground loop, which > > > > acts as an > > > > antenna, > > > > > > picking up noise from anything radiating it, and this noise can > > > > easily blow gates in the fpga on the 7i90. That bolt should also > > > > be the only place the 3rd rounded in the US pin in the power > > > > cord is connected, cut them off till you've only one left to get > > > > rid of ground loops, just don't cut the last one! > > > > ------------------------------------------------------------ > > ------------------ > > Check out the vibrant tech community on one of the world's most > > engaging tech sites, Slashdot.org! http://sdm.link/slashdot > > _______________________________________________ > > Emc-users mailing list > > [email protected] > > https://lists.sourceforge.net/lists/listinfo/emc-users -- Cheers, Gene Heskett -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author) Genes Web page <http://geneslinuxbox.net:6309/gene> ------------------------------------------------------------------------------ Check out the vibrant tech community on one of the world's most engaging tech sites, Slashdot.org! http://sdm.link/slashdot _______________________________________________ Emc-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/emc-users
