Two meters of a shielded CAT5e (FTP) cable are super cheap and provide 4 twisted pairs, so power transfer through the same cable would also be simple. The shield should solve any RFI/lightning EMP issues as well.
My main concerns are 1: signal integrity over the 2m distance; 2: not to kill the main FPGA if something goes wrong, e.g. the cable gets shorted. The driver's resilience should be comparable to what the CAN chips provide. But this is just belt and braces, nothing should happen in the first place, as this is a "connect and forget" application for monitoring purposes.
You could argue that the differential pairs should be routed on adjacent wires with a ground wire between each pair - it changes the characteristic impedance, but I can't be bother to dig out exact figures.
Sounds like a sensible idea. The geometry is simple enough and a simulation package could probably give better information.
Bill Sloman wrote in news: snipped-for-privacy@googlegroups.com:
Been in the industry for 35 years, Billy. Despite all your bullshit repsonses where you declare I do not know about something.
Don't need a primer from a decades disconnected dipshit.
Then you have the opposite problem. Ribbon cable crimp terminations are WAY more expensive than RJ45. Dollars each as opposed to cents each.
You lost either way. Singular or ganged, crimp connectors are out there, but due to the entire IT realm's embrace of Ethernet, the RJ45 crimp terminal is very likely the least expensive crimp for 8 conductors one can find.
I have trouble expecting you to understand economics, however.
The most likely situation that the customer detects RFI problems is when using a phone, especially GSM. Test the RFI behavior before shipping anything to the customer.
When the customer calls you and complains about some problem, the customer is then often asked to do this and that with the equipment, also moving the cellular closer to the cable and causing even more problems, making it impossible to solve the original problem :-)
You may have been there but you don't seem to have kept your eyes open.
What made you think that ribbon cable needed - or got - individual crimp co nnections on each conductor in the ribbon? All the crimp connections are ma de at once - or at least they were where I worked. Military clowns have bee n known to make life difficult for themselves.
Why the emphasis on eight conductors? There are wider ribbons available, an d Piotr Wyderski wasn't all that specific about what he wanted to do.
You have understanding what other people might understand - in the same way as krw does. You are happy when their thinking aligns exactly with your ow n, but when there's stuff going on you don't understand you imagine that th ey have to be wrong. There are other options, but you don't seem to have ha d much practice at finding out what they might be.
Bill Sloman wrote in news: snipped-for-privacy@googlegroups.com:
You seem to enjoy or have fun making this retarded statement often, and have seen you make it in your retarded responses to others as well. It only serves to make you look more like the ass we already know you to be.
My experience with connectors, ribbon cables, and individual contacts is far greater than yours. That is obvious considering your claim here.
There are several ways to terminate a ribbon cable. This person, however, is looking for cheap. First off, a ribbon cable with twisted pairs incorporated is not cheap. And the connectors and socket headers are also not cheap.
Cat6 cable and RJ45 terminations and connector headers ARE CHEAP.
Again, dipshit, I do not need a primer from a fool, whi still seems to think I have no experience and makes retarded cracks about not having my eyes open. You are a goddamned idiot, Bill Sloman.
Wow. Billy worked somewhere. And they used ribbon connectors that he walked past one day on the manufacturing floor. WOW!
What stupid shit are you mumbling about now, child?
PRICE, idiot. RJ45 IS 8 conductors, you stupid f*ck. So one or even two cables can be used to make the hop, and two would even STILL BE CHEAPER. There. Now it is 16 conductors and still cheaper than your solution.
No shit, and the wider the more expensive, idiot.
He does NOT want this link to have several dollars in cost attached to it, you dopey dipshit.
Here goes Billy being an abject idiot again. YOU HAVE NO CLUE WHAT I DO OR DO NOT KNOW, YOU RETARDED BASTARD.
Nice try, Billy. You have been OUT of the busines for DECADES and I have been IN the business during those decades you were OUT.
I can safely assert that you have little clue about what is in the channel today.
And if cost is not an objet then John's suggestion that he should use SFP is far better than LVDS.
But you have been gone so long, I have serious doubts that you even know the first thing about those links, and now you have to go look it up and get all familiarized with it.
I used sfp, sfp+ and xfp and qsfp, by the thousands. I built satellite baseband gateways, you pathetic worm.
I knew what LVDS was likely before you, because I WAS still working when it came out, and you were not. Nice try though, punk.
You seem to think that I do what you do. No, Billy. You are the googletard. I spoke from experience and have yet to google anything, and especially anything you ever brought up.
Yet you so deserve to have that and maybe even a ton of lead dropped on your lame ass.
There is a gang of fast ADCs monitoring something in a cabinet. The ADCs have plain CMOS outputs. There are too many of them and they are not designed to drive long cables. There is no room to put the full-blown data processing unit next to the ADCs. So the idea was to buy the cheapest possible FPGA and make it collect and serialize the multitude of slower data streams into one 200Mbps composite channel.
2 meters of CAT5e FTP costs nothing and is widely available. Hence the 8 (9, in fact, including the shield) conductors.
Piotr Wyderski wrote in news:quaedk$1a8$ snipped-for-privacy@gioia.aioe.org:
I would not use foil shielding as a conductor. Nor the carrier strand they include with it. You want braided full shielding if you intend to use that as one of the conductors.
Piotr Wyderski wrote in news:quaet4$3f1$1 @gioia.aioe.org:
Absoultely. Db9 connector is a bit more, but would certainly work. Keep the twists in place, right up to the pins almost.
But then at that point, my suggestion of individual pairs of SPC, twisted up by you to make a custom like would work with a regular DB9. They can be individually shielded each pair too.
The HDMI cable one person suggested seems perfect for that if the coaxes in there are not simply too tiny to work with easily.
I did a pretty extensive study of CAT5/6 cables, for a similar non-Ethernet application. I recall doing 100 MBPS up to 10 feet or so, with separate clock and data pairs, between FPGAs. I did a little transmit-end equalization. I'll see if I can find the report.
--
John Larkin Highland Technology, Inc trk
jlarkin att highlandtechnology dott com
http://www.highlandtechnology.com
There are no coaxes in an HDMI cable. only singles, twisted pairs and shielded twisted triples.
If you want coax (in consumer computer video cables) you'll need to use a VGA or DVI cable and they'll be 75 ohm coax. XVGA bandwidth is about 50MHz so don't expect UHF coax.
It is also worth considering USB-C and Thunderbolt 3 (which uses the USB-C connector). These have very high spec twisted pairs plus a few other signal and power connections.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.