Post by thread:[EE]: Ideas on a Wideish area Data network

In an industrial environment, and/or a home environment, has anyone had a fair amount of experience with Dallas 1-wire and/or a similar setup, over say 200 foot ranges using slow communications (1200 baud to 9600 baud)? I'd like to use RG58 coax for power plus comms for one design, unsure if it'll actually be practical for power plus data though; Thinking 9 to 12V DC power for this one, just thought I'd throw the idea out and see if anyone had used something like this. Lots of "hash" near the machinery, the cable can be in conduit for longer runs (not that THAT will save you!), it would just be really really convenient to have one cable (Ethernet-style) for data plus power - Alternately could use (cheap!) RJ11/RJ12 style cables, though they have resistance and noise problems, or RJ45 Cat 5 cables, a little costly but do-able and still with the problem of a noiceable resistance at each connection. (With those, go for say 14VAC power and opto-isolated RS422; proven, works, pricey at the PC end.) Just would like to go Coax if I can <G> Absolute requirement to be able to unplug each node, so just hard-wiring isn't an option; Could use a molex style connector, trying to use something quicker and easier to "plumb", if practical... Thanks! Mark -- http://www.piclist.com hint: The list server can filter out subtopics (like ads or off topics) for you. See http://www.piclist.com/#topics

----- Original Message ----- From: "Mark Willis" <spam_OUTmwillisTakeThisOuTFOXINTERNET.NET> To: <.....PICLISTKILLspam@spam@MITVMA.MIT.EDU> Sent: Saturday, October 21, 2000 3:01 AM Subject: [EE]: Ideas on a Wideish area Data network {Quote hidden}> In an industrial environment, and/or a home environment, has anyone had > a fair amount of experience with Dallas 1-wire and/or a similar setup, > over say 200 foot ranges using slow communications (1200 baud to 9600 > baud)? > > I'd like to use RG58 coax for power plus comms for one design, unsure if > it'll actually be practical for power plus data though; Thinking 9 to > 12V DC power for this one, just thought I'd throw the idea out and see > if anyone had used something like this. Lots of "hash" near the > machinery, the cable can be in conduit for longer runs (not that THAT > will save you!), it would just be really really convenient to have one > cable (Ethernet-style) for data plus power - Alternately could use > (cheap!) RJ11/RJ12 style cables, though they have resistance and noise > problems, or RJ45 Cat 5 cables, a little costly but do-able and still > with the problem of a noiceable resistance at each connection. (With > those, go for say 14VAC power and opto-isolated RS422; proven, works, > pricey at the PC end.) Just would like to go Coax if I can <G> I've found coax to be quite unhelpful in noisy environments. Being unbalanced it just soaks up those lovely impulses over long runs unless you ground it at frequent intervals. You should certainly consider balancing the 1-wire with RS485/422 drivers or similar. Unless it's balanced you'll not get any benefit from UTP. One solution I have used with some success in the past is to use a Belden foil-screened-twin audio cable (e.g. #8723) and studio type XLR connectors (Useful tip, don't use the chassis connection on the plug, connect the screen to one of the pins and apply equipment ground to mating pin on the socket.) This is much cheaper than UTP and gives good results for moderate speed data (I've used it up to 300k). It'd probably start to lose effectiveness at higher rates though. > Absolute requirement to be able to unplug each node, so just hard-wiring > isn't an option; Could use a molex style connector, trying to use > something quicker and easier to "plumb", if practical... The Belden cable is thin and easy to run in and it strips easily. If you want *really* quick and easy to install then there are solid core versions suitable for use with Neutrik IDC type XLR connectors. How much power do you need? Dallas have a suggested circuit for deriving parasitic power in their 1-wire data sheets - or you could use a phantom power scheme using centre-tapped transformers at each end, feeding power between centre-tap and screen. Hope this helps. Andy. . -- http://www.piclist.com hint: The PICList is archived three different ways. See http://www.piclist.com/#archives for details.

Another possibility is to use current-loop drivers. ----- Original Message ----- From: "Mark Willis" <mwillisKILLspamFOXINTERNET.NET> To: <.....PICLISTKILLspam.....MITVMA.MIT.EDU> Sent: Saturday, October 21, 2000 3:01 AM Subject: [EE]: Ideas on a Wideish area Data network {Quote hidden}> In an industrial environment, and/or a home environment, has anyone had > a fair amount of experience with Dallas 1-wire and/or a similar setup, > over say 200 foot ranges using slow communications (1200 baud to 9600 > baud)? > > I'd like to use RG58 coax for power plus comms for one design, unsure if > it'll actually be practical for power plus data though; Thinking 9 to > 12V DC power for this one, just thought I'd throw the idea out and see > if anyone had used something like this. Lots of "hash" near the > machinery, the cable can be in conduit for longer runs (not that THAT > will save you!), it would just be really really convenient to have one > cable (Ethernet-style) for data plus power - Alternately could use > (cheap!) RJ11/RJ12 style cables, though they have resistance and noise > problems, or RJ45 Cat 5 cables, a little costly but do-able and still > with the problem of a noiceable resistance at each connection. (With > those, go for say 14VAC power and opto-isolated RS422; proven, works, > pricey at the PC end.) Just would like to go Coax if I can <G> > > Absolute requirement to be able to unplug each node, so just hard-wiring > isn't an option; Could use a molex style connector, trying to use > something quicker and easier to "plumb", if practical... > > Thanks! > > Mark > > -- > http://www.piclist.com hint: The list server can filter out subtopics > (like ads or off topics) for you. See http://www.piclist.com/#topics > > > > > -- http://www.piclist.com hint: The PICList is archived three different ways. See http://www.piclist.com/#archives for details.

Andy: That's why I asked <G> Know in Apartment cable TV setups that even better-shielded coax seem quite good at picking up Cell Phone transmissions, was wondering if there was a trick I didn't know of. RG-58 Seems to work fine for Ethernet here, OTOH - Dontcha just hate it when experiences are contradictory?! I don't claim to be good in RF... I'll look at that Audio cable, 300k's more than enough for these jobs. Those XLR connectors aren't "free" though so POTS wire may be it due to cost for at least the home setup. Power: About 12-15 nodes in the home system, I'm guessing 40+ for the industrial setup and may be low (Could be 100?); Probably 'F84's at 4MHz on Resonators at a guess, guessing between 5 and 20mA (@5V) per node average power, sensors may "feature bloom" and drive that up - Some home nodes will be higher power use with the LCD display and all. (All guesswork so far, who KNOWS what customers'll end up demanding?! <SadG>) Current Loop, A good thought; Not enough power probably, though <G> Peter: Thanks, more I think of it more I think I'll do that, RS422, POTS unshielded is a good start and easy to convert to STP if needed. Mark -- http://www.piclist.com hint: PICList Posts must start with ONE topic: "[PIC]:","[SX]:","[AVR]:" =uP ONLY! "[EE]:","[OT]:" =Other "[BUY]:","[AD]:" =Ads

Alan B. Pearce wrote: > >guessing between 5 and 20mA (@5V) per > >node average power, sensors may "feature bloom" > >and drive that up > > Surely the way to deal with this is to insist that the "new features" go into an additional sensor on the network? That way they become another (chargeable?) additional item on the network which you can develop later, and get the original spec ones done "on time". You did design the network interface to be feature expandable? If they ask for TOO much at any given node, it'll require AC power to that node, is the plan <G> Better proper function and a PO'ed electrician than network failure from their not thinking. Design on this monster's starting from the electrical wiring, actually - Until they start payin' I'm doing it mostly in my head, minimal costs etc. It'll help once they present an NDA and I sign it, then I'll have lots more facts to work from... The home project is going to be a practice run for the other, network interface debugged and tested on that. THAT one is all really low current except for a few stations with LCD displays - those will have AC power probably <G> Did re-plan it all somewhat recently, got a neat gizmo at a local store and had a couple better ideas. I usually use ASCII based protocols (easier to read / debug, log to a file, etc.), have a lot of it pretty clear really from the last such one I did <G> RS-422 is NICE for Polling, probably will use that at 9600 or 14.4k baud I think, plenty of speed for their needs. All features will be "local per-node" - when I poll Node 123 for "ID yourself" it'll respond that it's type J, I can then ask it what packets it has and it'll supply a Type J version 0 packet initially {with version spec and checksum!} - Always Always Always leave expansion room as otherwise the client will break your poor brain! <G> Mark -- http://www.piclist.com hint: The list server can filter out subtopics (like ads or off topics) for you. See http://www.piclist.com/#topics

I was thinking along somewhat similar lines - formalizing an rs232-based ring network so that you can put in a good number of nodes (including things like PCs.) OTOH, I'm wondering if it's worth it, what with people here having done equally complex (?) things using I^2C, which is more "microperipheral friendly" but less "largish computer friendly" (which is easilly fixed via an rs232/I2C "gateway") Using I2C would get rid of a good bit of "protocol development effort", I think. Thoughts? BillW -- http://www.piclist.com hint: The list server can filter out subtopics (like ads or off topics) for you. See http://www.piclist.com/#topics

If we go for 3-wire I2C, how to power the PIC nodes though? Could use 6-conductor POTS wiring, use 3 wires for ground and 1 for Vin then the last 2 wires for SCL and SDA, might do nicely. Don't know if I'd want to try I2C as a Ring though? (I'd think it'd ring really really well without termination?) Mark William Chops Westfield wrote: {Quote hidden}> I was thinking along somewhat similar lines - formalizing an rs232-based > ring network so that you can put in a good number of nodes (including things > like PCs.) OTOH, I'm wondering if it's worth it, what with people here > having done equally complex (?) things using I^2C, which is more > "microperipheral friendly" but less "largish computer friendly" (which is > easilly fixed via an rs232/I2C "gateway") Using I2C would get rid of a > good bit of "protocol development effort", I think. > > Thoughts? > > BillW -- http://www.piclist.com hint: The list server can filter out subtopics (like ads or off topics) for you. See http://www.piclist.com/#topics

I missed the beginning of this thread, so this may be way off, but why not go with RS-485? Isn't that pretty much a standard for longer-that-rs232-distance? Talking about number of wires, have you looked into any of the data over power wire type technologies? I am researching using that (over the 110 mains) for communications and hitting lots of dead ends. A company called Metricom has a good product, but only sells the right to make the chip, not the chip itself. Oh well. Shawn -- http://www.piclist.com hint: The list server can filter out subtopics (like ads or off topics) for you. See http://www.piclist.com/#topics

RS-422 is easier, I've found, to deal with than RS-485 (Basically the same beast but no confusion on who's master thus talking) - Been a while and perhaps I'm mis-remembering / whatever. With a PC controlling a bunch of PICs, I use Poll / reply not multi-master usually <G> Mark Shawn Yates wrote: {Quote hidden}> I missed the beginning of this thread, so this may be way off, but why not > go with RS-485? Isn't that pretty much a standard for > longer-that-rs232-distance? > > Talking about number of wires, have you looked into any of the data over > power wire type technologies? I am researching using that (over the 110 > mains) for communications and hitting lots of dead ends. A company called > Metricom has a good product, but only sells the right to make the chip, not > the chip itself. > > Oh well. > > Shawn -- http://www.piclist.com hint: The list server can filter out subtopics (like ads or off topics) for you. See http://www.piclist.com/#topics

I've been down this road too. I first tried I2C. It worked better than expected, but I found that unless I used slow data rates I had problems. The system was being used in a noisy environment, and this caused me some problems too. I eventually switched to RS485 and was able to work at 19200 over a couple of hundred meters no problem. I used the snap protocol which is a free protocol available at http://www.hth.com/plm-24/ Joe ----- Original Message ----- From: Douglas Wood <EraseMEdouglas_woodspam_OUTTakeThisOuTARIUSA.COM> To: <PICLISTspam_OUTMITVMA.MIT.EDU> Sent: Wednesday, October 25, 2000 3:08 PM Subject: Re: [EE]: Ideas on a Wideish area Data network > RS-232/422/485/??? are electrical specifications, not protocol > specifications. Any problems you've had with masters/slaves/multi-masters, > etc. were outside of the scope of the RS-###. RS-485 is good for noisey > environments and (if memory serves) has a distance of 4500 ft. > > Douglas Wood > Software Engineer > > {Original Message removed}

>I2C I2C sucks for anything larger than a box with a few boards and multi mastering with collision detection is a medium large nigtmare unless you use a token ring type of discipline. So you are limited to one master and many slaves, or you will spend a lot of ROM on collision avoidance and recovery. This will be OK until you run into glitches RFI or ground loop problems. According to my experience this ONLY happens at far removed installation sites and causes comments of the type 'Oh, we didn't think about THAT. THAT cannot be. Their installation should prevent it. Noone ever heard of such a thing. Oh sh..'. There are other buses and protocols out there that were meant for use in exactly your situation, like Profibus, Modbus, star-wired RS232C (optical current loop). Search the net. All of these have in common the fact that they outperform the electrical immunity (as in ground loop resistance) specs of I2C by a few orders of magnitude and that data exists so you can calculate what might work and what might not. Peter -- http://www.piclist.com hint: The PICList is archived three different ways. See http://www.piclist.com/#archives for details.