Wire connections -- butt, IDC taps, solder, etc

[miamicanes]

Last night I attempted to wire up the two keypads for my alarm. The following wire segments were involved:

• m1g: keypad bus on m1g
• start: cat5e cable ~15 feet long running between the M1G and seg3
• seg3: the in-wall phone wiring between my computer room (where the m1g temporarily sits) and master bedroom.
• bed: cat5e cable ~50 feet long, eventually will be chopped down to ~15 feet, but long for now so I can have the keypad next to the M1G itself while debugging) running between the point where seg3 meets seg2 and keypad #1
• kp1: the "default" keypad for the M1G (with temp sensor and orange/green backlight)
• seg2: the in-wall phone wiring between the master bedroom and living room.
• liv: cat5e cable ~25 feet long running between the point where seg2 meets seg1 and keypad #2
• kp2: the surface-mount keypad for the M1G w/blue backlight
• seg1: the in-wall phone wiring between the living room and kitchen/laundry room

For last night's attempt, I used 3-lead IDC taps for everything (the things that you stick 3 unstripped wires into and crimp the circle to connect all 3 together).

Here are the wiring details:

m1g:red to start:orange pair
m1g:black to start:brown pair
m1g:green to start:green pair
m1g:white to start:blue pair

start:orange to seg3:orange
start:brown to seg3:orange-white
start:green to seg3:green
start:blue to seg3:green-white
(seg3 blue pair unused for now)

seg3:orange, seg2:orange, and bed:orange
seg3:orange-white, seg2:orange-white, and bed:orange-white
seg3:green and bed:green
bed:green-white and seg2:green
seg3:green-white and bed:blue
bed:blue-white and seg2:green-white
(seg2 blue pair unused for now)
(bed:brown pair unused for now)

bed:orange to kp1:red
bed:orange-white to kp1:black
bed:green and bed:green-white to kp1:green
bed:blue and bed:blue-white to kp1:white

seg2:orange, seg1:orange, and liv:orange
seg2:orange-white, seg1:orange-white, and liv:orange-white
seg2:green to liv:green
liv:green-white to seg1:green
seg2:green-white to liv:blue
liv:blue-white to seg1:green-white

liv:orange to kp2:red
liv:orange-white to kp2:black
liv:green and liv:green-white to kp2:green
liv:blue and liv:blue-white to kp2:white
(kp2 is terminated)

The wires for seg1 are dangling off the end. Eventually I plan to move the M1G from the "seg3" end to the "seg1" end, swap the terminators, and insert a M1XEP inline between liv and kp2 (leaving liv's green-white and blue-white intact, and inserting the M1XEP inline within the orange, orange-white, green, and blue wires of liv).



That said, it didn't quite work.

KP1 seems to work just fine. Terminated, unterminated, with or without kp2 present, nothing seems to bother it.

KP2 is another matter. It's simply not being seen at all. Not "unreliably" or "intermittently". Literally, never.

As a troubleshooting step, I cut the connections between liv:green-white & seg1:green and liv:blue-white & seg1:green-white (the A and B data lines) on the theory that having 50 feet of unused cable dangling beyond the end of kp2's data bus was screwing things up, with or without termination. It didn't make a difference. I tried it with and without termination on kp2 (kp1 was never terminated at this point). I finally severed the connections between kp2:green and liv:green-white & kp2:white and liv:blue-white to eliminate any trace of data lines extending beyond the terminated kp2. No effect.


Interestingly, last night was my second attempt. My first attempt actually ran everything in a literal line, including red and black, and used separate cat5e cables for the runs to and from each keypad; red used the cat5e orange pair, black used the cat5e brown pair. Green ran on cat5:green, and white ran on cat5:green-white. It can be summarized as:

m1g to seg1
seg1 to liv1
liv1 to kp2 and liv2
liv2 to seg2
seg2 to bed
bed to kp1. The wires physically ended at kp1, which was terminated.

What's surprising is that with THAT topology, kp2 didn't work... but kp1 did, which shows that if nothing else, the problem does NOT lie with seg1, seg2, or seg3 themselves since KP1 has always worked regardless of which direction I fed it from (last night its signal came via seg3; the first night, it came via seg1 + seg2). From what I remember, kp2 was detected by enrollment, and the problems I had were due to it having the same address as kp1 (at that time, I didn't know how to give it a different address).

So... for tonight... what should I try first? Replacing the 3-wire taps with some other kind of connector? Or recreating the first night's topology (running everything in a straight line, without trying to be clever and tap the red & black lines to reduce their total run lengths) using the same 3-wire taps?

For a straight-line run (with two whole cat5e cables running to each keypad... one to it, one from it), would I get better results with:

phonelinerange (keypad red) to cat5range PAIR
phonelinerange-white (keypad black) to cat5:brown PAIR
phoneline:green (keypad green) to cat5:green
phoneline:green-white (keypad white) to cat5:green-white
(brown pair available for some other signal)

or

phonelinerange (keypad red) to cat5range PAIR
phonelinerange-white (keypad black) to cat5:green-white and cat5:blue-white
phoneline:green (keypad green) to cat5:green
phoneline:green-white (keypad white) to cat5:blue
(brown pair available for some other signal)

Basically, in the first case giving +13.6v and ground their own pairs, and in the second case wrapping ground around the two signal lines on the total guess that it might be the "right" way to do it for shielding purposes.

 

[BraveSirRobbin]

I'm not even going to pretend I followed all of that , but let me just say you need to use a meter and ring out the cable(s) end to end with all those connections. Check for leakage and continuity. If you need help refer to my security install How-To.

 

[miamicanes]

I'm not even going to pretend I followed all of that , but let me just say you need to use a meter and ring out the cable(s) end to end with all those connections. Check for leakage and continuity. If you need help refer to my security install How-To.

Well, getting to the root of one of the questions... if I'm going to use 2 wires for +12v, 2 wires for ground, and one wire for each data line ("A" and "B"), am I better off putting both data lines on one pair, and both ground leads on another pair... or putting each data line on a different pair, and making 'ground' the second wire in each pair?

 

[dBeau]

I'm not even going to pretend I followed all of that , but let me just say you need to use a meter and ring out the cable(s) end to end with all those connections. Check for leakage and continuity. If you need help refer to my security install How-To.

Well, getting to the root of one of the questions... if I'm going to use 2 wires for +12v, 2 wires for ground, and one wire for each data line ("A" and "B"), am I better off putting both data lines on one pair, and both ground leads on another pair... or putting each data line on a different pair, and making 'ground' the second wire in each pair?

Listen to BSR. With all those connections, you really need to make sure that each path is clean end to end. Given a bundle of wires that dont work, I'll leave both ends unconnected then use my meter to verify that each wire is not connected to any of the other wires in the bundle. If all of that checks out, next I'd pick one of the wires to use as a "return" and at the far end, connect it in turn to each other wire. At the near end, I'd verify that the wires are shorted.

If both of these test pass, it's unlikely you have any problems with your wires... unless, one of your connections in the middle is flakey. The IDC taps are pretty hard to mess up. A visual inspection after the crimping will usually indicate success as you can see where the wires are grabbed by the little "fingers". Also, a gentle tug on each wire will tell you if you made a solid connection. Solder is a bit trickier but if you are good at, it makes a solid connection. Assumng you get the solder connection right, the problems there usually come from poorly insulating exposed wire. The jell filled bean connectors work well too but I dont like them for the hair thin wires in CAT5.

RS485 is very robust physical layer. It can go quite a distance using nothing but metal coat hangers so long as you are careful not to short them together Seriously though, untwisted telephone wire or even thermostat wire will work for short distances (like inside a house). But since you are using CAT5 you might as well use it as it was designed. Put A and B on a single pair. The data is passed as the difference between A and B. If both A and B take the same path (ie the same twists in the same cable) they will both be subjected to the same interference. If the interference were to induce a change, it will affect both A and B equally. So, the difference between A and B would be unaffected. As for +12 and ground, it doesnt really matter from an electrical perspective. I find that using a pair for each makes it easier to remember what I've done. Also, I'd suggest deciding on a single color code and sticking with it.

 

[forgets]

whoa, that's a heck of a lot of splices you've got there.
the advice i can offer is instead of going with your own wiring scheme, just use elk's throughout.

red (+12v) goes to cat5 brown / white striped wire (solid brown on some cables)
black (-) goes to cat 5 white / brown
green ( data A ) goes to orange / white and green / white cables
white ( data B ) goes to white / orange and white / green cables

the blue / white and white / blue pair are spares and it states in the manual they can be used to double up for power, but i've never found a need to (even with 3 keypads on one run)

an easy to read diagram is here http://www.elkproducts.com/pdf/M1DBH_Data_Bus_Hub.pdf
i'm not 100% sure i read your wiring right, but it seems you use both orange / white and white / orange for the data A line, while elk's instructions are slightly different.

couple of other things to remember, only 2 resistors total (and a lot of people forget there's one pre-installed on the m1 itself, so if you terminate 2 keypads without taking that one off, you have 3) .. and also power down the system before making the connections per the manual.

and if it keeps giving you problems just buy the data bus hub and run a seperate cat5 for each keypad, wired per elk's diagram, and then you can be sure it'd work, plus elk loves it when you buy accessories.

 

[dBeau]

whoa, that's a heck of a lot of splices you've got there.
the advice i can offer is instead of going with your own wiring scheme, just use elk's throughout.

red (+12v) goes to cat5 brown / white striped wire (solid brown on some cables)
black (-) goes to cat 5 white / brown
green ( data A ) goes to orange / white and green / white cables
white ( data B ) goes to white / orange and white / green cables

the blue / white and white / blue pair are spares and it states in the manual they can be used to double up for power, but i've never found a need to (even with 3 keypads on one run)

an easy to read diagram is here http://www.elkproducts.com/pdf/M1DBH_Data_Bus_Hub.pdf
i'm not 100% sure i read your wiring right, but it seems you use both orange / white and white / orange for the data A line, while elk's instructions are slightly different.

couple of other things to remember, only 2 resistors total (and a lot of people forget there's one pre-installed on the m1 itself, so if you terminate 2 keypads without taking that one off, you have 3) .. and also power down the system before making the connections per the manual.

and if it keeps giving you problems just buy the data bus hub and run a seperate cat5 for each keypad, wired per elk's diagram, and then you can be sure it'd work, plus elk loves it when you buy accessories.

All good advice.... especially the bit about using a "standard" color scheme and keeping track of the resistors. Just dont let the diagram fool you. Elk *is* suggesting that a single pair be used for the data.

The diagram shows two pairs connected because one, the orange pair, is intended to carry data out to the device from the hub while the green pair carries data back to the hub where it is then connected to the next jack in sequence on the hub. That jack then either goes out and back to the next device, or holds the terminating resistor. RS485 doesnt like a "star" topology but installers often do. So while the hub, in a physical sense, looks like it forms a star, electrically, it really just makes one long run (twice as long as the amount of CAT5 used) with each device hanging from the "middle" of each run of CAT5.

If, rather than use a hub, you wanted to daisy chain your devices and stick with the color code Elk suggests, the green at the device would be spliced with org/wht from the previous device along with the org/wht to the next device, while the wht/org from the previous, the wht/org to the next and the white from the device would all be spliced together. Follow the same pattern for +12 and (-). Notice that when daisy chaining we are ignoring the green pair. You could however, connect the green from the previous to the green from the next but it would take me another paragraph to explain why you might want to . It's also worth mentioning that for very long runs or to support many devices on a run, it wouldnt be a bad idea to use the spare (blue) pair to double up with +12 and (-) as this will ensure adequate current carrying capacity.

Because I cant help myself, I'll go on to say that you could still do without the hub and make home runs of CAT5 from each of your devices by using the green pair as suggested by the data bus hub instructions. You'll just be connecting the green pair coming back from device N to orange pair going out to device N+1. Back at the Elk, you'll have one large pile of connectors. So with that said, just get a hub. They're inexpensive and at least half of your terminations will be with RJ45s which I find to be pretty fool proof. Whenever I have to strip CAT5 I find myself wishing I could just terminate with an RJ45 and plug it into the device. Keeping the number of individual connections down helps.

In short, just think of the data bus as two long wires running in parallel with a resistor connecting them at each end. Each of the devices connects to those two wires somewhere between the two ends and with as short a lead as possible. Note also, that the Elk itself is just one of the devices.