Track feeder wire gauge

[KenS]

Interesting idea. BERT testers are a bit expensive (the cheapest I found was around $800), but if you had access to one, it might be a good way to check wire out. But what you really need is one designed for telco use that can go down to 56K, which would probably be much better at testing layout-grade wiring than one designed for 10 Megabit Ethernet speed tests.

 

Still, the problems that are going to mess up a DCC waveform are not subtle. An ohmmeter will do a pretty good job of identifying loose unijoiners or bad solder joints.

[inobu]

The ohm meter gives resistance readings but I think we need to look at it from a line quality perspective. Just a thought or opinion.

 

To answer the OP question, I think the gauge can vary based on your load requirements. The number of feeders per zone or block and the distance to the feeder plays a role in your design requirements.

 

I went off on the CAT 5 tangent because the bus not only distributes power but it also carries data. CAT5 wire has twists int them which aids in the data carrying properties of a wire itself. A proper twist in the wire will give you a clean running power bus.

 

Inobu 

[KenS]

I don't want to belabor the point, but I think it's important to point out that DCC is rather different from typical networking technologies, despite both being serial digital data transmission systems.  DCC operates at a signaling rate of 17 Kbps or less (the bit widths are variable, so the rate is too), not hundreds or thousands of Kbps. Even plain-old telephones used 56 Kbps (the old DS0 standard) for a voice signal when converted to digital. Seventeen Kbps is slow, with long bit durations, and that means most noise sources can't do much to it, because they're too short-lived.  I used to spend a lot of time looking at coax and twisted-pair network wiring with various test gear (I did LAN support for a number of years), and spent a lot of time reading about DCC and wiring when working on my old HO layout, so this is a topic I've thought about quite a bit.

 

And I don't mean to throw stones.  I've been known to get a bit carried away with my wiring too (you should see the control panels I did for my old HO layout; a masterpiece of unnecessary complexity that actually did very little). If wire that could be used for networking is what you want, go for it. But realize that you're probably removing problem sources that would never affect anything you'd notice.

 

Note: if you plan to use Railcomm, so decoders transmit back to the command station, there is a bit more reason to be concerned about wire quality. Railcomm uses 250Kbps signaling, and the higher-frequency signal will be more sensitive to distortion from wiring flaws that would not affect ordinary DCC.

 

There is, apparently, some value to twisting the wires, to keep them together and reduce inductance, according to this article. And, as also noted there, twisted wires are less susceptiable to, and generate less, electrical noise, which is the usual reason for twisting communications lines. But they're recommending twists in turns-per-foot, not turns-per-inch, something you can do by hand or with an electric drill and a screw-eye clamped in the drill bit. Off-the-shelf twisted wire is a lot more rigorously constructed (in terms of constant number of turns-per-inch) because the high-frequency signal is much more susceptible to distortion by noise.

 

That said, in network wiring there are basically two classes of problem sources: impedance mismatches, which are caused by resistance changes of any sort, and wire that had been untwisted too far, reducing its ability to reject noise.  Since DCC wire will operate without twists, too little twisting isn't likely to cause problems unless you are in a very electrically noisy environment (like a factory).  Of course you can also have problems with cross-talk from adjacent wires if, you were bundling DCC and some other low-bit-rate digital signal, or an AC power line, together (a high-rate signal would just raise the noise level, and since the DCC protocols are pretty robust, it would take a lot of that kind of noise to do anything).

 

Impedance mismatches are basically "mirrors" that reflect part of the signal back on itself, generating noise at the same frequency that swamps the actual signal.  (I used to spend hours with an electrical time-domain-reflectometer hunting down bad wiring in coax Ethernet, and I've probably seen every possible way to create an impedance mismatch, including a filing cabinet that only caused problems for a couple of hours every mid-day, when heat expanded it against a wire wedged behind it). But that assumes the wire is acting like a waveguide (as Ethernet wire does).

 

The end of the wire is a big mismatch; old coaxial Ethernet wouldn't work at all without an appropriate resistor across the end of the cable to "terminate" it. I'd be very surprised if that were an issue with DCC, given the low bit rate.  And the lack of any formal requirement for terminating the DCC bus or track reinforces that opinion (I've seen informal recommendations for termination as a cure for signal distortion in long bus wire runs).  As an open end is a perfect mirror, the lack of any concern with these in the standards indicates that of themselves, impedance mismatches aren't a major problem.

 

That's not to say that the things that cause impedance mismatches can't cause problems.  Impedance mismatches mean a change in resistance, and extra resistance degrades the voltage of the signal, making it more susceptible to noise. That's also the problem with long runs, as resistance in the wire adds up, reducing the signal level.  If you know how many ohms-per-unit-length your wire should be, a multimeter is all you need to find severe problem sources (wire that's been nearly cut through, or broken around a sharp curve, bad solder joints, loose terminal screws, etc).

 

A BERT tester is going to look for exactly that kind of degradation in the signal, and thus it will find those problems too, and probably a lot faster than an ohmmeter would (it will also find problems that wouldn't have any significant effect on a DCC signal). Twisting your wire has some value, and a BERT tester would be a useful tool with wire that was sufficiently twisted for its rate (one reason a low-bit-rate tester would be preferable to one designed for something as sensitive as Ethernet).  A BERT tester can also be left running, while you tug on wires and do other things that might reveal a transient problem that a quick check with an ohmmeter would miss.

 

But DCC layouts work fine with untwisted wire unless you have very long bus runs (and then you're likely to have problems from signal attenuation anyway) or exceptionally noise conditions. And an ohmmeter is a good, basic, tool that anyone can afford. I think it's important not to underestimate its value, or overestimate how "clean" your wire needs to be.

[inobu]

Creating a reliable network is based on the ground work one lays. Yes, yes worked on everything from DS0 to OC192. Ciscos 7500 to the first CRS routers. So I know what you are saying.

 

Lets put it this way I use 25 pair cable color coded for all of my turn out switches and that was my temp DC layout. 

 

[smg id=610]

 

Inobu

[KenS]

That's a good way to keep them organized.

 

My temporary layout was just a mess of switch wires.

[quinntopia]

Alllllllllright.  You convinced me.  I can no longer ignore this whole 'wire bus' thing.  I suppose back when Quinntopia was a simple loop, most of the above was not relevant.  Now that I've been expanding and expanding, my main "main" line now runs a total of at least 60 feet!  I think that's pushing it for a rudimentary system of using Kato 3-way adapters in series around the layout (with some of the ends cut off and wire extensions added on).  I see that I now have a terrible wiring system for DCC as there are many points on the layout where I expect the DCC signal has signifiant differences in distance from the 'command station' (which, I surmise, is one  of the reasons for the main bus in the first place!)!

 

Yes, this is a little embarrassing.  The Captain knows what a Luddite I am when it comes to electrical stuff.  I confess my sin.

 

In reality, I've been noticing some weird behavior when I go to try test some loco's on the new expanded track....they don't respond very good at all to commands.  Could likely be dirty track, or ongoing challenges with my Minitrix Mobile Station, but I at least want to try and rule out poor wiring.

 

So today I went out and got some 14awg stranded wire, will be using some of those automotive crimp connectors and then wire those to female Kato/Tamiya connectors which will feed the track.  Anyone see any holes in this method?

 

On the other hand, it has taken me several years to figure out just what y'all were talking about with this whole 'bus' thing! 

[nik_n_dad]

We've used the clip-on / suitcase / IDC / insulation displacement connectors on our little "temporary" layout and I just got done wiring our new "bigger" layout last weekend using these guys.  Unless you order them in bulk, they get expensive ("bulk" means boxes of 50-100).  But, it makes wiring a layout a breeze, especially if you invest in the crimper for IDCs instead of using pliers.  No problems with them, and like I said, a fast way to wire reliably.  As always, YMMV. 

[quinntopia]

Well, I just finished converting my old 'ad hoc' wiring system to a main bus using 14awg wire and the suitcases.  I was surprised at how fast it went; Nik_n_dad is right about how easy they made it (I ordered a bunch in bulk from eBay about a year ago anticipating I might want to do something like this)!

 

I did a quick locomotive test, and it looks like my 'dropped' command issue seems to have been solved! 

[KenS]

So today I went out and got some 14awg stranded wire, will be using some of those automotive crimp connectors and then wire those to female Kato/Tamiya connectors which will feed the track.  Anyone see any holes in this method?

 

I use stranded 14 AWG as a "U"-shaped bus to distribute track power to each "scene" on my layout (there are effectively four of these), with the command station at about the middle of the "U", so each leg is under 20 feet. Suitcase connectors drop the bus on each table to a local terminal strip, and other terminal strips are connected to the track 24 ga feeders (various types of Unitrack feeders).  Right now, there are just wires (16 ga for ease of wiring) from the bus terminal strip to the feeder strips.  Eventually the circuit breakers/occupancy detectors will go between the strips.

 

But what you're suggesting makes sense.  I don't see any problems with it.

 

BTW, a good crimp tool sounds like an excellent idea.  I used a wide-jawed pair of electrician's pliers, but I probably destroyed about one connector in ten due to failed crimps.  If you're doing a lot of these, the right tool is probably a sound investment.

[nik_n_dad]

As an fyi, if you want the crimper, it's the Scotchlok crimping tool, #84039.  Micromark has it, but if I recall, I found it elsewhere cheaper.  i think. 

http://www.micromark.com/SCOTCHLOK-CRIMPING-TOOL,8952.html

 

It is a pricey tool, and may be hard to justify.  You'd need to break something like 200 of them to break even.  I just found that it was much faster going with the crimper and much, much easier on the hands.....

 

and yes, I have insomnia

[The_Ghan]

I have no idea what a "suitcase" connector is ... someone please show me a picture.  However, I use the computer molex connectors where I need up to 4 wires.  Examples include my Peco point motors and feeder wires.  Here is a pic of what I mean: http://www.elitebastards.com/pic.php?picid=hanners/ocz/silencer-610w/images/molex.jpg  They are available in a variety of m/f configurations.  They make good feeders for double-track.  I use the red for outer loop positive and the yellow for inner loop positive ... etc ... On my point motors I remove one of the blacks and use the remaining black as a common on a three wire seup.

 

I use motorcycle 9-way and 6-way connectors between my boards when I need to connect more than 4 wires.  I can colour-code for zones, points, etc.  Here's what I mean:  http://www.kojaycat.com/2.8mm%20connectors.htm ... They are available in 6 colours. 

 

Cheers

 

The_Ghan

[KenS]

The term "suitcase connector" gets used a bit generically at times.  What I mean by it, and what I've seen others mean by it, is the 3M Scotchlok 558/560 Self-Stripping Dropwire connector, which comes in two sizes of interest to model railroads, red (or "pink") for 16-22 ga wire, and blue for 14-18 ga:

 

558 Red:

http://www.jimsmodeltrains.com/-strse-676/25-3M-ScotchLok-ETC/Detail.bok

or

http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=920074-07-ND

 

560 Blue:

http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=920074-08-ND

 

There's also a yellow version for 12ga and 10ga wire.

 

I primarily use the blue ones, and found them at a local electronics shop, but they seem to be harder to find online (and may have been discontinued by 3M since they aren't listed on the newer data sheet I found, or perhaps Scotch just considers them as separate from the Scotchlok connectors, although everyone else calls them that). However, Digikey seems to have a good stock.

 

The red ones would work well with 16 ga bus wiring, as they can support a 22 ga feeder rated for 3 amps.  And a 16 ga bus is fine for a smaller N-scale layout using a 3A or 5A command/station booster.  You really only need to step up to 14 ga for long wire runs (loss in 16 ga is double, but it's only an extra quarter-volt in 10 feet of wire).

 

There are other connectors in the family, the Scotchlok Butt Connectors, that are more useful for lighter wire, although I haven't used these:

 

Scotchloc IDC Butt Connectors

 

Here's a link to the data sheet for the family:

 

http://www.digikey.com/Web%20Export/Supplier%20Content/3M_19/PDF/3M_Scotchlokconnectorsandtools.pdf

 

You can find datasheets listing the 558/560 connectors on the Digikey pages linked above, but they're older ones, from 2003.

 

And there are similar connectors from other manufacturers.  Here's a post from someone who's written a fair bit on Unitrack (and whose writing is generally worth reading) and who uses a different kind of connector but still calls it a "suitcase connector":

 

http://www.thewhistlepost.com/forums/n-scale/11475-posi-tap-connector-connection.html

 

The problem with all of these is that they have a limited range of wire gauges, so connecting track feeders of 20 to 24 ga directly to a 14 ga bus isn't possible (which is one reason I used 16 ga as a local distribution wire between the bus and my feeders).

[The_Ghan]

IC ... thanks Ken.

 

I think the motorcycle connectors are a better solution for what I want.

 

Cheers,

 

The_Ghan

[quinntopia]

These are similar to what I used:

http://cgi.ebay.com/100-3M-T-Taps-Male-Insulated-22-18-Ga-Wire-Connectors-/190469035430?pt=LH_DefaultDomain_0&hash=item2c58d6d5a6

[Guest Closed Account 1]

As promised here is Kato UniTrack 20-043 Double Feeder track with parallel wiring.  Pretty good for too much coffee :)

 

I ended up using desolder braid inside shrink tube and removed a little of the support to provide separators for the individual wires.  The blue wire is representing the Kato blue wire path. Clear silicone for insulation and insurance that nothing moves.  Tough getting in there to rough up the track before soldering.

 

Bonus is that the feeder plug can be inserted in either receptacle and one controller for a double loop (V11 set).

[gavino200]

Funny you bring this aspect to the table. I was going to post my thoughts but felt it might be considered way out there.

 

I'm going to use CAT 5 on my next build because I see DCC as data communication more so than power distribution. Treating the rail as a data link will aid you in achieving optimum operation. When we look at if from the power perspective we just look for continuity. When we look at it from the data perspective it is signal quality which can drive you to creates a better layout medium resulting in a solid power distribution. 

 

Inobu

I curious, did you ever do this? Do you still agree with this?

[gavino200]

I've spent a lot of time lately staring at DCC traces in my oscilloscope. 

 

I'm curious, what can you learn from that?

[gavino200]

Alllllllllright.  You convinced me.  I can no longer ignore this whole 'wire bus' thing.  I suppose back when Quinntopia was a simple loop, most of the above was not relevant.  Now that I've been expanding and expanding, my main "main" line now runs a total of at least 60 feet!  I think that's pushing it for a rudimentary system of using Kato 3-way adapters in series around the layout (with some of the ends cut off and wire extensions added on).  I see that I now have a terrible wiring system for DCC as there are many points on the layout where I expect the DCC signal has signifiant differences in distance from the 'command station' (which, I surmise, is one  of the reasons for the main bus in the first place!)!

 

Yes, this is a little embarrassing.  The Captain knows what a Luddite I am when it comes to electrical stuff.  I confess my sin.

 

In reality, I've been noticing some weird behavior when I go to try test some loco's on the new expanded track....they don't respond very good at all to commands.  Could likely be dirty track, or ongoing challenges with my Minitrix Mobile Station, but I at least want to try and rule out poor wiring.

 

So today I went out and got some 14awg stranded wire, will be using some of those automotive crimp connectors and then wire those to female Kato/Tamiya connectors which will feed the track.  Anyone see any holes in this method?

 

On the other hand, it has taken me several years to figure out just what y'all were talking about with this whole 'bus' thing! 

 

I'm in a similar situation now as you were way back when you wrote this. 

 

I have a singer command center with no boosters, and a wiring 'system' consisting only of Kato 3-to-1 and 1-to-1 extensions. This is from a combination of my electrical ignorance and a desire to keep the connections kid-friendly. I set this up when we had a 4x8 layout. We now have around a hundred feet of track - just a guess. So I'm trying to work out what if anything I need to update.

 

The funny thing is that I don't have any of the system communication noise issues that you had. My layout is L-shaped, and by chance my command center is roughly in the midpoint. My 'tree' of wires (by pure chance) seems to go out in all directions for approx equal distances (give or take). I wonder if that's why I don't have noise issues. 

 

My question to you is - would you have gone from a Kato 3-to-1 tree pattern to a bus, if you didn't have signal noise issues? What are some other reasons I might want to consider a bus over my "Kato web method"?

 

Thanks.

 

I really like your blog btw

Edited January 22, 2017 by gavino200

[gavino200]

Well, I just finished converting my old 'ad hoc' wiring system to a main bus using 14awg wire and the suitcases.  I was surprised at how fast it went; Nik_n_dad is right about how easy they made it (I ordered a bunch in bulk from eBay about a year ago anticipating I might want to do something like this)!

 

I did a quick locomotive test, and it looks like my 'dropped' command issue seems to have been solved! 

 

Did you use just one bus? Or one bus per block?

[gavino200]

I use stranded 14 AWG as a "U"-shaped bus to distribute track power to each "scene" on my layout (there are effectively four of these), with the command station at about the middle of the "U", so each leg is under 20 feet. Suitcase connectors drop the bus on each table to a local terminal strip, and other terminal strips are connected to the track 24 ga feeders (various types of Unitrack feeders).  Right now, there are just wires (16 ga for ease of wiring) from the bus terminal strip to the feeder strips.  Eventually the circuit breakers/occupancy detectors will go between the strips.

 

So U-shaped 14G bus

Command center at center of U

Various 24G wires from the bus to the track

 

A single block system?

 

Later this will be a multi block system? You'll add insulated joiners and breakers between the bus and the track (on the 24G sections)?

 

Am I getting this right?

[gavino200]

As an fyi, if you want the crimper, it's the Scotchlok crimping tool, #84039.  Micromark has it, but if I recall, I found it elsewhere cheaper.  i think. 

http://www.micromark.com/SCOTCHLOK-CRIMPING-TOOL,8952.html

 

It is a pricey tool, and may be hard to justify.  You'd need to break something like 200 of them to break even.  I just found that it was much faster going with the crimper and much, much easier on the hands.....

 

and yes, I have insomnia

 

Looks like a vice-grip would work just as well. No?

[cteno4]

Gavin,

 

Captain isn't around much anymore. He was designing some custom decoders for production by a local train seller.

 

Jeff

[gavino200]

Gavin,

 

Captain isn't around much anymore. He was designing some custom decoders for production by a local train seller.

 

Jeff

 

Pity, seems like a very knowledgeable guy.

[cteno4]

Yep but they moved west and he got a new job and life sort of changed directions a bit, I'm sure he will be back when he has the time. He was one of the main creators of the forum!

 

Btw ken s and quintopia are also not on the forum much for quite a while.

 

Folks come and go with the forum as life, needs, interests, time, etc change.

 

Jeff

[gavino200]

Folks come and go with the forum as life, needs, interests, time, etc change.

 

Jeff

 

Fortunately, they've left a wealth of knowledge and ideas here in their posts, to guide a new generation. :)