Have been wanting to write about this, and now is the time in the middle of the winter. I started building my layout back in 2003 and went with DCC from the start. I wanted to try powering the whole layout with a single set of wires from the command station. It works well - I have never looked back, and have not been disappointed in the setup. Not saying this would work on a large layout.
Here are my caveats:
My HO scale layout is 14’ x 21’. There are 2 mainlines that go around the room, with a large ‘peninsula’ that the inside mainline uses. There is a single crossover between the 2 mainlines. The inside mainline goes thru a small yard with 5 tracks, the yard is about 9’ from end to end. Both mainlines go thru a staging yard in an adjacent room. There are 7 storage tracks in the room, and they range anywhere from 13’ to 20’ long. There are several turnouts for various industries. There are separate wires to power the turnouts.
Was very meticulous in laying down the trackwork. Made sure there were no gaps in the connections, most of which are with rail joiners. On some curves, I have soldered the flex track pieces together.
DCC system is Digitrax DCS200 8 amp. From the station, I have run a single set of feeder wires up to both of the main lines - that’s it. A friend of mine brought over his multimeter, and we found close to full voltage in numerous spots on the layout.
With the exception of a geep that does switching duties, I always run trains on the mainline with 2 powered locos. In the very rare event that a head light will flicker from a piece of dirt on the track or at a critical axle joint, the other loco will push/pull it along.
I don’t have sound locos. If I do ever get them, will probably get decoders with 'ke
Mark, several people here and on other forums insist that they do the same as you, with the same results, and that this has gone on for quite some time, years. What I haven’t thought to note is which of them were in DCC and how many said they were still running DC. But some claimed to be strictly DCC.
If it works, great. I think what most of us recommend, the multiples of feeder pairs, is due to the changes that happen over time. Joiners just don’t have a great record in the hobby of enabling robust voltage throughput over time.
The other thing is, assuming the joiners and rails over great distances, clean metals like those should not suffer huge degradations over distance. N/S rail isn’t bad stuff for transmission, maybe not the best metal, but it’s not bad. And, if the rails and the joints between them are well supported on strong and near-planar roadbed, and the joiners don’t have to deal with flexing joints, and don’t get badly contaminated with whatever glues and bits surround them, I don’t see why a single pair of feeders won’t work. The proof is in the quarter test at any one spot, especially the most distant ones.
I had a double track 8x12 which I initially powered with just 2 wires, until I hooked up the bus.
BUT - and since you are using an 8 amp booster, this is VERY important., because 8 amps is enough to melt things in spectacular fashion - 8 amp systems for HO frankly are silly, most layouts don’t need that much power and if you do, it’s better to distribute the power with multiple 5 amp boosters.
Measuring the voltage with a meter doesn’t prove anything - you need to apply a load and measure the voltage. Even better - at the point furthest from where your 2 wires attach, set a quarter across the rails. If the circuit breaker on the booster fails to trip - you are living life on the edge, as a derailed loco that shorts will also likely not cause the breaker to trip. 7 amps of draw will not trip an 8 amp booster, but 7 amps at 15 volts is 105 watts. Consider how hot a 100 watt light bulb gets - yes, this is why doing this is a bad idea.
8 amps for HO and smaller scales should never be fed directly to the rails. You should feed the power into circuit breakers and at a minimum divide the layout into individual sections with their own breakers set to something more reasonable like 3 - 3.5 amps for HO.
Technically, bus wires are redundant to the conduction provided by the rails. There’s no difference between two bus wires and the two rails. Nickel silver is slightly less conductive than pure copper but even Code 83 is pretty heavy gauge wire.
My layout was built as DC cab control and I inherited (bought dirt-cheap from a friend) a Digitrax Super Chief back in 2005.
For the first month or so I simply wired the Digitrax output to the former DC cab buss, all bundled together. I didn’t skimp on wire size since I could bring home scraps from work, THHN 10, 12, 14 and some MTW 18-20.
When I got serious about the inevitable “DCC conversion” i.e. removing the block control switches, dividing the layout into eight power districts, installing the early version of the Power Shield (2-4 unit breakers) and two reverse units, I did run eight pair of #12 THHN wire.
I never had any problems with power distribution but what was perplexing me was that after a brief short the breaker would not reset. This got bothersome after a while. Now I had dozens of lighted passenger cars and maybe an average of six to ten sound engines (HO) in any one power district at a time.
On a marathon crusade I doubled the number of feeders to each “block” over the course of two-weeks or so. This eliminated the reset problem and my layout has been running “electrical-glitch” free ever since.
I have lots of turnouts on the main. These (Shinohara) can be a problem area for electrical continuity. 95% of my joiners are soldered but the addition of feeders is a good “belt-and-suspenders” policy.
As mentioned by others, rail joints are not very robust, and vibration may break what appears to be a well soldered joint. If your track is easily accessible and scenery is yet to come, then going with only 2 feeder wires can be fine. Any problem can be easily fixed. But elecrical problems with track that is ballasted, scenicked and burried in tunnels are a pain to fix. I have verified Murphy’s law several times at our club layout … Multiple feeders are added suspenders to ensure long-term reliability. I also like to have a bus to facilitate the connection to sidings, and to sections track that are only held loosely to make room for expansion due to heat.
Simon
PS: Some forum members have mentioned that nickel silver is known to be a poor conductor of electricity.
I started my 10’ x 14’ layout in the late 80s as DC only mainly because DCC wasn’t in my crystal ball back then. I’m an old school guy and wired my layout for block control, a single pair of solid #19 bell wire homerun to each block. Longest run is about 15’, #19 bell wire load loss is .1 volts at 2 amps. My max load was under 1.5 amps (two locomotives).
Everything went super well for 17 years when I cutover to DCC. I followed the DCC guru way of rewiring my layout to buss wiring only to find out my signal system occupance detection no longer worked, strike one. I rebuilt my signal system for DCC detection operation, too much screwing around to get it to work as good as my original DC signaling system, strike two. I have a loop on my layout and purchased an automatic reverser and it was more trouble than it was worth, strike 3.
I removed the DCC buss wiring and reinstalled my original block wiring and everything works great. I have a relay that switches the track power from my DCC controller to my MRC DC Sound and Power power pack when the Sound and Power is turned on to prevent problems between DC and DCC mode, a fail safe interlock. I operate my layout dual mode, DC or DCC.
I ended up replacing my occupance detection with optical detectors that I now prefer over any current detection method. No resistors on the axles are needed for the detection.
My DCC operation works perfect off my DC block wiring!! NO special wiring for DCC.
I have replaced the motors in all of my locomotives with low current can motors and my max load per block is well under 1 amp so about .05 volt loss, perfect as far as I’m concerned.
If you don’t have any problems with the way your layout works more power to you!!!
The reference contains the procedure used to take these measurements.
So while a code 100 rail (with soldered or tight joints) is a little over 4 times poorer as a conductor than a 16-gauge feeder, it is imcomprehensible to me that someone would term it a ‘poor conductor of electricity’ as that term has any particular relevance in wiring layouts …
This table has actual measured values, not calculated, using a very precise measuring system. As it shows a 1m length would be 76 milliohms (~2mΩ/in), equivilent to #26 wire.
Agree with you 100%. All good points. In my mind multiple feeders are more important to ensure that the conductivity is good and that circuit breakers will function properly. In other words a safety issue. In addition I make sure that I have feeders after the frog rails on all my turnouts as I found that resistance increases after my Peco code 83 insulfrog turnouts. By the way I do not feed every rail on every length of track but do feed at frequent intervals.
I’ve never checked the current capacity of my HO track so today I did a test. I had 4 sections of new Atlas code 100 track and soldered short pieces of #12 awg solid wire as an electrical joiner to end up with a total of 12 foot of track.
I used my bench power supply set to 10.0 volts (at the rails) and a 1156 bulb as a load. I measured 9.94 volts at the lamp at 1.66 amps using my Fluke 179 meter. Somehow I don’t think .06 volt loss at 1.5 amps in 12 foot of track is something to worry about.
I don’t have 24 foot of #12 to compare it to but I’m happy with the track findings.
I have high current test leads (#12 superflex) for my Fluke so very little loss in the test leads.
Yes, but that is as he said the equivalent to #26stranded wire (composed as he further notes of 7 wrapped strands of #32). Does anyone use that for feeders?
Why he did the test at 60Hz instead of something representing practical DCC modulation frequency, I can’t really say. Perhaps it is relevant to non-digital-control Marklin modelers (their older power was 50/60Hz AC IIRC).
In any case his results are reasonably similar at ~1.93milliohm per inch instead of the DC 1.45 … I suppose this is like the Collaborative Ocular Melanoma Study in pointing out that any of the reported values, DC resistance or impedance at DCC modulation, are good enough conductance not to be critical. To put this in perspective consider the voltage drop produced at .00193 ohms per inch vs. .00145 ohms per inch over any of the rail distances from feeder to feeder represented by one of the mentioned runs of track.
I am disappointed he chose to round the number off for so long a reference length as a meter of rail, especially at a nominal criterion of ~“50ppm” for the accuracy and his feeling the need to specify that he used ME code 100; I think he could have measured the effective impedance of a loading resistor in-circuit to be able to use it for shorter test pieces (say 10cm which is comparable to the 5" in the test I cited) to get an accurate result with the methodology he adopted. But that precision is unnecessary for our purposes of predicting voltage-drop effects in practical operation.
For the record: some of the idea of ‘negative impedance’ is described here.
Simple reason for not testing at DCC frequencies - very few even high end meters read correctly at those frequencies, even using a sine wave signal. Related to Marklin - if it’s not Marklin command control, Marklin trains are 3 rail AC, like an HO version of Lionel. Just with a much less obtrusive center ‘rail’ - more like little bumps in the center of each tie. The locos have long sliders, not short shoes, so they touch multiple center bumps.
We all can (and probably have done so at some point in the past) defy good practice and report success. Does not make following good practice obsolete.
I installed bus wires and don’t regret it. It was easily and painlessly done at the early stage of my building my layout. If I have ever any problem at any location on my layout, I can quickly install a feeder to solve the issue. I also installed a second bus for lighting, which can be easily tapped into when needed.
However, the issues that would concern me more are the issues that Randy mentions in his initial response.
The cross sectional area of Code 83 rail is so clearly far greater than that of 26 gauge wire the irresistible inference is that the information in the article linked to is just wrong. More accurately, useless whether it is wrong or right.
Besides, the article itself states that accurately measuring low resistance is very difficult. I’m not sure what the reference to the margin of error as being 50 ppm could mean since the common meaning of that acronym is parts per million.
The inference the author appears to invite us to draw is that the rail is less effective than bus wires in transmitting the DCC signal. No proof of this inference is provided nor is a method of proving it suggested. DCC operates at very low signal strength and very high frequency relative to the standard AC frequency 'tested" for the article. The article begs the question: would 26 gauge bus wire be sufficient to eliminate any material degradation in the DCC signal?
End result? This link provides exactly zero useful information about whether bus wires work or not.
The fact is that whether the bus wires work or not they are simply redundant to the rails. Bus wires and rails are parallel conductors, on all layouts and at all times. Of course adding conductor cross sectional area is going to reduce resistance whether by increasing rail code or adding wires in parallel.
The OP suggests this redundancy is not necessary. Since his layout works that is irrefutable. Mind you, 8 amps of available current would probably overcome most minor resistance problems!
This is, I think, important enough to highlight: one use of a ‘feeder architecture’ is to preclude Too Much Power being delivered through a single set of wires, even if the “losses” doing so are acceptable. Sectionalizing the layout can be desirable; using smaller or faster or more responsive breakers on the sections can also be desirable.
It seemed to me that the impedance calculation in the example involved an approach not dependent on meter accuracy; it is a balancing operation then subject only to the accuracy with which the negative matching impedance is determined (and that, while interesting to radio nerds, might not be too critical for most model railroaders especially given the tiny practical distinction between the observed results for DC and 60Hz AC)
To imply that DCC suffers from small effective signal strength is one of those laughable things perhaps that might be expected from anyone deficient in electrical knowledge. One of the major basic points of the DCC modulation is to ensure relatively strong effective signal integrity – and say what you want about the concept, or the relative ‘existence’ of square-wave modulation, it works at least as well as any practical digital alternative at providing strong effective SNR at any meaningful railborne traction power.
Yes, I think he should at least have used an “AC frequency” closer to typical DCC modulation if ‘impedance’ were of much practical importance, and he should have adopted a square-wave modulation together with this to see what effect that had on the magnitude or characteristics of the observed impedance, either averaged or at peaks of interest. But as the results would likely be just as relatively insignificant as the practical difference, over practical rail lengths, of DC resistance vs. any of the measures o
