Do we really need heavy gauge wires for track feeders?

[peteski]

I figured that since this is a generic layout construction info I would post it here (not in the DCC/Electronics forum).

I have been seeing many layout builders using what I consider heavy gauge wire for the electric track feeders on their N scale layout.  While overbuilding things is not a crime, I think that heavier gauge wire is more difficult to solder and to deal with in general.  I did some research to see how much actual voltage drop will various gauge feeders have.

I decided to use 6" and 12" feeders as examples (since most feeders should be within that range).   6" feeders in the chart are actually calculated using 12" of wire and the 12" feeders chart is calculated using 24" of copper wire (since feeders consist of a pair of wires).

Average modern N scale DC or DCC engine (like Kato or Atlas)consumes around 0.25 A (250mA) at full throttle while slipping, and less with lower throttle settings.  Engines with DCC sound decoders use more current at full volume and throttle - around 0.5A.  I created the charts using 250mA, 1A and 2A currents to show what I think will be maximum realistic currents the feeder will have to pass.  The chart shows, for each gauge, the total wire resistance and the voltage drips for those 3 current values.

------  PAIR OF  6" FEEDER WIRES  (12" TOTAL)  -----
====================================

WIRE	RESISTANCE	CURRENT	CURRENT	CURRENT
GAUGE	(OHMS)	250mA	1A	2A
---------	---------	---------	---------	---------
20	0.0101	0.0025V	0.0101V	0.0202V
22	0.0162	0.0041V	0.0162V	0.0324V
24	0.0257	0.0064V	0.0257V	0.0514V
26	0.041	0.0103V	0.041V	0.082V
28	0.0653	0.0163V	0.0653V	0.1306V

-----  PAIR OF  12" FEEDER WIRES  (24" TOTAL)  ----
====================================

WIRE	RESISTANCE	CURRENT	CURRENT	CURRENT
GAUGE	(OHMS)	250mA	1A	2A
---------	---------	---------	---------	---------
20	0.0202	0.005V	0.0202V	0.0404V
22	0.0324	0.0082V	0.0324V	0.0648V
24	0.0514	0.0129V	0.0514V	0.103V
26	0.082	0.0205V	0.082V	0.164V
28	0.1306	0.0327V	0.1306V	0.2612V

As we can see, the voltage drop across the feeders is negligible, even with 2A of current passing through the 12" feeders and using 28 AWG wire.  I would say that 24 AWG or even 26 AWG wire should be plenty robust for feeders in N Scale.

[bdennis]

Peter,
Thanks for this.
I agree with your finsings.
I tend to use cut down Cat5 or Cat6 network cable separated into individual wires for my droppers.
The wire inside the Cat5 or Cat6 cables are 23 or 24AWG and are solid copper.
I had tended to just make the dropper 100mm to 150mm long and then solder on a much thicker wire (21AWG) than then goes back to my Digitrax BDL168 block detectors.
For the common rail I use a common bus wire and connect the 100mm to 150mm droppers to 21 AWG or 18AWG wire. 21AWG is rated at 4.5A and the 18AWG is 7.5A.
As my layout is separated into power districts and the BDL's and circuit breakers are fairly close to the blocks so the 21AWG is fine..
I use the 18AWG for the main track power bus from the DCS100 to the circuit breakers.

I too have seen some fairly heavy gauge wire being soldered to track in the past.

[kondor]

Peteski,
Cool post.  I'm sure a lot of modules, including mine, are over built.
It's funny to think of the  other tiny factors, such as strength of the joint.  This shouldn't matter theoretically but I have accidentally snagged a tool or my hand on my wiring in the past.  Or consider solder joint quality as a variable in the voltage drop equation. Or solid vs. stranded and different strand counts.  Pre-tinned or bare copper?

How does the DCC signal react to a 14 to 24 mV drop (or greater) ?

[C855B]

Agreed that small-conductor feeders are mostly a non-issue. Not only that, but unless you do the DKS method of not using joiners between track sections by sliding the adjoining rail under the spikes, provided you're doing the recommended one-feeder-per-flex, you have overkill in parallel feeders. IOW, in a block consisting of two or three hunks o' flex, no single feeder should be taking the entire load.

The only grimace I have with very small conductors like #26 or #28 is solid wire. I've fixed a fair number of others' feeder breaks, especially when they do the old-fashioned 90° bend to solder along the outside of the rail web.

[Ed Kapuscinski]

I think we need a new forum category for "N Scale Scientist".

Great post and info!

[mmagliaro]

Do we really need heavy gauge wires for track feeders?
No.   

Seriously, if you are running DCC, I can't see how any of the wire gauges shown in Peteki's chart could ever be a problem.
The decoder is going to keep the voltage to the motor pretty darn constant, regardless of little dips in the track voltage like 0.1 or 0.2 volt.

But... for running DC, I would stay away from the lightest gauges shown in that chart.   I know that I can see an engine visibly slow down with a drop of 0.2 volt because I experienced that on previous layouts (and added more feeders to fix the problem).  And on DC, if the track voltage dips, the voltage to the motor dips.

I am totally with you on not soldering big honking 14 or 16 gauge wire directly to the track.  It requires lots of heat
and it's much harder to conceal the wire.

I'd say 20 or 22 is a happy compromise.

[peteski]

Do we really need heavy gauge wires for track feeders?
No.   

Seriously, if you are running DCC, I can't see how any of the wire gauges shown in Peteki's chart could ever be a problem.
The decoder is going to keep the voltage to the motor pretty darn constant, regardless of little dips in the track voltage like 0.1 or 0.2 volt.

But... for running DC, I would stay away from the lightest gauges shown in that chart.   I know that I can see an engine visibly slow down with a drop of 0.2 volt because I experienced that on previous layouts (and added more feeders to fix the problem).  And on DC, if the track voltage dips, the voltage to the motor dips.

I am totally with you on not soldering big honking 14 or 16 gauge wire directly to the track.  It requires lots of heat
and it's much harder to conceal the wire.

I'd say 20 or 22 is a happy compromise.

So we seem to agree. 

But even with a 1A load and 12" feeder length, 26 AWG feeder wires only drop 0.082V (that is much less than 0.2V). And if you are running a single loco and not racing at full throttle with the model slipping on he track, then the voltage drop will be much less than 0.02V (that is one tenth of 0.2V you are worried about).

And of course if your model has a geared coreless motor which is probably consuming around 40mA (at full throttle) then 12" feeders using 26AWG wire would just drop 0.003V!  That should be undetectable. And that is the 26AWG wire (not 20 or 22).

Also, I think that the speed variations are the most noticeable when the engine is running at very slow speeds (and not so much when it is running at road speeds). At slow speeds, current consumption is low, so the voltage drop will aos be minimized.

[milw12]

Very helpful Peteski- I'll have to keep this in mind for the next one. I use 22ga currently, but I can see how thinner would be better.

Is there a chance to make this a Best-of thread? This is good data and I'd hate to see it lost.

[peteski]

Very helpful Peteski- I'll have to keep this in mind for the next one. I use 22ga currently, but I can see how thinner would be better.

Is there a chance to make this a Best-of thread? This is good data and I'd hate to see it lost.

Thanks Lucas.  If you use different length feeders, it is easy to calculate the resistance and voltage drop from my charts.  If your feeders are 3" then just divide the resistance and voltage drop values in the 6" table by 2.  If your feeders are 18" length then just multiply those values by 3.

[ednadolski]

I decided to use 6" and 12" feeders as examples (since most feeders should be within that range).   6" feeders in the chart are actually calculated using 12" of wire and the 12" feeders chart is calculated using 24" of copper wire (since feeders consist of a pair of wires).

Average modern N scale DC or DCC engine (like Kato or Atlas)consumes around 0.25 A (250mA) at full throttle while slipping, and less with lower throttle settings.  Engines with DCC sound decoders use more current at full volume and throttle - around 0.5A.  I created the charts using 250mA, 1A and 2A currents to show what I think will be maximum realistic currents the feeder will have to pass.  The chart shows, for each gauge, the total wire resistance and the voltage drips for those 3 current values.

What did you use to measure the current?   Since the DCC signal is time-variant, are these numbers representative of a DC equivalent, or RMS or such?  (I don't  know offhand what the conversion is based upon, but generally if you use a DC meter on a DCC signal it will tell you the wrong answer.  I know I should look it up myself but I am too lazy  )

What about feeders that will pass thru detector coils for DCC current sensing?  The DC wire resistance is probably negligible in that case (unless one starts getting into long runs of small gauge wires) but it would be interesting to know about any possible effect on the sensitivity of such circuits.

Thanks,
Ed

[jagged ben]

Peteski, I think I agree with your conclusions but I'd stress that in the larger picture the point is to have a feeder for every piece of rail, not so much the size of the feeders.  I'd probably be more interested in the math on bus wires. 

Average modern N scale DC or DCC engine (like Kato or Atlas)consumes around 0.25 A (250mA) at full throttle while slipping, and less with lower throttle settings.  Engines with DCC sound decoders use more current at full volume and throttle - around 0.5A.  I created the charts using 250mA, 1A and 2A currents to show what I think will be maximum realistic currents the feeder will have to pass.  The chart shows, for each gauge, the total wire resistance and the voltage drips for those 3 current values.

Well, I routinely run 4-6 unit consists at nearly slipping so I do look to the 1-2A range.  And moreover, I have thoughts of running SP drags with 10 units, including possibly 2-4 sound units.  So that'd be up to like 3.5A.    I think I'd stick with 22guage for that.  The last place you want voltage drop is on the steepest grade at the top of the mountain, farthest from the power source.   

Also, BTW, is the chart based on copper?

[C855B]

... The last place you want voltage drop is on the steepest grade at the top of the mountain, farthest from the power source.     ...

What? And miss the opportunity to operate well beyond stress limits like the real SP, breaking knuckles, pulling drawbars and smoking (traction) motors? Pshaw!

[peteski]

What did you use to measure the current?   Since the DCC signal is time-variant, are these numbers representative of a DC equivalent, or RMS or such?  (I don't  know offhand what the conversion is based upon, but generally if you use a DC meter on a DCC signal it will tell you the wrong answer.  I know I should look it up myself but I am too lazy  )

What about feeders that will pass thru detector coils for DCC current sensing?  The DC wire resistance is probably negligible in that case (unless one starts getting into long runs of small gauge wires) but it would be interesting to know about any possible effect on the sensitivity of such circuits.

Thanks,
Ed

I didn't measure any of the currents. It is from empirical data and my own past experiences.  The main point I'm trying to make is that with the currents consumed by average N scale trains even smaller gauge short feeder wires do not really drop any appreciable voltage. I really hate to see someone trying to solder 20AWG wire feeder to c40 rail.    You really  don't need it (especially if every piece of rail has a feeder on it).

Friend of mine has the DCC RampMeter (it measures DCC current). I'll have to borrow it someday and run some tests with some of my sound-equipped locos.  But I really don't expect the actual readings to be higher than my estimate.

Current-sensing detectors do not care what gauge wire is fed through them - they sense the strength of the electromagnetic field radiating from the wire, which is proportional to the amount of current going through the wire, and since the wire is in series with the load (the locomotive) the current through the wire will be the same as the current passing through the locomotive.

Jagged Ben: Yes, the values I used are for copper wire (I clearly mentioned that in the original post). Why would I use specs for other materials since in my experience the layout wiring is done using standard copper wire?  I simply Googled for AWG wire chart and used the info I found in one of those charts.

BTW, I envy your layout if you can not only run 10-unit drags, but also load it with a train heavy enough to run them at full throttle (12V) and almost slipping. If you can do this without mid-train helpers, and without couplers coming apart, then you have a superb quality layout and rolling stock.  Don't heavy trains like that usually travel at fairly low speeds (maybe 40MPH)?  If that was the case then the locos would not run their motors at maximum voltage, so the current would also be lower than you would expect it to be when the loco is slipping at 12V.

[nkalanaga]

Brendan:  I also use recycled computer cable for feeders.  Working in a data center for years, one to four foot lengths of multi-pair cable were common, and every pair has a different color combination.  Very handy!  And, yes, it works fine for my layout, where the throttles will only put out about an amp.

[OldEastRR]

My feeders are about right, but I've got 1/8" thick solid copper wire as my buss lines -- need heavy-duty pliers to bend it. But it was free, so ... (I think it came from 120v BX).