Does the DCS signal still need star wiring or can we do buss wiring?

Everyone has their personal opinion.  I'm a fan of looped 12-gauge stranded common ground and each of many blocks fed direct from a central control panel via 14-gauge stranded.  My 38x16 layout works fine.  Others say you've got to have star.  I ignore them and run my layout. 

If you want to run conventional also, toggled block wiring is essential.

Recommended? Yes.

Required? Not necessarily.

Different lengths of wire feeders can cause individual digital commands to arrive at the track at different times, resulting in digital echoes racing about on your track and possibly causing unexpected behavior of locomotives. Placing third-rail insulators between feeders alleviates this avenue of chaos. The star wiring method was intended to minimize signal duplication by having all track feeders (and thus control signal paths) as close to being the same length as possible.

I added a feeder to the far end of my 11x17' test-track loop, and sometimes if I hit the "two horn blasts" softkey, the locomotive will give two horn blasts--twice. Placing an insulating pin between the two connection points would solve that particular issue (but it hasn't become an operational problem so far)

---PCJ

I stick to star wiring as is recommended by MTH.  I tried a shortcut with some more remote switch tracks by jumping wires to their rails (Atlas switch tracks appreciate this).  This created a problem that bugged me, and I couldn't figure it out for quite a while.  I run DCS with TMCC.

When more modern Lionel electrocouplers went over these tracks they would sporadically open.  For a while I didn't realize that it were the switch tracks at all.  That's probably because since there are some clusters of switch tracks it was hard to discern the particular locations.

When I did I couldn't understand why.  One day I got the notion to look under the layout and see if I had failed to do star wiring.  Well, I did.  I rewired it in a star configuration, and I haven't had a problem since.

I'll skip the sermonizing.

Alan

[The star wiring method was intended to minimize signal duplication by having all track feeders (and thus control signal paths) as close to being the same length as possible.]

This myth still comes up every now and then though it was proven years ago  that the wires being equal or unequal have no effect on DCS. But having more than 1 track feed per  block can cause signal duplications.

Joe,

[The star wiring method was intended to minimize signal duplication by having all track feeders (and thus control signal paths) as close to being the same length as possible.]

This myth still comes up every now and then though it was proven years ago  that the wires being equal or unequal have no effect on DCS. But having more than 1 track feed per  block can cause signal duplications

Exactly.  

I've tested buss-style wiring and it seems to work well, but there's a BIG caveat. You must have a break in the track between feeders to prevent signal cancellation. Even with star wiring, two feeds within the same uninterrupted block will cause feedback problems. At the ends of the buss itself, you should consider installing "snubbers" ("Deats Filters")  to prevent bounce-back at the end of the buss. The ends of the buss wires should not be connected into a loop or there will be feedback through the buss.

I used to be a malevolent systems administrator and the digital signal feeding the engines in some ways is similar to the old "thin net" ethernet. You have targeted packet traffic picked up by various nodes on the buss. If the packet isn't picked up, it has to terminate or it creates feedback problems on the buss. The filters/snubbers are similar to the Ethernet buss terminators.

I stand corrected. 

But what exactly (i.e. technical reason) was behind the recommendation for star-pattern wiring then?

---PCJ

I agree with Joe and Barry that the equal length theory was balderdash and that there should be only one hot feed per block.

Exactly what constitutes star wiring is a matter of discussion.  I have heard it said that you only have one split if you use a terminal block, and at each split half the signal goes each way (I haven't seen that proven).  But if you look at a 12-post terminal block, you have 11 splits, each about 3/4" apart.  Using a control panel as I do, where the TIU output goes from toggle terminal to toggle terminal, I have the same number of splits but 2-5" apart.

As to common, even with running pairs from terminal blocks, there are innumerable return path combinations possible.  SO a common bus works fine.

Star wiring just radiates out from a central place (I use a 12 port terminal block) and each port that is used, a pair of wires runs to each isolated block. Even the pairs of wires don't have to be equal.

Maybe a better term than star could have been used to describe this form of wiring.

But what exactly (i.e. technical reason) was behind the recommendation for star-pattern wiring then?

Star wiring, with center rail-insulated blocks, is implemented primarily to ensure that a DCS engine (PS2/PS3) sees only one incidence of a command, rather than multiple incidences separated in (not very much) time.

Regardless, DCS is considerably more tolerant than one would suspect and, consequently, some different wiring schemes will yield acceptable, if not perfect, results. When adding DCS to an existing layout, the best advice is to just hook it up, see what you get and work from there.

This and a whole lot more is all in “The DCS Companion 3rd Edition", available for purchase  from many fine OGR advertisers and forum sponsors, or as an eBook or a printed book .

Gentlemen,

   I have always like Barry's direct engineering approach for retro fitting, 1st you just hook up the DCS, and see if it work before rewiring your layout, if everything works pretty well then IMO just add some Magic lights in isolated low signal areas, and your old layout is set to go.  If however the signal is poor all over, then consider rewiring.

PCRR/Dave

That's what I did when DCS was released.  Plopped the TIU between transformer and track, and ran.  Since the entire layout was in toggled blocks with one 14-gauge drop per block,  it was easy through turning off blocks to see which sidings were dragging down signals.  They got lights in their bumpers and I've been running ever since.  Wiring the bumpers was the extent of rewiring. 

Ok, some of this (but not all of it) is coming back to me, kind of like calculus, once you resume it after years hiatus you find yourself remembering various tricks, and some other aspects you're back to the books. 

Correct me on this: if I run a buss wire then I should install center-rail insulators?

If I do expand the existing wiring from central terminal strip (star wiring) in the layout expansion - around the room - and because have some decent long runs ~ 45 feet between a lift out bridge on one end and 45 degree crossings on the other (this is by its nature a block) - can I install just one feeder wire (from terminal strip) to that length of track or how many (old rule of thumb was feeders every 7-8 track connections)? And, from what I am reading thus far, center rail insulators are a good idea but may not be necessary?

Matt said: "At the ends of the buss itself, you should consider installing  to prevent bounce-back at the end of the buss. The ends of the buss wires should not be connected into a circle, but should be terminated using "snubbers" aka "Deats Filters".

What are these things, never heard of them?

BTW: am using Ross track.

Paul, the center rail insulators, aka fiber pins, are to assure that each block has only one feed.  They are most probably necessary.  I do have one block where, due to a poor connection in a UCS track, some 20 years ago I installed a second drop.  Haven't had a signal problem and never got around to repairing it.

I have no idea what Matt is speaking of, since I don't use a hot buss.  My common buss is a loop, startung at the U terminals, going around the layout, and ending at the U terminals.

Rev L TIU, scale mile of track, only using fixed one and fixed two outputs, one for the main line one for the yards and side tracks. With bus wiring all tens around the layout.  Just make sure you have a power feed to the tracks every 3-4 feet,

The term 'star wiring' seems to be a little confusing in what is suggests versus what is meant for trouble-free DCS operation.  The term suggests that each feeder should be run to a central location, but I don't think that is true.  Perhaps some other term should be used such as 'isolated feeders'.  From what I've read, a buss wire will work just fine as long as any one isolated block of track only has one feeder wire connected, AND the buss is not looped, ex it is connected to a transformer on one end, and nothing on the other end.  The goal being to make sure that the DCS signal can only reach an engine by one path at time.

Is this correct?  

JGL

I tried to follow Barry's book for wiring my current and more permanent layout. I used MTH terminal blocks and paired wires (OGR wire), but the lengths are not close to being the same and the terminal blocks are not centrally located as I had planned. The isolated blocks with a single feed per block is what made all the difference in DCS performance for me.

Before, I had a few different temporary layouts and no blocks (didn't think they were large enough to matter, wrong), while they worked, they all had problems from time to time. The new and properly blocked layout, now about 1-1/2 years old, has yet to have a single DCS problem, DCS has worked flawlessly. I have never really had a 'bus wired' layout, so I have no experience there. However, my outside rail is not isolated and has a connection to the outer rail where each power feed is connected to the center rail. As RJR mentions above, I believe I would have a 'common' bus with the outer track rail being the common wire around the layout. Many paths for the DCS signal to return to the TIU.

JGL, I'm inclined to agree with you.  I can see no reason whyt power should be sent out to the center of a layout and then distributed.  Certainly, it's not feasible for operating conventional locos if you want maximum flexibility.

There is no problem with connecting the common (outside rail) buss as a loop.  That is optimal for power distribution. 

Gentlemen,

    Having retro-fitted my 1st DCS Layout I learned that using the Star wiring does have its advantages for DCS, and Legacy & Conventional run well with it also.  However the myth that all the wires in a star wiring pattern, must be of equal length, I disproved very quickly, on that particular 1st retro-fit multi level layout.   I also found during my 2nd DCS Christmas layout, that the DCS signal deteriorated due to track joins, not just length of track,  this along with the Magic Lights MartyF found about that time, increased my DCS signal big time, and the layouts started to run with all 10's thru out.  Now that the Rev L TIU increases DCS signal, the Magic Lights are needed a lot less.  Lots of different factors determine how to set up a DCS/Legacy layout.   The DCS O Gauge Companion provides great instructions for engineering a DCS/Legacy layout, however if you start to engineer in a different manner, some of the things tested will work, and then some will not.  IMO the O Gauge Companion gives the owner/builder exacting instructional information that will work every time,  when followed.  If you take the engineering further altering it as I do, you learn some of these engineering alterations work well at times, some don't work at all.

PCRR/Dave

Another question came to mind, is it necessary to run ground (return feeders to both sides of Ross or Gargraves track? I am thinking that just one side is needed since train wheels will automatically ground the other side?

Paul, the superguru from New England, Marty Fitzhenry, insist Gargraves outside rails should be jumpered

Paul,

   This non-super Guru is advising you in the same manner, when using Gargraves Track jump the outside rails for best results.

PCRR/Dave

Ok, I'll jump in....
I'm going down to do some track laying & wiring right now.

From what I hear from this thread, the REAL importance is that the HOT or CENTER RAIL of a block section, no matter how long, should only have ONE feeder per block. But the GROUND for the whole layout could be a bus without messing up the DCS signal?

Am I understanding this right?  Can Barry or anyone confirm?

Rich, I confirm that.  I've been running that way for 15 years.

RJR posted:

Rich, I confirm that.  I've been running that way for 15 years.

I agree.

All that you need to promote strong DCS signal strength (all "10") is to have at least one outside  rail of Gargraves track with a strong Common at every spot on the railroad. While it may be necessary to connect both rails at some spots  to accomplish this, it is by no means a "rule" that one has to do so.

Further, it is also not necessary to remove the blackening from the center rail of Gragraves (or Ross) trackage. It's only necessary to remove the clear, oily protective coating from new Gargraves track.

This is how my layout, which is all Gargraves Phantom (black) center rail is built and, I can assure you, DCS runs quite well on this layout.  

My outside rails are jump together at the switches and I left the blackening on the center rails of my Ross switches and Gargraves  Phantom flex track. I just wiped the oil off the track with an old hand towel after installation.

 I have the same results as Barry.

Joe,

I use a rag with some old "Life Like" track cleaning fluid. That stuff has got to be 20 years old!

While I think I understand the nuances of the DCS signal, it seems that isolating center rails between TIU output connections seems counter-productive with regards to maintaining good voltage. Track looses voltage at weak connections - joints and pins. Also, Ross switches have the center rails "jumped" or wired underneath so that center rail voltage is carried from end of the switch to the other, and I was always taught that the more connections the better - from a voltage standpoint. From what I am reading here it sounds like cutting these jumps on the switches would be a good break point to isolate center rail connections ~ otherwise I'd have to install plastic pins or leave a center-rail gap every __ft?

Thus, to maintain both a good DCS signal and strong voltage there must be a criteria of how many center-rail isolated blocks to install? I recall the old pre-DCS days it was suggested to install feeders every 7 track connections - and it doesn't matter how long the individual track lengths are - since its the joints were voltage is degraded. What do you guys do who have good DCS signal and voltage - every __# of joints or __ft of track?

Paul, I have some 50 switches on may layout, all but 2 have the underswitch jumpers severed to end 3 blocks.  This two are because there's an adjacent switch and I want to power the connecting track. 

You said, " it seems that isolating center rails between TIU output connections..."  That has always been essential, because you then have 2 signal generators talking to the same track.  What we're talking about here is to also break the center rail between drops from the same TIU output.

Another reminder, when switches are and each end of a section of track, it's easy to overlook having the outside rails connected to ground.

In my opinion, using adequate wire is a must.  When building a small 8x4 layout for a grandson, witrh a centered control panel, I used 16-gauge, but on my layout, all feeds and drops are 14 gauge.

Paul,

Thus, to maintain both a good DCS signal and strong voltage there must be a criteria of how many center-rail isolated blocks to install?

The guideline is:

• Up to 12 track sections per isolated block
• Up to 14 blocks per TIU channel for pre-Rev. L TIUs and up to 24 track blocks per channel for Rev. L TIUs.

Your mileage may vary.

I never cut the outer rail jumpers on the switches. All outer rails are connected, even for different channels. Center rail connections have to be isolated for TIU each channel.

I always heard a near maximum  of 12 drops per channel per channel of the TIU or less before the signal starts to deteriorate. It could be less depending on the layout.  It might be different with the REV L TIU.

Thanks Barry,

I wasn't sure how many drops for the REV L TIU. 

I never cut the outer rail jumpers on the switches.

Joe, I wasn't clear.  I was only speaking of the center rail jumpers.

RJR,

I was just making it clear for Paul.

Joe, Jamie, and Barry, I understand that center rails  to be isolated between different TIU outputs (switches are logical break point for different TIU controlled on my mainlines)..I meant to specify breaking them between a terminal strip output (controlled by one TIU channel) seems counter-productive from maintaining strong voltage because in doing so you have a wire going to a block and then that has to supply voltage to 6 track joints in both directions (if you're using the 12 track joint guideline). Whereas, if you didn't have to isolate the center rail blocks you'd have continuity from one feed to another. If I don't do the center rail isolation I am taking a big gamble in DCS signal degradation?

I also recall something years ago about using paired wiring - such that every feed connection to the track should be paired? Is that still applicable - I suppose that's integral to the star wiring approach?

Paired wiring not necessary.

From a purely electrical power distribution perspective, you observation of multiple feeds being optimal is correct.  However, with DCS, you also have a higher frequency signal being imposed on the conductors, and the behavior of that signal differs from the behavior of 60-hertz power.  On my large layout, power distribution problems are minimized by using separate 14-gauge conductors from control panel to each block for hot feeds.  The blocks are constructed of Gargraves track, and I doubt that there would be more than 3 joints between drop and end-of-block. 

In one buried storage area, I did use Lionel 10" track with soldered jumpers at joints, but no block there is more than 3-4' long, end-to-end.  When the layout was built, well before DCS, to save wiring I used multi-conductor 18-gauge cable to feed these blocks, even though the rest of the layout was 14-gauge.  I fugured since I was running into a storage area, I'd always have a hand on the throttle anyway and could compensate for voltage drop.  When DCS came out, signal wasn't as strong as I wanted it, so I replaced the cable with separate 14-gauge THHS wire feeds to each block, and signal strength came up.

Paul,

If your track connections are that bad, then you could jumper each track section within the block with wire I suppose.  My track joints are very tight and clean. Just run a PW engine around the loop and you will soon see if you have a problem with voltage drop.

I find Lionel pins/track tend to rust after 10-20 years, making for poor contact.  Gragraves pins are also very thin, so I don't know what gauge they are.  I also don't know about the behavior of DCS signals running through steel rails connected by thin pins.