DCC Overview

This page is about the original planning for DCC on Sumida Crossing, but has been updated to reflect the final system configuration. For more about the wiring and systems, see the other pages in this section.

The Digital Command Control (DCC) system used for the railroad is Digitrax, with their Loconet control bus. This means that RailCom, which can identify specific trains on the track, is not presently an option, but Digitrax “transponding” may be used in its place. Digitrax’s proprietary “transponding” capability allows for reading CVs from a locomotive on the track (which can be used to supplement block-occupancy systems).

Digitrax was selected because I already had a Zephyr (as well as a couple of handheld throttles) and was familiar with their products, and because it’s going to be most compatible with the Kato trains I own (Kato decoders are made by Digitrax, and include trnasponding. However, this won’t be cheap.

Initially I added circuit breakers for roughly each scene to the Command Station, and a large power supply, along with occupancy detectors for the commuter and subway lines, and turnout controls. That cost about $1,700. I could have reduced it given the initial design, but it’s going to be easier to build the power distribution for the later phases now. And it will probably be a bit more reliable and easy to debug if I spend the extra money to separate the turnout power from the track power, and install the DCC turnout controls up front.

Command Station: Digitrax DCS 100

A command system takes control information from throttles, and encodes it for transmission, before passing the signal to one or more boosters, which supply power to the rails. All Digitrax command systems incorporate a booster.

I originally planned to use the Digitrax Zephyr I had from my previous railroad. Given the limited output of the Zephyr (2.5 Amps) and the desire to run several (4-7) trains, each of which may draw half an Amp or more when equipped with lighted cars, it soon became obvious this was going to be inadequate. The Zephyr was assigned to workbench duty, and a new DCS 100 was purchased.

Note: the above was written (and the DCS 100 purchased) before I’d measured an actual train on DCC. Without sound, power draw is likely going to be under 0.2 Amps per train even with interior lighting, so even my Zephyr would have been up to my needs.

The original Zephyr, incidentally, is limited to 10 “slots” (trains it knows about). If you have more than ten trains, you may need to make it “forget” one to run another. Digitrax makes to controllers with larger capacity, including the DB150 with 22 slots, and the DCS 100 (or 200) with 120 slots, and this was another reason I chose the DCS 100. In late 2010 they introduced the Zephyr Xtra (DCS51), which has 20 slots. Had it been available when I bought the DCS 100, it would have been a much harder decision to make, although at $225 list the savings aren’t that large, and I might have gone for the DCS 100 anyway.

Cost: $230 ($285 list)

Additional Boosters: None initially

The Digitrax DB 150 can operate as a simple booster, providing 5 Amps of track power and copying commands from some other command station (e.g., the DCS 100). It can also act as a command station, but lacks the ability to read CVs (which I regard as a fatal flaw in a command station). If/when power becomes a problem, I’ll supplement the DCS 100 with a DB 150 acting as a booster or something else, but I don’t expect that to be needed soon, if ever.

Future Cost: $179 list

Power Supply: Digitrax PS2012

The PS2012 is a hefty regulated DC supply that can provide four 5-Amp outputs at 12 volts DC (actually 13.98). I’m using one output for the DCS 100, one is reserved for a future DB 150, and the other two are being used to power the DS64 and SE8C/PM42/BDL168 systems.

Or not. I’m going to have to isolate the PM42’s local power from the PS2012 that sends power through them, which means I can’t use the PS2012 for the “accessory” power, or that I need to separate the PM42 from the accessory bus powered by the PS2012. I’m probably do the former, which wastes some of the capacity of the PS2012.

Cost: $144 ($180 list)

Circuit Breaker: Digitrax PM42

I need at least six circuit breakers per scene (plus the unsceniced curve) to properly isolate one line from another. My original plan ended up needing a total of 20. I’ve been through several iterations, and with the selection of the PM42/BDL168 combination, I’ve decided to deploy one PM42 with each BDL168, which gives me only four per scene (or eight if there’s one per table, but that doesn’t help). These four will isolate the two commuter and two subway lines. Separately, the Express lines will be fed from a central supply, with two circuit breakers (one per line) to isolate each from the main booster.

The result is that a short on any line is isolated from the other lines in the scene, and from other scenes, with the exception of shorts on the express lines, which are isolated from each other and the other lines, but a short anywhere on one of the express tracks takes down that whole express track’s circuit. Given that only one train per circuit is expected there, and switches are only used to add and remove trains, this is an acceptable limitation.

Before I decided on the PM42, I did some research into possible circuit breakers. One possibility was the PowerShield X (PSX) by DCC Specialties. They make models with 1-4 circuit breakers (approx $30 per breaker), and a separate line with (PSX-ARFB, $52) or without (PSX-AR, $48) integrated occupancy detectors. The PSX can be used with a Digitrax DS64 for reporting block occupancy. Another option is the Digitrax PM42 (4 blocks, $63). For tables with more than two circuit breakers, the PM42 is more cost-effective, but it uses mechanical relays, which are both potentially slow to trip and a moving part that could fail. In the end, I went with the PM42, saving nearly $300 (in the original configuration).

With a PM42, there are a few considerations: first, trip current should be set to 1.5 Amps when used with a Zephyr, and either 3.0 or 4.5 Amps when used with a 5 Amp booster. However, the default is 3 Amps, and no single section is likely to need that much, even if there were three trains in it, so I’ll probably stick with the default now that I’m using 5 Amp supplies, unless this causes a problem with inrush current detection on sound decoders or car lighting, in which case I’ll step up to 4.5 Amps. Another reason to leave the trip current at 3A, is that this is the maximum current that Unitrack is rated for (probably a limitation of the Unijoiners). Also, I expect to set trip speed to “fast” (recommended for N scale, but the default is “normal”), again unless this causes an inrush-current problem.

Secondly, I’m using the PM42 (which has four sections that can be driven independently) with the BDL168 (which has four sections that can be externally powered independently). I could have arranged these is several ways, but with the decision to keep the Express tracks separately wired (using Bus #3 and Bus #4), I’m going to arrange my two power buses (#1 and #2) to power two circuit-breaker sections each, and use one circuit breaker section for each of the four tracks. This implies I have no more than 4 occupancy sections per track per BDL168, which is problematic.

The PM42 can be connected to Loconet to report trip/clear events (the PSX would have used a second DS64 input per PSX for this purpose).

Right now I’m leaning towards using each PM42/BDL168 for just 8 transponding/detection zones, and that means two of them for each of the main scenes, plus one for the River Crossing scene and two for the return loop at the far end, or seven total. I’ll use part of one of the “return loop” PM42s (without the BDL168) to provide protection on the two express tracks when powered from DCC.

Cost: 7 @ $64 ($80 list ea.) = $448

Mounting Digitrax Circuit Boards: the CARDEDGECON

The PM42, like the BDL168 and SE8, uses a 44-pin card-edge connector (you solder wires to the connector, and plug the card into it). This is limited to about 18-gauge wire with stranded wire (I haven’t tried it with solid), and even that isn’t without problems, so perhaps 20-gauge would be preferable. Even 22-gauge can safely handle 5 Amps in open-air (although the smaller sizes may get warm), but I’d prefer to use heavier wire than this connector allows; one of several design flaws that make me dislike the PM42/BDL168 combination, but not severe enough to prevent me from using them.

The edge connector can be used, with a bit of drilling and some angle-brackets from a hardware store, to mount the PM42 to the underside of the layout or another wood surface. This part is available separately if you manage to destroy one.

Digitrax’s documentation is self-contradictory. It would appear than individual terminals have a limit of 3 Amps, and in some cases you need to solder a wire to two terminals to support a 5 Amp booster. It’s unclear if this limit is due to the card-edge connector or the circuit board. The BDL168 manual clearly says the card terminals are limited to 3 Amps, but the PM42 manual shows wiring for up to 5 Amps off a single terminal. It’s possible this is a BDL168 board issue, as they need to fit a lot of wires in a small space, and may be using smaller circuit-board traces as a result. It’s also possible the PM42 manual is wrong, and 5 Amps is only supported when you wire it to permit use in Reversing sections (the Reversing configuration uses both terminals, but it would apper to use them either/or, not together).

I’m assuming the output of a single PM42 breaker section will not exceed 3 Amps (and in practice I’d be surprised if it ever hit even 1 Amp), and wiring to support that design decision.

Cost: none (one comes with each board), retail $8 for a set of three, list $10/set.

Occupancy Detectors: Digitrax BDL168, with RX4 transponding

There are other ways to do occupancy detection with Digitrax: The Digitrax BD4 ($30) or the occupancy output of the PSX can be connected to the input of a Digitrax DS64 ($60) to provide block occupancy indication via Loconet. The BD4 provides occupancy detection to up to four blocks on a single power supply, and is limited to 3 Amps of continuous power through it (5 peak). It is thus a good match for a 5 Amp booster.

Instead, I’ve chosen to use the Digitrax BDL 168 ($150), which can be optionally configured with up to two RX4 sets of Transponder Receivers ($60) to add Transponding support. The BDL168 has a capacity of 3 Amps per detection section (or 6 if wired for that), and provides for up to 16 detection sections (four separate zones, each with four detection sections). The BDL can be connected to LocoNet (required for occupancy detection or transponding), and can be set to addresses from 1 - 999. If Transponding is desired, the BDL168 is required, and can work with either 4 or 8 RX1 sensors (each RX4 contains four RX1 sensors). Thus, the BDL168 can perform transponding on at most 8 of its detection sections, but it can still provide detection (a yes/no presence indication) on all 16.

I want transponding, which pretty much forces me into using the BDL168/RX1 combination. I have to say that I’m not impressed with this system. It’s terribly expensive if I want to make maximum use of transponding (and not much cheaper if I trim things a bit). And the wiring is rather awkward (lots of soldered connections and wires that need to either be twisted or kept away from each other; hard to do when working in a cramped space under the layout). Why couldn’t Digitrax build the transponding sensors onto the board (which should be possible with modern signal-processing electronics, even if they ended up with less than 8 sections per board) and put in screw terminals rather than the silly card-edge connector (why solder to something that’s just friction-fit to the board anyway)? This is really badly designed, and not up to what I consider their usual standards. And for all that, I end up with only 8 transponding zones, which makes the cost all the more painful.

If it weren’t for the fact that Kato’s use of Digitrax for their commuter EMU decoders somewhat forces my hand to use Transponding rather than RailCom, and that I’d already sunk a bit of money into these things before realizing what a pain they were, I’d probably be using something else. Digitrax is usually a fairly intelligent company, but this system is a real kludge, and quite a let-down.

Cost: 6 @ $120 ($150 list) = $720 + 6x2 RX4 @ $40 ($60 list) = $1,280

Transponding: using RX4

The RX4 is a set of four RX1 coils on a ribbon cable, used to detect transponding signals from Digitrax decoders. It is used to identify which train is in a detection zone, and for other kinds of “read-back” information transfer. Transponding pre-dates the NMRA RailCom standard (which is still in draft anyway) and is not compatible with RailCom; you need to use one or the other, and support for Transponding is pretty-well limited to Digitrax’s decoders.

The RX4 can be placed between the BDL168 and the track (8 zones) or between the PM42 and the BDL168 (four zones, but each zone covers all four BDL168 outputs for that sub-district so all 16 do “transponding”). The RX4, which is just a simple inductance coil, uses the BDL168’s LocoNet connection, and does not have one of its own.

If the BDL168 is used for occupancy detection, a pair of RX4 ($50 list ea) may be used to add transponding support to 8 (four per RX4) of the 16 zones. The RX4 requires a fairly large space (nominally 8x5.5 inches, but in practice a pair needs more like 18x6) for isolation from other wires, and must be within 12 inches (the length of the ribbon cable) of the BDL168. A “verbose” transponding mode (set on both the BDL168 and the RX4) is used to support the same ID appearing in more than one zone. This will likely be needed for multi-decoder Kato EMUs (presumably the end car lighting and mid-train motor will use the same address; however, it’s possible that only the motor decoder actually does transponding).

Digitrax recommends mixing transponding zones with simple detection zones, to minimize cost. I was originally going to do that, but then I realized that I wasn’t able to use 16 zones per BDL168 without a lot of cross-board wiring (it would be different if my layout wasn’t designed to be semi-mobile for moves) and that the actual number of devices needed didn’t go up a whole lot if I restricted myself to 8 zones, so now the plan is to do occupancy detection and transponding for every separate block.

Cost: $40 ($50 list) ea.

Turnout Control: DS64

The Digitrax DS 64 can switch up to 4 “turnouts”, where each turnout can be up to four actual turnouts (the Kato double crossover counts as four turnouts). I expect to use one output to throw both ends of a siding, to reduce the number of DS64 outputs required. Each DS64 requires 300ma @ 12v, so a 5 Amp supply will provide for about 15 of them. While this could be taken from track power, a separate power bus for DCC systems will be needed with the large number I’ll have. Each DS64 uses four device addresses, from the range 1 - 2048.

Note: one advantage of the DS51K1 over the DS64 is that while the DS51K1 can only throw one turnout, and thus can’t be used on the crossover or to throw both ends of a siding, it is actually small enough to install inside the turnout itself, eliminating any external wiring. I have a couple of turnouts I might do that with on the elevated station, just to simplify the wiring.

Cost 3 @ $48 = $144 ($60 list)

Grade Crossing Control: DS64

The DS64 can be used to drive a Tortoise slow-motion switch motor to lower crossing gates. Since a dedicated DS64 is required for this (at least as I read the manual), up to four motors can be controlled, which would allow independent control of entry/exit gates on both sides of an intersection (with four Tortoise’s, but I have a pile of those from the old layout).

Cost 1 @ $48 = $48 ($60 list)

Computer Connection: RR-CirKit Locobuffer-USB

To program, monitor, or control DCC decoders, a computer needs to connect to the control bus. With a Digitrax system, the connection to the LocoNet can be done several ways, but the LocoBuffer is widely reputed to be the best, and the current version of this supports USB (older ones needed a USB to serial adapter). One of it’s big benefits is that it’s opto-isolated, so an accident that zaps the LocoNet (i.e., shorting it into a power line) won’t fry the USB circuitry of the computer. Given the possibility of electrical accidents on a model railroad, that’s a compelling argument in its favor.

Digitrax does make a computer interface of their own. The older MS100 has been replaced by the new PR3, which does a number of things including the MS100 (USB to LocoNet) interfacing. This is a versatile and useful interface for a workbench environment, since it can be used to program sound decoders as well as being a computer interface.

The computer I plan to use is an old (2001) iMac that’s a bit slow for other uses, but should still be plenty fast for what I need it to do. I’ll use the JMRI software (which I fooled around with briefly on my last railroad, but never really used).

Cost: $63 ($70 list)

Signal Control: Digitrax SE8C

My use of a SE8C to control signals is more fully described on the DCC Japanese Prototype Signals page.

Cost: $100 ($125 list)

System Cost

The minimum system would have been to add circuit breakers to my existing Zephyr, with one circuit breaker per scene, each with four zones. With a PM42 this would have a cost of $64 per scene (or $256). This is, however, a very minimalist approach. Note that this does not include power supply, control system, booster, or turnout controls.

Adding simple block detection, and assuming four blocks per table (with Shinkansen and subway merging two adjacent tables into one block) would require either one PSX4 and DS64 per table ($180) or one BDL168 per pair of tables ($120). The BDL is the lesser cost, at $480, added to the above $256, for $736. It also has additional capacity to double the number of detection blocks per pair of tables at no cost (for comparison, the PSX method would cost at least $1,260, assuming the river scene tables were treated as four blocks total and ignoring the helix table, for $1,512).

A reasonable initial system would have been one PM42 and BDL168 per scene, with one each on the river, helix and staging tracks, for a total of five of each, or $920. This would provide for 16 detection zones per pair of tables (or roughly one per line per table, which is probably sufficient, even when a grade crossing is present).

As noted above, I went with a slightly larger system, although I’ve omitted the booster and systems for the Helix, at least for now. Total cost (not including the computer, or the LED lighting systems) is US$2,489 retail. Not cheap, but then I’m doing a lot with it, and it should last without change for many years. I also spread that expense out over about three years, as I gradually added things to the layout or bought them in advance of use when I had some spare budget.