Clearance Points

When laying out a CP I tried to ensure that insulated rail joints that separated the CP from the connecting block were placed a sufficient distance from the turnout so that a train could not foul the turnout without being detected. In prototype railroads this is called a clearance point. Rolling stock outside the clearance point will not foul a turnout. I placed the signals adjacent to the insulated rail joints that define the clearance point. Creating clearance points required that I attach small segments of track to most turnouts. It's tedious work and I used a lot of rail joints; regular and insulated. In a few cases I violated that principle and in hindsight I regret it. If the signal is placed after the insulated joints then the indication will change to stop before the train passes the signal. If the signal is placed before the joint then the signal has a delayed change to stop. The lesson learned is Don't Do That. 

System Design Parameters

When I started my design I fixed the design around the following parameters.

1. Number of turnouts: 64
2. Number of block occupancy sections: 64
3. Default signal: 2-head, 3-aspect (Digitrax SMBK)
4. Two track main
5. Each CP could include up to three turnouts that would correspond with 3 routes through a CP. A typical CP would include one crossover between Main 1 and Main 2, and a separate turnout diverging from the CP. The three possible routes through a CP are

• Normal; the crossover is lined normal  (no crossover or branch)
• Reverse; the crossover is lined reverse allowing movement between Main 1 and Main 2,
• Branch; the diverging route is lined. The diverging route can be before or after the crossover

When I added signaling to the reversing loops (Shortline Hill and Linden Branch), and Main 3, I modified the definition of a CP to allow for no turnouts. The only route through the CP is normal. Turnouts related to the CP are handled as a special case during CP logic processing.

Hardware and Development Environment

I chose the ATmega2560 Microcontroller board for the signal system because it was inexpensive and readily available, had sufficient RAM and flash, and had lots of discrete IO with sufficient drive strength to light LEDs. I did not need complex interfaces such as Ethernet, USB, or a file system. I chose Atmel Studio 7 for my Integrated Development Environment (it's free) and a JTAG debugger because I was familiar with their usage, thus minimizing the learning curve.

Atmel is now part of Microchip and they have a new IDE MPLAB X IDE that is also free. I have tried it with the ATmega2560 board but the integration and support did not seem as reliable as with Atmel Studio 7. The ATmega2560 is an ancient board but still  popular so I am not surprised that supporting this board in their new IDE is not a top priority.

I power the board through the USB connector and use the supported serial port to assist with debugging. I print out turnout and block occupancy changes, and I have a bare-bones command interface that allows me to query  signal system values such as block state, switch state, CTC authority, display status, and route status.

I use a second serial port to receive LocoNet. In theory I can transmit on LocoNet but I have not developed that support and have not needed it.

The Big Decision

The biggest design decision is the premise for this entire project: to replace dedicated hardware signals (wires; not track-side signals) for turnout state and block occupancy with LocoNet. In general this choice is a win. Occasionally a switch machine will malfunction and that will not be detected by the signal system; most often with solenoid machines. My longer term goal is to replace the solenoid switch machines with Tortoises or servos. My now more than 10 years old home brew block occupancy system occasionally fails to clear after all rolling stock exits the block (using resistor wheel sets). I have the block occupancy sensors on the panel and can clear occupancy manually. In addition, the occupancy defaults to Clear when track power is removed. I added a Digitrax BDL168 about 5 years ago and I have not seen it malfunction. Plus the BDL168 goes to Stop when track power is removed. 

I am happy not to have the extra wiring for turnout state as I already have enough wire under the table to make a pinball machine blush.

My dream layout would have hand-laid turnouts to eliminate all the extra rail joints and some turnout controller board that would handle turnout control, frog polarity, monitor turnout state, and detect block occupancy. This would replace most of the wiring with two connections; track power and LocoNet.

In spite of my gripes the current technology is light years different than the HO of my youth. 

Another question I ponder is how applicable this scheme is to other layouts. Clearly LocoNet is the price of admission for this whole layout signaling system.

Nancy

LCC?

Have you looked at LCC instead of loconet?

Jim B. I have not. I bought a

Jim B. I have not. I bought a Zephyr DCS50 15 years ago and that hardware as served me well; until it bit the dust last week :-(. Do you have any suggested pointers to LCC documentation? So long as the signal system can monitor block occupancy and turnout commands then it should not care where the data came from.

Nancy

LCC

The first stop would be RR-CirKits, because they make the hardware and Dick is an amazing resource for these thigns (especially for complicated signaling/BOD setups).

Detlef Kurpanek has put up videos on his LCC Channel with the basic setup, although he hasn't done any signaling yet:  https://www.youtube.com/channel/UCuRRF7W2tX4UjFhDAontT4Q

He also has his own layout tour with LCC implementation at 

I'm also heavily vested into LCC, but still far from signaling so can't help there. Right now using LCC in my staging as I do not have a mainline built yet.

Thanks for the pointer to

Thanks for the pointer to RR-CirKits, trainzluvr. I use Dick's original Tower Controller to interface my block occupancy detectors with LocoNet. LCC using CAN is a more robust bus than LocoNet and compatible with my signaling scheme. The LCC-Signal modules could also replace the Charlie-Plex LED drivers.

I am heavily invested in LocoNet with eight TeamDigital 8-channel turnout controllers and a BDL168 block occupancy board. I am in the process of separating out the local bus interface and signal driving code to allow for using other hardware options such as LCC.

The primary requirement for applying my signaling system is to isolate the track within a control point from the blocks that connect the control points. Signals are placed at each track entrance to a control point. You don't have to place the physical signal but the system will compute the signal indication. Each control point and block needs block occupancy detectors. If the local bus can report block occupancy and turnout positions then the signaling system can compute the indication for each signal based on the layout information that relates how the control points connect with each other.

Layout Wiring; A Cautionary Tale

I did not have a plan for layout wiring. I had a huge roll of two-strand orange and white #24 solid wire with which I wired the layout. The bottom side of the layout resembles a pinball machine. The bulk of the wiring consists of

• Turnout switch machine controllers to turnouts: 8 Team Digital SMD82 switch machine controllers
• Track blocks to block detection hardware: 6 home brew block detectors and one Digitrax BDL168
• RR-CirKits TC-64 Tower Controller
• Signals to signal controller. The controller is still on the bench so the wiring is shoofly
• Miscellaneous wiring; LocoNet and +12V to modules that need it

There's no getting around having lots of wire. The wiring is as simple as it is abundant. 

Each block feed has a series light bulb. I chose T5 bulbs because they are available in 4-18 W range suitable for N-scale locomotives, and I found bulb sockets cheap at Ax Man (local surplus store).

In my first iteration I mounted the turnout controllers and block occupancy hardware on a single hinged panel. The problem with that was all the wiring concentrated at that panel. I then decided to distribute the various devices to the middle of the layout to minimize wiring length and spread out the wiring over a larger area.

I don't recommend my implementation but I also do not know how much it can be improved upon. The wiring is mostly trouble-free except for the occasional bad splice or unsoldered connection.