Computers and Trains

Hello:

I recently rewired part of my layout, mostly the DCC decoders controlling the signals. Since most of my signal are semaphore signals from Fleischmann, I used regular DCC accessory decoders (DS64 from Digitrax) to control them. The plan was to use the local inputs of the decoders and reed contacts to control the signals, alas the local inputs did not react with 100% certainty to the reed contacts. I feel that this has something to do with the programming of the inputs since they can be set to respond either to buttons or occupancy detectors.

I ended up installing panel decoders by MGP, a Swedish company. They work perfect, however it took some time to deal with the electronic noise under the train coming from the DCC wiring and other decoders. The owner of MGP was extremely helpful dealing with this.

The logic of the signals was to be controlled by relays, however I decided to abandon this idea since it required the wires between contacts and decoder to be too long. Therefore, I finally took the dive, and used JMRI for controlling the signals. This is a Java based program, and runs on virtually all platforms. I used the built in logic function (Logix) to control the signals, and it works extremely well. Apparently, JMRI also has a signaling program, but it is designed more for American signaling system. In particular, the aspect of a signal is affected by the aspects of the next signal. It can be adapted to reflect German practices if one uses the new KS-signals. However, since my railroad is based on the H/V-system, I decided to create the logic myself. The H/V-system does not rely on the next signal, while the KS-signal does.

Ulrich

Comments

  • I did not have any sort of reliabilty using reed switches with semaphores. It looks great on paper, but in practice a passing locomotive over the reed switch would trigger the semaphore only about 75% of the time. Adjusting the reed switch, the magnet, or both did not seem to have any effect on reliabilty.

    I may get back to working on the semaphores on the layout but for now they are stationary with lamps lit for a clear block.

  • BR42BR42 Auburn
    edited November 2021

    Michael:

    Do you switch the semaphores directly through the reed switch? Which type of semaphores do you use? On my old analogue layout, I threw the signals (Fleischmann) directly through reed switches, and had very high reliability. With the DS64 I got about 90% reliability, using the panel decoders close to 100% just as in the analogue setting. Team Digital SMD8 (out of production) and NCE Snap-It decoders had the same reliability as the panel decoders and analogue setting.

    Ulrich

  • edited January 4

    I tried operating the semaphores directly from the magnetic switches and when that didn't prove reliable, I tried using relays with two different sources of power. Neither was reliable and there was a lot of contact bounce in the relays. I should have tried SS relays and see if it made a difference.

  • BR42BR42 Auburn

    Michael:

    The problems driving signals or other devices directly with reed contacts are two-fold: Either the reed contact impulse is not long enough, e.g. for slow moving signals like Viessmann signals, or the currents are quite high using regular semaphore like in Fleischmann. The most reliable approach I found is the following:

    The signals are controlled by DCC accessory decoder which allows button control like the Digitrax DS74 or NCE Snap-It. My favorite decoder for control through a reed contact is the Snap-It. Connect the reeds where the buttons are to go, and it switches the signal absolutely reliable. The only drawback is that it is somewhat messy to have complicated signal logic unless you have an NCE system. In this case, the NCE-miniboard can control the logic for you.

    If you do not have a NCE system, then you could control the logic through relays as long as you have only NCE decoders. I had installed such a system with NCE decoders and Teamdigital SMD8 (which work on the same principal, but have routes for signal combinations), and it worked absolutely reliable.

    I then tried to do the same with DS64, and found that their inputs did not operate the routes required for signal combination reliably. They operate single turnouts and sensor without problems using reed contatcts. I switched over to controlling the signals through JMRI and use the reed contacts as sensors. The DS64 work reliably sending the sensor information on the the computer.

    I would recommend using JMRI or a similar software because it allows a more realistic approach to signalling. I did not use its signalling feature, but created the logic myself using their built in logic function called Logix. Another advantage of computer control over relays is that changes in the system or errors do not require a soldering session under the layout, just reprogramming.

    JMRI does not put high demands on the computer. I run it on a 4GB laptop running Linux. It would have run fine on Win10 on that laptop. However, it was an old laptop my daughter used in high-school, and she had forgotten her password, so I installed Linux.

    Ulrich

  • I use the NCE Powercab for control so I might have a look at some of their accessory boards.

  • BR42BR42 Auburn

    Yes, their minipanel seems to be a wonderful controller. If it were available for Loconet, I woul not have used JMRI.
    Ulrich

  • BR42BR42 Auburn

    Yes, their minipanel seems to be a wonderful controller. If it were available for Loconet, I woul not have used JMRI.
    Ulrich

  • BR42BR42 Auburn
    edited May 11

    I finally finished adding computer control to my layout. The goal has been to achieve automation of some tasks, not automatic running of trains. The latter would, in my opinion, be somewhat boring. Instead, the goal was to automate some functions that distract from enjoying running the trains. These include:

    a) Controlling the signals through the running trains.
    b) Automated block control so that two trains running around the layout will keep safe distance.
    c) Changing trains in the station in mode b)
    d) Automated running in opposite directions on the single track main line. Trains will stop and start in the main station and in the small station in the back of the layout. The latter has rather short tracks, so longer trains will pass through this station without stopping. There are four commands which will start the running in opposite directions depending on the length of the individual trains:
    Command 1 runs two long trains. Command 2 and 3 run on long and one short train (2 for the long train counter-, 3 for the long train clockwise), Command 4 runs two short trains.
    e) Automated switch from block control to opposite direction mode with automated turning of one train by running through a loop.
    f) Automated switch from opposite direction mode to block control with automated turning of one train by running through a loop.
    g) Interlocking turnouts and signals so that trains do not have a go aspect if the turnout is set against them.

    This operations are achieved by JMRI Logix statements of the form if..., then... All automation runs without reference to particular locomotives. Smooth braking is made possible using a feature of the programming track of my Digitrax DCS200 which allows it to be used as a brake generator when not in programming mode. The output of the brake generator is fed into the track via an old NCE DB3 booster. Altogether, 164 Logix statements, 56 Digitrax routes and 55 JMRI routes are needed to make this work. The layout is connected to an old 4 Gb laptop running Raspberry Linux via a LocobufferNG.

    Ulrich

Sign In or Register to comment.