Train Controller Bronze AutoTrain

The AutoTrain Feature

The Train Controller Bronze Modelling package allows you to control trains via the Uhlenbrock Intellibox IR.

One way to control a train is to use the AutoTrain feature. The AutoTrain feature allows you to move a train between two blocks. Train Controller will automatically search a route between the two blocks, turn all switches accordingly and drive the train from the start to the finish.

AutoTrain takes away your freedom to specify the route in detail. It is a convenient way to move locomotives if you only care about start and the target of a route but not which blocks are passed in between the start and the end.

First put a train on a block. Click on the block and choose the option “Assign Train”. A list of locomotive opens, select one of the locos and also select one of the orientation boxes. The ‘OK’ button is only activated if a loco and a orientation is selected.

Once a loco is on a block, click on the AutoTrain symbol (‘A’ Symbol) or select AutoTrain from the context menu.

AutoTrain now opens a small dialog window. This window is used to configure what role the blocks have in the route. One block has to be given the start role, the same or another block has to be given the target role.

On the layout select the block you want to start from. This is the block that you put the loco on in the prior step. In the AutoTrain dialog, use the green arrows to assign the start role to the block. There are two arrows which define the direction that the loco will drive when starting the route. Once you clicked the arrow button, the block will be marked green at the direction that the train will start from.

Now you need to assign the target role to a block. The target block can be the same block as the start block. If you assign the start and end role to the same block, Train Controller Bronze will find a circular route. The train will leave and arrive at the same block.

You can also assign the target role to another block. Train Controller will then drive the loco from the start block to the target block. Train Controller will automtically choose a route.

In order to assign the target role, use the orange buttons in the AutoTrain dialog. The target block will also be marked orange in the layout view. The orientation matters. If you want a train to enter the target block from the left, you have to select the target button that puts the orange marker on the right side of the target block. The orange marker marks the end of the route.

After the start role and the target role have been assigned, click the ‘A’ button in the AutoTrain dialog to start the route. The loco will move over the layout and AutoTrain will control all switches to open the automatically choosen route.


What if nothing happens? What if the route does not work?

  1. Verify that one of your locomotives is working correctly. Put that loco onto your layout. Use the Intellibox to manually control that loco. Now you know that the Intellibox can send commands to the loco.
  2. Check that the PC is connected to the Intellibox. Use the device manager to find the COM port (on windows) that the intellibox is connected to.
  3. In TrainController, edit the settings of the Intellibox. Use the correct COM Port. Control a switch manually in order to verify that Train Controller can talk to the Intellibox and the Intellibox sends commands to the layout correctly.
  4. In Train Controller, control a locomotive manually, that is without AutoTrain. I do not know how that works but you should be able to have a manually controlled throttle for your locomotive and the locomotive should accelerate and brake correctly.
  5. Try out AutoTrain after all steps above did work correctly.

Sometimes a locomotive does not run correctly. Try to put oil into the locomotive and try to push the locomotive over the layout manually in order to help it gain speed until it is able to run itself. A case was observed where TrainController and the Intellibox were working 100% correctly but the loco just was stuck on the rail and would not start the route!

You should test with your most trustworthy locomotive. Use a locomotive that never fails. Basically you want to exclude all sources of errors to narrow down the real issue.

Building OpenCV on MacOS


These installation instructions are taken from:

Download the sources from

Extract the zip to /Users/bischowg/Downloads/opencv-4.1.2

Create two new folders inside of the openCV directory, one called StaticLibs and the other SharedLibs.

Build the Static Libraries

Open cmake-gui (type cmake-gui into the console)

Click Browse Source and navigate to your openCV folder.

Click Browse Build and navigate to your StaticLib Folder.

Click the configure button.

You will be asked how you would like to generate the files.

Choose Unix-Makefile from the Drop Down menu and Click OK.

CMake will perform some tests and return a set of red boxes appear in the CMake Window.

You will need to uncheck and add to the following options.



it will look something like this “/Applications/”.
—> Option on my system: /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk

Add x86_64 to CMAKE_OSX_ARCHITECTURES, this tells it to compile against the current system

Uncheck WITH_1394


Click Configure again, then Click Generate.

When the application has finished generating, Open Terminal and type the following commands.

– cd <path/to/your/opencv/staticlibs/folder/>
– make (This will take awhile)
– sudo make install

Enter your password.
This will install the static libraries on your computer.

Build the Shared Libraries

Open cmake-gui (type cmake-gui into the console)

Click Browse Source and navigate to your openCV folder.

Click Browse Build and navigate to your SharedLib Folder.

Click the configure button.

You will be asked how you would like to generate the files.

Choose Unix-Makefile from the Drop Down menu and Click OK.

CMake will perform some tests and return a set of red boxes appear in the CMake Window.

You will need to uncheck and add to the following options.



Add an SDK path to CMAKE_OSX_SYSROOT, it will look something like this
—> Option on my system: /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk

Add x86_64 to CMAKE_OSX_ARCHITECTURES, this tells it to compile against the current system

Uncheck WITH_1394


Click Configure again, then Click Generate.

When the application has finished generating, Open Terminal.

– cd <path/to/your/opencv/SharedLibs/folder/>
– make (This will take awhile)
– sudo make install

Enter your password.
This will install the shared libraries on your computer.

Problem During Building the Example Application
OpenCV 4.x+ requires enabled C++11 support

Into the CMakeLists.txt add:


Then call

cmake .



Sequelize describes itself on the sequelize homepage:

Sequelize is a promise-based Node.js ORM for Postgres, MySQL, MariaDB, SQLite and Microsoft SQL Server. It features solid transaction support, relations, eager and lazy loading, read replication and more.

Sequelize is an ORM (Object Relational Mapper) which allows you to interface with a SQL database without writing SQL but purely using the domain objects of your node application.

Sequelize shields you of any SQL to the point where Sequelize will automatically generate and execute table creation statements and all the insert, update, select and delete statements for you! You do not have to create tables in your database server, Sequelize will do it for you.

While reading this article, you can look at this repository. It is not good code by any means but maybe it helps you to see a running example which you can modify and do your own tests on.

Defining Models

A model is an object that has attributes which will be stored in the columns of a database table. Models can be connected to each other using associations.

An example would be an application that manages bank accounts. There is a model for an account, each account has amounts of money in them and money is transferred between accounts by transactions. You could model Accounts, Amounts and Transactions. Sequelize will allow you to perform CRUD for those objects without writing a single line of SQL.

The model folder and index.js

In order for Sequelize to manage your objects, you have to define the models first so Sequelize knows about their structure and can generate SQL for you.

This approach is taken from and it works very well.

One way of managing your models is to have a separate folder to store all your model definitions in. A model definition is a node module that exports a model. A model optionally has a function that defines it’s associations to the other models.

Inside that model folder, you will also have a index.js file. This index.js file will first scan for all the models defined in the models folder, set them up and then export an object called db. db will be the object your node application uses to interface with sequelize. db contains a connection to the database and all the model definitions that you can store in the database.

index.js will perform the following steps to set up models:

  • It will connect to the database and store the connection into the db object
  • It will scan the model folder and load all model definitions it finds
  • It will call the associate() function on model (if defined) so that a model can define it’s association to the other models
TypeError: defineCall is not a function

One very import thing to remember is the following: The error “TypeError: defineCall is not a function” is thrown if your model folder does contain any files except valid Sequelize Model-Modules or the index.js file! If you put any non-Sequelize code into the model folder or if you even comment out the entire contents one of your Model-Module files, Sequelize will get confused and throw the “defineCall” error! So do not comment out any of your Model-Modules and do not put any other files into the model folder!

An example Model-Module (models/account.model.js) for accounts is:

module.exports = (sequelize, DataTypes) => {

    var Account = sequelize.define('Account', {
        'id': {
            type: DataTypes.INTEGER(11),
            allowNull: false,
            primaryKey: true,
            autoIncrement: true
        'name': {
            type: DataTypes.STRING(255)

    Account.associate = function (models) {

        // optional, foreign key is stored in the source model 
// (= Account has foreign key to RealAccount) models.Account.belongsTo(models.RealAccount, { foreignKey: 'realaccountid', targetKey: 'id' }); models.Account.hasMany(models.Amount, { foreignKey: 'virtualaccountid', targetKey: 'id' }) }; return Account; };

This module defines a RealAccount model and it’s associations to a Account Model and several Amount models.

The index.js file looks like this:

// inspired by

var fs = require('fs');
var path = require('path');
var Sequelize = require('sequelize');

var basename = path.basename(__filename);

const sqlConfig = {
    user: 'root',
    password: 'test',
    server: '',
    database: 'cash'

var sequelizeConnection = new Sequelize(sqlConfig.database, sqlConfig.user, sqlConfig.password, {
    host: 'localhost',
    port: 3306,
    dialect: 'mysql',
    logging: false,
    pool: {
        max: 5,
        min: 0,
        idle: 10000

var db = {
    Sequelize: Sequelize,
    sequelize: sequelizeConnection

// collect all model files in the models folder to automatically load all the defined models
    .filter(file => {
        return (file.indexOf('.') !== 0) && (file !== basename) && (file.slice(-3) === '.js');
    .forEach(file => {
        var model = db.sequelize['import'](path.join(__dirname, file));
        db[] = model;

// if a model has an associtate methode, call it.
// The associate method will define the relationships between the models.
Object.keys(db).forEach(modelName => {
    if (db[modelName].associate) {

module.exports = db;


Using the db object

To store and load into the database, you have to use the db object. Import the db object:

var db = require('../model/index.js');

Now the db object is ready.

For Testing: Erase and recreate all the tables for all models

Note: Never have this in your production code! All data will be lost! This is usefull for testing!

During the testing phase, it is usefull to let Sequelize erase all tables and recreate them for you. To do so execute sync() with the force parameter:

// delete the entire database
await db.sequelize.sync({
    force: true
Inserting a Model

The functions that operate on models are added to a mode_services folder into the service.js file.

It is a mistake to insert the services.js file into the model folder as Sequelize will get confused and produce the “TypeError: defineCall is not a function” error.

The module to insert a model looks like this:

async function addAmountByIds(db, realAccountId, virtualAccountId, amount) {
    return db.Amount.create({
        amount: amount,
        realaccountid: realAccountId,
        virtualaccountid: virtualAccountId

module.exports = {

    addAmountByIds: addAmountByIds


To use this method:

var services = require('../persistence_services/services');
var db = require('../persistence/index.js');

var realAccount = await services.addAmountByIds(db, 1, 2, 299);

You can see that the function addAmountByIds() is defined async because Sequelize is inherently async and uses Promises for everything.

addAmountByIds() will not return the created Amount object instantly but it will return a Promise. The Promise is your handle to an asynchronous flow of operation which is creating the Amount object in the database asynchronosly. As soon as that flow finishes, the Promise will return the actual result which is the created amount object.

Declaring addAmountByIds() async makes it a non-blocking call which means that when you call it, it immediately returns a Promise and your normal program flow continues while a parallel thread starts in the background.

This way of dealing with asynchronicity is awesome but my brain just can’t handle it. I cannot write a working application dealing with hundreds of Promises instead of real objects. Because of my own imperfection, the examples will always call the service functions with the await keyword.

The await keyword turns the non-blocking calls into plain old blocking functions. The program flow will block after calling the service functions until the real object is returned from the promise. That way, you are sure that the result returned is a valid object that is no persisted into your database. You can now write sequential code and sole your problems without thinking about asynchronous code.

then() – chaining

An alternative way of dealing with asynchronous Promises is to chain Promises together with calls to the then() function.

The callback specified in the then() function is called as soon as the Promise returns the real object. Besides await, this is another way to block until the result is ready. then() chains are a very akward way of writing sequential code in my humble opinion. I maybe have never seen code that is high-quality and easily readable using then() chains, but I cannot imagine easy to read code using then() changes as of now.

Accessing Associations via Properties

If you have objects associated to each other such as a BlogPost and all Comments people left on that BlogPost, you can access all Comments if you have a reference to the BlogPost object just by calling a property on the object.

This means, you do not have to explicitly query for the Comments instead Sequelize will lazy load the Comments for you in the background.

var comments = await blogPost.getComments();

Again, the Promise is resolved by the await keyword and the comments variable contains the loaded comments.


How do you connect a BlogPost to a Comment for example?

Remember that index.js will call the associate() function of each model during initialization? The associate() function is where you define the associations for each model.

You should consult the Sequelize documentation to find the right type of association that describe the situation you want to model best.

As an example, if a BlogPost should be associated to many Comments, define:

BlogPost.associate = function (models) {

    models.BlogPost.hasMany(models.Comment, {
        foreignKey: 'blogpostid',
        targetKey: 'id'

Not creating Objects in Associations

I have not fully understood why but calling Setters will automatically insert new objects into the database unless you use the save: false parameters:

transaction.setSource(targetAccountObject, { save: false });
Updating Objects

If you want to change the amount in an account, you have to change the value and trigger an update:

amount.amount += 299;
await => { });

Here, the amount is increased by 299 and an await is used to wait for the result of the save() call which updates the object in the database.

Another way is to use update():

    amount: newAmount
}).then(() => { });
Querying Objects

The findByPk() convenience method allows you to query by primary key.

var result = await db.RealAccount.findByPk(amountObject.realaccountid);

General queries are executed using findAll() and a where clause:

var result = await db.Amount.findAll({
    where: {


Creating an Angular Application

Generating the Application using ng new

Angular uses the ng tool to generate all sorts of code for you. In fact ng is used to generate all of the angular items from components to the entire application.

First install the current long term support version of npm:

nvm install --lts
nvm use node --lts

Alternatively use the latest release:

$ nvm ls
->     v14.17.3

Now install and use the lastest version:

$ nvm install v16.1.0
$ nvm use v16.1.0

ng is added to your system by installing the Angular cli globally using the node package manager npm:

npm install -g @angular/cli

You can check the angular CLI version:

ng version

At the time of this writing, the version Angular CLI: 12.1.3 is current.

The global ng installation is used to generate a new application:

ng new <APPLICATION_NAME> --style=scss --routing

This will create a folder called <APPLICATION_NAME> in the current working folder containing the new project. It will use the sass processor for CSS stylesheets using the scss syntax. It will automatically use routing.

You can also go through a interactive process where the angular CLI asks you about all options before creating the project.

ng new <APPLICATION_NAME> --interactive

Once a project was generated using the global version of the angular CLI, you change into the project folder and from there on out, the local version of the angular CLI as specified in the package.json is used. This means even when the global Angular CLI is updated to a newer version, your application will not break because it locally uses the version as specified in the package.json file.

This helps project stability as you can update the global angular CLI version for new projects and keep the local angular CLI version to prevent the angular application from breaking due to version differences.

Additional Dependencies

For the NgRx store architecture:

npm install @ngrx/store --save
npm install @ngrx/effects --save

Starting the Application

You can open the application folder using Visual Studio Code. Withing Visual Studio Code, open a terminal using the Terminal menu item in the menu bar and the New Terminal Item within it.

To start the application, type npm start which will internally call ng serve.

npm run start

You can also call ng serve directly

ng serve

The application is running at http://localhost:4200/

Adding a Module

Angular is particularly useful because it has a rigid scheme for organizing code which benefits structure and ultimately the quality your application code will have in the long run.

That scheme consists of the use of Typescript Modules which contain components.

There are two types of modules: modules and feature modules.

Initially, ng generates a module which is the app module containing the app component, which is used as a starting point for the application (called bootstrapping).

Starting from the main property inside the angular.json file, a main.ts typescript file is configured. WebPack will execute this main.ts / main.js file when starting the application.

Inside main.ts, a call to bootstrapModule() is made and the app module (AppModule) is specified as a parameter.

Looking into app/app.module.ts, you can see the bootstrap property of the @NgModule decorator. It contains the AppComponent. That means, when angular starts the app module, it will initialize the AppComponent first and use it as an entry point into the application.

Feature modules can be added to the application to extent the application with new functionality. For each separate feature, you create a new feature module so that the feature modules stay independent from each other (strong cohesion, weak coupling). A module will contain all components, services, interceptors, decorators, pipes, model classes and everything else to make a feature work as a independent unit.

Move common functionality such as utilities into their own modules to reuse them from several modules.

The angular CLI allows you to add a module:

ng generate module <MODULE_NAME>

To generate a module that houses all user-related code, execute

ng generate module user

Note that the name was choosen to be user and not user-module or anything. Angular CLI will automatically generate a user folder and a user.module.ts file. The CLI will postfix the module identifier to the generated files for you!

Lazy Loaded Feature Modules

Documentation is:

A word of advice: lazy loaded feature modules have been updated in newer versions of angular. This article shows console output from version 12. If your output differs, consider migrating to the newest angular version. 

When you want a lazy loaded module, do not import the module into the app module, instead use the router. The idea behind a lazy loaded module is to only load it into memory, when the router navigates to the module.

Therefore the router is used to load the module when a certain route is visited. To setup the lazy loading, update the file app-routing.module.ts

const routes: Routes = [
path: 'projects',
loadChildren: () =>
import('./projects/projects.module').then((mod) => mod.ProjectsModule),

Here, once the path ‘projects’ is followed, the router will execute the import() function which loads the lazy loaded projects module in this case.

The question is, which component will be rendered when the path is visited? The code in the app-routing.module.ts file does not specify a component to load within the projects module! The projects module itself will again contain a routing configuration file called projects-routing.module.ts which specifies all the routes and components.

The file looks like this:

import { NgModule } from '@angular/core';
import { Routes, RouterModule } from '@angular/router';

import { ProjectContainerComponent } from './components/project-container/project-container.component';

const routes: Routes = [
path: '',
component: ProjectContainerComponent,

imports: [RouterModule.forChild(routes)],
exports: [RouterModule],
export class ProjectsRoutingModule {}

One last change is necessary, in the lazy loaded feature module, import the ProjectsRoutingModule from the projects-routing.module.ts file and add it to the imports of the FeatureModule so it partakes in the routing:

import { NgModule } from '@angular/core';
import { CommonModule } from '@angular/common';
import { ProjectContainerComponent } from './components/project-container/project-container.component';

import { ProjectsRoutingModule } from './projects-routing.module';

declarations: [
imports: [
CommonModule, ProjectsRoutingModule
export class ProjectsModule { }

When you start your app, you should see a listing of Chunks. Chunks are the files containing the application code, that are ultimately downloaded to the client to run your app. You should see your lazy loaded module listed as a Lazy Chunk File as opposed to the Initital Chunk File option for eagerly loaded modules.

Initial Chunk Files                    | Names         |      Size
vendor.js                              | vendor        |   2.39 MB
polyfills.js                           | polyfills     | 128.55 kB
runtime.js                             | runtime       |  12.51 kB
main.js                                | main          |   9.50 kB
styles.css                             | styles        | 118 bytes

                                       | Initial Total |   2.54 MB

Lazy Chunk Files                       | Names         |      Size
src_app_projects_projects_module_ts.js | -             |   5.81 kB

Adding a Service

To add a service into a module, you can use the Angular CLI.

ng generate module auth
cd src/app/module
ng generate service services/auth

This will create the auth.service.ts file inside the auth module.

import { Injectable } from '@angular/core';

providedIn: 'root'
export class AuthService {

constructor() { }

@Injectable providedIn: root means that, the provider for this service is added to the root provider. This is documented here:

A provider is responsible to retrieve an instance for a dependency used in dependency injection.

Building an Angular Example App

We will build an application called ‘cash’ that helps you organize your finances. The code is available on github.

The cash application manages accounts.

Real Accounts mirror an existing account that you own within your bank institution.

Virtual Accounts are accounts that you can create within cash and which do not exist in your bank institution.

Virtual Accounts can have zero or one real accounts connected to them.
That means virtual accounts can exist, that are not backed by a real account. A real account can be connected to at most one virtual account.
That means it is not possible to connect a real account to two virtual accounts.

If a virtual account is connected to a real account, the real account is hidden by the virtual account. The amount of money that is stored in the real account is now accesseable only via it’s virtual account.

The way the cash application is structured is that whenever you add a real account to the cash application, cash will automatically create a virtual account for you and connect it to the real account.

You basically only are working with virtual accounts when you use the cash application. The layer of virtual accounts hides the layer of real accounts beneath it. You only interface with virtual accounts.

You can create any number of virtual accounts. For example if you want to structure your income and if you want to save up some money for a new PC or anything, you can create a virtual account for your savings and transfer money in between virtual accounts onto your savings account.

That way your income is subtracted by the amount of money you want to save and you can clearly see how much money there is left to spend after putting away the saved amount of money.

Money is transferred between accounts in transactions. A transaction starts at at most one virtual account and ends at at most another virtual account. A transaction transfers a non-negative amount of money and has a date as well as a description. Transaction between the same virtual accounts are not allowed.

For a transaction, there are the following cases:
A) Source Virtual Account (SVA) and Target Virtual Account (TVA) both are not backed by real accounts.
B) SVA is backed by a real account but TVA is not.
C) SVA is not backed by a real account but TVA is.
D) SVA and TVA are both backed by real accounts.

There are special cases:

Incoming amounts of money are transactions without a source account.
Expenses are transactions that have no target.

E) There is no SVA. The transaction adds income to your bank accounts.
F) There is no TVA. The transaction denotes an expense you made.

If SVA and TVA are both backed by real accounts (case D), then the money of the transaction is also transferred between the real accounts.

If SVA and TVA are both not backed (case A) the real accounts are not altered.

If there is a sequence of transactions of cases B – A* – C, then the money that made it from the real source account to the real target account is also transferred between the real accounts.

B – A* – C means that the sequence starts with a transaction of type B, then there are arbitrary many transactions of type A, then the sequence ends with a transaction of type C. That means, money travels over several hops over virtual accounts and in a global perspective between real accounts.

The amount of money in an account over a time frame can be displayed as a graph.

Transactions are stored in a transaction log which you can open and inspect at any time.

After every transaction, ‘cash’ will compute the amount of money in all real accounts and the money in all virtual accounts. The two amounts of money have to be the same at all times.

Let’s implement this as an angular application. It might be a bit ambitious but it is a usefull example that you can really make use of in your real life.

CRUD for Accounts

The first step will be CRUD for accounts. CRUD stands for Create, Retrieve, Update and Delete. I will use the acronym for an application that provides a user interface and backend code that allows the user to manage items (Create, Retrieve, Update and Delete). The user interface will consist of forms that allow the user to input data for accounts or edit existing accounts. The backend code will persist the accounts to a SQL datastore using Express and the excellent Sequelize SQL-ORM mapper.

First a SQL schema is created to manage real and virtual accounts. Then an API is created for the account CRUD operations in the backend. Once the backend and data storage are functional, a Form is needed to add new accounts. A list is needed to display existing accounts for editing, inspection and deletion.

Accessing Virtual Account via the API

First, lets create a service that covers the API calls. Creating services is done via the angular cli.

ng generate service <service_name>
ng generate service Account
ng generate service AccountDetails

If you want an AccountService, you have to generate with the name ‘Account’. ng will add the name service for you.