47137

I have gone back to the wiring so I can have a working layout. All I needed to do was make a lever frame for the turnouts, and this is it: It hangs on the front fascia of the layout, held with two screws installed from behind: The box for this is 6.3 mm ply for the top, 25 x 15 mm pine for the back, 3.2 mm ply for the front and base, and 3.2 mm basswood for the ends. I was determined to make it from bits and pieces to hand and not buy fresh material! The 'levers' are a job lot of genuine C&K toggle switches with long dollies, bought for a song from a seller on eBay. Hugely more reliable than the Chinese copies. These switches were intended for pcb mounting, so they have plain bosses (no threads or retaining nuts) and have gone into the ply panel rather neatly. This is the connection from the lever frame to the layout: I am not a great fan of multi-way plugs and sockets because they introduce new places for wires to fall off, but doing things this way meant all of the soldering for the lever frame was on the bench, away from the layout. I need barely half of the pins on the 25-way connector here, but I have spare pins if I ever try for a more sophisticated control panel. And ... it all works! I might indulge myself in running some trains before tackling the scenery. - Richard.

Yes - the metal will stretch so you can create shapes like hemispheres, upwards or downwards. - Richard.

Yes. A customer bought a modern industrial unit and installed a first floor. He has a spiral staircase inside the front of the building, it was just about the only architectural feature he could have. But there is a second staircase, and this had to be a straight flight. In practice, customers and suppliers use the spiral staircase (I told him I would swear he put it in to disorientate people), and staff use both. - Richard.

The landscape of "Shelf Marshes" is supposed to be reclaimed land (an excuse if ever there was for a closed-top baseboard!), so the scenic treatment is going to be quite modest and gently undulating. I've got about +10/-0 mm to play with to get the effect I want. This is my test piece built yesterday: Photo lost, cannot be restored The Sculptamold seems to hold well, including on the aluminium mesh where it has gone through and locked itself into the holes. I want to let it dry out, and then try drilling it and sanding it down. So far it has lost 30g (note on the corner of the plywood board). - Richard.

Hi Simon - nice to see you back here. Have you actually used Sculptamold, and if so how do you feel it compares to your 'mash'? - Richard.

I think my buildings should be "attached" so I can take the layout to shows one day. But the fixings needs to be undoable - screws going up from underneath or maybe some small magnets. Actually, one of the buildings needs to lift off easily, because I built a voltage and current meter into the baseboard. I wanted an ammeter available as I loaded up the power unit, but I didn't want to be looking at it all the time. An electricity substation building would be good. The chemical plant will be harder. Perhaps I should make it "detachable if desperate" - fix its sub-base onto the baseboard with screws but not glue, and lay scenery over the join. I don't suppose anyone reading this has tried Sculptamold? The obvious thing to do next is give the landscape a bit of shape, and try to keep the weight down. - Richard.

Over four months have passed since I ran a train. I have put the layout back into its alcove and shuttled my Class 66 up and down along it - good. I have tried filming this with the Magnorail running at the same time, with some success too. I painted the fascia of the layout a dark grey a few weeks ago. Curiously this seems to make the front of the layout recede, and the baseboard look more shallow from front to back. I was expecting the opposite to happen. I need to have a rethink about the model buildings on the layout. The chemical plant has got to stay (of course) but the area behind the Magnorail might look better with a single-storey warehouse or similar (low relief of course), not the block of flats. I do rather like being able to move buildings around the layout to photograph them in different locations ... and with this, I can feel a reluctance to fix anything down. - Richard.

I've been fettling the Arduino software for my Magnorail system. The car now accelerates to a higher maximum speed, holds it for longer and brakes to a lower speed for the roundabout. This looks good, the overall impression is about right. The chain still manages to trip every reed switch every time, but the smallest window for this to happen is now 33 milliseconds, between waypoint 4 and waypoint 1. During the initialisation sequence, this is the free time after setting the servo for the route of the bike; and during the normal sequence it is the free time after decelerating from a pleasing maximum speed and using a realistic braking rate. So really, I have reached the limit of what I can achieve in this rather confined system. On the bright side, I have managed this using a purely procedural approach to programming where one thing happens after another. I do not need to alter the speed of the car and move the servo at the same time; and this is fortunate because it would need programming techniques beyond my present ability. /cut (... and so I have reached the limit of what I can achieve using a purely procedural approach to programming where one thing happens after another. Anything better needs me to alter the speed of the car and move the servo at the same time; and this is somewhere I don't want to go with my present Arduino expertise.) / The next video on YouTube has had to be delayed, but I am sure it will come along one day. In the meantime, I've attached a log file from a run yesterday. I've now promised myself to the leave the software untouched until I get back to work, and I'm hoping this will be around four weeks from now, when non-essential retail shops in England can prepare to reopen. - Richard. log.pdf

The adaptors I bought from A&H Models of Brackley originally came from AMF 87 of Wavrin (northern France). The adaptors are still on their web site, they come in a pack of four: https://www.amf87.fr/prestashop/attelages/917-a265-interface-d-attelages-nem363-kadee-4000000012078.html?search_query=Kadee&results=2 However, if filing down and reshaping the prongs on a Kadee #17 works then go for it! - Richard.

Well, I was looking at their page of shipping rates and I couldn't find the UK. That's my excuse anyway! I'm glad you are receiving parcels ok; I've always found them an excellent supplier. - Richard.

Modellbahnshop Lippe are listing them at just shy of 100 Euros with 'reservation possible'. This is of course just fine for you expats, because in the current Brexit fallout the UK remains as just about the only territory in the World they won't ship to! On a more serious note, the Brawa catalogue shows the models with containers of a much more modern design and in keeping with the prototypes. The containers supplied with the earlier releases were rather old-fashioned and to be honest just "wrong". - Richard.

I have barely an inch of space behind the fascia ply here (barely enough room to manoeuvre the display in), so a saw wasn't going to be very practical. I rather like bendy ply, although I do not understand how it works! It ought to be useful for structural things where a curve makes for a stronger job, and for model-making like a curving retaining wall. So far I've only tried the 3mm version. It can shrink lengthwise very slightly. I have three panels of the stuff for my backscene, these were butted up tight against each other last summer, fixed with Evostik Timebond and then painted, and now there are narrow cracks opened up between them. Perhaps 0.2 mm movement on a 1.2m board and 0.1 mm on a 400 mm board, giving a 0.3 mm gap between them. - Richard.

Going back to the Magnorail, I remember the length of the electrical connection to its display is supposed to be limited to a metre or so. Thinking about reliability and reducing the number of pins on multi-way plugs and sockets, I've decided to put this display on the fascia of the layout, not on a demountable control panel. If I paint the fascial black instead of white and arrange for the display to switch off when it's not in use, it should hardly show. The fascia is two layers of 3mm bendy ply glued together and I think it is worth posting this photo because it shows how easily this stuff cuts with a craft knife. No need for a saw, so it was straightforward to carve out a rectangular hole: The hobby room is now out of bounds while fresh undercoat "goes off", hence me sitting at the computer posting a photo of a hole in a bit of plywood. - Richard.

It gets better. This morning I sat down to debug the PIR sensor on my "scenics processor". This processor is another Arduino, but dedicated to driving a text display with a description of the layout plus two animation servos. Well, the sensor kept on tripping, even with me consciously keeping arms and hands well out of the way. Then I saw my cup of coffee on the table about a yard away from the front of the sensor ... - Richard.

Yesterday evening, the laptop fell off the edge of the table (where it was rather precarious to begin with) and landed on my foot which I somewhat stupidly but instinctively pushed out to deflect its fall. The laptop is a ruggedised thing, built like a metal brick, and is unscathed; the foot felt the impact but is unbruised; and the upload cable ripped itself out of the Arduino. This has left the Arduino in perfect working order but without an upload port. Perhaps Fate is trying to tell me something - no more software tweaks! - Richard.

It was incredibly satisfying to see the car run for the first time last week, and watch the log file spooling itself out in real time. Since then I have tweaked some of the details, but I know the code is basically "right" and I can now add or remove features without expecting to break it. Here is a log file from this morning. The system powers up and then the car does three laps. The total of three laps is decided by the software; it gives me either two or three chosen at random. You can see the emulated gear change at line 25 (08:37:04), and the amount of free memory is the same at the end as it was at the start, which is some reassurance of robustness! - Richard. log.pdf

This is my installation of an Arduino to control my Magnorail system. The Arduino is a clone 'Nano' version sitting on an expansion shield in the middle of the photo: This photograph should be portrait format (the lighting will then look more natural!) but whatever I do, the RMweb software insists on rotating it counter-clockwise. Above the Arduino (i.e. "left") is a small pcb carrying a MOSFET to drive the Magnorail motor, and below ("right") is my panel of Veroboard connecting the inputs from the four reed switches. The bunch of cables exiting to the right ("top") are temporary connections for the start button and its associated LED, and the display. The idea is to put a start button and LED on the fascia of the layout. The display will go on a control panel separate from the layout, with a duplicate start button beside it. The MOSFET pcb does not have a flywheel diode to protect its transistor from back emf from the motor, so I put a diode across the motor terminals: Putting it here means I can't lose it or forget it if I ever try a different driver stage. I added the small capacitor too, maybe this will suppress some motor noise. I am convinced, if you want to control an electric motor and it needs to go in only one direction (no reverse capability) the MOSFET is the way to go. I bought five of these boards for £5, this is cheaper than buying the transistors on their own. The Veroboard panel worked out nicely and dropped in to connect to my existing layout wiring: This is my Meccano enclosure to protect the motor from impact damage, and the wiring to three of the reed switches: The software is still behaving itself. The car accelerates and brakes, and the servo changes its exit line from the roundabout, but I need a video to show this. - Richard.

Things are looking up a great deal. Earlier this week, I installed my new Arduino prototype (on the wooden tray above) onto the underneath of the baseboard, and connected it to the four reed switches. I used a separate analogue controller to drive the motor, and proved a great deal of the software, especially its logging and text display. Yesterday morning I realised I had almost certainly destroyed the H-bridge, so I built a very simple circuit using a power MOSFET and a couple of discrete components to drive the Arduino motor. This gave me successful control of the motor, so in the afternoon I rebuilt my Magnorail project to use this MOSFET. Around 5pm, the car covered a few laps under control of the Arduino, and I realised I had cracked the project. The software puts the car into its garage to initialise the system, and sends the car out for two or three laps after the press of one button. The software gives the car one rate of acceleration and three rates of braking, and sets up different exit routes from the mini roundabout. There is even a simulated gear change from first to second gear when the car sets off from a standing start on a lap. There is logging for fault-finding and tweaking, and a small text display showing debug information on one row and a user dashboard on the other. The struggle is over, this is going to work :-) Photos to follow of course, but at the moment the MOSFET lash-up is too fragile to let me move the layout. I can now polish up the software and build the hardware to a more permanent standard. - Richard.

At the end of the day, I will want an Arduino with some kind of software if only to drive the servo moving the Magnorail channel sideways. So I have put my problematic software aside and returned to the hardware side of the project. This is my third version of Arduino hardware for my Magnorail installation: The main differences/improvements are: An Arduino Nano instead of a Uno, because this is so much easier to connect to layout wiring. A separate pcb instead of a shield to drive the Magnorail motor (I could be using a simple transistor here but it seems easier to buy a ready-made circuit with an H-bridge, even though the chain always runs forwards and I don't need the reversing capability of the bridge). Inputs from the four reed switches using dedicated digital inputs to the Arduino instead of being combined onto one analogue input (I have built a new interface on Verobard to use the existing layout wiring). A spare servo connected to emulate the servo hidden under the Magnorail roadway. At this stage, there are only two wires between the Arduino and the layout, these being the connection to the Magnorail motor (the pink and violet wires). I am simulating the progress of the Magnorail chain around its circuit by pressing the buttons on the piece of Veroboard in the right sequence. I have also been pressing the buttons in the wrong sequence to show myself how robust the software will be in the face of electrical noise on these inputs. The piece of Veroboard in the foreground here is my first version using an analogue input on the Arduino for five inputs and is no longer part of the system. The Arduino Uno is doing nothing; I am simply using its breadboard to hold two push buttons. The spare motor top left is from an old Epson printer. This has let me try out the system in my office where I have been writing the software. This motor runs on 12V, which is good (it's the same voltage as the Magnorail motor) but its driving characteristics are completely different, which is not so good. I have let this project become an occasional activity during the last couple of weeks, and turned to sorting out old family photographs. This is fun and a welcome break. - Richard.

Yep. Clyde Humphries (of Magnorail-Oz) has described my Magnorail as a "proof of concept" and I think this is very fair - there is a visible effect from the servo moving the chain but to my mind it's a bit low-key. I do have some functional software in the Arduino, working mostly at a debug level. It sets up the servo for the bike and the car, and can use a display to tell me things I can see with my own eyes like the last waypoint passed by the car and the number of laps completed. It does many things but does not drive the car. Me trying to get a finished system probably needs a fresh and larger Magnorail track, and this in a less confined location, and a partnership with someone with a natural flair for software. - Richard.

I have good and bad news to report! The new prototype with its Arduino Nano and new motor driver circuit are up and running. The Arduino motor made a fair noise being driven by this circuit, especially when starting, and I have figured out how to solve this. I have altered the frequency of the PWM signal I am using to drive the motor, and shifted this up to 32 kHz, way above my threshold of hearing. Good. Unfortunately, I have hit two snags with the software. One is, because my Magnorail track is so short, it is going to be hard to program visible acceleration and deceleration into movement in the small length of road available. A mate who is something of a computer guru has suggested developing an adaptive algorithm. I would want to check and respond to a reed switch whilst applying acceleration or deceleration, and I've not worked out how to start. It may be too difficult for me. Fortunately, this is only a hobby. More importantly, I have managed to fill my brain with so much programming I haven't left any free space to actually think any more. I have therefore put the Arduino project aside, to restart after I have tackled and completed a few jobs around the house. I can't help thinking, it might be sensible to assemble the Arduino hardware onto the underside of the baseboard, and get the baseboard back into its alcove. I haven't run a train for ages. - Richard.

I am developing a home-made controller for my Magnorail installation. I have an Arduino Uno with a display, and a set of five push buttons representing my four reed switches and a 'start/stop' control button: I have written a good sized chunk of the software for the Arduino, but hit a snag with the layout wiring to the reed switches. Essentially, the motor drive shield (sitting on top of the Arduino in the photo) consumes all of the digital I/O pins. I have built a ladder network using a series of resistors to make a simple digital to analogue converter, and this was intended to connect the four reed switches to one analogue input on the Arduino. This circuit is on the piece of Veroboard. Unfortunately, this arrangement works fine on the bench but suffers from a great deal of electrical noise and false trips when installed under the layout. I have worked around many of the false trips in the software, but I get stuck when the code is waiting to see a specific switch and it sees noise representing the value of the same switch. My plan now is to rebuild the controller using an Arduino Nano instead of the Uno, with a completely different motor driver circuit. This will free up all of the I/O pins I could possibly want for this project. At the moment, the motor driver circuit is somewhere in the post. - Richard.

No decent ideas I'm afraid. These are masonry nails so brittle and prone to shattering, and I've put mine into the bin as I've pulled them out of their clips. Brass pins are very cheap in "layout sizes" and I usually use these if I want a pin. - Richard.

I think this comes with the territory for some of us here :-) A working traverser is always going to be an appealing project to many people, and the concept will inspire fresh ideas. I wish you all success with your own solution. - Richard.

When power is restored your traverser will be in one of three states: - at one end or the other, with an engine across the gap - at one end or the other, but free to move - somewhere in between You need a visual inspection to tell whether there is an engine across the gap, and to my mind this asks for a sort of "permission to start" push button to let the DCC system take control. It is good the decoder doesn't lose its status, but I'm not sure whether you can rely on this where there are delicate models involved. If you are going to have such a button, this leaves you to decide whether the traverser controller should also need a push on it when it powers up between end stops. I suppose, the hardest decision is whether your traverser controller should be committed to the control system for the layout. If the DCC system were to provide just a pair of clean on/off contacts to tell the traverser to move one way or the other, then you can build the traverser controller in isolation, away from the DCC controller, give it the logic it needs to restart in a predictable if possibly long-winded way; and test its operation with a toggle switch. - Richard.