Maximum slope

Thank You Markm.

Some problems are not yet solved ... In France we have a proverb (by Jean de La Fontaine) : "Don't sell the bear's fur before You killed it". So my layout is in the state of a pre-pre-project. The first problem was in fact the slope ... now solved !

Now there are still several other problems remaining ; first the cutting of the supports for the curves : I will probably try to let a cabinetmaker make this, because he can cut all the curves (of 5mm plywood) at the same time by assembling all the layers together, in my design there are only curves with 400 mm and 500mm radius. The curves will be "flat", it does not make a significant difference for the appearance (I have read the topic about Rokuhan tracks) but a very big difference in precision of laying (including the "helicoïdal" transitions to straight portions of tracks ! )

Concerning the nails, I found some made of brass instead of steel, so no corrosion can occur, the same diameter as those by Märklin (0,5mm), the length is 10 instead of 8mm (no problem) ; they are made by Amati, Italy, for historical wooden ship models (together with the tool to set them) and seem to be strong enough, as I tried to bend one between my fingers. As I wrote, the noise of the trains is no problem for me.

And there are some other problems with the electronics (in my case, DIY) : some dealers continue selling the items they have in stock without any warnings, and if one looks to the datasheet on the net, he can see that some components were declared as obsolete at least 5 years ago ... No comment, perfect to get spare parts in the future ! But the electronics define the "cuttings" in the track, so I must know first what I will use. But definitely no DCC !

As You see, there is still a big amount of preliminary work to be done.

Thank You for Your answers. Concerning the problems with temperature changes, I have read on other forums that a gap of 0,75mm per meter is recommended for nickel-silver rails (when the layout is "cold" ) ; so I think 0,5mm per track section (660mm) would be ideal at every rail joiner, it should not cause any problems (each section with its own feeder, all connected in parallel).

I think that block system is interesting on a section with 2 tracks, I built it into my former N-layout (4 block sections of 1,5m each per track) ; but with single track, it seems to be less interesting ; before starting any train in opposite direction one must wait for the whole track to be "empty" : so if block system is used, the main danger (with an automated and "optimizing" program) is to use immediately a free section to send a train from the station to the track, and so to create permanent trafic in the same direction !

For automation, I will try for the turnouts the so-called "capacitive discharge" with relays, they can switch the power to the corresponding track with the remaining contacts. I think I will run only one train at the same time. For the power, I will use PWM with a 1A transistor, polarity change via a relay. The PWM signal can be generated by the main microcontroller of the system. To run 2 trains at the same time, I found out that I would have to use 4 supplementary (small) microcontrolers, one for the PWM of each train (to perform it with 1 microcontroller, at least 2 interrupts would be necessary to manage the progressive acceleration-deceleration, which is not the same for all trains, one value for passenger trains and another for freight trains) and then there is the problem to switch the power to the different sections (preferably the PWM signal with a booster+relay per section, in my case 7 items) according to the position of the trains. And to allow simultaneous train movements in each station, the turnouts on each end must be independent ! So I calculated :
- For one train only, one microcontroller, 6 relays with 4 circuits and one with 2 circuits (unfortunately 12v relays with 6 circuits are obsolete, one with 4 circuits and one with 2 circuits could be merged).
- For two trains simultaneously, 1 main and 4 secundary microcontrollers, 7 boosters (amplifiers) with a 4-circuit relay each, and 6 relays with 2 circuits and 4 with 4 circuits for the turnouts. And a serious risk of short circuit in case of a "bug" or a crash of the main controller. For me it's what I call a "gaseous plant".

I think seeing one or two trains running does not make a big difference in the appearance of the layout, but it would make for the complexity (and non-reliability) ; one version of Murphy's law says that if You are looking for trouble, eventually You will find it !

Alberich,

Since our last posts crossed, may I give you some thoughts using my automation project as an example. I'm using an Arduino with the power shield and DIY breadboard shield. The power shield will give me 2 independent power channels with direction control and if I want to use it, current monitoring. It seems that using PWM on most microcontrollers uses up all the interrupts. So my DIY part is to build status registers for the layout: edge-triggered latches to monitor track blocks and two registers to drive the layout, one for track control and one for scenery control. All input registers trigger a single interrupt. It's sort of an 8255.

I think you might to able to use a similar scheme for your layout. With the processing power in microcontroller these days, I find it difficult to justify multiple units. Another possibility to consider is something like the Parallax Propeller microprocessor which is essentially multiple controllers and a supervisor in one chip.

Finally, speaking of change, I just got a note about a new Arduino using the Intel Curie: 32MHZ plus Bluetooth.

Mark

Sto_o_o_o_ppp !

1) No DCC please.
2) My old N-layout was sold for buildings and rolling stock a few months ago, after some years of inactivity. This was the example of what I should NOT do again : when it worked, it worked fine ; but if not ... much time to find and solve the problem. In 99,8% this was caused by an Arnold turnout (37 were installed) : either it was locked, the contacts didn't switch off, it became hot, and overheating destroyed it ; or the contact for the power supply to the tracks didn't work properly. Or some dirt on the tracks (a drop of sweat was enough) triggered the detection of a "ghost" train (too sensitive, but necessary to detect freight cars with graphite paint between the wheels). And measuring frequently voltages under the layout is no problem for a youngster of about 30, but todays I am almost 64. I think You understand the problem ... My new version for the detection system (lower value for "the" resistor) will detect the locomotive and nothing else !
3) The system of "source transfer", switching the power to the correct section according to the position of the train, was a technique already described in some publications in the 80's, in conjunction with the block system ; the main difference was that it was performed via relays, and no amplifiers were needed because the output of the transformer was switched directly ; todays it would be possible to select the proper PWM signal for the final power stage via multiplexers at CMOS levels, under control of the main microprocessor.
4) I am not afraid by multitasking programming : One task is affected to each train, with a pointer to the present state (via a CASE instruction) ; when the corresponding condition occurs, the pointer is incremented, an exit from the task is performed, and the next phase is selected when the task is activated again ; the tasks are called cyclically and communicate via common variables (occupation state of the tracks in the stations, for example). I used this style of programming in the past (in Basic on the good old C64, R-I-P ! ), to manage the different phases when a freight train entered the station, then pushed back to the depot, a turnout was changed and another train came into the station and was started instead of the first one, the next time when it was possible (I don't know if my explanations are clear enough), so the freight trains were used alternatively "between" the passenger trains.
5) Perhaps the "gaseous plant" would work ; but I think the possibility of using 2 trains simultaneously instead of one alone is not worth the supplementary complications in hardware, wiring and programming.

That's all Folks ...

 

About the complexity of the control system for 2 trains simultaneously (without DCC), and the associated problems, I gave the details of a comparison in a former post in this topic.

Because we are all speaking about our former "career", I can say that I began with an Univac 1110 system (and perforated cards, Fortran V) from 73 to 77 as a student, then I worked as a maths teacher for more than 15 years before I came back to computers again, network administator with Novell and then Windows servers, and worked for 17 years at the "academic school's administration" (in french "Rectorat") in Strasbourg as a software developper with Excel VBA (because it was available without any supplementary costs ! ) and Data Base requests developper with Business Objects (now owned by SAP). I retired in 2014. Before I got the C64 in 83 I used the Kim-1 and the Pet-2001 (with the "Datasette" ) too, and they allowed to develop some experimental sequencers for minimal railway layouts, to test if there were no risks of crashes due to electrical interferences ; I remember that I crashed the Kim once, because the line to the turnouts laid unfortunately ... over the processor ! I used the IEE-488 interface only for an Axiom "electrostatic" printer, this was the best a hobbyist could afford, a Centronics 7 pin printer was more expensive than two or three colour GDI laser printers today ...

I think that the main advantage of running only one train at a time is that there is only one traction current source (with stop sections simply switched off, unfortunately that system excludes multiple traction ... and a train with 2 Budd's ! ) for the whole layout, and that seems for me to be the safest way to avoid short-circuits by countiguous sections of opposite polarity ... due to a programming bug !
About multiple traction : I remember that in the 80's Atlas (and perhaps Lima) made in N-scale dummy versions of the locomotives and the B-units, without a motor. Why not in Z ? This would simplify some technical problems ... and at a lower cost for multiple units !