Guy_Walker

Good link -I’ve not heard of Ig Nobles before -I’ll keep a watch out for next years topic ;-) Btw Peter, if you don’t have enough of the old (sprung) points to make a counter layout, then you can emulate them using new (lazy) points. However these need to be gang linked in pairs so both points switch together. This is an important train track component. So much so that I’ve used it as my photo (see left). Here is the Duplo layout: and if you run out of cross-overs then you can mimic these too: The train is automatically routed to the opposite side. You can also construct a cross-over with old type points. This might help anyone building a perpendicular cross-over :-) I’ve added a page on equivalent circuits here. The DigiComp emulator is now finished a few pages further along. Now working on a multiplier layout -not as easy as I had expected...

Yes - web security has stifled so many programmers. Not sure what the answer is. Anyhows, back to trains, I’ve redrawn my previous counter circuit as a Duplo layout, in case anyone ever wishes to build it. Here is a single stage: And this is all 3 stages. It needs 10 points (3 new and 7 old) It counts up to binary 7 (111) and then rolls over back to 0 (000). Add more stages to count higher. The loop on the right just serves to return the train. The solid red points are ‘new’ lazy points. They show the data count. The others are ‘old’ sprung points. I’ve drawn the diagram so that the train always follows the track with the red tongue. The layout is APV. Maybe this answers cortex31 request for a “train that will go everywhere by itself” You could add this layout to any existing layout to display the number of times the train had passed a certain point.

oh OK hold on -the terminology has already got confused !! I’m using ‘remember’ to mean the switch remembers the last line the loco arrived on. It has a memory - which is important for computing. The new switches do this. Whereas you mean the switch ‘remembers’ how it was set by human operator. Which is what the old ones do. I can see both views. Maybe best if I avoid the term ‘remember’ to describe old and new switches. The Fibonacci layout requires around 60 points plus around 40 points per stage at a quick count. The layout is modular so stages are just duplicated to allow for larger binary numbers. The complete image will bloat this post badly. But here is an example of a counter function. The train increments the counter each time it visits before returning back along the same track. It needs 1 new and 3 old points per stage. All new points start in position ‘0’ as shown and are switched by the train to show a binary ‘1’. Old points always switch to the straight track. Track 'e' indicates an overflow error. I played around with LiveCode for a while but then stuck with Javascript because I could build code straight into a web page. You might like this. http://www.cr31.co.u...wang/stage.html If you select ‘Rail’ from the ‘2-edge’ pull down menu you get random rail tracks. Click the lower ‘Maze’ button for a more joined up layout. I need to add wandering train sprites. I will tidy up the complete Fibonacci layout and upload to the website. A 5 stage circuit will fit on a web page. I really want to add some explanation as to how it works, otherwise it can all end up a bit meaningless. UPDATE: Uploaded triangular and Fibonacci calculating layouts as well as a random number generator to cr31 site. Use the above link and click the 'Trains' tab. The DigiComp emulator will be added soon.

Ah ha - thankyou Robert. I think I remember reading about old points some time ago This makes the terminology kinda neat. We’ve got ‘new’ ‘lazy’ switches that remember and ‘old’ ‘sprung’ switches that do not So no need for elastic bands. Just use the old type of Duplo points to make large APV layouts BTW if you mix old and new points you can build computational traintrack layouts. I have some track layouts that can calculate triangular and Fibonacci numbers. Takes quite a lot of points though…

An APV layout was one of my dreams too, but no it cannot be done However, you do suggest the way forward… You need to make 'Sprung' points In theory … Duplo points (which are correctly called Lazy points) can be modified. Some kind of spring or elastic band needs to be added to keep the points in one position while allowing the train to pass over undisturbed. The points move just enough to let the train pass before springing back. In practice it’s difficult to get right. But I feel sure it could be done. Here is a circuit using only Sprung points. The spring ensures that the train only exits on the straight line, never the curved ‘branch’ line The train joins the main loop and then visits each ’sub layout’ 1, 2, 3 in turn. The train always returns from each sub layout back along the same track. There is even a bell to alert the operator to the start of the next loop :-) I think you can see the similarities with a computer program, where each sub layout can then ‘call’ other sub layouts etc etc. So you end up with a multiple branching tree-like structure So this technique will allow you to build large APV track layouts...

I did investigate this problem a few years ago. The circuit you describe is this: The train traverses all the track automatically, without human intervention to switch the points. I call this an ‘All Points Visited” (APV) circuit. The train sets off and covers all the track, visiting all the points, eventually returning to its starting position. As Robert says this is quite a different problem than APR (All Points Reachable) As far as I can tell, no other circuit than the one above exists. Adding more points cannot create an APV circuit. Either: 1/ The train enters a trap. At best this will require human intervention to switch a point and allow the train to ‘escape’. The circuit may be APR but not APV. Or the train does manage to traverse a large loop but 2/ Some points are never switched by the train. They will have track that is never traversed. The points and associated track are redundant and can be removed. The complex circuit quickly reduces to the layout above. Large APV circuits can be built but you need to also use some different types of point (sprung, flip-flop etc)