Søk etter:

Started 05Apr2026. Updated 12May2026. This note is in group Models

This is sharp and short wye switch #2

See sharp and short wye switch #1 described in My sharp curve wye switch... Observe the very important chapter Read this first in that note – it’s about when such a short and strongly curved switch is not usable!

This new switch is basically built with the same recipe, and much of the detail is therefore not repeated here. It expands MSRS v.5.1 with another side track. See here:

MSRS v.6.0

My Shelf’s Railway Strip (MSRS) is an extension of the previously described 224:[MSRS v.5.1]. Listing of the descriptions at 201:[My Shelf’s Railway Strip (MSRS) versions]

The v.6.0 main purpose is to be able to have three locos (perhaps with a car or two) swap their track positions, between  Li  (Left _inner),   Lo  (Left _outer),   Ri  (Right _inner) and  Ro  (Right _outer).  Ri  with the extension is the longest; it helps  lot. There would be two basic scenarios: with and without the rightmost part inside the wardrobe. Therefore I needed to build a short and max curved wye switch #2.

MSRS v.6.1

Trying to explain the SBB Integra-Signum automatic train stop system system (Wikipedia) with a coil inductance and magnets (which was installed on the SBB Ce 6/8 II and III in the nineteen-thirties) I soon realised that it would be a matter of time before I’d see the handle broken by myself.

So I installed two solid posts, made from stainless steel tubes and 6 mm machine screws which go through the shelf, and there washer and nut. I filed away the screw head’s slot, to make it somewhat assemble a post.

Aside: I think it’s actually easier to explain some more or less fictive working principle, inspired by the German Indusi / PZB (Punktförmige Zugbeeinflussung) system (Wikipedia), which in 1992 I read was installed also on the remaining SBB locos and Swiss track (Wikipedia: ZUB 1xx). I base this fictive system on simply using the 16 ⅔ Hz on the train coils meeting  a matching coils in the rail – shorted («red light») or open («green») – then to detect the difference in current in the loco’s coil. The European Train Control System (ETCS) system (Wikipedia) now is the prevalent safety system in Europe.

Discoveries since wye switch #1

Not ideally curved track?

I have discovered that the curvature of ¾ length of Märklin 59035 rail is a little more straight than the ideal circle arc! (See Wikipedia Circle). The middle of the inner rail it is about 2 mm «too straight», being some 2 mm closer to the center. (Meaningless wording since an arc would have zero «straightness», but that was the only way I was able to describe it.) I have scanned a rail and compared with a perfect curve sector made by the correct radii in my text editor (Pages) for the same rail.

I wonder why Märklin has done this – or how this might be. I have compared with a rail that still remains in a track, with sleepers, with the same observation (Second look, the two rails I have cut are not equal. One is almost correct. May the observed straightness be within tolerance?). If I connect eight of the full 59035, I get half a circle, and it looks like it easily might be persuaded to become a full correct curve of diameter C-C (center-center) of 2 * 1020 mm = 2 040 mm. (I don’t have 16 of them, for a full circle).

I will not bow the rails. (Maybe I will, since one is better than the other).  Meaning that Fig.2, which show the ideal situation, may not map the real situation. I had to choose between correct mapping with faulty tracks or two «faulties». Because, my first switch already may had suffered as a consequence of following the landscape, so to say.

Ideal wye switch diverge more

I discovered that the first wye switch I made is a little narrower than it should have been, by 4.0 – 3.25 cm = 0,75 cm. I will now make it correct, meaning that the two will differ some. It would be ok for my track, but it would make a loco and a car even more vulnerable to hook ing onto each other’s buffers. See Too narrow for some loco-wagon combinations (below).

Sleeper void

My switch’s end piece  is built more like Märklin’s 59092 (R=1394 mm, 15° ends several different pieces) and not like the ends of the obsoleted 5976 (left turnout R=1020 mm, 22.5°) or 5965 (left turnout R=600 mm, 30°). See Some switches out there (below). See figure, where the three rightmost are cropped pictures from the Märklin pages.

I don’t know what a prototype for my switch would look like, since  there is no R=1.02 * 32 = 32.64 m switch out there. When will it go from sleepers that cross the whole structure (double width) to individual sleepers for each branch? There is a huge difference in when Märklin has placed the last with double width. Naturally, in the most curved it’s even before the frog heartpiece!

What should I do with the open space in mine? The «sleeper void»? Could I just add a sleeper that crosses the inner rails? Would it be too toyish? But then, I guess that an axle with much force from might even break the rail on my switch? Guess what. I’ll just build it like the first, since it in my opinion looks nice. Whether I’ll fill the void, stay tuned 1. Update: I’ll look at the spread of the sleepers, and see whether it may be pushed a little upwards. Stay tuned 2. Update: See the new Fig.4. I decided to move the last with double width up. This made the distance between the sleepers 11.1 mm instead of 10.5 mm. The straight track from Märklin has about 14 mm, but then the curved ones varies wildly. Fig.2 is updated. (Another reason for this solution is that I don’t want the sleepers to connect at an angle like in the rightmost figure. Ok for one piece plastic, but since I want the sleepers to mechanically be a skeleton, gluing them together is not something I would even consider.)

Flangeway filler v2

See Flangeway filler v1 (224:[flangeway_filler_v1]). I’ll go for the NEM 110 [2] 1.6 mm (‘H’) value for my second switch. The Märklin rail that I use is 5.2 mm high. This means that I need to fill up 5.2 – 1.6 = 3.6 mm from the flangeway’s bottom. Update: in the final end the ‘H’ value became 1.8 mm, which also goes for a smooth ride.

The switch’s basic drawing

If you, like me, don’t have an A3 printer, here is the A4 PDF with the same.

The text which I have removed from Fig.2 is:

All measures in cm Outer curve rail marked with red dot (inner in switch) Straight (gray) chord between ends of edge is 29.1 cm Rail top 0.25 (shown), rail base 5 mm Inner rail gauge: 4.5. Across rail and 0.5 mm outside of base of rail, since red and black are 3 mm each (to align) (red-black) 16 sleepers, width 0.85, height 0.4 + «spacer» 0.1 = bottom of rail from floor 0.5 + rail heigh 0.5 = total height floor – top of rail 1.0. Märklin 59035 (22.5°) radius of track 1020 mm. This switch uses ¾ length (16.875°). The arcs are mathematically correct. Pull the rails carefully out of the 59035 sleepers. Any rail may differ from the mathematically correct curvature (by up to 2 mm?). If needed bend or straighten to correct curvature carefully with your hands. Upper part of rail drawn. Mathematically correct arcs:  99.5  (⌀ 199.0) black (left rail outer)     ( 0.0 )  99.75 (⌀ 199.5) red   (left rail inner)     (+0.25) 102.0  (⌀ 204.0) theoretical radius of track (+2.25) 104.25 (⌀ 208.5) blue  (right rail inner)    (+2.25) 204.5  (⌀ 209)   lilac (right rail outer)    (+0.25) Pictures will always outrule this drawing! If no A3 printer then export this A3 file as PDF and import into the Pages A4 document and print out as two A4 directly from Pages, then glue them together by the line.

Switching

The animated view

This is my first ever animated gif:

Fig.5 – Two views, animated

Building it

These points do not add up to a complete working description. But add them to the description of switch #1, and it’s approaching, at least.

Photos 1-7

According to 224:[1(DccWiki)] the term «heel» is used for the «arrow» part of the frog, but also about the hinge part of the switch. I will use frog heel and switch heel (or heel spread) here.

1. This time I used small «pillars» to hold the rails. I made them from white POM plastic material. This made it much easier to fasten and remove, fasten and remove
2. Some part of the rail would be fastened on one side only. On some of these bored holes and soldered copper wire, and bent them below the sleepers. Since most of the aluminium have rail fasteners have wires through the rail body, only the sleepers, I thought that this would make it easier to handle, on fastening and removing the switch-to-be time after time. On each guard rail I actually made the wires go though both the rail fastener and rail, on two places for each. I did not want them to be bent on use. (I think, but cannot find any proof, that for switch #1 I did this below almost all fasteners)
3. This time I shaped the frog heel by slowly pushing the parts together as I smoothed the metal (by removing and doing it in the vice). When finished it was just to cut of what was to become the heel. I then soldered the two pieces together, and bored some holes in them, to fasten it on the frog’s base. I tried to use gas soldering because I thought it necessary, but I failed. My small electrical soldering iron sufficed. Metal saw and diamond files were the main tools. Plus some judgement and risky filing
4. Making the profiles for the rail fasteners I had printed out a template, to get the parts of the millimetres right. That document is here (PDF, 1.4 MByte, scaling when printing out: check that 10.0 cm ends up being exactly 10.0 cm). Using my Proxxon «FET» table saw, see 27070.
5. I cut the individual base plates from the strip that I had cut with the saw, by hand. Then filed down the space for the center piece. The base plate build 1 mm
6. Having cut off the heartpiece, ready for the next round, I had made the steel and aluminium work for the point blades, for the switch heel. I managed to find two suitable screws which I could screw from the aluminium base below the point blades, up into the rail, to swivel on. The steel lists «inside» the rails hold the point blades in place. I made them from some spring steel I found in a box. Hard to saw and hard to bore, but fine in the end
7. This is my office’s working space. In addition I have a basement workshop, which is where I keep the table saw

Photos 8-18

8. The guard rails certainly needed some fastening. The base is 1.8 mm instead of the standard 1.0 mm base, to raise the top 0.8 mm above the other rail. The template piece of point 27 I used to define the distance between the rail and the guard. It’s about 2.8 mm. As mentioned in point 1, the guard rails were also fastened with soldered wire. It’s important to make the rails start so early that «the other wheel» soon to arrive at the heart is safely positioned so that it will be taken safely through the heart and not collide with the heel. I think the guards are only needed driving in that direction
9. I built the frog unit from the frog heel heel, a base plate, a plate to hold the wing rails and a plate between the thinnest part of the rails. See Flangeway filler v2 (above). I had to cut slightly in the sleepers the get the exact height
10. This is the frog’s right side (as seen in 9)
11. And this is the frog’s left side (..)
12. Here are the the bottom of the rail and the bottom of the two parts of 9. You can seen that I cut and bowed the nails
13. The finished (but not painted) frog from the top..
14. ..and from the bottom. I mounted the frog on the on the sleepers with screws. The other track parts I mounted with nails that I cut and bent on the underside of the sleepers, see 31. But the frog I thought, would perhaps need to be fixed after the first run test – screws are it. But I never had to unscrew it and file or anything
15. I filed the switch blades, filed and filed and filed over and over again, using the full length of the bent rail. Observe that these are made from the outer rail of ¾ Märklin 59035. I filed until I was sweat enough after fearing that I’d have to go back to start, and the click was gone, from pushing the blade onto the main rail. No click, it just silently slid into it. I did not touch the main rail with the file. Even if it some times was tempting
16. This shows how loose the the blades are, when held between the iron plates and with the screw. But they don’t, in any weird position, fall off
17. The switch basically follows the same scheme as 224:[The switch]. But there are some main differences. I did start of with the brass connector from the discarded 25A plug. I wasted one plug by not being accurate, so I had to make an accurate drawing. The two were leftovers from switch #1. If I were to make a third switch, I’ll have to rethink, because the plug aren’t designed like this any more. Or ask friends if I could perhaps see whether their plug is of the old type..
18. I decided to make the handle and the switch stand much different than with switch #1. This one should not need the repair that the first needed. And it does not need the wire crossing the Y. Photo 18 in particular shows how I made the axle to the correct 3 mm ⌀. Just by filing it carefully and equally carefully rotate it inside that 3 mm hole

Photos 19-28

19. Now the base of the switch stand has parts that both reach (and embrace each other) across the sleepers and thus locks it sturdy. It’s difficult to make the Y and its holder. But I only needed to make the Y twice. The handle weights 5g and pushes on the Y by I think, the half at 45°. Enough to make sure that the blades keep in position, in addition to the friction and the fact that the passing wheels also would push the one needed where it should be.
20. The Y and its holder now are connected onto a slider which slides between the switch stand base (which also holds the connecting rod) and
21. a bottom plate. When I first tried it it felt kind of hard to push. But when I learned that a drop of oil (or less) was needed on the slider’s top and bottom, it feels beautiful. After some months I haven’t needed to re-oil. This is the main difference with switch #1. The purpose of the Y is to allow some movement before the connecting rod is pushed or pulled, and to allow for a greater movement to allow the smaller movement
22. This shows some of the material I used. First 1.3 mm copper, for the Y. Then 1.6 mm epoxy / glass fibre printed circuit board for the part of the connecting rod that only goes between the blades. This is connected by a single nail to the full length connecting rod, which is made of 2.0 mm PCB board. (I keep a box of some of these, taken from units on their way to becoming electronic waste.) This makes it easier to unmount the whole switching mechanism without touching the blades. The handle is also screwed onto the sleepers. The bottom connecting rod also has a tiny handle at the other end, so that it’s possible to switch with the finger if some rolling stock hides the main handle
23. I guess this photo just summarises up much of what’s said above
24. This time I only pained the aluminium fasteners. I did not try to weather the track, simply because the other track isn’t, which is fair enough. Again I used the Humbrol enamel paint #173 track colour matt. (According to a EU regulation from 2022 a component of that paint has been characterised as a worse carcinogen (MEKO (Methyl Ethyl Ketoxime)), so in the future I guess that this paint will come without that component. The one I bought this time was not of that sort; it was oil – not acrylic, as the new generation is is made of)
25. I used my 229:[GwLBay96 baggage car] as a template to make the template to get the switch’s switch rails correctly positioned. With 27 I could now push the wagon (sort of) through the track. I did nit have this for switch #1 – too bad. It made it so much easy. I was surprised to see how much slack there was
26. I also made two more templates for the top of the track proper. I learned that having three of them was reassuring when I bored holes and fastened the connecting plates and nails to the sleeper. It ok here, but still ok there and even also there. Nice
27. See above
28. ––〃––

Photos 29-31

29. Here is the final switch. It ran flawlessly the first time I ran the crocodile or the E71 with the wagon over it. And it has since. Its weight is 201 grams. It is very sturdy. It is actually not possible to handle it so roughly that it gets out of shape, or so that it would twist in any way
30. This is most of MSRS v.6.0. What I hope for next is a small third locomotive, see Some short scale 1 electrical locomotives. Plus som street lamps – even if I have no plans of making it a layout as such. But it’s too dark there when it’s dark outside.
31. This is final wye switch as scanned on my smaller A4 scanner – with some corners lost. You may notice the four quite important wires making every part except the frog having electrical contact with the correct track. The frog electrically floats, it’s neutral, to avoid shorts. Therefore an axle will not get fed at the moment it’s passing the frog, in which case the other axles will. The blades should be electrical through the aluminium base and steel screw etc., but this also is not critical. Running stock certainly are longer than a blade, too

Finally, why?

Because, the switches I could buy would have been muuuch longer. MSRS v.6.0 would not have been possible without these two short wye switches.

It feels like the 70-100 hours work was quite meaningful. Same good feeling as with switch #1 😊