HOx-Trak

[EB421]

Hmmm. Just spent half an hour stuck in one place with a laptop and got to overthinking HO module track.

 

Atlas makes a pre-bedded clip-together track like Kato stuff, "True-Track". Buuuut it's not great. The connectors appear to be way more fragile, which isn't great when your track sections are large, heavy boards. Also, it's code 83, so limits rolling-stock compatibility - which is more important than visual perfection on a system like this. For one thing, I couldn't run my Lima Shinkansen! And worst of all - they only do a 90' crossing; so you can't do a crossing on the hex tile, only curves, straights, tables and points (/switch/turnout).

 

Their Snap-Track Code 100 does have a 60' crossing that's equal to 3" of straight, and 3" is one of the pieces needed for a straight-across on "Solvo Mi Velico"'s diagram. So it looks like the perfect fit, except for the lack of bed. HO Hex-Track will need to account for that in any kind of standard.

 

I think the sub-hex tiles will be a lot more important to HO-scale modules than N-scale, if just for the ease of storage.

 

 

You end up with five possible pieces - a "scenery-only" triangle piece, two quadrangles (essentially a pure-straight and a pure-curve), a pentangle which fits a left or right turnout, and the full hexagon which is only needed for 60' crossings and turntables. It's obviously not as clean as always-hex like the N-scale original, but still keeps the intercompatibility of sections and actually gives a little more freedom - triangle tiles also fill gaps between hexes to make a straight edge, which means you can "pad-out" your layout with more scenery tiles that can also be mixed-and-matched to give variety. Which is great, because the dioramas and scenery are the fun part, right?

 

Definitely playing with this!

 

Edit: I'm missing the obvious solution for the trackbed, which is just to print it. That way it can easily be made to a repeatable standard, better joiners can be designed, and it should still be easily hand-replicable for those without access to a printer. Plus; using PLA it's basically carbon-neutral, which is nice when talking about using a bunch of plastic for something.

 

And doing it this way means that people can use Code 100 flexitrack to mimic the Atlas stuff if they don't have access to it - though they would still need to find 18" turnouts from somewhere.

Edited February 10 by EB421
Extra content

[gibet_b]

I played with this idea some month ago, and a good alternative of Kato Unitrack in H0 scale is the Trix C (same as Marklin C, but for 2 rails).

[EB421]

7 hours ago, gibet_b said:

I played with this idea some month ago, and a good alternative of Kato Unitrack in H0 scale is the Trix C (same as Marklin C, but for 2 rails).

 

I don't suppose you did the math on this, did you? I took a quick look at the Trix track pieces:

 

https://www.trix.de/en/products/trix-c-track/programm

 

The curves line up, including the points (5.7 + 24.3 = 30) - but they don't list the sizes of their straights so I can't work out if there's a kit to make up a perfect straight section. There's a bigger problem though (at least for me): "The cross section of the rails is only 2.1 mm / 0.083" and is the same as Code 83."

 

Most of my stock won't run on that. No 60' diamonds either. It's all the same problems as Atlas True-Track, other than the fragile clips. Pity, because obviously 437,5 mm < 457.2mm; which (again, obviously) means smaller tiles; and it's easier to get here. Still though, I'd love to know the exact height of C-track from the bottom of the road to the top of the rail. It would be neat to try and match the height of a custom printed Snap-Track roadbed so that Atlas and Trix modules could be linked cleanly.

[gibet_b]

Here you have all my research :

 

The hex is 36.1 x 36.1 cm.

[Ander88]

Sorry if I am missing something, but those are not hexagons, right?

[gibet_b]

14 minutes ago, Ander88 said:

Sorry if I am missing something, but those are not hexagons, right?

 

OMG, you're right ! Poor me 😨

[EB421]

19 minutes ago, gibet_b said:

 

OMG, you're right ! Poor me 😨

 

I mean it still works as a concept, and you can use 45' (2x22.5') turnouts instead of 60' (2x30') which means you could actually use KATO Unitrack! But it won't tesselate perfectly like Hexagons do. 🙂

 

Hmmm. I still think my Snap-Track and 3D-printed bed is the best way to go for me. For why; I'm looking at all these things though and seeing that every single one is Code 83. I kinda wanna get a piece each of HO Unitrack, HO True-Track and HO C-Track and see how the base to rail height differs.

 

If anyone has one of these and a pair of calipers and can save me 20EUR by measuring it, that would be appreciated. 😄

Edited February 11 by EB421

[EB421]

On 2/10/2025 at 8:12 AM, gibet_b said:

I played with this idea some month ago, and a good alternative of Kato Unitrack in H0 scale is the Trix C (same as Marklin C, but for 2 rails).

 

So I'm still playing with this and decided since I'm in France I should try it your way! So I have enough C-Track for four experimental 502mm HOx-Trak tiles.

 

 

Next challenge: make the tiles!

[EB421]

Ok, big update: My experimentally-derived numbers were close but no cigar. Pity, since it means some of my track is no longer useful - but I'll find that a home. I went on the internet and I found this: 

 

https://www.reddit.com/r/nscalemodeltrains/comments/1giwh00/hextrak_tile_variations_for_216mm_single_track/

 

Where someone worked out the ACTUAL math for this, which gave me a inradius of 505.181mm. And boy do those 3mm make a difference in both helpful AND annoying ways. Helpful first: It now means the tiles are way more accurate to track length, with straights made up of 64.3+64.3+188.3+188.3 = 505.2mm. Pretty much perfect, and means useful stuff and unuseful stuff.

 

Because track is now in even numbers, we can now make a custom turntable! It's a BIG turntable, at 188.*2 = 376.6mm = 32.76m = 107.5ft. Not the biggest (some in the US are about 130ft), but way bigger than most (about 70ft to 90ft). You can sub in pairs of 94.2's instead of 188.3 though within reasonable tolerance, and using 94.3 access tracks, you get a 64.3+188.3+64.3 turntable of about 90.5ft equivalent, which is much more prototypical. You could probably also get away with using the Trix C-Track turntable, which (with two 64.3mm access roads) is only 1.3mm too big, or 0.65mm on each side. So if you want to spend 500EUR on a DCC sound-equipped turntable, crack on!

 

The un-useful stuff - my 236.1mm segments no longer fit on the precise tiles. Which is moderately annoying, but I'll find a use for them. The much bigger problem is if you want to run multiple lines. Since Trix and Marklin (did I mention this also works for three-rail?) make both 360 and 515mm curves, you CAN in theory assemble a well-aligned triple-track HOx-Trak module including curves - heck, you can do ridiculous things like this:

 

 

Entrances on four sides, with two bypass tracks, four buffer stops and a Wye. The problem comes when you try to switch tracks.

 

 

There's no way to cleanly get from one track to another without making custom pieces.

 

 

The two possible pairs (you need both for a double switch like this) are 13+51mm and 18+46mm. All very doable with flexitrack, but means doing multitrack is more of a pain than I'd like. When I get 5, I'll model some trackbed-fillers to these sizes, but I'll need to pick up some non-trix Code 83 to get the height right. Based on it's ready European availability, I'll go with Peco.

 

I drew out (in 3-track form, just strip the extras for a classic Hex-Trak vibe) the possible combinations:

 

With the exception of the irritation on the interchange points, I think that's honestly pretty good. As a side note, on the rectangular part-module, the triple-straight leaves 50mm on each side - so I think unless you're really big on the triangular tiles that's not going to be your bread and butter. Still, a double track piece leaves 125mm on one side, or nearly 44x11m to scale. Plenty for a halt, since (as an example - GIRT7020 Issue 2), a UK platform need only be 915mm tall and 3m wide. (For an express. For sub 100mph lines, 2.5m!) As a side note, I now also know that (most) UK platforms should be between 73 and 74.5cm away from the nearest edge of the adjacent rail. 8.6mm in HO! I should really check that against other loading gauges before I implement it. Regardless, 11m - 3m = 8m, which is wider than my house is. Pretty manageable.

 

That's it for now, unless anyone here works at Marklin/Trix - in which case, please can I have tiny 13, 18, 46 and 51mm segments made?

Edited March 3 by EB421

[brill27mcb]

Welcome to the frustrating world of section track design theory! You can see that it takes a lot of thought to come up a family of straight track lengths, curve track radii and degrees, turnouts and crossovers, etc. that work well together in a lot of situations. Imagine having to dream up the dimensions as the original designer at Kato, Tomix, Maerklin, etc.

 

Rich K.

[EB421]

4 hours ago, brill27mcb said:

Welcome to the frustrating world of section track design theory! You can see that it takes a lot of thought to come up a family of straight track lengths, curve track radii and degrees, turnouts and crossovers, etc. that work well together in a lot of situations. Imagine having to dream up the dimensions as the original designer at Kato, Tomix, Maerklin, etc.

 

Rich K.

 

Being fair - they did a pretty good job; the system works pretty much perfectly (the turntable is 1.3mm too big, that's all) on the tile layout - even the points (switches). It's only the crossing tracks that don't work - and only because I need to be unrealistic and keep a consistent distance between parallel straights AND curves.

[EB421]

So for a bit of fun, I tried to model this for a 3rd or 4th radius centre-rail, as to avoid ridiculous 360mm 1st radius in exchange for a little extra size.

 

In 3rd radius, the switching sections become possible. You can EVEN use the double-slip, which can't be used on an R2 HOx-Trak because 171.7+171.7+188.3 > 505.2mm. But the trade-off is you can no-longer use the Turntable. For an R4 centre everything is just kind of broken.

 

Then I took another look at my slip math. First up - I'm an idiot. Two of the small pieces are unnecessary, because 236.1 and 171.7 fit in those spaces. The two remaining gaps are roughly 82mm and 145mm, but they're edge pieces so we need to add some real track to fit them. Guess what! If you subtract 64.3 from 145mm, you get roughly 82mm. So. What we actually need is two of a single type of missing piece (And i did the math properly for this) - 80.8mm. And if you want to do a half-straight? 171.7+80.8!

 

In practical terms - I still need to 3D print a section, and it's (bless the gods of Math) ....16.5mm long. I need a tracksquare! The reason is that one of the 80.8mm sections is edge-adjacent, so needs to be a real Trix/Marklin piece (depending on your system). Since there's no point designing two types of part if one will do... the simple answer is to use a 64.3mm piece, and make up the gap.

 

Part of me wonders if I draw up diagrams for all this and include the math, they might actually consider making it (80.8mm, not 16.5mm!). The other 99% me realises that's silly, and is gonna go printy. Unfortunately, I checked my local store's stock on their website yesterday and they're out of Peco Code 83 flex-track. And the place that seems to have the best stock of Trix track doesn't carry Code 83 Flex. So; I've put a stock notification in on my local place's website and we'll fall off that bridge when we trip over it. I'm not doing a separate order.

 

The basic plan is "Replicate the dimensions of the road bed, match the rail heights, and approximate the coupling mechanism so it friction-fits". I'm not actually going to try and match the rail joiners or electrical mounts - Four spade connectors, 20cm of wire and a dab of solder will perform no worse and is still vaguely in keeping with Trix's rustic aesthetic. For Marklin - sorry - anyone will have to mix their own 3-rail tracksquare. For later, anyway.

[EB421]

Properly wrecked myself a couple days before I posted the mention above that I'd gotten enough track for a few test tiles. Was hoping it would heal quickly - it isn't. So; not been doing a lot project-wise lately. This weekend though, I got the printer running - someone on Thingiverse had made a handful of custom non-conductive C-Track tiles (interestingly, not an 80.8!) so I had a bash at customising their work to give me my missing piece and -

 

 

Here's what those look like - and as a result, here's three single-line complete HOx-Trak segments. Mind you, please note the 80.8mm isn't necessary for a straight or turnout tile - due to my previous math error I bought some of the wrong pieces, and the 80.8 happens to add up to the correct length with some of them. Side note, (my previous half-straight math) 171.7+80.8 = 252.5 = 188.2+64.3. So you don't actually need 80.8's for those either.

 

The person who modelled the short straights also modelled Radius-1 and interestingly "Radius-0" curves - which I found appealing as it potentially allowed for a five-track curved tile again, but it's worth remembering that A: their segments are inert and B: the minimum radius for (for example) a Trix Class 66 or Gresley A3 is 360mm. Ok, fine - just means you have to stick to 4-wheel on there... but unless all your curves go in the same direction, you're mixing your largest and smallest radius in a way that is pretty wasteful of the former. It's not something I want to encourage in any hypothetical standard - so my view to that mind is kind of "if you want to put track there, it can't protrude over the edge of the tile so that it doesn't violate compatibility". Means people can do custom stuff like (relevant to our interests) add a 12mm H0m track to the edge of the tile for the sake of doing a Shinkansen + Mainline.

 

While spitballing multi-gauge stuff, I was thinking about how best to do H0m/H0j. As far as I can tell, the only manufacturer of clip-together track in "TT" is Tillig, their "Bedding-Track" series. And what do you know? It's actually H0m. I did the math - it IS possible to get both a curve and a straight out of their system in 505.2mm - to within half a mm on each side, anyway. But it's pretty horrendous. The straight is 166+83+36.5+36.5+36.5+36.5+36.5+36.5+36.5 and the curve has a pair of 36.5's on either end and only works with radius 310. For just making TT-Tiles rather than keeping a standard H0 tile size - best match is 396m radius which is 0.2mm under each side with 166+166+41.5+41.5+41.5

 

For now - my to-do list on HOx-Trak:

 

• Work out the maximum loading gauge for: OO-British, H0 British, H0 European, H0 Japanese and H0 American.
• Extrapolate that into a design for a 3D tile - with the tile stackable in multiple heights.
• Contact Marklin about the 80.8mm track segment*.
• Order some extra track.
• Finally make a test tile.

* So - the 80.8mm segment. Is it necessary? No. Would it save me a lot of stress if it existed? Also yes. As it stands, the only possible straight tile uses 2x 64mm and 2x188mm; and you can optionally sub 2x94mm in for the 188's. With an 80.8 tile, a second straight option opens up - the 236/188/80; and a third - the 64/188/171/80. At least one of these two is needed for any two-lane or three-lane switch, and since they both rely on the 80.8mm segment; it means without that part - HOx-Trak can't have lane-switching. But wait, you can just have a curve out without a second switch, right? Nope. Still needs the 80.8mm track (there's almost certainly some magic mathematical formula that perfectly explains why - but I have no idea what it is). 80.8mm is the curse of HOx-Trak. One last question then. Do I expect Marklin to actually start making them based on my request? Of course not - but if I don't ask, then I didn't even try. What I'm probably still doing is plan A. The biggest hold up on that (injury aside) is that I'm still short of Code 83 Peco locally.

 

** A bit like track radii, layer 2 will be the default - but it will be an intended feature that mixed-height and low-only tiles can be built as well - the reason I'm sticking with a default minimum height (as opposed to Hex-Trak going flat only) is that I want to have the ability to do tiles with a well-turntable (especially given that Trix and Marklin make one that fits) or a traverser - as well as have things like subway stairs going down, and rivers or canals, and have a metro without all the mainline trains turning up on it whenever you join someone elses' tiles.

 

One side note on my 12mm experiment - I want to make this a universal tile system for anything I could be tempted to do in 3.5mm. In theory this means doing TT-gauge Cape and EM-gauge Irish on the same size tiles, but A: no models exist to run on the latter and B: those curves are getting very tight. In practice it means doing remote control cars instead. To-do list is to work out the best way to make road-based scenery tiles join neatly into rail HOx-Trak so I can drive around a W Beetle in 1/87. Bonus points for Magnorail (or clone) on one or more tiles so there's traffic to contend with!