In lieu of actually building something while I wait for moving day, I've been playing around with track layouts (more specifically, track elements for a layout) and I've noticed something off about the Rokuhan 90 degree crossing. I've tried to replicate the 'straight zones' shown in the 90 degree crossing instructions: R025 90 deg Crossing track instruction manual.pdf (627.66 KB)
for 195mm curves in SCARM, but when I do so I get a gap like so: My best guess is that SCARM is off, likely because the 25mm crossing length given in the instruction manual isn't quite right. But the straight zone length given in the instruction manual of 190mm for 195mm radius curves doesn't seem right either, given that it works out at a crossing length of 10mm (assuming that 195mm rad for the curve is taken from the centre of the track, whick it may well not be and the photos of the crossing on the Rokuhan website look larger than that.
Has anybody actually built this track arrangement, and gotten it to work?
I don't use SCARM, but I would wager that if that is all the gap it is showing, you should be more than fine there. I've had gaps like that in XTrkCAD and there was zero issue with it in practice since there is always some leeway with how the track fits together. XTrkCAD has a "slop" value but it doesn't always seem to work so I stopped worrying about small gaps like that.
Frank I appreciate you pointing out Cubbins blog post. But the real issue is understanding the limits of using sectional track in CAD.
Each piece of sectional track has a bit of wiggle room in it's placement. Unfortunately no CAD program I'm aware of has a "close enough" option that can replicate this. So we can build many operational layouts that will not work in the CAD program. No fault of the program. I've found one can get about a 1mm and 1 degree wiggle per section. With a bit of experience one can adjust CAD components to get the wiggle.
With published track plans, it's helpful to understand the geometry as many will assume wiggle room. For the example that started this thread, the length of straight track to intersection is equal to the radius of the curve:
Since the distance to intersect on the crossing is 12.5mm, there is no accurate solution of straight sections that will work. Although it wasn't mentioned, it would appear that the gap is about 5mm, which can be "wiggled" out of the 5 curve sections. To represent the figure "8" in CAD I would suggest using a straight section of length R-12.5mm.
Once you get away from ovals, these sort of problems will crop up in CAD layout of sectional track.
Unfortunately no CAD program I'm aware of has a "close enough" option that can replicate this. So we can build many operational layouts that will not work in the CAD program. No fault of the program. I've found one can get about a 1mm and 1 degree wiggle per section. With a bit of experience one can adjust CAD components to get the wiggle.
XTrkCAD has a wiggle room setting but it is a little finicky. It seems to work well or not work well with every revision. I typically ignore gaps as long as the angle looks reasonable. Then I will go back once the plan is "finished" and try to close them.
Doug thanks for the info about XtrkCAD. I've been using 3rd PlanIt software for a number of years now, but maybe I should look into other packages.
There are actually a number of situations where CAD and reality don't quite match up. If one wants to model track easement (the prototypical spiral of curves) using varying radii curves there can be problems. particularly when using 30 degree curve pieces. A siding (or "Y") can be built perpendicular to the main line track, but is hard to draw.
Of course I have a certain advantage over most people in that I learned CAD using Computervision and Calma systems (they're in Wikipedia) and with them there was always a good deal of thought necessary to prevent drawing oneself into a corner.
Hi all, thanks for your replies. Given the need to accommodate 'wiggle room' when laying set track I'll definitely do some experiments before nailing anything down. I can see issues with kinks being created that lead to poor running and jerky movement, and gaps in the premoulded ballast looking less than great.
You make an interesting point about CAD and geometry Mark, I work as a support engineer for a SOLIDWORKS reseller and I quite often find myself having to explain that some bit of geometry a user is trying to build is in fact impossible.