Kadee produce a useful application note for their uncoupling magnets, from which I quote:
"It is highly recommended to install the uncouplers in a straight
and level piece of track, not on a curve or a grade. A curve
would not allow proper alignment of the couplers to open and a
grade would not allow the needed slack between the couplers".
This is sound advice, and especially pertinent for Shelf Island, where most of the track seems to be curving or climbing or doing both at the same time. A gentle curve may have no effect on uncoupling for a short-wheelbase wagon, but will prevent uncoupling (and coupling) for a bogie vehicle with a frame-mounted coupler.
Lightweight Rolling Stock
It is important to accept, the magnetic uncoupling features of Kadee couplers are designed to work with rolling stock conforming to the NMRA recommended practices for car weights. These practices are in NMRA RP-20.1, and in essence the expected weight of a wagon or coach or whatever should be a basic one ounce (28 grams), plus another half an ounce (14 grams) for each inch of body length. The lightest imaginable American car weighs 115 grams. Many British wagons are simply too short to have enough weight, and too free-running to let the magnetic uncoupling feature work properly. The larger Kadee uncoupling magnets especially can cause these wagons to lurch and appear to try to centre themselves or their couplers over the magnet.
One solution is to increase the drag of such wagons with a small piece of sponge glued to the bottom of the floor and rubbing gently against one axle. This will reduce the free wheeling of the wagon, but I haven't done this on Shelf Island because the 1:20 gradients are bad enough without making the trains even harder to haul. It may be possible to bring the weight of some wagons up to 115 grams, but this is difficult with open wagons, and to my mind it places an unwanted load on the two axles.
It is well worthwhile trying out an experimental uncoupler before committing to install them around the layout.
Locations for Automated Uncoupling
At the moment, Shelf Island has four possible destinations for wagon loads: the processing plant at Fairport, the dismantlers, the quayside and the fiddle yard. Uncoupling will happen at all of these. In addition, it would be good to exchange wagon loads at the switchback beside the passenger platform. I want to arrange for magnetic uncoupling to suit popular train movements.
Uncouplers at the Processing Plant
The operations at Fairport are too varied to know where to install an uncoupler, and half of the siding is on a 1:50 gradient. Because of this, I haven't tried to install any magnets here.
Uncouplers at the Dismantlers
The track for the dismantlers has one turnout serving two sidings and the logical place to put an uncoupler is before the heel (blades) end of the turnout. This would allow spotting into both sidings.
The bridge before the turnout is on a relatively gentle transition curve between a straight and an 18-inch curve, but also on a gradient which is a transition between a 1:20 climb and level. It is probably the worst imaginable place for a magnet. I tried three locations along the span of the bridge. The location furthest from the turnout worked perfectly for one wagon, and never for any of the others. The location nearest the turnout gave guaranteed uncoupling for all train movements in both directions, and the location in the middle of the bridge rarely worked at all. I did think about lifting and realigning the track to make the gradient steeper and the bridge level, but this would spoil the look of the track - I think the present track has some aesthetic appeal and I want to keep it.
Moving towards the level track before the turnout, I tried two small neodymium bar magnets, one under each rail and opposite each other. I concluded that these would uncouple a wagon, but not necessarily set the couplers into their delayed positions, and so these magnets were not reliable enough to serve the sidings:
However I have used this arrangement elsewhere - see later.
I settled on a Kadee #308 under-track magnet below the straighter part of the turnout, and I have written up the installation separately. This will let me me spot a wagon like a tanker out of arm's reach, at the far end of the siding beyond the flat crossing. With hindsight, a hinged under-track magnet might have been possible closer to the heel of the turnout, but an installation here would mean reversing the installation of the Tortoise motor and I really cannot face doing this - it would more than likely interfere with another motor below and it's not worth the effort. I think getting operation on both routes of the turnout (however close) would be difficult.
The shorter siding here can only ever hold one wagon, and so the requirement is to uncouple on the spot, and be able to push the wagon away from the magnet to couple up again. Here I have used a pair of neodymium bar magnets. I used a scrap of balsa wood to align the magnets before I set them in a bed of glue. This has cost me one sleeper, but I can reinstate something cosmetic or hide this area under the ballast:
These magnets seem to be about the minimum length for straightforward uncoupling with good slow running. Their dimensions are 2.5 x 3.5 x 10 mm, and so they fit neatly under the rail of Peco track. I should mention, all the sleepers on my layout are spaced out to represent a scale 30-inch pitch at 1:87 scale.
These magnets would work better if I had offset them to match the offset of the coupler trip pins, but having Araldited them into place I cannot work out how to move them.
Uncouplers at the Quayside
If a train leaves the fiddle yard hauled by the loco (and not propelled) then the train will reverse twice and the loco will haul the train into the quayside. The same sequence happens if a loco hauls a train down the branch from the dismantlers to the quayside. It is best for trains to arrive in the "loop" (the centre track) because this will leave free access to the maintenance depot. So, for example, one loco could leave the depot and the train loco could enter the depot without having to move the train.
The train loco will uncouple, run round its train, haul the train back through the archway and then propel it into the inset siding beside the quayside. The first uncoupling is a simple operation, which does not need the delayed action of the Kadee couplers. So it does not need a particularly strong magnet. The baseboard has a support running straight through the location for uncoupling here, and this makes it difficult to install a magnet from underneath, below the sleepers. The alternatives are a surface magnet from Kadee (their part #321) or a home-made alternative. I could build a foot crossing here to disguise the Kadee magnet, but this would really spoil the look of the model - it is tiny location and I want the visual lines to be flowing in sympathy with the track, not across it. I have settled on another pair of neodymium bar magnets, placed opposite each other below the rails:
After uncoupling, the loco will draw forward, run round and then haul the train back through the archway. It is then necessary to uncouple the train again (but with a delayed action) and let the loco propel the train into the siding. I have used a Kadee #321 coupler to do this. This coupler is long enough to give some flexibility in uncoupling location which I could not achieve with the small Neodymium magnets:
This magnet is located inside the archway and on a gradient. This is obviously not ideal but it seems to work. Amazingly, I can hear the uncoupling happening when the archway scenery is in place. The loco is always uphill from its train, and so the if the gradient has an effect it lets the train roll away from the loco. This arrangement would not work with the loco downhill the train. The magnet is sitting on a bed of "No More Nails". I used the Kadee #334 gluing jig to set the location, it is too easy to get it wrong with all the sleepers removed (this magnet is for code 100 track, my track is code 75):
There is one problem here, in that the trip pin on the leading coupler of a long bogie van snags the magnet. This is because the length of the van straddles the transition from the level track at the bottom of the gradient to the gradient itself, and the coupler is fixed to the body not the bogie. So the trip pin ends up too low here. The options were to ban this van from the quayside, or bend up the trip pin and forgo magnetic uncoupling for this one vehicle. I settled on bending the trip pin. It is the only vehicle of its type I have. This particular uncoupler location will end up as either innovative or stupid, I don't think there will be a half-way outcome.
The track in the fiddle yard is offcuts of C&L 00 gauge track (bullhead rail on moulded chairs). I found, I could glue a Kadee #321 coupler directly on top of the sleepers and the top of the magnet sits near the perfect 1/64 inch above the rails. It is necessary to shave off two rail fixings on one sleeper to let the magnet lie flat (at the left end in the photo), and I used the gluing jig again to make sure the alignment was central.
The curvature of the track here means this uncoupler is only suitable for shorter wagons, but I had the magnet to hand and it seemed sensible to use it up.
This is all the uncoupling magnets the layout is likely to get. I haven't planned out the uncoupling operations at the passenger platform, but I think it will be best to use manual uncoupling here because of the gradients nearby and the short length of the headshunt.
Thank you especially to AndrewC for pointing out the importance of car weights. Revised 5 February 2019.