N-scale modelers almost always seem to go in the magnet direction when it comes to uncoupling operations. A few enterprising souls have built their own uncouplers out of coils, iron cores, mounting plate, field plates and plate dividers. I prefer such things pre-made, so I looked around and found only the under-the-track and over-the-track permanent magnets and the powerful under-the-track electromagnets that are expensive and difficult to make do delayed uncoupling. But I insist upon flawless delayed uncoupling, as I often push cars to their destination once I uncouple them and this only works if I can reliably get couplers into their delayed positions. So I kept looking until I found something that's close to right: HOn3 Kadee #708 magne-electric uncouplers. I got these mail order at considerably under $10 a shot. I'm very pleased with them—they're easy to modify and work flawlessly. I'll get into the modifications and how you use this uncoupler, but first let's look at the basics of uncoupling.
I have talked to dozens of hobbyists about this and they like the reliability of permanent magnets, but don't like the nonprototypical appearance. But they strongly dislike the accidental uncouplings that a permanent magnet over or under the track ties often causes. I won't use permanent magnets on my layouts for that very reason. You'll find directions to hinge-mount and pulley-operate large commercial permanent magnets in Micro-Trains literature, but I don't like the unreliability factor when it comes to the delayed uncoupling operation and any under-the-track magnet, or the crudeness and hassle of hinge and pulley operations. Note: one under-the-track magnet maker I talked to says that it doesn't matter that his uncouplers are so powerful that they pull couplers beyond their delayed position—he says you can still do delayed uncoupling and car pushing even if the coupler or car body is pushing against the body of the next car rather than the coupler. I tried that in under-the-track magnet experiments and all is well until the body of the car the coupler is pushing against encounters a curve or a turnout. Either encounter will sometimes result in a derailment, at least in N scale—I can't speak for other scales.
Hobbyists have relied upon restraining springs in cabooses to keep slack out of their train's couplers in order to avoid the frustration of accidental uncoupling from mainline magnets. Some have good luck with this—others don't. For one thing, during operations, sometimes no caboose is present at the end of a given train. For another thing, the more you keep couplers taut, the more you increase the chances of certain types of derailments. And finally, caboose springs only help when your train is going forward. It happens that many operations require forward-reverse-forward sequences, and during the direction change from reverse to forward, all couplers are slack and any coupler over a permanent magnet is doomed to uncouple. Yes, you can try to position your train so that you have a car center rather than a car end over the uncoupler at such times, but that's not only a hassle but sometimes even an impossibility, since you may be precisely placing or uncoupling a car somewhere else on your train so the train position may not be flexible.
You can choose to design layouts and operations so that magnets will be less troublesome by keeping them off the mainline, but when you read operations books and see why mainline uncoupling is needed, you're left unsatisfied. But you can prevent this dissatisfaction by a good uncoupling plan that uses electromagnet uncouplers wherever your operations need them rather than designing a layout around the weaknesses of magnets. Just as Kadee comes to the rescue in HO and HOn3 scales, it come to the rescue in N scale as well (but don't tell them that—they're allergic to the term N-scale). Read on:
Here's a nutty statement if you've ever seen one: The HOn3 Kadee #708 magne-electric uncouplers fit perfectly between N-scale track rails and can do perfect delayed-action uncoupling every time, with no coupler trip pin bending; and the upper track mounting plate fits the assembly perfectly as well; but let's take this out-of-the-box perfection and modify it so the upper track mounting plate is useless and must be discarded and then let's modify all couplers you want to use with this uncoupler. In essence, then, let's make the thing work worse and look worse and let's bend all the pins on couplers we want the 708 to uncouple. And since you may think I've lost my marbles, I'll explain why this is necessary:
The two field plates that stick out through the rails and are parallel to the rails—and at rail height—are over 7 mm (O.D.—from the outside of one plate to the outside of the other) apart as purchased (when assembled into an uncoupler), but you'll need to bend them in a vice so they are only 6.2 mm (O.D.) apart when assembled. I did the bending after assembling the field plates, core, coil, bolt, and field plate divider—you may bend the plates before assembly if you like.
The bending will preclude a fatal flaw in using these uncouplers for N-scale—actually two of them: (1) Unbent plates will short circuit your locos when they go over the uncoupler, and (2) unbent plates may cause derailments because there isn't room for the flanges of some/most/all wheels, depending on your wheel-sets. Unbent plates will be so close to your rails that they'll be contacted by loco wheels, which wouldn't hurt except that the flanges won't have room and, worse yet, the two field plates are—electrically speaking—one continuous conductor. So the distance between wheels (I.D.) on cars and locos (A in diagram 1) has to be considerably more than the distance apart (O.D.) of the field plates (B in diagram 1).
The good news about all this is that N-scalers can end up with easy-to-assemble, inexpensive, great-working uncouplers that are reliable for both regular and delayed uncoupling, look better than magnets, and are run by pushbuttons, so you can kiss the accidental uncouplings goodbye.
The bad news isn't really all that bad, but it's necessary to deal with, so here goes: You need to bend your coupler trip pins ("air hoses") to make using the 708 succeed in N scale. Some couplers won't even uncouple without this while others will uncouple but won't do delayed uncoupling. With regards to the trip pin bending, don't even think about using only one pair of needle-nose pliers. You can't just grab the trip pin and bend, whether pliers or not—you'll just wreck the coupler. Instead, grab the uppermost half of the pin with one pliers and the lowermost half of the pin with another pair of needle-nose pliers and bend the pin so the lowermost portion of the pin doesn't angle out to the side so much but is instead virtually parallel to the rails. See the correct bend in B of diagram 2 (A represents unbent couplers). Do NOT exert force on the coupler! I'm assuming Micro-Trains couplers here.
Here (A of diagram 3) is why unbent 708s work well in N scale even without coupler bending, if you ignore the two fatal flaws (that are illustrated in diagram 1), why they don't work worth a hoot once the plates are bent in to 6.2 mm (B of diagram 3), and why they again function great even with the narrow plate positioning as long as you bend the trip pins appropriately (C of diagram 3). The figures are all top view representations. The two long lines are the field plates and the two short lines are the trip pins. Magnetic forces don't pull the pins in B apart because the pins start out virtually where the plates are trying to pull them to—there's no reason for them to approach plates when they're already there.