Several solutions can be envisaged, such as intervening on the drawbar — complicated because it is difficult to dismantle — or use height-adjustable dovetail couplers — which is much easier, but which produces a larger gap between the buffers (about 3 mm: this is much).
I chose an intermediate solution: make the tail of the couplers thinner in their upper part, and compensate underneath with a wedge. These are Fleischmann Profi models. Their total thickness is 1.7 mm. If 0.7 mm is removed, value moderate enough so as not to weaken the coupler too much, a 0.75 thick plastic card (7 × 3 mm rectangle) will compensate. The problem is to maintain the coupler during the milling operation. Hence the making of a machining assembly: a 2 mm aluminium plate with two ø 2 holes to allow the movable — and fragile — part of the coupler to pass through, and two others, tapped M2, for securing a flange to immobilize the head.
The external dimensions are not critical, nor that of the spacing of the tapped holes. However, the width dimension of the flange must not be exceeded, since it must not press the rear part of the coupler head.
Coupler after machining. The dismantled flange can be seen with its two M2 screws and the coupler placed on the support through which the protruding parts of the moving pre-uncoupling piece pass. This allows the head to be fixed firmly during machining, without risk of deterioration.
Assembly of the thinned coupler with the polystyrene wedge, before completely engaging the assembly into the NEM box. The coupler thus arrives at practically the same level as that of the “curved-walls”.
After modification, I find the result rather satisfactory, even if the “curved-walls” is now slightly too high, thing easy to correct with raising wedges slightly thinner.
There remains an unknown factor: the retaining force of the coupler, which is probably lower due to the smaller thickness of the tail. If there are coupling breakages, you will have to insert a wedge to lock the clips on the back of the coupler.