Andy rightly makes a good and important point. He says that worm gears 'can' be just as efficient as other gear arrangements.... They can be surprisingly efficient - some manufacturers quote just over 90% which is remarkable - however those are very good bits of kit, and worm efficiency varies dramatically. The ones I get to shove in my models certainly aren't anything like that good, sadly. Having said that, I'm in no way anti worm, having fitted and used countless of them on countless models.
In theatre, we use a variety of hoists for lifting and flying. The simpler ones are motor/worm/gear arrangements, and one of the very important safety features is that they do NOT feed back (some early manual hand winches did, and it would on occasion end up very nasty!) . The worm and gears are of course specifically specified and designed not to feed back for this application.
In the interests of collective experience, the reason I made comment about the poor efficiency of worms was one particularly 'dramatic' empirical example I had when building one chassis- which may be of interest as it concerns the N20s.
I build a chassis for a 7mm narrow gauge Bagnall (for radio control) with a white metal body- so relatively a heavy beast..... because I had used many N20s on vehicles and been do impressed with them, I thought I would use one on this as well - but because I needed to off-set it to hide it in the Firebox, I decided to use a 5:1 ratio N20 in conjunction with a 20:1 worm. This proved to be very weak and easy to stall, giving poor slow speed control and overall disappointing performance. I removed that installation, and replaced it with the very much smaller K20, driving through a modified 0.3M brass bevel gear. This was superb, giving the same track speed but much 'stronger' at the wheels, and thereby giving much better, smoother control.
Both installations delivered the same nominal RPM at the axle, both were properly meshed. Control etc. remained the same and un-altered.
The N20 was a 6v motor (12 x 10 x 20) and the K20 a 5v motor (8 x 6 x 16) with a planetary box..
My experience would suggest that N20s are most effectively utilised with a 1:1 final drive if possible, and avoid a high ratio final drive (like wot I tried..)
5:1 N20 with 20:1 worm and gear
03B65D83-16E9-46C2-88CB-7D1F63F78992_zpsgmmk0e2a by giles favell, on Flickr
K20 with bevel gears
2019-02-08_07-14-46 by giles favell, on Flickr