Thanks. The acacia and wenge discs are 50mm, the walnut disc is 43mm. The small ones are all 25mm except for the maple one which is 15mm. If I make another that small I would insert a piece of metal rod in the center, because it can't spin for long.
Welcome, Jan-Pieter! Lack of a lathe clearly didn't get in your way. The disk-on-cone design is simple but elegant. The 3rd from the top is my favorite. Like the string-pull starter as well. From the woods you mentioned, I gather that you do a lot of non-lathe woodworking.
Some thoughts on the smallest top's short spin time. Turns out that the correlation between spin time and overall size or weight is surprisingly weak, so your smallest top need not be the first to fall. The following applies to tops of all sizes.
Spin time is of course just the time it takes to spin down from release speed to "critical speed" under the influence of air and tip resistance. Below critical speed, the top's no longer stable against gravity and hence falls at the slightest provocation. Thus, your opportunities to improve spin time generally fall into 4 main categories. Most of these are pretty obvious, but the details might not be.
1. Increase release speed. Your starter certainly addresses this one but may be less effective with your smallest tops. When twirling by hand, practice helps a lot, too.
But also experiment with stem diameter and knurling, as these determine how efficiently energy transfers from your fingers to the top. Generally, the smaller the top's rotational inertia, the thinner the stem. But only to a point, as the stem still has to be thick enough to keep your fingers from rubbing during the twirl. That happens to me below about 3 mm diameter.
2. Reduce critical speed. Mass distribution is the key here, and "down and out" is the ideal when it comes to spin time. Practically speaking, the lower to the ground, or the farther from the spin axis the top carries its mass, the lower the critical speed.
The "out" part of "down and out" also increases the top's ability to resist the combined effect of air and tip resistance. If you try another 15 g top, consider a thinner disk with a larger radius and mount it closer to the ground. Or redistribute mass to the periphery by hollowing out the disk near the stem and building up the rim.
3. Reduce spin decay rate by reducing tip resistance. Tip resistance generally increases with weight and contact surface area but varies little with speed. And friction isn't the only process involved. Hard, polished, rounded tips with a small radius of curvature tend to give the longest spins. But not too sharp, lest the tip dig in or deform. You're already on the right track here.
Of course, the materials in contact at the tip are also important. A wooden tip on a hard table might benefit from a hard, low-friction coating.
4. Reduce spin decay rate by reducing air resistance. Air resistance generally contributes far more to spin decay than tip resistance. But the latter grows in relative importance as the top approaches critical speed, so both are worth addressing. Polished streamlined top surfaces are the ideal when it comes to spin time. Since your tops are already pretty aerodynamic, doubt that you'll find much more spin time here. But rounded or beveled disk edges might help.
All that said, it's easy to get hung up on spin time to the detriment of play value and visual design. Measures that maximize spin time also tend to suppress precession, travel, and other behaviors that make a top fun to play with. To see what I mean, experiment with different disk-tip distances without changing the disk itself.
Sacrificing some spin time to an interesting visual design or behavior has a long, rich, and fruitful history in top-making.