@meng Aahhh... I think I know the issue now - we are using different independent and dependent variables to test the similar thing.
Yes, and I'll explain why better...
@meng I am using the gear ratio as independent variable and then observe both the torque and speed of the drive.
There are three problems with this as I see it:
1) Gears and sprockets/chain are not the only way to effectively mechanically build in the equivalent to a gear ratio. There are multiple other ways. Keeping the gear ratio as an independent while not similarly managing these other things means you're not fully keeping what you want to be independent actually independent. For example, what if I were to say tank drives are faster when you use 100 rpm on both a tank drive and an x-drive? I'm I wrong? You might think so, but maybe I'm saying it because with a tank drive I am using 6" wheels, while my x-drive's omni-wheels are limited to 4". Different wheel sizes as well as different wheel alignments are ways to mechanically build in the equivalent of gear ratio without using gears nor sprockets/chain. x-drives included a mechanical equivalent to gear ratio, essentially gearing for speed. While it's important to know about this to use the design, it's not useful to make comparisons.
2) Gear ratio is not a performance (can't think of a good word there) part of the robot. For example, you do things like deciding to climb a hill. You need to make sure the robot has the strength to climb the hill. Then the fastest design is the one that provides the needed strength and carries the robot along the fastest. Or you have some speed you want your robot to be able to maneuver at in the field. You design tank and x-drives to perform at your desired speed. The one that can push stronger while doing so is stronger. Performance values are what are needed, so they are really what ought to be used.
3) We've seen repeatedly that people have trouble understanding what this method means. Many statements have been made with a caveat like "with the same gear ratio" when comparing the drives. Yet repeatedly people have misunderstood what is actually being said, entirely losing track of that caveat. So if approaching it this way hasn't worked so well in conveying the message, maybe we should use a different way that will show the actual gains and losses better.
Really, we should say it all differently anyway. X-drives have lower output power linearly and higher output power rotationally than do tank drives. While you probably don't need the same rotational power, x-drives are holonomic drives, which provides additional advantages, while their configuration makes things like driving over a bump harder. Once everyone understands the power part (I think the second is well understood.), the whole faster/slower/stronger/weaker thing and making choices based on performance will become clearer.
This is why I really hate the statements I so often hear that x-drives are faster, which is based on this gear ratio comparison. Are they really? So if you're going to design robots for a straight-line race, would you choose an x-drive? We can even disallow multi-speed transmissions so that there is no need to worry about those complexities getting in the way of a comparison. If x-drives really are faster, shouldn't we use an x-drive for this? Or maybe we shouldn't be saying x-drives are faster?