Thanks for that informative answer. Good to know! Though it feels like something else is at play here. Because the problem isn't that I have trouble not bottoming out, but rather that the second actuation frequently happens long before the bottom out.
The software has a slider that lets you choose at which point the actuations happen, and goes from 10% to 90%. It's not actually clear to me how the slider works (whether the % value refers to the 1st or 2nd actuation), but I did many trials at 10%, 50%, and 90%, and I would have thought that one of those would give me a result where the first actuation comes relatively early while the second actuation doesn't happen until on (or very near) the bottom out. Unfortunately, I couldn't reliably get this behaviour, and often both actuations seem to happen somewhere around the middle.
I've actually given up on the aimpad function of the ControlPad, as it just seems to flaky or hard to get a handle on. For the sake of any future Googlers, I'll relate what I've found in my efforts to get it working.
The aimpad technology is very picky when it comes to switches. It's not just the stem color that matters, but the stem construction and the bottom housing material. Anything with an opaque bottom housing flat out doesn't work. The only switches in my collection with a 'clear' polycarbonate bottom housing are dragonfruits, but they're quite dark magenta, and they don't seem to work too well. A whitish/yellowish milky nylon bottom housing (like that of the Cherry reds that came with it) seems ok.
The actual length or shape of the stem pole probably doesn't matter (unless it's so long it itself is the point of bottom out - see below). The infrared sensors are on the side, and appear to measure the light bouncing off one side of the stem rails:
This seems to have several implications. Firstly, the rail must be able to come all the way down (A stem with a very long pole may prevent this). Secondly, the area where it comes down must be clear of visual obstructions. This happens to be same area where the extra pins are in 5-pin switches. If there's a little bumpy irregular knub where you've snipped off the pin, that may degrade the infrared accuracy. That seemed to be the case in a few of my tests. Maybe filing them down till smooth would help, but I didn't have the patience to try that. Finally, the stem needs to have a large, flat area in that spot. Some stems have a hollow slit in this area, or are very thin, or may be weirdly angled.
And as noted above, the stem needs to be a bright color that will effectively reflect the infrared. This doesn't seem that straightforward though. In the video I linked above, white stems underperformed red ones, while yellow ones outperformed them, which doesn't make a lot of sense to me. Though perhaps that was due to other stem attributes, such as those I just mentioned.
Of the switches in my collection, I only found one or two that fit all the criteria. And these worked pretty well, but once I tried to add a heavier spring things went downhill.
One of the big problems is the way the ControlPad callibration seems to work (or not work). It seems to require frequent recallibration; If I unplugged it then plugged it in, sometimes the aimpad accuracy seemed to become poor and require recallibration. The problem is that you can only callibrate all the keys at once, and if just one of them fails to calibrate then it stalls the entire process. Since I wanted to use a mixture of switches, including some tactile/clicky ones with black housings, this was a problem. I thought that I could get around this problem by installing all 'analog friendly' switches first (eg. ie. milky nylon housings, red cherry stems), calibrating, then swapping out those switches that were going to be just digital. This is easy enough, since you can callibrate switches without even soldering them, as the aimpad sensors work inependently of the metal pins. This approach worked initially, but then like I said, as soon as I unplugged it and replugged it in, it felt in need of calibration again, which was impossible because of some of the black-bottomed switches that were now soldered in (the black soaks up so much of the infrared that the callibration never completes).
So I've given up on trying to get the analog side of this to work. To be useful and not super annoying, the dual-stage activation had to be super accurate and predictable, and to get it to be like that seems to be too much trouble to be worth it. I'm just using the ControlPad as a run of the mill macropad at the moment. At least the two wheels on top of it are useful and work great, so it still feels like a worthwhile purchase, and not just an overpriced numpad.
