Hi Community!

I'm using the ProX AIO Keyboard (Which is the ProX you can either use as touch or mechanical keyboard).
Today, I decided to give touch a try and see how I fare with it in comparison to mechanical keys.

After playing some time, I noticed, that I successfully tapped some burst circles even though my fingers felt like having tapped one time less than required.

I decided to look into that some more and long story short: When touching the metal plate with a very specific area of my finger (that is much lower than the usual area I would touch it when tapping normally), this can cause a very fast burst of key inputs.
Having a result of over 10 taps/s by just slightly touching the metal plate was easily achievable. I sometimes even got 20 to 30 (after realizing what's happening I tested this in graveyard, of course. I also did not pass any maps that affect my rank using touch).
I have my theory what's happening here on an electrical level, so I also tested touching the edge of the metal plate et voila: I could sometimes achieve over 50 ghost taps/s (physical explanation for that behaviour at the end).

Conclusion: Touch function of ProX AIO may lead to additional ghost taps from time to time making it unsuitable for ranked plays. Addionally it could be abused to cheat higher taps/s rates (even though it's hard to hold a steady rate, it could probalby be abused with proper training). Personally, I will not use the touch functionality for ranked maps anymore.

I hope that any automated reports that may have been triggred by the built-in anti cheat will not be used against me (as I have not abused it to achieve anything).

King regards,
Tauronan


Edit: I will now inform the creator of the ProX. I also have an idea on how to fix this issue (may only be possible for future generations of the ProX) which looks as following:
Instead of determining between "touch" and "not touch", the Chip should distinguish between multiple levels of "touch" (no touch, very light touch, light touch, medium touch, etc.). One could then program a hysteresis that requires a minimum decrease of the touch level followed by another inimum increase to register another tap. The right hysteresis should be hardened by sophisticated testing to ensure, that the various effects discussed in this post are all much weaker and can not cause ghost taps.

By programming a relative "touch level" hysteresis on could achieve an adaptable system that works reliable with different finger sizes and tapping styles.


Physical explanation (only for fellow nerds who really want to know):

After measuring between USB-GND and metal touch plate, I could observe a ~5kHz pulse frequency. This is being generated by the Atmel-Chip within the ProX AIO to check, if something touches the plate or not. Unfortunately, when only slightly touching the plate, the measured signal is blurry, so it's somewhat random whether a touch is being registered or not.

One could easily imagine that in this state, anything could influence the result, such as: Human heart pressure, 50Hz net frequency and many more (depenging on a lot of factors the human body may act as a weak antenna for electrosmog and other signals. This is, of course not something special about the human body but a natural effect that any elecitrcally conductive object can have).

Concerning the increased effect at the edge/corner of the metal plate, there are two possible explanations (which may both apply at the same time):
1. It's easier to create a very specific contact area/skin resistance when touching the edge compared to the middle area.

2. A voltage level is caused by a difference between the eletrcial field strength within two objects. In other words, voltage and electical field is physically connected to each other. Now remember, that the Conctroller within the ProX AIO pulses the metal plate with an electrical signal of ~5kHz. Hence, the electrical field inside the plate follows this change.

However, Corners and edges causes the electrical field to increase (which is why a lightning protector works best, if it's a sharp spike on top of the roof: it concentrates the electrical field in that spike kinda drawing the lightning strike towards it).

Hence, a number of electrical effects influencing the actual ghost tap frequency are increased as well.

Please not, that this is only a very simplified explanation not taking all physical effects into account. I did not want to write a novel, though.

i got some problems using the prox touch too
when i hold down keys
it will be registered but it will release on it's own
which cause slider break or slider end
but i never met the problem

Your problem sounds even worse to me as it should be quite easy to design and program the circuit in a way where this does not happen. Tried out if my exemplar has your issue as well but at least when I just let the fingers rest on it for some time the key stays "pushed" without any breaks.
However, the ProX touch should have a different circuit and programming than the AIO so that could explain the difference.

I have two of those keyboards (both are AIO but one for middle and the other for ring finger playstyle).
Tested the second one as well where the issue, while being present, is much less apparent. It took me quite some effort to trigger it in a way where it creates multiple taps on it's own but it was still possible to do so. Conclusion for me was that it's not just bad luck with one exemplar.

As I didn't want to create autoclicker-like stuff inside the OSU-client I used the following site for further testing:
https://keyboardchecker.com/

Now, the following may sound a little bit harsh, but I don't like the layout style of the PCB within the ProX. (couldn't resist voiding my warranty by taking a look inside I'm designing PCBs for a living and I see multiple rules not being followed here (which may cause decreased reliability). However, I do really like the mechanical and outer design (plastic case, arrangement and design of the C-Key, etc.). So I would really like to see a fixed version of those ProX touch-type keyboards. Maybe a variant using a more sophisticated capacitive touch approach?

gsdeman wrote:

i got some problems using the prox touch too
when i hold down keys
it will be registered but it will release on it's own
which cause slider break or slider end
but i never met the problem

Didn't have this problem, i'd say it was way too sensitive. But i just didn't find device itself comfortable. Bought it in summer 2019 so maybe new revisions became worse.

PS My words only apply to clear plate. When it's a bit dirty some random events may happen, so try to clean it with wet wipes.

killian wrote:

Didn't have this problem, i'd say it was way too sensitive. But i just didn't find device itself comfortable. Bought it in summer 2019 so maybe new revisions became worse.

PS My words only apply to clear plate. When it's a bit dirty some random events may happen, so try to clean it with wet wipes.

My plates were clean when I started the OSU-session where I first tried them seriously. After about two hours I realized there was a problem. So I guess a small amount of body fat or whatever the skin leaves on a metal plate after touching it was there but if it's really that much sensitive I think the algorithm/circuit should be reworked.

However, your first sentence gave me some inspiration. Next months look quite packed for me but at some point I will probably create my own touch key seeing that the existing ones seem to lack this and that. Already have a concept in mind but let's just say, there are reasons why you probably won't find a commercial variant of my idea. Should I really find the time and a well working solution I will look for a way to either make it obtainable or at least share it online. However, should I find a well working already existing product til then I'll stick with that instead. No need to reinvent the wheel.

this has been a known issue since at least btmc's video in 2017 here is link and timestamp https://youtu.be/vcdNyXSTVgE?t=155

i got the pro x aio touch keypad (standard, not ring finger) back in january (and have been consistently using it ever since despite most ppl dropping it for being too used to keys) and have not encountered any issues with holding down the keys as gsdemon mentioned or unwanted ghost inputs as you did. i was however able to replicate the ghost inputs both when i first got it and now by intentionally holding my fingers very slightly above the copper plates at a level much lower than i usually do, however i actually had some trouble replicating it now and was only successful once because of how close my finger needs to be. i wanted to get it a second time but there's no need and after 3 mins i give up lol

it's possible that since youre used to resting your fingers on the keys from playing with keyboard that you hold your fingers very close to it so it's basically touching but not quite but i doubt it since my first instinct was to raise my fingers quite a bit (~1.5cm) above the plates as there is no actuation distance or weight so if i held it closer my tapping and especially streams would be really awkward and ineffective. holding it closer is rlly good for jitter clicking and therefore spamming doubletaps though but i almost never do that lmao
though this isnt mutually exclusive with what is almost certain, which is that it seems there is a quality control issue even after all these years and (since of them are custom made by the creator afaik) they have gotten better but some mistakes are still made. i find it pretty plausible that they saw the issue especially when btmc mentioned it in their review after they sent them their prototype and left it assuming it wasnt worth a redesign considering it shouldnt come up in gameplay but it obviously does if they make mistakes during production
still despite this i think they shouldve redesigned it and dont see why they wouldnt test each keypad for how sensitive they are with this uhhhm... "feature". assuming they test it that is. i suppose they probably do test them individually but are simply always busy and never considered testing for that specifically so quality control might be slightly lower there

i dont know of anyone other than you who has had an actual issue with this though, it seems everyone is able to replicate ghost tapping but only intentionally, and yet you got ones which are defective in this way twice in a row. i also suggest you email the creator if you havent already, although probably not about a refund unless you already recorded footage of the defect before voiding your warranty haha (personally i (am pretty sure) i voided my warranty when one of the times i took my sticky nanopad which ripped the sticker in half. i put it back so ill just pretend it's not ripped. while im talking abt the sticky nanopad, they can actually stretch and tear quite easily if you dont take them off slowly after theyve been stuck for a while or handle them carefully while wet which i found the hard way and both were why it ripped/snap so i ordered the pack of 4 from the website and it was no biggie)

dont have much else to say hope this wasnt too long and hope this helps you even though im 6 months late
slight edit: the plates are actually brass it seems, but the original design years ago was with copper im pretty sure