Syringe full of lube on my desk at all times editionPrevious: >>107055317>Keyboard recommendation template:https://pastebin.com/n220xk9V>Find vendorshttps://www.alexotos.com/keyboard-vendor-list // Up-to-date list of reputable vendors with brief descriptionshttps://keycaplendar.firebaseapp.com // Tracker for current and upcoming keycap group buys>This keyboard stuff is so expensive!https://aliexpress.com (or Taobao if you know how)>Learn about MX-type switches ("mechanical keyboard switches")https://rentry.org/mkg_switches // Introductory guidehttps://www.theremingoat.com // Switch reviewshttps://www.switchesdb.com // Compare force curves>Split and non-standard layout resourceshttps://compare.splitkb.comhttps://keyboard-design.com>What does ______ do to a keyboard's sound?https://blacksimon.tv/science (Google sheet)>Practice typinghttps://monkeytype.comhttps://www.keybr.comhttps://thetypingcat.comhttps://play.typeracer.com>How Cherry switches do backlighting, and why it's not idealhttps://rentry.org/mkg_backlight
>>107099559I came.
nth for zakus
Resurrected the thread to say that I finished modding my V6 Max and the spacebar still sounds fucked. I think it's just very difficult to make an ABS MT3 spacebar sound good. I've decided it doesn't need to be clicky like the other keys, I don't think it even needs to be tactile necessarily. So I just need to go through all the 70 or so types of switches I have and find the right one.I think it might be the top-out that sounds particularly bad so a "semi-silent" with a silenced top-out might be the right way to go. I'm trying a JWick Semi-Silent right now and it's not awful. But the resonance of the MT3 spacebar is SO thin and metallic. I almost wonder if I need to put foam in it... yuck
>>107099858Nothing wrong with putting some foam in the spacebar. Hell the leopold FC900R I bought as a gift for someone has it stock.
>>107100685It's very rare that I like the sound of foam in anything. But this could be a first.
There doesn't appear to be any way to get KBDFans spacebar foam that doesn't involve spending $30+ because of shipping. Fuck.
>>107100969I'm just gonna have to make my own foam shit I guess
>>107095032ok I put the scope on the hall effect sensor and it's weird. it's open collector output, normally pulled high. so when you press the button voltage output goes down.the problem is with the low voltage difference. when switch is installed there's like 2.5Vout, and fully pressed it drops to 1.65V or so. the diode doesn't matter, thought there would be some dead zone with 0.6Vdrop but in open collector mode it works fine no gain from replacing it.so adc is reading 2.5V to 1.65V, hence the shit resolution. but I still don't get it, the micro has a 12bit ADC, means 4096 steps. keyboard advertises 0.1mm resolution in 4mm travel, so max 40 steps, which is insanely low.even with under 1V from total 3.3V, that's still some 1000 steps of resolution that micro would be capable ofthe whole hall effect switch is a total clusterfuck based on a plethora of switches and output options combined with a fuckload of sensitivity options. add unknown specs for switches magnet and you got a total mess. which might explain why some hall effect switches don't work on some hall effect keyboards.what I noticed is that generally south pole should point towards pcb, from switch. but if you're working with unknown magnet strength you're pretty much fuckedsee https://www.ti.com/lit/an/slya093/slya093.pdfandhttps://www.ti.com/tool/TI-MAGNETIC-SENSE-SIMULATORbut you need an online account to use it. so you need to get something which has the sensitivity you're looking for, based on the magnet you use in the switch. if it's not sensitive you'll get shit output, if it's too sensitive you'll quickly max it out. so you need something which gives high output resolution for those 4mm of travel, and then build your circuit based on whatever tf you settle on.and again, sensitivity is not up to magnet size/strength, larger = higher flux = activates it sooner. you want something with high output range for those 4mm of travel. you won't get higher sensitivity with stronger magnet
>>107101103>even with under 1V from total 3.3V, that's still some 1000 steps of resolution that micro would be capable ofif anyone has any ideas on how to increase the resolution do tell, might be something that can be done in software? 0.8V or whatever should still translate to way more steps than fucking 40, at 12bit. even 6bits has more values than fucking 40
what's better? typeplus or knight v4s?between the two, I've only tried typeplus and I liked them. I used only 205g0 and very little of it and got away with 0 rattle and no tuning
which keyboard are you bringing with you to a deserted island
I'm now practicing typing random numbers on keybr. Possibly the most boring thing I've ever done in my life.
I just found out C64 8bitdo model is the only one that comes with ABS switches.Is it because of SA profile and original C64 using ABS (assuming)?Any of you have experience with them, are they slippery or nice in touch?I like C64 color scheme but would prefer PBT - although don't have much experience with mk anyway
>>107102180I've never heard of PBT SA, which makes me wonder if it can't be done for some reason
One of the amber switches on my evo75 already died
The masking tape method for making a template to cut fabric/etc for inside your keyboard really works btw. ChatGPT was right
>>107102778How early did it die?
>>107101103Maybe simplifying to 40 ends up being faster to process or something. I'm not sure what benefit you'd actually feel by going up to 1000 step resolution on the switch.(I don't really have anything useful to say. Just felt the need to reply since you're effortposting)
>>107103419they are using x10 more steps on TMR, so 400 steps. which is also not even 9 bits. doubt it's processing, must be something related to implementation or future marketing since they can sell the same shit with higher resolution or smth.>Just felt the need to reply since you're effortpostingsome people are interested in building their own HE so figured I'd share my findings, maybe it's useful
>>1071033062 months in, the I key.I knew the switches weren't great before buying and anticipated them breaking quickly but geesh
>>107102218aliexpress seems full of double-shot PBT SA keycap sets
https://blackdrop.net/projects/ccrx-seYou guys gonna jump on this project? Looks interesting
>>107103875Aliexpress """PBT""" is what the rest of us usually call ABS
>>107102218Signature Plastics does PBT SA, but only for sets you basically have to cobble together on your own if I recall. They're not doubleshot either, but dye sub only.The issue, besides designing new molds for PBT, from what I've gathered is that manufacturers often don't let the caps rest long enough to prevent warping and thus they usually get fucked up tolerances. My MTNU sets (pure PBT from GMK, not SA) all came with warped spacebars, and they're a lot shorter of a profile than SA is.
>>107105168Even ABS spacebars will sometimes warp; shit happens. At least it's fixable with a hair dryer or hot water.
>>107105194Maybe, but both SA and GMK charge more for keycap sets than entire keyboards cost, so it's a bit inexcusable at that point. For <$60-80 you can expect keycaps to start being consistently wonky, but for $150+ they should, for all intents and purposes, basically be perfect. Also, it would take a few seconds to QC each spacebar or other longer stabilized key by placing it on a level, flat surface to verify if it's deformed or not, which should come with the price they charge, too.
>>107105250>SAMeant SP* here for Signature Plastics.
>>107105168>My MTNU sets (pure PBT from GMK, not SA) all came with warped spacebars,Scary stuff. Fortunately my MTNU spacebar was OK.
>>107105250Yea I get it, it's pretty fucking annoying especially for the price paid. I'm just glad I don't have to deal with warped spacebars (or stabilizers) because the largest key my board has is 1.5u.
>>107101103I can't tell if you're asking for help or blogging. What adc are you using to measure the hall effect? Are you using an adc on a mcu? An external adc? If mcu, What mcu are you using? What are you setting the vref to? Are you using an external vref? Where is the source code of the adc setup and read?
>>107106356I don't think blogging is a crime. It's not like he's just giving us a daily update about his life, he's detailing his keyboard DIY adventures.
>>107106356>come to a keyboard thread>"why are people posting about their keyboard projects?"
Asian anons... what do you call the sort of Asian guy who's into this
>>107106356>I can't tell if you're asking for help or bloggingsharing info I found on my P1 HE for others that want to build their own HE PCBs. some people want to do it, HE or TMR.>What adc are you using to measure the hall effect? Are you using an adc on a mcu?one in micro>What mcu are you using?not my choice, Keychron chose it. stm32f402 which is a chinese version (compatible) with stm32f401>What are you setting the vref to?vref seems connected to a 3.3V supply through a Schottky diode. micro is not powered form 3.3V but from 3.0V. opamps feeding the ADCs are also powered from that 3.3V supply.>Where is the source code of the adc setup and read?>https://github.com/Keychron/qmk_firmware/blob/hall_effect_playground/keyboards/lemokey/common/analog_matrix/analog_matrix_scan.c>https://github.com/Keychron/qmk_firmware/blob/hall_effect_playground/keyboards/lemokey/common/analog_matrix/analog_matrix_type.h>https://github.com/Keychron/qmk_firmware/blob/hall_effect_playground/keyboards/lemokey/p1_he/mcuconf.h>https://github.com/Keychron/qmk_firmware/blob/hall_effect_playground/keyboards/lemokey/p1_he/halconf.hthere's some 6 bit stuff in there tho not sure if it's limited by something else or what
Need a wireless one with hall effect switches
if I buy a 60he V2 with a split spacebar in ISO layout will I get cucked by aftermarket keycaps due to the split?>The split spacebar is from left to right a 2.75u, 1.25u, and a 2.25u key. These are common keys on an ANSI layout. It's the same as Left Shift, Ctrl, and Right Shift key.
>>107107042so keychron?
>>1071070482.75+2.25u is standard for alice so it’s pretty well supported. alice space was born from using left over shift keys so in a worst case you can always do thatdon’t buy chinkshit on ali for keycaps and you won’t have an issue. an hhkb 6u or 2x 3u is where you might start running into issues if whatever you’re buying doesn’t have an add on spacebar kit, and even then many kits only include a single 3u
>>107106929i dont this the limit is set by something else. I just think they chose 6 points of actuation. Could probably be expanded with little effort. You could also always ask in a github ticket.
>>107107113looking at ISO keycaps, most don't seem to include more 2.75u capssay you go full rgbtq+ and buy RGB caps like aura or pudding caps
>>107107061seems like it's only them, not very expensive either
so what's the point of anything other than buckling springs?
>>107106406I wasn't saying it was bad if he was blogging. I just didn't know if me looking into the issue was something he actually wanted or if he was just documenting progress
>>107107048If you want a convex 1.25 key (1.25c) they can be harder to find.
>>107107297considering normie stores have them the k2 he and whatever feel very good to me compared to keyboards that feel like pure shit from the mainstream gaymen brandson sale pretty cheapbut the software seems to be worse?>>107107297I remember when people shilled the IBM boards as the holy grailis that still a meme>>107107348that seems like a normal function key thoughI'm thinking the middle function key would be pretty gud on a 60% board no? if you gonna use layers with your thumb idk never used something like that
>>107106929Looking at the datasheet, I can't tell why they implemented it the way they did. You're right, the code caps out at 40 steps which makes no sense. The only thing I can think of is that they implemented it this way for convenience because 40 steps is still technically a lot. Just way less than the 12 bit resolution of the adc. Might be worth changing the firmware just to see what breaks if you were to arbitrarily change the 40 to 140 or 240 or whatever. It shouldn't be hard to do because that 6 bits looks arbitrary as well as the random > 40 checks they do as a cap.> if (profile[i].global.act_pt == 0 || profile[i].global.act_pt > 40) profile[i].global.act_pt = DEFAULT_ACTUATION_POINT;
>>107107385>>107106929Ok nvm. The reason why they cap is at 40 is because they're using units of 0.1mm commonly. The HE sensor saturates heavily (according to their datasheet) at 4mm. So 40 is the cap because 0.1mm is the unit of choice. You'd have to change the common unit from 0.1mm to the smallest resolution your adc can handle. In your case your Vh is 3.0 and your Vl is 0, so you have 0.000732421875 V per bit. However, the HE sensors low is 1.65V and high is 2.5, so you'd have to take that subset and then find the smallest unit you could use given your range is 0.85V with a bit resolution of 0.000732421875. (My math could've been wrong there. Fact check me...)Not sure why they chose 0.1mm as their unit of choice though.
>>107107382>is that still a memeyes because it's true and every other keyboard i've tried feels like a toy
>>107107297You miss typed beam springs
>>107107474probably sucks huge dongs for gaming so seems like a meme at this point clinging to relicsnot sure I used on back in the daymemory failing
>>107107264I don’t know about ISO but shine through caps are always lower quality. Pudding in particular are cheap trash.
>>107106929I can dig into this more later once I figure out how to compile qmk lol. I just need to figure out where they're converting samples to actuation counts and doing those comparisons.
Can I lubricate 123 Cherry Reds with a single dish of 205g0?
>>107107297people who want linear switches and/or quiet boards or to gameotherwise buckling springs and beamsprings are great tech
>>107107644>but shine through capsseems like abs yes>Pudding in particular are cheap trash.pbt tho?idk, not like im gonna waste money on thatok I will take the custom keyboard meme and buy ultra premium non see through capsmaybe
>>107108009>or to gameIf you think you can't game on buckling springs you are a moran. If you are playing games that require NKRO, you are a moran.
>>107108064If only wooting had proper facing leds
>>107108104>jew kaysadly, I don't have the loicense to purchase thiswooting standard bitch caps are already see throughtalking more rgb though
>>107108065>If you think you can't game on buckling springs you are a moranI mean you can, but I doubt you can actuate repeatedly as fast as on a linear switch looking at my friend who uses a Model M13 who I play with/against
Hot take; shine through keycaps suck; LEDs are cool but they are silently deadly to your eyes. Having bright keycaps with bold legends is more healthy than LEDs which dazzle and strain your eyes slowly.
>>107106929Thanks anon, you've inspired me to dig out my TMR and HE sensors and give making them work another shot please stay around so you can hand hold me with your brain ;_;
>>107108493most people think that anon O_Oshine through is heavily frowned upon from what I've seenits the hall mark of shitty gaming keyboard(I unironically love jelly and pudding keycaps tho, but I'm white trash)
WS Aurora Clear (UHMWPE bottom) seems to attenuate the bad MT3 spacebar sound significantly on my Keychron V6 Max. Swapped in the JWick semi-silent stem for silenced top-out, and it was nearly there.Cut a bit of felt to stick in the space bar and that helped, so I stuck it in a bit more permanently with double-sided tape. It's not perfect but it's way more tolerable now. I can probably just be happy with this keyboard for a while before I feel the need to fuck with it again.And I didn't have to spend $30 on having spacebar foam shipped to me, so that was nice.Also I'm glad I removed the battery, which was a pain in the ass to do because it was so thoroughly glued on there.
>>107108009>>107107633lmao weak ass niggas have a hard time pressing a buckling spring key lol!!!! why the hell would it be an issue gaming
>tfw people finally realize full size keyboards are for trannies
>>107107576I'm super tempted to get a beam-spring keyboard with my next tax return (assuming the gubberment isn't still on fire and they aren't processing that) but I already have a buckling spring keyboard from the same guy.
Im going to totally btfo the USB C faggot, preliminary tests show even tho they are shielded im able to get a ton of data from them with a 3 axis magnetometer and a linear HE sensor. Prepare to get absolutely cooked fahgpt
>>107107216I think there's 40 points of actuation, they used 6 bits for the actuation, at least as far as I can tell. 2^6=64 points. and they advertise less than that ,0.1mm ticks for 4mm travel, so 40 steps, which fits the 6bit in their files
>>107107385>>107107463thanks anon. so I wasn't crazy thinking it can do more precision. I'll start experimenting after I assemble it backI want test something else before putting it back together, switch sensitivity. going to move it to the side of the sensor and see how it picks it up.I'm not convinced on the whole TMR is better thing. what I know is that ye they can be more sensitive and they draw less power, which is useful for battery applications. but if I can increase the resolution for HE then TMR is pretty much irrelevant for me, I don't care about battery, took it out anyway.
>>107107463ah yes, high is 2.5V or so and low is 1.65V so there's 0.85V for full travel. and 3.1V for Vref means ~0.00076V step at 12bit. that still means some 1100 steps, ideal, but even 800 or so would still be double than TMR is advertised these days (0.01mm).
ok did some more measurements and found some interesting and rather good news.I took a switch apart so I can play with the magnet stem. I wanted to offset it to HE sensor just to see what happens, but centering pins kept the switch way too much above the pcb.so I tried a few positions, I put the magnet centered with the centering holes, on each side, and I still got a reading. there was about 300mV deflection in those areas. which you know...is usable, with higher resolution sampling. but this was some extreme caseI also tried it somewhere in between the center of switch and one of the side centering pins, trying to emulate the HE sensor being mounted to the side of the hole like TMR are, and I still got a reading, like 500mV or so (2.15V).so I don't think they absolutely have to be right under the switch, they can be mounted to the side, with lower sensitivity, but they work. the question becomes if 500mV delta is usable.another thing I tried which I found interesting was reversing the switch, and that seemed to work just as good as normal position. so if you get some north leds HE keyb and want to reverse your switches, it works fine (at least for those switches with these sensors). you'd need clear body switches if you still want the light from leds but sensor reads just the same when switch is 180 degrees rotated.I tried playing with the magnet in all kinds of directions and angles and sensor read goes the other way around, outputs higher voltage signal.so sensors are powered by 5.5V, hence why the ~2.7V read, they output 2.7V DC when no magnet around. depending on magnetic pole, it either goes up to 5.5V or down towards 0V.the magnet strength in the switch is such that when installed on PCB the sensor barely starts registering it. I think the calibration function in the software reads this small offset and cancels it. this way there's no dead zone up top when you start pressing.if you wanna build a keyb first experiment with the footprint on a small pcb
>>107110681>another thing I tried which I found interesting was reversing the switch, and that seemed to work just as good as normal position.but will try this again when the keyboard is assembled back, don't know if the plate allows it to be mounted in reverse. but on the PCB works just as well, registers the same
>>107108493>but they are silently deadly to your eyes.*stares into a 4000nit peak oled*nothin personal
PREPARE to get FUCKED haters
> Try to buy a simple TMR development board>125 dollars>ok Ill just make my own!>14-step production making a 3-slit dogbone trace >need a ceramic layer in the PCB> multiple differential amplifiers Yeah.......
There isn't a dedicated mouse thread so I'll use this one to say Keychron's mice are surprisingly really good. I bought an M5 vertical mouse and I'm very pleased with its quality. Also came with USB-A to USB-C adapters in both directions, C to female-A and C to male-A.
>>107110242If I could get the keychron fork of qmk to compile I would mess around with just making the step size .001mm instead of .01mm and then making the cap 400. But I can't make it compile! I tried it on my mac and on my gentoo. Oh well
>>107112955you might be fucked tho. +/-15mT or 150G is on the low side, it's probably saturatehttps://divinikey.com/collections/hall-effect-switches/products/gateron-magnetic-jade-emerald-heavy-tactile-switches>Initial Magnetic Flux: 120±8GS (PCB1.2mm)>Bottom Magnetic Flux: 700±30GS (PCB1.2mm)>>107113449wat lol. you don't need any of that nonsense, just put the damned sensor on a small pcb and connectors so you can add power and read output and a switch footprint so you can install one and that's literally it. and choose a chip that fits the switch magnet flux
>>107114268I'll try building as well see what happens, didn't yet try it
>>107113822yeah the m5 feels like an endgame mouse for methe only path forward is to make it lighter even though it's only 100g
>>107114640my savior anon ;_;I know how to get the first pololu sensor working thats why I'm using itI'd rather use this thing https://www.digikey.com/en/products/detail/allegro-microsystems/CT452-H06MRSN08/22114396but it looks so complicated to get running and its development board is 125 dollars
