🍃 Grovemade are celebrating their 15 year Anniversary with an awesome Sale running for the entire month of June! Check out their premium accessories and enjoy discounts on some of the classic designs before they're retired. 🔗 Shop the sale here → bit.ly/45gBGjy
Let it grow at its rhythm. 38K with only 4 video, that's performance, man. Don't push Salim to play the "youtube" game by click-baiting (GreatScott admitted it), doing stupid joke and storytelling and useless scenery (like Sean Hogdins unfortunately :/) DO NOT FORGET: the making and learning are the "interest" of these video, not your "film making" skill. @Salim: if you do that or create a thumbnail with a wide open mouth, I unsub right away! :P
@@salimbenbouz more detail would be loved. im trying to work out how to do things like a modular keyboard because i think its a really cool project and wanna have a go one day but WOW electronics are hard
This is why i love open source projects, i've had this idea in my head for years but never got around to designing and making it. Now i dont need to and can just tailor these designs to my needs. Awesome project!
Hi, may i ask what kind of knowledge I need to learn inorder to do everything He did in the video? Right now I'm studying C++ in college first year and try to selftaught Python.
Yeah dude, the editing is top notch and the project is even better. Maybe a work space tour for an upcoming video? Looks like a relatively small space but it’s action packed with great stuff.
The interrupt fix can be that you buffer data on the module side. Meaning if there is a press, you put it into a buffer, and when the main modules comes to read it, you are not losing any data. Yes, you will have a small delay, but still with no data loss. I would suggest the next version to have CAN Bus, as its a multi-master system and you can send messages directly to the main host, but also to the modules as well. Awesome project!
The CAN Bus protocol might be appropriate actually for this project, I had read a bit about it in the past, yeah it's probably the right choice. Your buffer fix will probably also do the trick as well, if he doesn't wanna go the extra step and add CAN Bus to all this.
I agree the solution for the interrupt line is buffering it, set a flag and then exit the routine asap. Check the flag when you have time. Another possibility is to simplify the I2c data you are collecting so that you read all the devices every cycle. Also, I2c had different speeds available and dma might be possible to offload the task.
😂 the funny thing is I was going to make a bracket when I was designing the enclosure and I refrained and decided not to waste more time and just use good ol hot glue 😂
I graduated back in the day in electronics but for the past 12+ years I've been pursuing art as a career. Seeing stuff like this makes me want to pick up the soldering iron again
Regarding your performance issues: you may be able to speed things up by writing your signal interpreter in C and compile to a dynamic library (.dll on windows, .so on *NIX). Using the Ctypes (uctypes in circuitpython) module , you can import it as a python module and call your functions while benefiting from C-level performance.
@@joe-skeen Rust is great if you already know Rust and need it to do something. It is a terrible idea if you dont. Learning curve is Dwarven Fortress like.
Dynamic libraries aren't a thing for circuitpython (a loader probably wouldn't even fit in memory). You need to statically compile it into with the runtime. I couldn't find any up to date guides, but I think you just have to copy what the other modules in `circuitpython/shared-module` do.
Just have to say, I love that idea with the golden pads on the side. Being made from small PCB's that slot in and then solders. That is brilliant! Thanks so much for that idea!
Man, I love your content. This video was suggested to me by the algorithm and I clicked it because it is a very specific topic that is of interest to me but also I vaguely remembered watching your video about the first macro pad. As usual when someones content impresses me I checked out your channel and my mind was blown - "omg, it's the same guy that made the DIY space mouse" I thought as I had that video in my 'watch later' for some time now but never got around to actually watching it. Well, now I got around to it as my adhd brain hyperfixated on learning Fusion the last 3 days and already wants more control so it was a perfect combo. In that video I noticed your CRT Rapsberry so naturally I had to watch that video as well and was impressed as well. Your workflow is great and you do a great job presenting it. As well as documenting it on instructables and github? I must be dreaming :o Do you have a background in design? You pay attention to details, your creations are great in function but visually beautiful as well. The idea with a PCB as a front plate instead of a sticker was amazing. You really are an inspiration for me and my projects!
3:52 so I was mildly interested in this video because of the modular aspect, and I was curious how you planned the board layouts, … and then saw this right here 😮 Instant subscribe. This is a masterclass in “it’s not the tool, it’s how you use it.”
We need more videos on RUclips with this quality. 40K subs?! You deserve more man. This video got the editing of a premium commercial video. Keep it up.
I say forget magnets, they're clever, but I'd rather have a physical lock under the pieces to securely keep them together, maybe a slide mechanism or like legos. Realistically, users will setup the parts once, and rarely move them around. Just my two cents on your amazing idea.
Sounds like someone's never used magnets nor a modular platform before. I say you teach yourself some things before you start spewing nonsense. When you do actual work using inputs that aren't a basic keyboard, your setup will evolve as your workflow evolves, literally the entire point of a modular setup, and modular setups are rarely setup and never changed. As for your issue with magnets, you clearly have no basis in reality for that statement, neodymium magnets are more than strong enough for this application and the modules aren't just going to break apart as you use them, not that they even would break apart without intentionally separating them anyways given that they'd only realistically separate under linear opposition as there's no point in the mechanism to skew along the other axes. You clearly lack the capability of logical thought given your stance on this, and you clearly base your stance in combination with lack of logical thought on your lack of experience with anything relevant here considering your lack of understanding. If you choose to be this mindless, then learn to sit down and shut up instead of spewing your mindless shit.
A magnet might not be as strong as a slide lock or lego style stud (which is much harder to disconnect), and I'd personally love some kind of push latch mechanism so you're not sliding on pogo-pins but anything mechanical is going to be a potential breakage - especially with brittle resin printed parts. The magnets might be less secure but there's no sliding pins against each other and no small parts to snap
You can use CAN to make the communication faster so you don't need interrupts, can is really simple on hardware side and can go up to 1Mb/s. I2C is usually 400Kb/s which is not great and is pretty noisy on long bus topologies such as yours. I'm not sure if the rpi mcu has CAN but it most probably has SPI so you can put an SPI to CAN ic to make it work. CAN is exterelemy noise resistant and has many many features that would work great in this project
CAN is a more robust bus system that i agree, but for such a system CAN is too much. I2C has the benefit that it is build in almost all uController. It simpler and cheaper. And if the interrupt system works the I2C bus will be good enough
This is awesome and absolutely hilarious since I've been developing my own modular macropad line based on seeing the creativity console - all the steps you went through are legitimately all the steps I went through as well, even down to chucking an extra interrupt pin. I didn't have the thought to make custom pogo connectors though, that was a super clever design. Nice work! Also to add onto your part at 18:00 - my approach uses a single I2C controller device with some extra logic that auto-assigns I2C addresses after going through a discovery/scan process, which involves the peripheral device responding with its unique chip ID and a module type identifier. All keypresses or actions are buffered and then cleared on an I2C read, but latency issues like you mentioned start to pop up when you get 8+ devices. Cranking up I2C speeds worked for a bit but I am currently redesigning since I started hitting I2C bus capacitance issues at higher speeds.
haha! that's amazing. I'm glad i'm not the only one to think this way. I received a lot of suggestions that I might try and buffering the interactions is one of them. thanks for the taking the time!
A great option for combining all of the modules would actually be to connect those as a USB bus. All modules would be independent, the number of modules would mean nothing and you would get rid of any protocol definition and programming.
@@lanepemberton8886 Well, maybe uneven number of pins can help with reverse connecting options so that you don't need to think how to connect all of those.
Wow glad I clicked on this video. Criminally underrated. Great job on the project! As some who also has built custom macro pads I can only imagine the hours of blood sweet and tears into this and glad to see you got it working. Keep it up!
that is a bananas project. amazing work. also the things that we can now do in personal labs is crazy nowadays. the polish that the resulting modules exhibit is superb!!! also great video editing. somewhat overshadow part of the video since the project is by far the most captivating thing in it.
the only sub 500k channel i watch because of the standard of the content. Your content and production has the qualities of a million subscriber channel.
not trying to tell you what to do since i don't know what you have in mind but just my opinion : don't go hard on the meme-fication stuff. you and your content give off a classy professional vibe and i felt it doesn't suit too much meme-fication. but all upto your taste and the majority of your viewers. i personally have weird tastes sometimes.
I sketched an near identical design concept back in 2012-ish (when 3D printing was in its infancy and I had an early Makerbot Replicator which wasn't really up to the task, plus my embedded knowledge was more rudimentary). Well done for getting so far! If I was going to tackle it today, I would probably program it in Mecrisp Stellaris Forth on a STM32F411, perhaps using a overclocked MIDI scheme or I2C. Forth is ideal for this sort of project - fast real time incremental compiling and testing with instant debugging feedback (unlike C), plus it is an order of magnitude faster than Micro Python.
This is a really impressive project! I also started building custom keyboard because of the super inspirational Mirage keyboard that Zack is working on!
I love this. I would dig faders and even motorized faders alot. Many modern composers use similar tools like the monogram to control the expressions of virtual instruments over midi commands, and its really useful and would be a great cheaper alternative to the monogram.
personally i really like these the only thing i would do differently however is instead of having everything go through one main board have each set of pogo pins connected to a slave controllers that convert the button presses into unique hashes that the main board then sends feedback to the computer.
Will 100% be adding this to my list of future projects. I had the OG Pallet Gear before they changed to Moment and I loved how it worked. Design(s) look beautiful and being able to have those hotswap keys so if I am working in a shared space I can have quieter keys or if I decide to use it for something like gamin I can have a more tactile or clicky key is just genius.
I would investigate an interference joint as well as the magnet. Maybe 2 holes and two posts? Something to give it structure along those weird load paths that knock the magnets out
Brilliant project! Can't find any suggestions as to how to solve the interrupt problem? Here's my take on it: - Every time the master detects a new device, it will assign it an seven bit number as an ID and send this back on the IRQ line. First bit will always be a 1 and you'll never send more than 8 bytes, so you know if an address is incoming. - The device will save the address and use this as it's I2C address. On each event, the device will send it's I2C address on the Interrupt line. - This approach works well with SeeSaw devices since they can take on any I2C address. A nice side-effect is that you don't need to hard-code I2C addresses and thus won't get any conflicts. Now it obviously takes longer to send 8 bits than to just raise a line, but it will still take less time than polling many devices. Using a UART pin to receive the interrupt address, the MCU should be able to buffer whatever address is incoming so you don't loose data. I've formerly used a similar approach with multiple RS485 devices to set up a master device that can talk to multiple slave devices. An alternate approach that scales less is to assign each new device an analog voltage. It's not as solid or scalable, but would allow for near-instant detection. Message "jenschr" on X if you want to discuss?
the fact that you are not an expert in electronics, 3d modeling, 3d printing or product development makes this so much more interesting and inspiring. Thanks for sharing and dedicating your time to these projects.
Cool project and great video! For interrupt issues - instead of pulse count for modules, modules can send one pulse as interrupt and followed by unique module address in 4bit binary..😉
Brilliant project :-) To help with stability and the magnets disconnecting you could print a thin frame that clips around the entire assembly once the modules have been magnetically snapped into position. Different configurations would need different frame sizes but that can be done parametrically in F360, to make it easy to generate then print. I also like the idea of making a v2 that fits Gridfinity: then it could be combined with other Gridfinity modules.
The best way I can think of resolving the issue (and many others down the line) is for each module to have it's own "command buffer", that pools the triggered commands and sends them to the agregator on a "when available" basis. It should decrease latency and avoid ignored commands, as modules that didn't execute any commands will simply return an "empty command", while modules that have several commands can send the command even after you triggered it, so it doesn't depend on the user still be holding the button when the agregator tries to read.
This concept is unreal! Ive yearned for this kind of platform for such a long time to build up a MIDI controller for a portable recording studio Ive built. Using a handful of the faders and the button/encoder combo units would nail that. Also the resin based printing is incredible. It looks flawless compared to extrusion driven prints.
I rarely write comments. Really rare. But given the quality of Salim's content, how good his videos are, how imaginative and expert he is, it would be a sin not to leave a comment here. There are so many DIY RUclipsrs. And then there is Mr. Benbouziyane!
I have been wanting a fader system for a long time now. I was an audio engineer so was very used to them, but became introduced to them on PC for general use with the GoXLR. Loved the ease of use to quickly change the audio volume of an app for chat. The GoXLR just doesnt have quite enough faders, it would be perfect IMO with 6. Sadly Elgato wont do faders, as they arent as powerful as knobs. Very happy to see you make a macropad with faders, and I hope more people get interested in them. I really want a product with 6 faders that present itself as 6 different audio devices in windows like the GoXLR or Elgato devices do so I can map apps, games, browsers, etc to specific output devices and have the faders control that device volume.
I think that you've learned so much in a short amount of time, on top of having good production value makes you a very underrated channel in my opinion.
That chonky dial is definitely the star of the show. What kind of costs all in are we? Just curious what you could expect in this kind of project. Looks great either way, would be pretty cool if that also incorporated a headphone DAC amp :).
For your next project I suggest a variant of this. Make a magnetic board/s of this style for keys, dials, whatever to be popped on top of. Give it gravitational liquid resistance by having the magnets and pins on raised protrusions instead of a flatly aligned with the boards main surface, this will allow liquids to just fall off. The keys, dials, etc should have rubber overhangs that give further liquid resistance by preventing splashes from making their way up to the raised magnets and pins. I'm sure others will have ideas for bettering this one, slap on your comments so salim doesn't need to hunt for them.
Great job on the video! Your explanation of the development process and design decisions is clear and easy to follow. I appreciate that you highlight the weaknesses and suggest improvements. Keep up the excellent work!
Great solution for the magnetic connectors. I think that extending the connectors out, farther proud of the main pcb on each module would give you the ability to thicken the walls on the modules. That way you can have stronger magnets. I have a design for push-to-click square, small oled display buttons and other similar modules as yours but I didn't like the magnetic pogo pin connectors available - now I think that can be solved. To implement some of my ideas, I was considering SPI instead of I2C between modules but was concerned about the interconnect connector voltages and amperage limitations that would accumulate quickly with each subsequent added module as you get further away from the main power source. I wonder if CAN + I2C might be a better strategy?
You’ve become my favourite RUclipsr, Salim. Another really great project and the time and effort you take to explain your successes and failures is refreshing. Looking forward to your next project.
Jesus you made that look so easy, as someone who is completely out of the sphere, I really want to build this stuff but I also would definitely was my money if I just dive head first into something crazy like this
Just a thought - Interrupt means "Someone has a message". Then host says "Who has a message". Client says "I do". Host says "Client 001 - Give me your message". Host could receive multiple "I do's" and process them, and if interrupt stays high after reading messages, it might be because another client is holding it high, so host continues asking until interrupt goes low.
This project looks absolutely cool! The modular idea is really great and makes it easy to experiment with different modules. This could be really interesting and helpful for beginners. It's also nice to see someone showing the problems that occurred while making the project. You often see such great projects with no problems whatsoever. And if you try to make it on your own it doesn't work as easy as it seems on other videos. Really great! 👍
Great, amazing project. Nice work. One thing: I've had a few of those OLEDs and leaving them on consistently (like with your "OCREEB" logo) will eventually burn out some of the pixels (or reduce their brightness compared to other pixels). I would recommend turning the OLED off when the module is not in use so you don't have the same problem.
WOW this was one of the most incredible DIY projects i've seen related to macropads and stuff.. I had a similar project a few years ago, but I lack some technical skills and left this project aside. Now, with your project in mind, I've got inspired again. Thank you! And congratulations for this awesome project!
One thing you'll want to keep an eye on is the the mating contacts of the pogo connections. You'll probably be fine with the HASL coating on those contacts for a while, but over time the pogos will wear through it and overall it will have a lower cycle time than a harder coating. It is a little more expensive, but an ENIG coating will be harder and will last significantly longer. Great video, thanks for sharing!
You are 100% correct. I started seeing some markings where the pogo pins lands. If I do this again I will upgrade at least the landing pads to ENIG. thanks
Wow impressive! Just the amount of time, money and energy you put into this project, just to put it out there for anybody to use for free, is insane! You earned a new subscriber, looking forward to upcoming projects! If you are looking for new module idea´s for this project, i would love to see motorized sliders for use in lightroom.
Get the ESP32 versions of the XIaos (or even better an Espressif WROOM module) and use the ESP-NOW wireless protocol. It's not WiFi and it's not Bluetooth but it is so awesome! That way you can even add batteries to your boxes and simply stick them together when you need a charge or wanna draw lots more power for fancy RGB.
Love the project. Suggestion I had for aesthetics could solve the strength issue in a single vertical 1x2 unit. Create some caps for the sides. You have them click into the magnet slots, and the default layout is a wide side on the magnet side and a 45 miter to a shorter outer side (trapezoid with the magnets on the long side) this will take care of the inside corners where this issue is most serious. You can then have pieces that are flipped trapazoids (2 outer corners) and a left and right side rhomboid (1 outer 1 inner). If you don't want so many different pieces. Purely have the first trapazoids (2 inner corners) and then 2 end caps, one that connects 2 inner corners side by side and another that connects 2 inner corners on an outside corner (270 degrees if an inner is 90 degrees) All these end pieces should have a 20-40 degree bevel from the inside to the outside. Could also just make these all 3D prints that just have a clip in magnet holder on the underside near the snap points.
I would love to convince you to make these on commission because theres a hole in a certain market and these would be very very welcome additions to that market which is already used to spending money on large projects...I don't have the tools or experience to put this together even for myself...
I love how this project is one any one can benefit from but I do have some stuff that want to ask for a possible inclusion. 1 CAN bus it use in car it have protection against the interrupt system long check time so the module can react instantly. 2 hard mount as mentioned in another comment the project can be benefited from a hard mount system via screw or the gridfinity system or similar. But with all that said the project is very cool and I might consider building this.
Just discovered your channel. You are developing and tinkering with things, just like I would love to do if I had the time these days. Keep up the great work 👍
One option to fix the problem with the I2C polling missing the event is to reprogram the seesaw module to save the current state of each attached device and push that when requested. Buttons could also be programmed to latch when pressed until the I2C bus can send the info that it was pressed. Those seesaw boards have an ATtiny on them so the possibilities are basically endless!
USB keeps track of various IDs (device IDs, manufacturer IDs, positional IDs, etc), as well as a fairly hard limit as to how many devices can be connected to a bus. How about use hardcoded IDs on the devices and using a size-limited (ring?) buffer keeping track of only the most recent connections/disconnections rather than polling every single device in the chain every interrupt? That's how we usually do it in software synthesizers when we need to "steal notes" for MIDI note playback. Admittedly, the most immediate downside I can think of is the small possibility of the buffer list of attached devices getting out of sync with the actual set of attached devices, but maybe you handle that forced resync in a once-a-minute idle refresh poll or something.
you are such an inspiration!!! i can't believe how many disciplines you have under your belt to be able to make such a bespoke hardware interface. i just subscribed
Head this idea in mind for a very long time as well, I'm very impressed with your work and passion. Thanks to show it and mostly, to make it even open source. Thanks a lot
🍃 Grovemade are celebrating their 15 year Anniversary with an awesome Sale running for the entire month of June!
Check out their premium accessories and enjoy discounts on some of the classic designs before they're retired.
🔗 Shop the sale here → bit.ly/45gBGjy
The content on this channel is WAY too good to only have 36k subs. Hope your channel blows up, my dude. Sweet project, as always.
Agreed!
What my sub isn't good enough for you?
Yess ❤
Just subbed. Luckiest find of 2024 so far!!!
Let it grow at its rhythm. 38K with only 4 video, that's performance, man.
Don't push Salim to play the "youtube" game by click-baiting (GreatScott admitted it), doing stupid joke and storytelling and useless scenery (like Sean Hogdins unfortunately :/)
DO NOT FORGET: the making and learning are the "interest" of these video, not your "film making" skill.
@Salim: if you do that or create a thumbnail with a wide open mouth, I unsub right away! :P
"I'm not an expert." Made me chuckle to mask the tears of my failures
I’m glad people are enjoying the process. I'm always reluctant to include a lot of detail for fear of producing a boring video
@@salimbenbouz more detail would be loved. im trying to work out how to do things like a modular keyboard because i think its a really cool project and wanna have a go one day but WOW electronics are hard
@@salimbenbouz maybe a second channel with the same video but longer? :D
This is why i love open source projects, i've had this idea in my head for years but never got around to designing and making it. Now i dont need to and can just tailor these designs to my needs. Awesome project!
Same feelings!
Hi, may i ask what kind of knowledge I need to learn inorder to do everything He did in the video? Right now I'm studying C++ in college first year and try to selftaught Python.
Yeah dude, the editing is top notch and the project is even better. Maybe a work space tour for an upcoming video? Looks like a relatively small space but it’s action packed with great stuff.
Thanks! it's a small space indeed but I try to make it work. Maybe some day we do a tour yeah!
The interrupt fix can be that you buffer data on the module side. Meaning if there is a press, you put it into a buffer, and when the main modules comes to read it, you are not losing any data. Yes, you will have a small delay, but still with no data loss.
I would suggest the next version to have CAN Bus, as its a multi-master system and you can send messages directly to the main host, but also to the modules as well.
Awesome project!
The CAN Bus protocol might be appropriate actually for this project, I had read a bit about it in the past, yeah it's probably the right choice. Your buffer fix will probably also do the trick as well, if he doesn't wanna go the extra step and add CAN Bus to all this.
I agree the solution for the interrupt line is buffering it, set a flag and then exit the routine asap. Check the flag when you have time. Another possibility is to simplify the I2c data you are collecting so that you read all the devices every cycle. Also, I2c had different speeds available and dma might be possible to offload the task.
Everything was so clean. I almost got depressed about my DIY until you added those tiny bits of hot glue) What a relief)
😂 the funny thing is I was going to make a bracket when I was designing the enclosure and I refrained and decided not to waste more time and just use good ol hot glue 😂
I graduated back in the day in electronics but for the past 12+ years I've been pursuing art as a career. Seeing stuff like this makes me want to pick up the soldering iron again
To say I'm impressed would be an understatement, this video inspired and sparked a motivation in me.
Thank you for making this video
Thank you! I appreciate it
Regarding your performance issues: you may be able to speed things up by writing your signal interpreter in C and compile to a dynamic library (.dll on windows, .so on *NIX). Using the Ctypes (uctypes in circuitpython) module , you can import it as a python module and call your functions while benefiting from C-level performance.
Thanks for the suggestion! I will look into it. You're right maybe this will work way faster for the custom address pulse I was tinkering with.
Or even Rust
@@joe-skeen Rust is great if you already know Rust and need it to do something. It is a terrible idea if you dont. Learning curve is Dwarven Fortress like.
Dynamic libraries aren't a thing for circuitpython (a loader probably wouldn't even fit in memory). You need to statically compile it into with the runtime. I couldn't find any up to date guides, but I think you just have to copy what the other modules in `circuitpython/shared-module` do.
Just have to say, I love that idea with the golden pads on the side. Being made from small PCB's that slot in and then solders. That is brilliant! Thanks so much for that idea!
Man, I love your content. This video was suggested to me by the algorithm and I clicked it because it is a very specific topic that is of interest to me but also I vaguely remembered watching your video about the first macro pad. As usual when someones content impresses me I checked out your channel and my mind was blown - "omg, it's the same guy that made the DIY space mouse" I thought as I had that video in my 'watch later' for some time now but never got around to actually watching it. Well, now I got around to it as my adhd brain hyperfixated on learning Fusion the last 3 days and already wants more control so it was a perfect combo. In that video I noticed your CRT Rapsberry so naturally I had to watch that video as well and was impressed as well. Your workflow is great and you do a great job presenting it. As well as documenting it on instructables and github? I must be dreaming :o Do you have a background in design? You pay attention to details, your creations are great in function but visually beautiful as well. The idea with a PCB as a front plate instead of a sticker was amazing. You really are an inspiration for me and my projects!
3:52 so I was mildly interested in this video because of the modular aspect, and I was curious how you planned the board layouts, … and then saw this right here 😮
Instant subscribe. This is a masterclass in “it’s not the tool, it’s how you use it.”
thank you! I appreciate it
We need more videos on RUclips with this quality. 40K subs?! You deserve more man. This video got the editing of a premium commercial video. Keep it up.
I say forget magnets, they're clever, but I'd rather have a physical lock under the pieces to securely keep them together, maybe a slide mechanism or like legos. Realistically, users will setup the parts once, and rarely move them around. Just my two cents on your amazing idea.
I think it should be combined with gridfinity
Sounds like someone's never used magnets nor a modular platform before. I say you teach yourself some things before you start spewing nonsense. When you do actual work using inputs that aren't a basic keyboard, your setup will evolve as your workflow evolves, literally the entire point of a modular setup, and modular setups are rarely setup and never changed. As for your issue with magnets, you clearly have no basis in reality for that statement, neodymium magnets are more than strong enough for this application and the modules aren't just going to break apart as you use them, not that they even would break apart without intentionally separating them anyways given that they'd only realistically separate under linear opposition as there's no point in the mechanism to skew along the other axes. You clearly lack the capability of logical thought given your stance on this, and you clearly base your stance in combination with lack of logical thought on your lack of experience with anything relevant here considering your lack of understanding. If you choose to be this mindless, then learn to sit down and shut up instead of spewing your mindless shit.
A magnet might not be as strong as a slide lock or lego style stud (which is much harder to disconnect), and I'd personally love some kind of push latch mechanism so you're not sliding on pogo-pins but anything mechanical is going to be a potential breakage - especially with brittle resin printed parts. The magnets might be less secure but there's no sliding pins against each other and no small parts to snap
Dude, I am blown away by your creativity and technical follow through. Huge inspiration. Thank you!
This is such fantastic production quality! The editor is genius and the way they are structured are absolutely fantastic!!! Well done, Salim!
You can use CAN to make the communication faster so you don't need interrupts, can is really simple on hardware side and can go up to 1Mb/s. I2C is usually 400Kb/s which is not great and is pretty noisy on long bus topologies such as yours. I'm not sure if the rpi mcu has CAN but it most probably has SPI so you can put an SPI to CAN ic to make it work. CAN is exterelemy noise resistant and has many many features that would work great in this project
CAN is a more robust bus system that i agree, but for such a system CAN is too much. I2C has the benefit that it is build in almost all uController. It simpler and cheaper. And if the interrupt system works the I2C bus will be good enough
CDBUS is similar to CAN, but much faster and simpler than CAN, we should use UART controller like CDCTL01A.
@@dukelec This looks super cool, but sourcing the chip seems like a pain, I could only find alibaba links to it. I would love to get a reel of this
That will definitely blow up. Almost full coverage of subscribers is 19 hours, amazing quality and cool b-roll. Keep going and you'll get it!
This is awesome and absolutely hilarious since I've been developing my own modular macropad line based on seeing the creativity console - all the steps you went through are legitimately all the steps I went through as well, even down to chucking an extra interrupt pin. I didn't have the thought to make custom pogo connectors though, that was a super clever design. Nice work!
Also to add onto your part at 18:00 - my approach uses a single I2C controller device with some extra logic that auto-assigns I2C addresses after going through a discovery/scan process, which involves the peripheral device responding with its unique chip ID and a module type identifier. All keypresses or actions are buffered and then cleared on an I2C read, but latency issues like you mentioned start to pop up when you get 8+ devices. Cranking up I2C speeds worked for a bit but I am currently redesigning since I started hitting I2C bus capacitance issues at higher speeds.
haha! that's amazing. I'm glad i'm not the only one to think this way.
I received a lot of suggestions that I might try and buffering the interactions is one of them. thanks for the taking the time!
A great option for combining all of the modules would actually be to connect those as a USB bus. All modules would be independent, the number of modules would mean nothing and you would get rid of any protocol definition and programming.
Also one less pin in the connector
@@lanepemberton8886 Well, maybe uneven number of pins can help with reverse connecting options so that you don't need to think how to connect all of those.
Wow glad I clicked on this video. Criminally underrated. Great job on the project! As some who also has built custom macro pads I can only imagine the hours of blood sweet and tears into this and glad to see you got it working. Keep it up!
that is a bananas project. amazing work. also the things that we can now do in personal labs is crazy nowadays. the polish that the resulting modules exhibit is superb!!! also great video editing. somewhat overshadow part of the video since the project is by far the most captivating thing in it.
I won't build something like this. But your dedication on this project amaze me man.
awesome!!!, brazilian mechatronics engineering student here, you are soo good doing these things, bro haha
@im_Carlisle yeah, something in the middle of two others, isn't so recognized cus is "something new" but, at the same time, is in a rising way :)
the only sub 500k channel i watch because of the standard of the content. Your content and production has the qualities of a million subscriber channel.
not trying to tell you what to do since i don't know what you have in mind but just my opinion : don't go hard on the meme-fication stuff. you and your content give off a classy professional vibe and i felt it doesn't suit too much meme-fication. but all upto your taste and the majority of your viewers. i personally have weird tastes sometimes.
I sketched an near identical design concept back in 2012-ish (when 3D printing was in its infancy and I had an early Makerbot Replicator which wasn't really up to the task, plus my embedded knowledge was more rudimentary). Well done for getting so far!
If I was going to tackle it today, I would probably program it in Mecrisp Stellaris Forth on a STM32F411, perhaps using a overclocked MIDI scheme or I2C.
Forth is ideal for this sort of project - fast real time incremental compiling and testing with instant debugging feedback (unlike C), plus it is an order of magnitude faster than Micro Python.
This is a really impressive project! I also started building custom keyboard because of the super inspirational Mirage keyboard that Zack is working on!
I love this.
I would dig faders and even motorized faders alot.
Many modern composers use similar tools like the monogram to control the expressions of virtual instruments over midi commands, and its really useful and would be a great cheaper alternative to the monogram.
personally i really like these the only thing i would do differently however is instead of having everything go through one main board have each set of pogo pins connected to a slave controllers that convert the button presses into unique hashes that the main board then sends feedback to the computer.
Will 100% be adding this to my list of future projects. I had the OG Pallet Gear before they changed to Moment and I loved how it worked. Design(s) look beautiful and being able to have those hotswap keys so if I am working in a shared space I can have quieter keys or if I decide to use it for something like gamin I can have a more tactile or clicky key is just genius.
I would investigate an interference joint as well as the magnet. Maybe 2 holes and two posts? Something to give it structure along those weird load paths that knock the magnets out
Brilliant project!
Can't find any suggestions as to how to solve the interrupt problem? Here's my take on it:
- Every time the master detects a new device, it will assign it an seven bit number as an ID and send this back on the IRQ line. First bit will always be a 1 and you'll never send more than 8 bytes, so you know if an address is incoming.
- The device will save the address and use this as it's I2C address. On each event, the device will send it's I2C address on the Interrupt line.
- This approach works well with SeeSaw devices since they can take on any I2C address. A nice side-effect is that you don't need to hard-code I2C addresses and thus won't get any conflicts.
Now it obviously takes longer to send 8 bits than to just raise a line, but it will still take less time than polling many devices. Using a UART pin to receive the interrupt address, the MCU should be able to buffer whatever address is incoming so you don't loose data. I've formerly used a similar approach with multiple RS485 devices to set up a master device that can talk to multiple slave devices.
An alternate approach that scales less is to assign each new device an analog voltage. It's not as solid or scalable, but would allow for near-instant detection. Message "jenschr" on X if you want to discuss?
the fact that you are not an expert in electronics, 3d modeling, 3d printing or product development makes this so much more interesting and inspiring. Thanks for sharing and dedicating your time to these projects.
Wow, That's a great idea. Just Subscribed
Incredible production value, both of the build and the video. What a gem!
Cool project and great video!
For interrupt issues - instead of pulse count for modules, modules can send one pulse as interrupt and followed by unique module address in 4bit binary..😉
I have hope for the world.
There are so many incredibly talented people out there. 👏
Brilliant project :-) To help with stability and the magnets disconnecting you could print a thin frame that clips around the entire assembly once the modules have been magnetically snapped into position. Different configurations would need different frame sizes but that can be done parametrically in F360, to make it easy to generate then print. I also like the idea of making a v2 that fits Gridfinity: then it could be combined with other Gridfinity modules.
Wow this is fantastic!
Thanks Matthew!
The amount of knowledge and abilities put into this is just ovewhelming, awesome project. You just got a new subscriber.
The fact that you made these all by yourself is impressive. Great work!
Amazing build!
The best way I can think of resolving the issue (and many others down the line) is for each module to have it's own "command buffer", that pools the triggered commands and sends them to the agregator on a "when available" basis.
It should decrease latency and avoid ignored commands, as modules that didn't execute any commands will simply return an "empty command", while modules that have several commands can send the command even after you triggered it, so it doesn't depend on the user still be holding the button when the agregator tries to read.
This concept is unreal!
Ive yearned for this kind of platform for such a long time to build up a MIDI controller for a portable recording studio Ive built. Using a handful of the faders and the button/encoder combo units would nail that.
Also the resin based printing is incredible. It looks flawless compared to extrusion driven prints.
I rarely write comments. Really rare. But given the quality of Salim's content, how good his videos are, how imaginative and expert he is, it would be a sin not to leave a comment here.
There are so many DIY RUclipsrs. And then there is Mr. Benbouziyane!
You earned a subscriber. Congrats on the production quality and product quality!
holy hell, HeyGears is amazing. I cannot believe I've never heard of them until this video!
I have been wanting a fader system for a long time now. I was an audio engineer so was very used to them, but became introduced to them on PC for general use with the GoXLR. Loved the ease of use to quickly change the audio volume of an app for chat. The GoXLR just doesnt have quite enough faders, it would be perfect IMO with 6. Sadly Elgato wont do faders, as they arent as powerful as knobs. Very happy to see you make a macropad with faders, and I hope more people get interested in them. I really want a product with 6 faders that present itself as 6 different audio devices in windows like the GoXLR or Elgato devices do so I can map apps, games, browsers, etc to specific output devices and have the faders control that device volume.
I think that you've learned so much in a short amount of time, on top of having good production value makes you a very underrated channel in my opinion.
That chonky dial is definitely the star of the show. What kind of costs all in are we? Just curious what you could expect in this kind of project. Looks great either way, would be pretty cool if that also incorporated a headphone DAC amp :).
For your next project I suggest a variant of this. Make a magnetic board/s of this style for keys, dials, whatever to be popped on top of. Give it gravitational liquid resistance by having the magnets and pins on raised protrusions instead of a flatly aligned with the boards main surface, this will allow liquids to just fall off. The keys, dials, etc should have rubber overhangs that give further liquid resistance by preventing splashes from making their way up to the raised magnets and pins. I'm sure others will have ideas for bettering this one, slap on your comments so salim doesn't need to hunt for them.
Great job on the video! Your explanation of the development process and design decisions is clear and easy to follow. I appreciate that you highlight the weaknesses and suggest improvements. Keep up the excellent work!
Great solution for the magnetic connectors. I think that extending the connectors out, farther proud of the main pcb on each module would give you the ability to thicken the walls on the modules. That way you can have stronger magnets.
I have a design for push-to-click square, small oled display buttons and other similar modules as yours but I didn't like the magnetic pogo pin connectors available - now I think that can be solved.
To implement some of my ideas, I was considering SPI instead of I2C between modules but was concerned about the interconnect connector voltages and amperage limitations that would accumulate quickly with each subsequent added module as you get further away from the main power source.
I wonder if CAN + I2C might be a better strategy?
Damn, where have you studied all that tech stuff?
Your projects are insanely good! Mind-blowing attention to detail from product design to video editing.
thank you!
Man, this is awesome! I currently use a shuttle and numpad + slightly macro'd 10keyless to edit and I dream of a custom set up like this - great work
You’ve become my favourite RUclipsr, Salim. Another really great project and the time and effort you take to explain your successes and failures is refreshing. Looking forward to your next project.
This is amazing. I don’t know half the terms you used but this was so cool. That printer is wild.
that little magnet jig was a big brain move, I love this build where do I order one LOL
Jesus you made that look so easy, as someone who is completely out of the sphere, I really want to build this stuff but I also would definitely was my money if I just dive head first into something crazy like this
You make things at a quality level and competence that I think most hobbyists only ever dream of
Thank you! it might seem that way in the video but i'm a total noob still :)
What an insane project to share with us, i loved every second of it, Thanks for sharing this with us.
I would love to use this for Lightroom AND Davinci resolve timeline or grading. This is so compact and came out looking awesome
Just a thought - Interrupt means "Someone has a message". Then host says "Who has a message". Client says "I do". Host says "Client 001 - Give me your message". Host could receive multiple "I do's" and process them, and if interrupt stays high after reading messages, it might be because another client is holding it high, so host continues asking until interrupt goes low.
Also, not sure if this multiple back and forth is even worth it vs. just getting messages directly.
This project looks absolutely cool! The modular idea is really great and makes it easy to experiment with different modules. This could be really interesting and helpful for beginners. It's also nice to see someone showing the problems that occurred while making the project. You often see such great projects with no problems whatsoever. And if you try to make it on your own it doesn't work as easy as it seems on other videos. Really great! 👍
Great, amazing project. Nice work.
One thing: I've had a few of those OLEDs and leaving them on consistently (like with your "OCREEB" logo) will eventually burn out some of the pixels (or reduce their brightness compared to other pixels). I would recommend turning the OLED off when the module is not in use so you don't have the same problem.
Thanks! yeah you are absolutely right and I was thinking the same thing. I will probably have it sleep when not in use at some point.
WOW this was one of the most incredible DIY projects i've seen related to macropads and stuff.. I had a similar project a few years ago, but I lack some technical skills and left this project aside. Now, with your project in mind, I've got inspired again. Thank you! And congratulations for this awesome project!
One thing you'll want to keep an eye on is the the mating contacts of the pogo connections. You'll probably be fine with the HASL coating on those contacts for a while, but over time the pogos will wear through it and overall it will have a lower cycle time than a harder coating. It is a little more expensive, but an ENIG coating will be harder and will last significantly longer. Great video, thanks for sharing!
You are 100% correct. I started seeing some markings where the pogo pins lands. If I do this again I will upgrade at least the landing pads to ENIG. thanks
Wow impressive! Just the amount of time, money and energy you put into this project, just to put it out there for anybody to use for free, is insane! You earned a new subscriber, looking forward to upcoming projects!
If you are looking for new module idea´s for this project, i would love to see motorized sliders for use in lightroom.
Thank you!
Damn this is really next level, fantastic project and quality content! - Happy to finally the YT waking up and promoting creators that deserve it.
This is remarkable. I am going to be tuning in deeply to this project.
Get the ESP32 versions of the XIaos (or even better an Espressif WROOM module) and use the ESP-NOW wireless protocol. It's not WiFi and it's not Bluetooth but it is so awesome! That way you can even add batteries to your boxes and simply stick them together when you need a charge or wanna draw lots more power for fancy RGB.
That's a Dream 3D Printing Setup. Loved the results !
Can't appreciate sharing your inspirations enough, brings much more depth to explaining the design thinking of the project.
Wow. Very high production value on this video! I'm impressed! You won a subscribe for that alone. Not to mention KEEEEEEEEB!
Love the project. Suggestion I had for aesthetics could solve the strength issue in a single vertical 1x2 unit. Create some caps for the sides. You have them click into the magnet slots, and the default layout is a wide side on the magnet side and a 45 miter to a shorter outer side (trapezoid with the magnets on the long side) this will take care of the inside corners where this issue is most serious. You can then have pieces that are flipped trapazoids (2 outer corners) and a left and right side rhomboid (1 outer 1 inner). If you don't want so many different pieces. Purely have the first trapazoids (2 inner corners) and then 2 end caps, one that connects 2 inner corners side by side and another that connects 2 inner corners on an outside corner (270 degrees if an inner is 90 degrees)
All these end pieces should have a 20-40 degree bevel from the inside to the outside.
Could also just make these all 3D prints that just have a clip in magnet holder on the underside near the snap points.
I would love to convince you to make these on commission because theres a hole in a certain market and these would be very very welcome additions to that market which is already used to spending money on large projects...I don't have the tools or experience to put this together even for myself...
Absolute Legend!
For the shape of the footprint you could also go the way ISO paper sizes are handled. That is, the aspect ratio between sides is the square root of 2.
Might be my favourite youtuber, legit excited for uploads with only a couple vids out!
Amazing as always Salim!!! Thanks for all the help with my projects recently too. Love your channel!
This was a really fun watch. I'm glad this popped up on my feed!
I've been planning to do this for sooooooooo long. Nice someone finally made it. Great work. Subscribed.
That's some of the BEST maker content out there. Keep it up mate!
the project is cool, but the videography is on a different level. keep it up!
What a killer project man, video was super in depth and well edited
Seriously a great video. hope you keep making content because you are bound to blow up.
I love how this project is one any one can benefit from but I do have some stuff that want to ask for a possible inclusion. 1 CAN bus it use in car it have protection against the interrupt system long check time so the module can react instantly. 2 hard mount as mentioned in another comment the project can be benefited from a hard mount system via screw or the gridfinity system or similar. But with all that said the project is very cool and I might consider building this.
Just discovered your channel. You are developing and tinkering with things, just like I would love to do if I had the time these days. Keep up the great work 👍
One option to fix the problem with the I2C polling missing the event is to reprogram the seesaw module to save the current state of each attached device and push that when requested. Buttons could also be programmed to latch when pressed until the I2C bus can send the info that it was pressed.
Those seesaw boards have an ATtiny on them so the possibilities are basically endless!
thanks for the suggestion! I will look into it
USB keeps track of various IDs (device IDs, manufacturer IDs, positional IDs, etc), as well as a fairly hard limit as to how many devices can be connected to a bus. How about use hardcoded IDs on the devices and using a size-limited (ring?) buffer keeping track of only the most recent connections/disconnections rather than polling every single device in the chain every interrupt? That's how we usually do it in software synthesizers when we need to "steal notes" for MIDI note playback. Admittedly, the most immediate downside I can think of is the small possibility of the buffer list of attached devices getting out of sync with the actual set of attached devices, but maybe you handle that forced resync in a once-a-minute idle refresh poll or something.
I just found this channel. I need a desk showcase, it looks so clean. Also really cool stuff, subscribing.
you are such an inspiration!!! i can't believe how many disciplines you have under your belt to be able to make such a bespoke hardware interface. i just subscribed
Head this idea in mind for a very long time as well, I'm very impressed with your work and passion. Thanks to show it and mostly, to make it even open source. Thanks a lot
you are quite good. with persistence and consistency (a quality video every couple months is great) you will continue to do well
Your videos,organitation and design are the perfection
Professional level production on both video and product.
Need more videos and the workspace setup is so on point!
Next step: motorize the sliders 😎