What I’ve Learned Using the DIYson - DIYson Lamp Build Log #11
HTML-код
- Опубликовано: 26 сен 2024
- Since the last build log I've been using the DIYson as a real-world task light and I've learned a lot. During my testing I uncovered a couple of hard truths about my lamp which highlighted some of my weak points as a maker. It's been a humbling experience confronting and trying to overcome these issues, and this build log documents that process.
I'll be at Open Sauce 2024! Please say "hello" if you see me. I might have stickers.
opensauce.com/
The DIYson is my attempt to build an open-source version of the Dyson Solar Cycle lamp using primarily 3D printed and off-the-shelf parts. The project repository and working parts list can be found here: github.com/ste...
Follow on Instagram for updates:
/ stevenbennettmakes
Join the Discord:
/ discord
My Favorite Filament: amzn.to/3HiAHnv
(Amazon Affiliate link)
Thanks for the thoughtful and friendly comments, everyone!❤
This has been such an interesting and educational project to follow.
Was hoping to find the next build vlog as it has been a while.
But more importantly, I hope all is well with you.
This is one of the best DIY series I've ever seen. The way you explain things is simply incredible! What truly sets it apart is how you walk through each step, including mistakes and less effective solutions, explaining your thought process throughout. By showing why one solution is better than another, you provide an excellent teaching method that highlights both the good and the bad. It gives the viewer a full 360-degree understanding of the solution. Great work! Keep it up - looking forward to your next video :)
We are back! 🎉
🎉
There aren't many channels or series I get excited about when I see a new upload. But this never fails to excite me!
The amount of passion and attention to details is just incredible!
came here to say this!
Have you considered a data-driven approach? During regular use, every time you manually set the brightness, have the DIYson record the ambient light and user-set brightness. After a week or so of use in various conditions, you’ll have enough data to plot and extract the ideal curve.
I've followed this series very intently because it shows how difficult making a "real" product is, and it encourages me when my first attempts at a project fail/suck. Thanks for not making it look easy.
Also, I really love the micro schematics you show to help us understand how everything works, as well as the code. You're teaching, not just sharing.
Great to see the project is moving forward. You've made it past an area where I tend to get stuck on my projects, I get a prototype working "good enough" to do what I want, and never truly finish the project after that.
Same for me, wires dangling included.
This is one of the best DIY series I've ever seen. The way you explain things is simply incredible! What truly sets it apart is how you walk through each step, including mistakes and less effective solutions, explaining your thought process throughout. By showing why one solution is better than another, you provide an excellent teaching method that highlights both the good and the bad. It gives the viewer a full 360-degree understanding of the solution. Great work! Keep it up - looking forward to your next video :)
So glad you're still working on the Diyson! I am excited to build it too, once you're happy with the end result!
This dev log series is excellent. An idea to consider for the auto brightness: 2 light sensors, the one you have on top, and add one underneath. The auto-brightness adjusts the light brightness to "match" brightness below with the ambient brightness above so your workspace is never lost in shadow. The curves for the top and bottom sensors won't match completely, else you'd have no light in a dark room, but the auto-brightness could converge when ambient light is bright, and remain close-but-separate when ambient light is dark.
Best part of this project is that I can try this out when I go to make my own diyson.
Awesome update! Keen to build one of these myself at some point ^_^
I think auto-brightness could be improved to be tune-able by user input. What I mean is, you say at 9:13 "I want to take the brightness of the room and add one to it". What would be cool is if you could use the buttons to say "hey I actually want to take the brightness of the room and add two, or add three". This way auto-brightness is still running and adjusting for ambient light, but you can also bias it towards whatever value you're feeling at a given point in time.
You could then have a routine where a long-press or double-tap of one of the buttons (or some other kind of special input) manually toggles on/off auto-brightness, allowing you to still manually control it if desired.
Yeah this is exactly what I would want in a lamp like that
No idea how I stumbled to this great video and build log. I suspect it's from my 3D Printing searches and other things like regular Adam Savage viewings. But I'm loving my stupidly expensive Tizio Artemide Lamp and I bought it second hand. It has huge range of motion and as I'm getting more and more involved with ESP32 and HomeAssistant, this DIYson Lamp is sure one to get me interested in another project. Can't wait for everything! Great work!
These updates are always such a dopamine hit! Thanks for taking us on the journey. 💡
Discovered this project a last week. Started from one I watched an episode every evening. I really like your calm and analytic way to address the obstacles in the project. That is really inspirational to take part in (albeit as a passive observer). Looking forward to see what is in store for the final phase of the project.
You got yourself a new subscriber. Keep up the good work!
Great stuff, can’t wait to build it myself once the project is ready ❤
Wow thank you so much, that is so nice of you! This will go directly into DIYson development. Thank you for the support
continual top-tier work as usual! hopefully ill run into you at Open Sauce!
OH I LOVE IT when you come out with a new video. You could never finish it and I would still watch every video. Entertaining, educational, and somehow relaxing. Keep up the good work my man!
Wow! Your really talented! Keep going man! I want to see whats its like when its finished!
Love when new Diyson videos come on my feed.🎉
Wooooo love these vidsss! Glad to see another one posted’
The analog brightness control is really great! I’ve always felt like LED drivers based on PWM for brightness were kind of taking the cheap and annoying way out. This driver seems to be doing something a lot more sensible!
wow, very cool! I use this Dyson lamp everyday and always marveled at it's clever engineering.
my favorite RUclipsr is back
Was just thinking about this project the other day, glad to get an update
Thanks, Jake!
I’m still impatiently waiting to build my own 🔥
This is my favourite youtube show now.
Thanks for the update, great to see you're enjoying it! You could consider switching to a Bridgelux Thrive COB for maximum CRI (at the cost of a little brightness of course). Also keep in mind that LEDs aren't really meant to be driven below their rated current, the output spectrum changes which influences CRI values. If you don't notice this then I'd say don't worry about it
Not meant to be driven below rated current? Are you sure?
Nowadays computer monitors use LED backlights driven by constant current, as verified by many test reports. This means billions of LEDs are regularly driven below their rated current in mass-produced devices. Given the color rendering is so dependent on the backlight, i would expect LEDs react in a predictable and fairly benign way to lower currents.
Hello, regarding the rotation of the desk lamp on the Z axis, you can use a DC plug and fix one end to the aluminum profile and the other end to the base. Of course, additional 3D printed parts are needed to fix the desk lamp. You can choose the diameter. The 60mm bearing helps fix the desk lamp. I wish your project goes well. I hope it can help you.
Welcome back! Hope you're doing well!
About the ambient light sensor, I think there should be multiple sensors to calculate the actual ambient light in the room. A number of sensors pointing at different directions should do the job. I am saying this because when I use auto brightness on my iPad or phone, it depends on how much light falls into the sensor only. if there is a light in the ceiling the brightness changes when I lay it flat on the table or when I tilt it to write. I find that a bit annoying.
Diamond age - love that book!
This series means so much to me. I was at opensauce, wish I ran into you!
When I see the pogo pins rubbing along the tracks, it makes me think they ought to be rolling rather than rubbing. I imagine like two little brass wheels separated by some plastic spacer, "strung" like beads onto some springy wire, that's then bent into shape so it can hold the wheels against the track with some pressure while hopefully allowing them to spin fairly freely (if the interface between the wire and the inside of the wheels were lubricated perhaps)
But idk how practical that is if the goal is off the shelf parts
Great pace on your videos! Amazing content!
One thing I've learned is the Solar Cycle is a shorter version of the Lightcycle. Good luck on your project. I still can't believe the only lamps I have found on the market that provide high quality lighting and continuous brightness and color temperature control are expensive and very few.
Now this, this is the way. Not a bodge, but a proper fix with the new LED driver.
See you at opensauce! love the lamp progress
Pleaseeee come back sooon for the final one !!!
Quick question, what app are you using to create diagrams such as the ones in 7:54 - 7:57? They look really neat!
Fav series on YT
I'll definitely build this when you finalize your work :)
I was using some fancy bearing that had zero deflection as it spun. It was pieces of Teflon instead of bearings. It was super smooth, didn’t need grease and whatnot. It was pretty fantastic. Maybe do something like that with the horizontal slider. A square tube with a Teflon bearing that fits perfectly to the shape of the bar with tracks.
FINALLY ! We are back on track! Thanks man i love this project
I’ve been enjoying my DIYson express. It was practically the first thing I made when I first got my 3D printer. The battery powered lights are a little annoying, though, because I keep forgetting to charge them, so hopefully I can upgrade to the full version at some point.
Already psyched for the next one!
Glad to hear it! I think it's going to be a big one :)
Great stuff! Always worth the wait mate, keep up the good work 🙌
Finally a new entry, thank you for this one!
Adjustment of the driving current to the LED can cause significant color shift at low levels, but may not be noticeable in your range of brightness... I did LED and LED driver development for 6 years, about 3 billion drivers shipped...
It would be interesting to experiment with having a distance sensor on the underside. You could make a "Focus"-function that automatically increases brightness when its getting pulled closer to a surface, like when you have to solder small wires etc. In this way you could go from cozy ambient light to work light without touching buttons. You could also flip the relation if you want a constant light intensity no matter the distance to an object. In this way, you effectively only alter the light radius on a surface. These features could also just be super annoying, but a test would reveal that. :D
Super awesome progress. I think buying aluminum plates and maybe metal V wheels to stiffen the joints will solve the pogo pin issue. I know that changes the 3d printed aspect, but...trade off might be worth it. Also, maybe small carbon brushes like from DC motors instead of header pins? Industrial Slip rings are typically carbon brushes
since you already have that circular design on the top, it looks like the perfect place to put a knob to control brightness instead of buttons, which seems more intuitive. and following that, some knobs come with a push down type button under them, so you could use that toggle the autobrightness. of course, you would need to deal with how to press the button without repositioning the lamp :V. if i missed a previous log where you considered this my bad !! but i love this design, it's great!
An easy retrofit way to fix PWM flicker is to add a pi filter after the PWM driver with a corner frequency about an octave or several below PWM frequency. The filter looks like inductor-capacitor-inductor. The capacitor is effectively parallel with the load, while the inductors are inline. You can also do just one inductor at a minimum instead of two.
A friend of mine who's working on hobbyists aviation machines once told me that he and his crew have started to work under "focus lighting" conditions. Illimunating just the areas of work, and storage/tool areas, everything else is kept dark. I've never heard of that before, but it kinda makes sense. And i guess thats what's happening with the DIYson in the first minutes of this video as well.
Paramotors perhaps?
Amazing update!
LED driver noise is often from the PWM switching causing the ceramic output capacitor to buzz...not related to switch frequency, but the flickering on camera is.
You need a driver that has the LED PWM switch after the capacitor, so that the capacitor stays charged between PWM cycles... You can probably add such a switch, a small N-FET on the ground side of the LED, driven by the PWM signal...
Diamond Age! I already thought you were cool, but now I know it for a fact.
This is so cool. Finally seeing it with the LED working of a wired power source constantly, the flicker being resolved and adding the autobrightness feature. For the autobrightness it could be cool to be able to set "profiles" to toggle with a button combination, for inverse (more brightness the darker the room), direct and maybe even exponential, linear and parametric. This could be done by a future app as well. Just leaving it out there as an idea that could surpass the functionality of the original thing (I know this is OpenSource and that I could implement it, but I still lack some knowledge to do so, I might try in the future). I wanted to ask what app are you using to create diagrams such as the ones in 7:54 - 7:57? They look really neat!
See you at Open Sauce! I'd love to chat about this project and I'd be interested in helping with development, too!
WE’RE SO BACK 😍😍😍
Could be fun if you added a learning function of some sort in the code, whereby if you have auto brightness on and adjust the brightness manually, the microcontroller would look at the room brightness, look at the LED brightness, and set a data point for those two values as axes on a graph. With enough points on that graph you can form a relationship between whet the room brightness is and how bright the LED is
I swear it was me who mentioned copper strips and pogo pins to you and you said it sounds interesting im claiming it. yay
great work, thank you for sharing!
Love it, thanks for sharing your progress with us!
welcome back!
One thought on Auto Brightness. Is have it calc when you turn it on for where to start brightness at. Even if its off for auto adjustment. It determines the Starting point of your brightness steps. Its looking awesome so far.
HE'S BACK!
Can't wait for the next one!
I have never thought about auto brightness as fully as I have after watching this video! My kitchen lights follow the opposite curve from your DIYson, in that they brighten as ambient light falls, and this is what I need and expect from them, but I also understand completely why you don't want that from the DIYson. Is this a difference between room lighting and task lighting in general, do you think?
If you feel like diving deeper into the electronics, I’d suggest looking into techniques to manipulate the output of switching regulators. It can be pretty easily done by biasing the feedback loop with a rc filtered pwm signal from a microcontroller. There is an appnote from TI on this. I will reply to this comment if I can find it.
Awwwww yeah! You're back!
Good work!
Was just wondering a couple days ago if you’d have an update. I love this project.
See you in 5 more months!
You make me want to put my phone down and make cool stuff!
What about using the human adjustment as the new baseline for auto-to-reference? I feel like that is how my Mac adjusts auto brightness. + or - 10% of whatever I adjusted it to unless there's an obvious change in the environment such as opening/closing of a curtain. Super cool!!!
The QT Py has 2 Analog Reference pins, A1 (VREFA) & A2 (VREFB), I'm not sure how exactly this is implemented but it's worth investigating. If it samples the voltage on these ports and uses it to do the formula you have already, then there is no value here; but if it uses the voltage on these pins in an analog process to scale the analog output, then that would allow you to use the entire precision of the ADC, instead of the bottom 79% of it's range
Hey!! i've been looking for a high amperage equivalent pogo pins, maybe more like a contact spring, but with the size of spring you have on contact switches and maybe a little longer ? I have searched far and wide, but my limited electrical engineering skills havent given me much help ahah! I have seen specialty companies for aerospace make some huge pogo pins for 50 or 100amps per pin, but they seem huge, heavy and expensive.
I'm trying to make hot swappable batteries for my drone. So i will make a base on which i'll stick a standard XT60 drone battery on a FR4 pcb with a female XT60 connector, and tracks either side (wider than you have for the light for appropriate amperage specs), that or tons of pads for individual pogos!
I'm note sure if i should go the plenty of ubiquitous 1A pogo pins or find something else, like what I imagined must exist.
The main issue is that drones are subject to high g force, tons of vibrations and can require 100A peak power. Maybe 20 continuous at times. So just like the problems you have, i'd have to make sure those pads and springs get good contact, all the time.
If you've ever seen something that looks like what i've imagined hmu! or if you have any suggestion or something that could help!!
Thanks for the videos, huge inspiration!
Dude bring that thing to opensauce get a spot! reach out to william, that thing is more together then most things!
Very nice. I feel a proximity sensor would be nice. Turn on with autobrightness on approach, and off after a period of no movement.
Will you be selling kits, or just plans?
I feel the boards would be a nice halfway.
IMO, four brightness settings seems like too few, hopefully you have thought of a way to expose more granular control over wifi/bluetooth. At night, I frequently run my desk or living room lights anywhere between 1-5% for silhouette lightning where you have a clear visibility of what's there but it's not bright.
Have you considered using a time of flight sensor to add gesture control of brightness?
How about hold both buttons for 3 seconds to toggle auto-brightness mode?
I was thinking the same thing! Even 1sec is enough
When’s your next update? Close to the end?
If the issue with the pogo pins is that flex in the extrusions (or flex in the various elements of the structure of which the extrusions are a part) can cause the pins to lose contact, could an answer be to introduce a curve/radius to the extrusion? Or longer pins!
lets go boiiisss
Having an ambient brightness mode and a user controlled mode as two different modes of operation seems clunky and limiting. I reckon it would be more intuitive and useful if the user changing the brightness up or down actually changed the brightness that the ambient adjustment indexed to.
i.e. pressing increase should increase the preferred ambient brightness by 25%, and the lamp continues to adjust to retain that increased brightness as conditions change.
Adding another button also betrays the simplicity of the design.
Is there a way to utilize the aluminum extrusion already in use as a heat sink for the LED?
I've been enjoying this project. I do have a question. I thought the QTPY adc was 12 bit, not 16 bit. Are you using a 16 bit ADC in the design? If it really is 12 bit that could explain the failure of the linear brightness being not so dissimilar. Perhaps you have a 16 -> 12 bit scaler function. Cheers good project. Ah the duty cycle of the PWM is 16 bit. ok that helped. Looking up the specs.
Ong new video awesome!!
I do hope that you will make a kit :-D
how do you make the desktop look like yours? looks awesome!
what is the circuit visualization tool that you're using? I love the look of it and would love to integrate something similar in my own projects in the future
I just made them by hand in Illustrator, nothing fancy :)
The *output* of the new driver still uses PWM though, right? What is its frequency?
Is it completely impossible to use 2 LEDs in order to be able to the light temperature (hint: maybe in auto mode with a color sensor right next to the light sensor, following the ambient lighting)?
How did you like the Hyperdrama album?
Good shit bro
What software is used to do the drawing at 3:30 ? Thx.
could you please tell me which software do you use to create their illustrations 3:32 and also where do you get the component footprint from
Adobe Illustrator is the answer to both questions :)
Thanks for this series! Thanks to you, I'm creating (and almost finishing) a kitchen light that has 6 cree cbx1512's, 3 pwm-capable drivers, all encapsulated in nicely looking wood. Refrigeration was a pita but I could find some noctuas @5v also pwm controlled which are pretty silent. As for operation of the light, I added a lux sensor and a presence sensor that can read out 3 different zones. It also has a rotary encoder for manual overwrite and all code is done with esphome, much easier to work with (imo) than direct Python. Once again, many thanks for the inspiration of your video series!
Yaaay finally
How are you making those sweet schematics?
I drew them in Adobe Illustrator :)