Just How Good Are Heat Pipes Anyway? - DIYson Lamp Build Log #10
HTML-код
- Опубликовано: 8 сен 2024
- One of the most celebrated features of the Dyson Lightcyle (now the Solarcycle) is the heat pipe cooling technology, which Dyson says can keep it running for up to 60 years. Can I achieve the same cooling efficiency on the DIYson using off-the-shelf parts?
I featured parts I purchased from SendCutSend in this build log. They didn't sponsor this video but they've given me discounted services in the past. I like them. Check them out here so they know I sent you: www.bit.ly/Ben...
Current DIYson and DIYson Express project repository and working parts list: github.com/ste...
Special thanks to Kiran Malladi for lending me the thermal camera. Here's Kiran's channel:
/ @kiranichiban
Thank you to Jake Brownson and his beautiful aluminum carriage:
/ jake.brownson
Join the Discord:
/ discord
My Favorite Filament: amzn.to/3HiAHnv
(Amazon Affiliate link)
Glad to see the DIYson continue!
This series is such a goldmine. Hearing your thought processes along the way. It would have been easy for you to gloss over the tiny detail of the wheels not rolling smoothly with the first copper solution, but it was really powerful to see the iterations towards perfection. I can’t imagine how much time it takes you to put these videos together and I (at the very least) totally appreciate it. Can’t wait for more!
Like all the previous videos, this is amazing content. Its clear, detailed and very approachable. Like good wine, it takes time to make good videos, I much prefer quality over quantity. Well done!
Countersink screws could be a good solution
With the screw held in compression, a set screw is better.
Could also shave some material off of the wheels.
@@scruffy3121 I would have changed the wheels holding the horizontal arm to grab it from the sides instead of top and bottom.
@@Flumphinatorin the new way the screw is to pull towards the center and counter sink screw is the right choice
I’d prefer counterbore holes. Countersink holes require higher precision in location and can over-constrain the assembly. When you don’t actually need the countersink to locate your part you’re better off with counterbore holes.
Electronics Engineer here: Try using carbon motor brushes running on the pcb tracks. Much less wear. You could use strips of spring steel (like a clock spring) to hold the heat pipe against the AL. The strips can be placed across the slot, pushing the heat pipe deeper into the slot. No screw heads.
Always cool to see people doing real quantitative comparison of thermal designs.
Looking at your data, it seems like the dominant thermal resistance is that thermal paste on the LED, and also between the copper pieces. Soldering them together would probably make a huge improvement.
One of the only channels I watch on RUclips at 1x speed because I want to soak up every aspect of these build logs. Great work so far!
Wow, thanks!
I can imagine the reason for the 200mm sensor beeing hotter, is because thats next to the screw that holds the heat pipe. Something I've learned from PC cooling is, that the contact means alot. If you could make the contact between the heat pipe and the aluminium even better by maybe more screws or something, it would be even better. But of course, that's just my guess!
Anyways, great video!
Agreed, to confirm he only needs to measure temperature on the cold end of heat pipe, I'm sure a lot of thermal resistance is between heat pipe and aluminum extrusion. Also screw LED right to the heat pipe, 2 layers of thermopaste is just a waste. This helps because every 10° after 65° C on the chip cuts LED life in two.
I think brazing would be a good fit here, for thermal conductivity and mechanical connection - It could also be removed with heat, for repair.
You can braze Copper and Aluminum together with just a regular torch
@@weeveferrelaine6973 I somehow think brazing with "just a regular torch" is not something done by anyone.
@@Gebsfrom404 Also, applying super high temperatures to a heat pipe will cause it to inflate and pop.
For the "copper heatspreader" I would have used just simple countersunk screws. Also for the heat pipe version I would look into a way, to be mounted such, that it directly comes into contact with the LED. In this way I think it would be even more efficient with it's heat transfer, therefor it wouldn't heat up the whole copper bracket. (If you look at a PCs heat sink with heat pipes you would see that the heat pipes directly come into contact with the CPU)
Like other people I also dig this project. It’s focused in scope but highlights the engineering process as a whole well and is interesting as someone who takes on similar projects to see how you approach a problem. Good design is just damn hard.
As a separate note you actually motivated me to buy a Dyson light cycle lol. The motion system that they developed and your diy version is pretty interesting I’ve been trying to come up with a secondary use for it. Maybe a video recording mount or magnifying glass for projects mounted to it?
I love these videos. If it's possible to move the LED to directly under the rail, you could add a modest press fit aluminum part and a little thermal paste or a thermal pad to use the rail as a nearly direct heatsink. The issue then is finding space for the PCB around the end of the extrusion without making the housing look silly. Aluminum is a good enough conductor that the surface to surface contact resistance is most of the problem. Possibly just a heat pad directly to the rail would work. Polishing the end of the rail a bit and using a small clamp should get you the answer pretty quickly.
Is the biggest issue with this suggestion that the height of the extrusion is currently being used in the 3d printed enclosure?
@@forked-lightning it's 3D printed so just change the model, and the aluminum bar can be cutted and filed if need space or more area of contact for the led
I fully agree, you could still slide a heat pipe into the rail slot under the LED for improved conductivity if needed. The entire top half of the rail (Within the housing) could be cut away to make space for the PCB, making the housing tiny.
This is the way. The copper sheet solution is incredibly over engineered. Just screw the led onto the extrusion, it's easily enough thermal mass for 15W of load.
Copper Bar or Heatpipe could also make your Ground for the Power Delivery.
If Nickel plating is available there should be no relevant Oxidation. Although bare Copper might be good enough.
With this the PCB only needs one Trace that can be wider.
(Same PCB with two Sides to match each Extrusion.)
But like some other Commentators I think Aluminum from the Extrusion might be good enough, although thats also depending on the Alloy, if you can connect the LED directly to it. For DIY it would be okay in my Opinion to have to change the LED after a Year because it got cooked.
Although less DIY friendly the End of the Extrusion could be milled down to give a Face for the LED Mounting. With some good thermal Paste or Pad it might work out.
if you wanna avoid the mess of thermal paste, you could try using thermal tape. it has lower thermal performance, so maybe still use paste in the high-heat-flux interface between the chip, bracket, and heatpipe, but use thermal tape to stick it down in the channel, where the thermal transfer is spread out over a much larger area. I'd also wonder about improving performance with some sort of super-slim spring clip that inserts into the channel and pushes the heatpipe against the aluminum (i'm imagining something kinda W shaped with a flat bit in the middle, so that it stays out of the way of the v-wheel)., the current setup looks like it would have uneven contact pressure along the length of the heatpipe (and i'm also a little worried about the screw at the end damaging the little sealing nubbin it's pushing on)
So glad you discovered the heat pipe! They really are like the closest thing to magic or a teleporter (for heat anyway) that you can just buy off the shelf, no batteries required. I bought a bunch of them about a year ago. Put some on my fridge's compressor, keep them on external hard drives during major back-ups/transfers, have them on the motor housing on tools that run too hot or large Ah batteries while they charge, etc. I have some in the kitchen junk draw to help defrost food if it hasn't completely thawed. Thermodynamics is a big deal if you're Goldilocks. A few well-placed heat pipes would have solved many of her problems.
The lamp and your work and how you document the process are just amazing.
But can I just appreciate how aesthetic your Excel graphs are?
Hey! Love this build series. I was thinking maybe countersink screws would work instead of the low profile screws. I think send cut send can do countersink operations on the holes and it’s not too hard to do it by hand after the fact. Especially in copper.
Ope! I commented before the video ended and you decided to go with the heat pipe 😅
@@clomads still good advice if something similar comes up in the future!
Great content. I have quite a bit of experience with long-term deployment of high power LEDs and also with heat pipes, but not together. My advice, though, is that absolutely anything you can do to keep the LED cooler must be done. The cooler you can keep it the higher it's output will remain over the life of the LED. Additionally, mounting pressure will drastically increase thermal transfer. The better you can clamp the LED to the heat pipe and the better you can clamp the heat pipe to the extrusion the better the whole system will perform. I might recommend a thin u-shaped shim that slides/wedges on top of the heat pipe and under the overhanging portions of the extrusion. With the "bowl" of the u facing upward, the wheels should glide right over it and the increased pressure will really help with thermals.
As always, a surprisingly great quality video, about a surprisingly great quality project, from a, relatively, smaller creator. I absolutely love seeing this project progress, and your channel grow! This series has been a well-grounded look at the engineering process, and it has an appropriate scope; all of that makes it perfect for exploring your design decisions, the factors to consider when choosing constraints for a project, and how to go about solving them. I think this is important to give those interested the context needed if considering a project of their own. Keep up the good work; this content has been phenomenal... and super entertaining!
Edit: Missed a comma.
I also think your testing of the two solutions is hilarious because I find myself doing the same thing from time to time. The heat pipe was obviously going to be more efficient, but you liked the solid nature of the heat sink; as did I. I find that, whatever it is that makes me prefer the solid nature of the heatsink, would overpower my assumption it'd be less efficient and know I won't make that decision... So, you force yourself to run the data: now confronted by the fact that it would be outright illogical; to an extent you can't justify; all for the sake of unexplained preference; you can't illogically pick the heat sink.
I do this literally all the time. Something about me is willing to compromise, until shown how little it makes sense; until seeing just how beneficial the obvious solution is. It seemed as though that was your motivation too. Maybe I'm wrong but it made me laugh all the same.
Edit: Fat fingers and improperly placed commas.
I just found your channel yesterday, and I could not stop watching your videos. It's really cool to see how other people create things and solve problems. I am very much looking forward to future videos.
So yeah, I did search yesterday in the evening on youtube for what to look out for when using high power LEDs (which led me to the 4th episode) and binge-watched the whole thing until now! Please take your time -- as long as you need to -- in order to keep producing these high quality videos! You've won an extra subscriber which seems so little compared to the goldmine of information you're putting out there, for free! Thanks a ton!
Man I clicked as fast as I could! Great video, I'll watch it later
A bit paradoxical that one of YT's strong contenders for overengineered project of the year will be as a DIY kit :P
why not go with counter-sunk screws with 3.18mm you should have enough meat to countersink the screws.
and by btw i like the heat pipe solution more, just the cooler physics.😁
Yup exactly, because the heat pipe is better anyway!
I love the thermal cat!
How about using small leaf springs bent to a concave arc to press the heat pipe or copper strip to your Aluminium extrusion?
oops, you said this way more concisely than my comment, have a like
I will probably never build a lamp like that. But the explanation and the measurements with the heat pipe alone are worth watching the video. Thanks for sharing with us.
Great to see you continue the series!!!
so glad this project is still going strong
This is my favorite series on RUclips. I love how you show us the entire design process.
not just saying this because of the time of year, but when a video from this channel drops, it feels like christmas. genuinely love this channel and everything it stands for. keep it up :)
Somehow, I am glad that it took so long. Amazing video. I look forward to seeing this project evolve.
I'm a bit worried about the wheels being perfectly 0.1mm off the screws at the start, but then they wear out after a while and start catching again. Maybe countersunk screws would work better? I also think that if you put a bit of thermal compound under the sheet you can make the piece a lot shorter as the heat would dissipate into the rail before "reaching" the end with the current design.
I love the heat pipe but I'm a bit worried about needing to glue it down, something simple to fix would become a whole redo and replace afterwards.
Ah I didn't notice I was just halfway into the video. Looks like my prediction was accurate haha
The graph for the copper bracket looks exactly like a 1st order differential equation, which is exactly what I would expect. The heat pipe is clearly nonlinear, which I would also expect. Good experiment!
This looks like an amazing project! RUclips suggested this, the tenth video in the series, to me, so I jumped right into the middle of it. I'll go back to the beginning now.
Why not try using countersunk flat head screws instead of the low-profile socket head cap screws? They would end up flush with the copper (or perhaps slightly below the surface). Not sure if the fab you're using could countersink them for you, but if not, countersinking 3 holes isn't that much work.
Countersunk would definitely work. Other option is to avoid the screws and use a spring shim or press fit shim to wedge the heatpipe in place.
Damn that's some beautiful data visualization.
i am in awe at how good your video is, just stumbled on it by chance. I am trying to develop hulls with built in heat transfer components. I just learned so much. thank you!
This guy's style really reminds me of the channel " wristwatch revival". Both talk about function, form, and feel in a smooth easy way.
Finally another vid! Missed this series!
I think you made a huge mistake with the copper heat sink. you made it quite long, but penetrated the material with the mounting holes. the cutting surface are is drasticly decreased by the holes. the material after the 1st hole basicly doesnt do anything. maybe try the heatsink version again with another mounting system? I think it might outperfom the heatpipe because of better contact to the aluminium rail. just my 2 cents :)
This is a great point - but if the holes are filled with the brass set screw does this mitigate the holes to a significant degree? If they are left empty, you’re probably right that these holes render the length effectively useless.
Solid work. Well shot you didn't talk down to your audience. So glad i found this channel.
Thank you :)
I think that comparison also needs active cooling. Passive cooling may be good enough and even more more elegant than active cooling but if a micro fan and a slightly bigger housing can cool the LED to less than 50 or something it would still be a better solution for the longevity of the LED.
And such fan could cost you even less than one of those screws you used.
Noise might be an issue though. I wouldn’t want my lamp making a sound
What a nice and detailed video! While comparing bare copper and the heatpipe I saw that you secured the heatpipe only with one screw at its end, which looks like too little mounting pressure from the middle to the led end. The copper piece had multiple screws over the whole lengh, giving it an advantage with heat transfer to the aluminium channel. If you could somehow get enough even mounting pressure on the heatpipe I feel like it can perform a lot better.
I didn't think the heat pipes would work in a horizontal configuration. Now I know thanks
If you want to further lower your LED temp here is something you can do : raise the thermal emissivity of your system . your aluminium profile is good at conducting heat, but aluminium radiates it poorly. in your setup, heat would be mostly evacuated by thermal conduction to air. You can increase radiation by painting your profile. the paint will radiate the heat. (for aesthetics reasons it could be varnish also, and maybe not on all faces nor tracks) might worth a try ! in all cases, great job !
I’ve been on the hunt for a lamp with:
-high CRI
-at least 800 lumens brightness
-good bulb position adjustability
-continuous brightness and color temp control
-long bulb life or cheap and easy bulb replacement
-small and recessed light emission source (not a large emission point that puts a lot of glare in the eye)
The best thing I’ve found that suits my needs is the $200 VariLum 30W lamp which doesn’t even have a small emission light point or recessed light source and doesn’t get anything close to Dyson’s level of adjustability.
People say this lamp is overpriced, but when I ask them to find me something that has the feature sets I want, they just can’t find any. Even I can’t find any. No one comes close to the feature set and engineering that Dyson has accomplished with this lamp which gives them no competition. The high end lamp market is under appreciated.
I look forward to this project of yours and wish you the best luck. You earned yourself a subscriber.
Annealed copper is very soft and easy to work with, and you can anneal copper very easily with a basic torch, perhaps even a gas range if you're feeling spicy. There's tons of videos on it but you get everywhere on the copper that you want to bend up to cherry red heat, it doesn't need to get there all at once, just everywhere at least once. Once it's annealed you can absolutely bend this copper with a 3D printed bending tool; and bend it into a ( shape, or {, maybe C, I'd go with a 5 sided C that follows the bottom of the slot. In an ideal build I'd want the copper to push itself against the the extrusion, staying far away from the V-groove. Hopefully this gets close enough to the extrusion that you can pack in a little thermal grease and get it all thermally bonded. In this ideal, no screws are necessary.
For the design of the tool, I'd go with a pair of form rollers that bolt on the end of some extrusion so that after the rollers the copper gets guided into the slot. I don't know that you'll be able to get all the way there in one pass through the rollers, so you might want a way to hold the form tool, that will help you keep it reasonably straight until you can mount it on the extrusion to force it straight.
If it takes multiple passes through the form rollers, you'll need one of them to be adjustable, and you may need to re-anneal the copper part way through. Working copper in this way is very likely to distort any holes near the bend. Any parts you don't anneal will remain the hardness they arrived at so avoid annealing near the LED mount area and it will keep it's shape, but also, the formed portion of the stick part, may be stronger than the flat stick, because of gaussian curvature, so you may want to curve it closer to the LED mount than you need to. Just, avoid any plans for forming the LED mount area in any way, unless you want to drill the holes later. That said, working copper is really easy and only takes a coping saw or dremel, a file, and the form tools; since it's a one-off part in the lamp I think it's totally reasonable to make this out of some copper sheet, by hand, as someone who isn't you and won't be doing the prototyping. I should only have to make this part once, and I'm willing to.
Should have had less faith in the copper. I could still make the heatpipe copper sheet bracket, I could probably make it more easily, and the 3D printed form tool would be simpler, either a buck I hammer the sheet against, or a sandwich to crush it into place
Science! also ... Engineering! This is an excellent example of "hypothesis meets experiment" type of workflow, everyone should be doing it, and talking about it more.
Very cool, as usual!
Adding to the comment about people cranking out videos. I have to go one step further and say, I don't understand how people can do these complex prototypes and produce videos. It is a ton of work. I've got a partially open source smart vent project and I can't bring myself to make videos. I also have a few too many hobbies and other interests.
But in any case, props to you!
I like the high quality you keep on these videos! Quality over quantity!
Really interesting. This is the first video I've seen of yours. I used to work for Dyson and in the early days, I attended a talk by Jake Dyson. He felt that the future of lighting was LED but all commercial products are flawed because the LED deteriorates over time, reducing light output and changing colour. That is certainly my own experience. They start great but just fade away! Most, if not all LED lighting pays little or no attention to cooling. They last long enough to cover the guarantee. Apparently, if the LED is cooled to the correct temperature range, the performance and reliability of the LED could be boosted to years, rather than weeks or months and the colour will not change.
Heat pipes exist for a reason.
The heat pipe solution also has another notable advantage, that L bracket used to mount it is providing a good thermal interface into the aluminum rail.
The algorithm brought me to your video (I've not seen your channel before) and I was very pleasantly surprised. I have been working on a rudimentary animatronic character and I'm attempting to put a 3W LED inside each of his 3D printed eyes, but I've never dealt with LED heat dissipation before. I greatly appreciate the "thorough yet concise" nature of your video. I haven't found any good info out there and was flying quite blindly. From your video, I learned that while not ideal, my solid sheet of cut copper solution will likely be adequate. Before watching this, I always just assumed a heat pipe was just a thick piece of copper. (Glad I didn't cut into one thinking I could just trim it that way, lol).
Also - thank you for including video of the seemingly straightforward steps of prepping the surfaces and applying thermal compound. In my mind I knew it SHOULD work (I've built dozens of PCs over the years and applied thermal compound to CPUs countless times). Seeing someone else do it in this sort of LED heat-dissipation application makes me feel a lot more confident that I'm not wasting my time (which is honestly a major barrier in moving forward on a project more often than I'd like to admit).
Thank you for taking the time to make this video. I know it takes a ton of time, effort, focus, and drive to make a video this cohesive ON TOP OF doing the actual experiment your working on. Your efforts are very much appreciated.
god i love seeing these videos, makes me want to pick up all the projects i half finished weeks ago and finish them
25:10 I would love a video on how you converted your photos into numerical data points. As well as how you turned those points into a well designed, and important education tool as a line graph.
Screw holes talking most of the copper cross section - don't help with heat transfer.
Would be curious to have this data compared to a modification where the heat pipe is replaced with a similar size copper bar (no screw holes, same mounting as heat pipe).
I really liked this, the creativity and learning heat pipes really do the piping. I dont have any current use for that information but boy is it cool.
Newbie to the channel here. Great video! A new subscriber was born ;) Looking forward to more.
Reminds me of my indoor plant grow shelf where I mounted the LEDS directly to the underside of the shelves, so it all worked like a giant heat sink. Worked perfectly fine until I turned on the 50W high power grow light and cooked all the parsley sitting right above.
Love this project, and really like my Diyson express! I’m going to make my next one with 750mm rails to better reach my work surface. I use tungsten shot in the counterbalance so no problem balancing the larger weight. Thanks!
watching you strive for perfection is really satisfying
No way! I was just thinking the other day I should ask if this series was dead 😅 I'm so happy to have a new video!
On the LED holder for the heat pipe, I'd have an upper tab that slots into the opposite side of the rail, with a screw; as an adjacent topic, I'd personally keep the power rails on the inside face of the rail-roller bracket-rail assembly, that way you can just plop a simple block of pogo pins into the roller bracket and it just floats as the rails move, only being dependent on the intersection point of the rails which is constrained by the roller bracket itself. Just to constrain the part from an upper flat and screw, and the perpendicular flat and screw. This improves alignment reliability and assembly while not adding much material or cost. On the other side of this, I'd also make it to where the heat pipe doesn't have to be bent, just for the ease of most people making this themselves, most people won't have the tools nor the experience and will likely crush or crack the pipe; this wouldn't be difficult, just keep the aluminum extrusion's base square and the thickness of the flat heat pipe as a keep-out zone for the bracket, then have the LED side of the bracket sit under the heat pipe. This in combination with the upper screwed-down tab will act as a clamp on at least the initial portion of the pipe, which will help constrain it in place, given that along the length of the rail the only constraint will be the adhesive and how straight the pipe is in combination with its resistance to deflection under its own weight. Though I'd personally also find a longer heat pipe and just constrain it at the opposite end of the aluminum extrusion.
Hard constraints and multiple non-redundant constraints will help the project a lot. It's like your discovery of the roller bracket flexing, instead of just using a beefier metal bracket you could also find a better constraint method, such as gusseting key areas to a perpendicular surface for better resistance to the plastic bracket bending.
Stoked this is back!! Still loving my new monitor mount too.
I'd suggest trying any kind of screws which are flush or below the bracket surface, and using bare non-sealed ball bearings. I admit I just dropped in the middle of this video without prior knowledge about the project, so I may be completely wrong. I'll watch the previous videos because it seems an interesting project. Black matte anodized profile, and maybe also sand blasted and painted black bracket should also radiate more heat away if one of the goals is cooling the LED as efficiently as possible...
There are a few possible improvements that come to mind.
One would be getting a smaller LED that fits inside the extruded aluminium arms. In that case you could then just add a copper heat-bridge between the LED and the aluminium and use the entire arm as a heatsink (aluminium is still a good heat conductor).
With the copper sheetmetal holder you would have more space and better contact if it was U-shaped - the lower-part going towards the inside of the channel, the outer edges contacting on the flanges of the channel. This should solve the clearance nicely while also giving some good contact due to the springiness of the U-shape. (that form would also allow for thicker/wider copper sheets).
If sticking with a heat-pipe: A thinner heat-pipe that fits at the bottom of the channel of the rail. Then the mounting could be a simple clamp that pushes the heatpipe against the inside of the rail.
Or use a heatpipe in the side-channel and have a 90° bend at the end. And then there have a copper-connector to the lamp.
And finally - a bit more custom.
Laptops, handhelds and other mobile devices often use small vaporchamber+heatpipe combinations. Those are great for picking up the heat at the source and then transporting it away. For the lamp even just a longer vapor-chamber might be enough to spread the heat so a passiv fin-cooler could be slapped on top. There are also various companies that offer custom solutions. They have some higher setup costs but if you plan on selling the lamp-parts that might also be an option. Definitively the most involved solution but also by far the best thermal performance you could get.
Thank you so much for the production work it took too so this and your choice to share with us! Great video and I'm going to go price this out now (Jan '24) and look forward to building it and having an awesome lamp at my desk! I may do an RGBIC circle though for color control with subtle animation!
Man this project is epic. This really shows what engineering is about.
Can I ask why you didn't use the heat pipe in the side of the arm, away from the rollers. Instead of on the top or bottom, under the rollers?
quick idea here, you could make a strip of spring steel with a crossection shaped like a " ) " that pushes the heat pipe down while "grabbing" from the edges of the aluminum extrusion. Also if it is the full length, you will have even pressure across the heat pipe, although i believe that smaller length but multiple "springs" would do the job just fine...
Glad to see an update on the lamp and as usual very interesting to see the development and testing process. Can't remember if you have mentioned or considered the aluminium to copper corrosion risk. Looking forward to seeing the next steps. Happy holiday season
Steven love your content. Such well thought out and concise videos that cover well thought out projects. Take care.
One idea I have for keeping either solution firmly pressed against the center of the rail is photo frame spring clips, or something custom made that's similar.
Since the heatipe is not even close to being uniform, I'd recommend swapping thermal paste to k5 pro, or thermal pads (but the k5 pro which basically is "liquid" thermal pad would be cheaper/better). Thermal paste is a good choice where variance between low and high peaks are few fractions of a millimeter, but with the heatpipe they are more like 10-50% of a millimeter. :) Alternatively thermal epoxy is also a good option to secure the heatpipe into the aluminium rail.
Just a thought, but I believe this would improve your cooling further.
Please make the final video already ! Can’t wait good luck!
Great progress, and very cool (pun intended) seeing how heatpipes work in practice.
If you thread the copper bar and drill through holes in the extrusion you could attach it from the other side. Also I think the copper part is too long, heatpipes work over a longish distances but with solid metal you have diminishing returns the longer the part. If you go with the second option ( bent metal + heatpipe) I don't think copper would be required, Alu would be more than enough.
Wow those findings on the heatpipe performance blew my mind. I guessed it would perform better, but it throwing all the heat away to the farthest point makes so much sense but is unintuitive for heat transfer though a material. Super cool. Makes me want to make a project that needs heat disapation just to get to use that information!
Btw I check for a new video from you periodically, thanks for sharing your design process with us!
I've been getting into microcontrollers recently, and I want to build my own 12vDC to 120VAC power inverter. I was just planning on a giant heatsink, but this had me wondering about moving heat away from the board, to get the space for a higher tech radiator/fin setup for cooling - Since for the MOSFETs, everything seems to be about how cool you can keep them, for how many amps you can pull.
if the copper had simple formed ridges on the "outside edge", it would be a press fit into the rail, and that would keep the copper part lower in the slot, plus, you would only need one set screw to hold it in place...if you were designing for production, you'd get the copper part stamped/formed, and have it screwed, (or a kind of press fit/clip into the slot) to the face of the channel, with as much copper/aluminum surface area as possible
been waiting for this for a while now
When you machine into the end of the horizontal extrusion just before you hit the conter hole you can attach the led directly to the highly thermally conductive and high cross section aluminium without needing any custom parts.
If you haven’t already (assuming you’re using something like PLA) consider some of the printed materials like ASA or ABS (or even CF/PC if your printer can get hot enough for them) for the outer housings
Could let you hide some big boi heat sinks and not have to worry about the housing warping
Looking forward to the solution you come up with, only just discovered this series and the powered superrail contacts alone were so cool to see
I appreciate your priority on other people being able to build it themselves, I’m honestly tempted too - it looks so satisfying already
Even though the heatpipe solution seems to be the most promising, and therefore you no longer have the same problem looking for low profile fasteners, I'll throw the information in anyway. The best solution for keeping the screw heads lower is to go with flat head cap screws, or sometimes called countersunk screws. With those, you can countersink existing holes in the copper heatsink, which is still a nice and simple operation with readily available fasteners that aren't nearly as expensive. There are still design limitations with them, like needing a part that is thick and/or wide enough for the countersink, but otherwise, allows for an extremely compact mounting solution, and I think you'd have enough room for them with the size of material that you showed in the video.
Just keep in mind one major thing if you're ever shopping for these fasteners: On most fasteners, the screw length indicated measures from the *start* of the threads, and the head of the fastener is not factored into that length. On a flat head cap screw however, the *entire* fastener is what is measured for the screw length indicated, which *does* include the head of the fastener.
You could also use a standard socket-head cap screw with a normal size head and counterbore the hole, but countersinks are a much more approachable situation that would be accessible to the largest number of hobbyists, and a little less demanding on the width of the material in order to get a good looking setup.
Either way, fantastic to see the video pop up in my feed, I was beginning to wonder if you were still working on the project! Thanks for sharing your process, and keep up the amazing work!
Awesome to see you back! I was looking for you the other day and was sad to see the channel seemed dead. Glad to see you again!
Take a look at electro etching. It's surprisingly easy with tooling that consists of household items. And it's surprisingly precise if you take the time to dial in the voltage. You can "machine" your single piece copper in all sorts of ways and it makes for a neat subtractive process that's incredibly cheap for how much material you can remove. The most difficult part will be making the resist pattern easy to apply (laser cut vinyl works really well though that thin tape used in airbrush masking is likely the best choice for general use).
And I am new so haven't watched all the videos, but why not run the heat distribution down one of the vertical slots, leaving the roller slots unobstructed? I'm guessing it's the 90deg turn adding cost. (in frantic recitation): then add peltier devices to harvest the heat back into electricty and feed it back into the system then, perpetual energy! (no, I am not even close to suggesting this is really perpetual energy).
Don't try and speed up the video for the algorithm, very well designed and researched part
For me the results indicate that the only place where the heatpipe has good contact is the end part where its screwed to the frame.
Maybe try using a thermal putty instead of paste for the frame to heatpipe interface, that should do better with gaps and maybe improve things even further.
I really like this project and I truly appreciate all the troubleshooting and methodology used to go through each issue. Please forgive me but I wonder if you may have overlooked a few simple solutions. For the heat sink screws catching the wheels, I noticed no countersinking of the holes. Perhaps a slight countersink...say 0.3mm would give you JUST enough clearance. As well, for both the heat sink and het pipe issue I'm curious why you limited to a single piece. Surface area would seem to be your friend here so the more heat sink/pipe in contact with the aluminium channel the more heat will be dissipated. Even a slight application of a carbon based thermal compound (MX4 comes to mind) might be just enough to make the difference.
Then again I might be completely off base on all of this so.......😊
You should be recording the temperature of the room and subtracting it from your other readings. The temperature delta is what should be used in this comparison. The absolute temperature will rise proportionally to the ambient room temperature which is inconsistent between tests. The heat pipe clearly performed better regardless. Great work.
Very nice experiment! Have you considered having your horizontal arm at some other angle? With the heat source (LED) lower the heat pipe should transfer heat more efficiently by allowing the steam to rise to the opposite end faster and the condensed water to fall easier allowing more cycles. Even better if there is a heat sink at the opposite/upper location.
cheers,
Dave
Nice work and great looking result. I'd really like to see you nail this properly, so I will share a little engineering knowledge with you in the hope it helps. I crunched the numbers for your Cu design on the basis of 300mm overall length and 1.6 x 10mm cross section to get a figure of 46K/W. So for every Watt of heat go into one end of that bar, the hot end will settle 46deg hotter than the cold end. Something you need to keep in mind as well, the temperatures you are getting are only relevant with respect to the ambient temperature. You need to consider them as temperature rise, not temperature. This is important because if anything tangible gets to 60degC or higher it is a burn hazard. Even 55degC is pretty uncomfortable.
My calculation above ignores the heat lost into the Al extrusion as well as the impact of the rise in temp around the length of the Cu bar.
The reason, as an aside, that the pipe version rises slower is because it is more effective at removing the heat from the lamp and dumping it into material further away. This effectively increases the thermal mass experienced in the area of he lamp, and hence the slower increase in temperature.
My advice here would be to replace the LED with a power resistor say in a TO220 package for convenience and return to the LED when you have the thermal design right. Keeping the LED below 70degC is ideal. Keeping it (typically) below 85degC is critical.
Using the power resistor eliminates that problem from the testing. And the first test should be if it is possible for the Al extrusion to lose the 10W of heat without excessive temperature rise.
On the topic of your power rail PCB, beware if your ballast (LED driver) is a constant current type and that it can handle an intermittent load (if the rail connections get dodgy). You could also use the Al extrusion as one of the connections provided the power to the LED is classified as SELV. It should say on your ballast if you are using one.
The thermal equation I used is (T1-T2) = length / (Gcu x Ax) :: Gcu is thermal conductivity of copper = 400W/mK, Ax is cross sectional area of the bar and l is the length of the bar. You could try developing a thermal model of the bar and extrusion or the pipe and extrusion. The concepts are the same except the pipe will have far far higher thermal conductivity than the bar.
I hope this helps and it is not insulting to you. I noticed you did not do any calculation type stuff and my being an engineer, I just had to try to help solve the problems.. I look forward to seeing where this project goes.
Will watch later, but just wanted to drop by, like the video, leave a comment and say I'm glad you're back :)
That LED is a hot boy. I've done some experiments with high power LEDs my self. A small heat sink with a fan goes a long way. I'm thinking, what if you drop a Heat sink on the rear of the LED, then use something silent like the Noctua nf-a4x10-5v to pull air through the lamp from bottom to top, so it exhausts the air up, and not down on to you or what you are shining the light on.
Anyway, love your project, and I hope to be able to buy a kit from you in the future! Keep up the great work and videos.
Love your build logs and how jou walk the viewer throughout your design proces and train of thought. It is a shame you’re not able push out more content faster. But it is such great content that it is worth waiting for.
This was awesome. The effort and care is inspiring AF
when building the diyson myself hopefully next year, i think i will go for a nice heatsink added to the top of the LED as i really like the look of them and it should help a bit with cooling. win win in my book :)
keep on going!
Would be cool to have the LED in the carriage area with a nice low profile heatsink and use fiber optic to send the light to the head.
What is the cost difference between a off the shelf heat pipe + custom brackets and the custom made copper part?
My guess is the heat pipe is 1/10 the cost. Copper is expensive.
Heat pipe is an off-the-shelf, super easy to get part. So I'd guess it's a much cheaper solution, and imo fits better with the original goal of easy to access parts anyways
@@eslmatt811 you still need to get the custom bracket to attach the heatpipe
good stuff man 👍🏼 Keep posting with this quality
you could add a tiny fan at the far end, blowing air along the profile, to utilize it's surface area for cooling.