• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

[PROJECT LOG] 3-stage Thermosyphon vastly outperforms quadruple radiators

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

Dr_Emmett_Brown

Registered
Joined
Apr 22, 2020
Location
Caprona
This project began when Intel made it known that their i9-9900KS chip was essentially pre-binned for overclocking excellence, guaranteed to be 5.0 GHz right out of the box. Press leaks and over-zealous overclockers were reporting running the chips at 5.1 and 5.2 GHz across all 8 cores and 16 threads. After all, the published TDP was a lowly 127 watts. Surely the CORSAIR H150i can handle such a heat load.

(Note: I independently confirmed at 5.0 GHz x 8 cores and 16 threads, the H150i does keep the temps at a respectable 98C max with the harshest of CPU tests running on my build with the fans running at 1500 RPMS, which is also remarkably quiet)

Cutting to the chase for the impatient: The chip was giving off way more heat when overclocked than the TDP rating suggests. The H150i was overcome at 5.1 GHz x 8 cores. A quadruple radiator was overcome at 5.2 GHz x 8 cores. The heat just gradually creeps up and thermal shutdown is the inevitable result.

Not to be discouraged, I decided to start over. Completely from scratch.

I was sick and tired of banging my knuckles inside even full tower cases, so the first thing to do was design my "dream case" for the finished product.


01_design.jpg

I should at least mention the "good part" here. I already prototped a new kind of heat removal system that should be capable of handling up to 350 watts of heat.

This is what the final prototype looks like.

02_thermosyphon.jpg

This is a 3-stage cooling solution that will become known as a thermosyphon.

The key takeaways:

1. There is a large condenser with copper fins for much better heat removal than aluminum.
2. There is a copper CPU block through which the vaporized refrigerant flows, not water.
3. Air cools the water which cools the vapor in the condenser. The condenser turns the vapor back into a liquid, which cools the CPU block.
4. This is the "gravity" version of the thermosyphon where there is no pump in the vapor loop. That is why it is so tall. More on that later.
5. Later versions I added a vapor pump so I did not need the large gravity drop.
6. The fans are large 4000 RPM suckers in push-only mode.

Here is a video of it in action:



Below is the copper-finned condenser compared to the length of my arm.

03_big_condenser.jpg

It is also pretty thick too. The thermal conductivity of copper within the operating temperature here is about 392 watts per (meter x Kelvin), an unusual unit of measure at first glance, which is much better than Aluminum at 240. That means you would need an aluminum condenser 1.6 times as large to get the same heat removal as the copper. Silver is only slightly better at 420, but it is so much more expensive, it is not worth the tradeoff.

And shown below, the copper CPU cold plate.

04_copper_block.jpg

OK, now on to the build. Skipping the boring measurements part, you figure out where you want stuff to go, either on the vertical case walls or the case bottom and top. I designed my own brass mounting brackets too. Like I said, I was "sick of" the way computer cases CONSTRAINED me, and since this was going to be a big system, including a built in monitor, a storage compartment for the keyboard and mouse, plus a storage compartment for the power cable(s), I was not going to limit myself in any way.

So once all of the dimensions are set, you order some custom aluminum pre-cut bars from your friendly local bar-cutter. Make sure they also sell the necessary fasteners and nice things like knurled rollers that you can just hand tighten instead of needing a screwdriver.

05_bars.jpg

I should add here that I could not resist going to Lowe's and strolling through the "bathtub aisle" and picking up some matching brushed stainless steel towel racks to be used for the case handles at the top.

06_handles.jpg

The first thing added once the frame was set was the CORSAIR power supply.

07_case_PSU.jpg

The next mini-side project concerns the motherboard. How many times have we had our "favorite case" all modded up, and the motherboard went south one day, or we could not resist upgrading to something better and faster? Right. For me, about a dozen times. So I decided I would build a SLIDING RAIL on the bottom of the motherboard bracket, so that I could just swing my computer case door open, detach the motherboard cables quickly, and slide the whole thing out! I would have my pre-built replacement motherboard + CPU + RAM etc already to "slide in" and take its place!

08_mobo_01.jpg

Other angles...

08_mobo_02.jpg

08_mobo_03.jpg

08_mobo_04.jpg

Fast forwarding a little bit, I added the "wireless antenna" to the exterior. But where to put this "shark fin?" I don't know why manufacturers do this. I mean, it's not cool or anything like that. Furthermore, it provides NO FUNCTION whatsoever.

09_wireless.jpg

You can also see the rather hastily-constructed grill for venting the condenser. And don't worry, I get rid of that shark fin, and re-route the wireless antenna inside one of the bars of extruded aluminum. The wireless hardware is completely hidden!

And time to show one of the most frustrating parts of the build. And it is kind of silly. But I wanted a unique "Power On" button. I literally looked at like 2 dozen different kinds. This one glows red when it is on, and, of course, powers the computer on when it is glowing red. It also "obeys" the software shutdown command issued by Windows, and likewise goes off and does not light when that happens. All "common sense" things we take for granted, BUT.... all of these behaviors require a specific wiring configuration, and the manufacturer makes it as clear as mud!

10_power_button_01.jpg

So here I am re-labeling everything, because the dang Chinese manual was of no help at all. And what was the reward after 4 hours of frustration?

10_power_button_02.jpg

There it is, something nobody will ever appreciate!

And, just in case you thought things were being "made pretty" at this point, the thing is actually as ugly as this now...

11_ugliness.jpg

Power button dangling. Lots of empty space. No wire management. The CORSAIR H150i triple radiator mounted at the top of the case for the "first boot test" (before hooking up the thermosyphon). Those orange things are 3D printed mounting brackets for the internal monitor, which is coming soon. Notice the brass bracket on the rear wall on the bottom left. That can house up to 4 SSDs. Also custom made.

12_less_ugly.jpg

And shown above is a less ugly perspective. You can see those awesome towel rack handles I bought at Lowe's, lol. It is easy to identify the motherboard, video card, hard drive, etc. Everything looks about done, except for the small part about it not being able to boot without a monitor.

OK, now I am even bored with the post, so let me just show the first boot picture.

13_final_01.jpg

There it is. Built-in monitor. Speaker bar. Compartment on the left for sliding in the wireless keyboard and mouse. And running a bunch of prime stuff to test the heat removal with the H150i.

And here I will pause for now. More to come later. Including the science of measuring the efficiency of the heat removal system, no matter what kind of system you have.
 
Last edited:
These projects are why I live here. Looks great Doc! I can't wait to see some final results.
 
These projects are why I live here. Looks great Doc! I can't wait to see some final results.

Thank you kindly. I updated it some more. The "Mark 1" build with the Corsair H150i. Then the swap to the really techy stuff to come later.
 
The case is impressive! Great Job!

Those 4,000rpm fans I could not deal with, way too loud, are you stuck on those, or planning something with lower noise levels?

That was the prototype.

The copper condenser is half the size of the aluminum one, and produces better results. I had 4 fans on the aluminum condenser, and that was extremely loud.

It turns out the condenser is doing most of the work, and even one fan "missing" 60% of the cooling surface will suffice for 5.0 GHz (but not 5.3 GHz as shown).

The latest build of the 9900KF chip (not the KS) will hold 5.2 GHz (not 5.3) with 4 smaller fans running at 900 RPM making virtually no noise.

5.3 GHz requires way too much effort. I'll be content with a quiet 5.2 GHz averaging 95C per core at full tilt.

I'm already working on a combined chiller + thermosyphon. That will be a separate build.
 
That was the prototype.

The copper condenser is half the size of the aluminum one, and produces better results. I had 4 fans on the aluminum condenser, and that was extremely loud.

It turns out the condenser is doing most of the work, and even one fan "missing" 60% of the cooling surface will suffice for 5.0 GHz (but not 5.3 GHz as shown).

The latest build of the 9900KF chip (not the KS) will hold 5.2 GHz (not 5.3) with 4 smaller fans running at 900 RPM making virtually no noise.

5.3 GHz requires way too much effort. I'll be content with a quiet 5.2 GHz averaging 95C per core at full tilt.

I'm already working on a combined chiller + thermosyphon. That will be a separate build.

Where exactly is the vapor coming from, it is hard to tell from the picture or the video?

You have a schematic of the case do you have a schematic of the cooling?

With my setup I have built a mostly modified setup, no case made from scratch simply because my needs for cooling housing change, and I already have numerous cases that fit the need at the time but are now irrelevant, so for myself building a case for my needs would be like getting a girlfriend tattooed on my body and then breaking up.
Kudos for your case design, and build efforts it is very impressive!
 
Last edited:
Where exactly is the vapor coming from, it is hard to tell from the picture or the video?

You have a schematic of the case do you have a schematic of the cooling?

My assumption is the vapour comes from the CPU block, then get condensed up top and travels down to start over again. Thats why the MB is upside down to increase the head distance for better flow into the block.

EDIT: yup, he says as much near the top of the OP.
 
Where exactly is the vapor coming from, it is hard to tell from the picture or the video?

In the original, unscaled video you can see the vapor boiling in the tubes. I zoom in on it at one point. You might have to go to vimeo to watch it.
The liquid is in the tubes coming from the condenser up top and descending to the cold plate on the horizontally mounted motherboard at the bottom of the case, at which point it vaporizes and rises up to the condenser again.

You have a schematic of the case do you have a schematic of the cooling?

When it comes time for the hard tubing I will.

Kudos for your case design, and build efforts it is very impressive!

Thanks. I built it mostly because I have large hands and I was sick of banging into everything when I was working. I built a completely open frame, complete with a "swinging wall" that rotates on a hinge to make it easy to swap out any components, including the motherboard, which slides out on a rail.
 
I like the slide out M/B tray, great Idea!

08_mobo_01.jpg

Yeah that was like a mini-breakthrough all on its own.

I was like "Why is it such a pain in the butt to swap motherboards? SSDs are hot swapable. Why not do the next best thing for motherboards?"

And the SliderMount was born.
 
lian li
Lian%20Li%20PC-A79%20moederbordplaat.jpg

cooler master
cmstack830se_open2.jpg

mountain mods
61AXK0RdNXL._AC_SL1024_.jpg

caselabs
caselabs_sma8_review_29.jpg


and there are more...
 
None of those come close to the innovation I introduced.

08_mobo_03.jpg

I can replace not only a motherboard in about 45 seconds, but an entirely different socket, chipset, RAM, and video card.

08_mobo_02.jpg

I slide it off on a rail, swap in the new one, and close a swinging door.

But thanks for being a contestant on "Somebody Else Did This Already." Stay tuned for your parting gifts.
 
The others can do the same thing if you had more than one mobo tray on hand by the looks of them.

Overall people think this is a really cool setup and are supportive of your build. Someone comments and you ask for proof then they provide it and then you act like they are stupid for not seeing the superiority of your design? I think you should just be more kind overall with your responses. I don't think anyone here is out to get you. Text can be hard sometimes to get intent or sarcasm, but I feel like this community deserves better.
 
But thanks for being a contestant on "Somebody Else Did This Already." Stay tuned for your parting gifts.

OMG, i cant believe this guy.

The stuff that i showed can do exactly what your stuff does and more, the actually sliding mb tray's i posted have a cut out to easily swap cpu coolers, and they can secure pci cards and i/o shields proper. They can act as a test bed etc...

What you should do is improve your stuff based on what we show you, but you seem offended just because you think you "invented" something better.

edit: and btw your method, to me, does not slide, yes you can adjust the weight that mb seats in the case and that's probably an innovation, but its a lock into place system.


Tired of your supposed superiority/know all/im always right attitude...

Good bye boy...
 
Last edited:
The others can do the same thing if you had more than one mobo tray on hand by the looks of them.

To me it looked like the motherboard attached to an L-shaped plate at the bottom of the case. That appears to require the disassembly of one or more "case walls" to get to it.

Overall people think this is a really cool setup and are supportive of your build.

I don't get that impression AT ALL. I get the opposite, with the exception of Blaylock's comment early on which was appreciated.


Someone comments and you ask for proof then they provide it ...

Noooooooo... not at all. "Got a pict?" was all I asked. I did not say "Prove it." You are inserting an incorrect misinterpretation of my words. I had never seen the cases before and just wanted to know what they looked like.


I think you should just be more kind overall with your responses. I don't think anyone here is out to get you. Text can be hard sometimes to get intent or sarcasm, but I feel like this community deserves better.

When someone says "Seen cases doing that long time ago..." how am I supposed to interpret that? It is certainly not praise. It seems like a swipe at me. "Somebody else is doing that already..." is how I interpreted it. And then, from what I saw, it did not appear to be the case.

I present my stuff on here and some of the remarks are derisive. I don't get it. Then I reply and I get the lecture from the friends of the person I replied to. Someone on here even called me a "kid" when I am a 53-year old grandfather.

I built a 5.0 GHz version of the i9-9900KS into a thinner form factor that fits into a stainless steel briefcase. It's basically a laptop that needs constant AC power, but you can close the lid and take it with you. I put analog temperature gauges on each side of the handle so you can see the temps without needing to launch a software utility. The power cord retracts on a self-capturing ratchet wheel. There's no way in hell I'm uploading that build log here now.
 
To me it looked like the motherboard attached to an L-shaped plate at the bottom of the case. That appears to require the disassembly of one or more "case walls" to get to it.
I presented 4 different case manufactures with sliding mb tray feature, you are just seeing what you want or suits your opinion... This is as simple as your method, open door (super easy no screws on this caselabs), remove cables and slides out:
Them you have some thumb screws if you want to keep that safely secured.

When someone says "Seen cases doing that long time ago..." how am I supposed to interpret that?
I'm assuming we are not robots and we can interpret things according to context, the context was i've seen cases with sliding mb trays long time ago, its not a knew thing.

Then I reply and I get the lecture from the friends of the person I replied to.
I'm not friend or enemy of anyone in this forum, including you. The fact that other users lean towards some side, should give you a reason to look again on what's presented and not jump to the conclusion
None of those come close to the innovation I introduced.

Someone on here even called me a "kid" when I am a 53-year old grandfather.
Again you're interpreting things as you want, if you are referring to "Good bye boy" i never intended to offend you and i'm sorry if i did. In my country (in a argument when two people disagree) it's just a way of saying you keep your opinion, i keep mine, lets move along without so many words.

There's no way in hell I'm uploading that build log here now.
I'm in no position to tell you what to do, but maybe it's best if you cant take criticism.

With all that said, i actually like the looks of your case.
 
Your stuff can be cool without being the best thing ever created and 100% unique. I would guess that most of us think it's pretty cool. I especially like how you used height to optimize the thermosyphon and then integrated a monitor and did other things to utilize the extra space. Nobody wants to deride your work itself, but you present it with this air of being totally unique things that only exist because you created them. If you really did arrive at these conclusions independently then congratulations, but I hate to break it to you, doing something well is absolutely something you can be proud of, but being first is not.

Lets go back to one of your first posts on the forum, you state,
A new type of heat removal was needed
. There are no fewer than five articles on this very website discussing thermosyphons, although you'd have to search for them as they are about 15 years old. New indeed. 5 months ago a youtube video was released with a prototype thermosyphon cooler that has been viewed 3.3 million times. Are we supposed to really think that you did all of this research and development yet never came across any of this? Or are we to believe that you didn't think WE had come across it. You take the same attitude towards an assertion that die size impacts cooling efficiency and now sliding motherboard trays. This is where you're attracting negative comments. Presenting things known to the users as brand new and your own invention.

Of course your implementation is different, it is impressive to see one that can manage such a high wattage CPU, but we are all a learning community. I mean you made us delve through a formula just to figure out what your load temperatures are. I can understand not wanting to have a 9900k compared to something like a 7700k temperature wise, but many here have a good grasp on what temperatures to expect from various CPUs and various voltage / clocks.

Nobody is criticizing your you build, just the presentation. It leads me to believe that either you believe the users here to be generally ignorant of the subject of which we are passionate about, or that you yourself are not aware of these past things or the body of experience that resides on these pages, in an example of the Dunning-Kruger effect. The latter is far more forgivable, but I fear you may find the implication insulting.

Either way thanks for sharing, and apologies for any misgivings on my part.
 
I presented 4 different case manufactures with sliding mb tray feature, you are just seeing what you want...

The video was cool. I wish I would have seen it first. I didn't understand what I was looking out in just the plain pictures. Now it all makes sense. My motherboard does not mount that way, so my own tunnel-vision could not picture how it was used. So thanks for clearing the fog.

5 months ago a youtube video was released with a prototype thermosyphon cooler that has been viewed 3.3 million times.

If you mean the one for Linus, I know all about it.

What you may not know is the turn-around time for a custom built condenser NOT being mass produced. Let me assure you, it's "back of the line, wait your turn" and in my case, it was 60 days.

For a custom-cut copper CPU block it is even longer. And I had to order 10 at a time at $ridiculous each. (My next 10 will cost me much less, like $500 each).

And the Intel blocks are not nearly the same size as the Threadrippers. More money out the window.

It took about 14 weeks for me to get everything I needed for the build. One week to do the prototype then convert it into something nice looking. One week to find a cute Asian girl for the video I did of it. And I didn't rush on here right afterwards to post about it.

I had all of this done by January 2020. So my guess is I started around August 2019.

I am a paid researcher, and I was working on another project that turned my focus back to computer cooling. I am not claiming it's an independent discovery. My guess is Linus had access to some Intel engineering sample chips, verified they ran hotter than hell, probably boiled the water in a large radiator like I did, and used whatever resources he had to find an alternative. Or maybe he was contacted by someone who already built a thermosyphon, maybe even the same person who paid me! Who knows. The timeframes are very close indeed.
 
Last edited:
Back