KMPKT Heatsink Discussion

[theGryphon]

Such tease you've been with this heatsink

I'm curious to see the design and whether it can be upscaled... any comments on that? I mean, can it be upscaled to a taller, beefier, more potent product?

 

[onelsonic]

Hi there, quite a unique product you have there. Congrats on the great work!!
I just ordered the dynamo 360 + combo, and wondering if this heatsink could be retrofitted in the future to the current dynamo 360 model?
I needed the unit now for testing, but my load will be quite demanding in terms of power and I suspect it might generate some heat (->I will be testing to fit a 1080Ti)
Thank you

 

[Kmpkt]

I'm curious to see the design and whether it can be upscaled... any comments on that? I mean, can it be upscaled to a taller, beefier, more potent product?

I can scale it for width or height very easily in order to make it run at a greater capacity (ie. 65mm tall variant). I am also hopeful that I could eventually scale it for length in order to cool GPUs.

I just ordered the dynamo 360 + combo, and wondering if this heatsink could be retrofitted in the future to the current dynamo 360 model?
I needed the unit now for testing, but my load will be quite demanding in terms of power and I suspect it might generate some heat (->I will be testing to fit a 1080Ti)
Thank you

This is a CPU cooler, not a cooler for the Dynamo 360. I wouldn't worry about heat dissipation on the 360 that much unless you're planning to run it well over the rated 360W.

 

[ECHO]

I am also hopeful that I could eventually scale it for length in order to cool GPUs.

That would be awesome

A complete high performing cooler ecosystem? That's great news for the SFF community!

 

[Kmpkt]

After two or more months of trying to get someone to make the jig and spacing shims for assembly I finally have everything. Should be putting this thing together on June 30th and will hopefully have testing results for all four prototypes at that time. Rough specs right now are:

Width: 92mm
Length: 92mm
Height: 31mm
Height with 14mm thin fan: 45mm

Variants:
1.) Alu Base / Alu Fins / 2 Heat Pipe
2.) Alu Base / Alu Fins / 3 Heat Pipe
3.) Cu Base / Alu Fins / 3 Heat Pipe
4.) Cu Base / Cu Fins / 3 Heat Pipe

Will be testing all four models with the AF-9x14 fan to determine the most effective combination of heat pipes and materials. Once I've tested all prototypes, then I plan to progressively shave away fin material until I've determined optimal cooler height given the ability of the base and pipes to move heat to the fin array.

Will be testing the Curve against the Cryorig C7 Alum Cryorig C7 Cu, L9i and LP53 + A9x14 combo.

Once I have encouraging results, I plan to file my provisional patents and once those are in place will be posting a comprehensive build log that goes from the concept stage, through engineering, simulation, manufacture, assembly and finally testing. Wish me luck guys!

 

[Nanook]

After two or more months of trying to get someone to make the jig and spacing shims for assembly I finally have everything. Should be putting this thing together on June 30th and will hopefully have testing results for all four prototypes at that time. Rough specs right now are:

Width: 92mm
Length: 92mm
Height: 31mm
Height with 14mm thin fan: 45mm

Variants:
1.) Alu Base / Alu Fins / 2 Heat Pipe
2.) Alu Base / Alu Fins / 3 Heat Pipe
3.) Cu Base / Alu Fins / 3 Heat Pipe
4.) Cu Base / Cu Fins / 3 Heat Pipe

Will be testing all four models with the AF-9x14 fan to determine the most effective combination of heat pipes and materials. Once I've tested all prototypes, then I plan to progressively shave away fin material until I've determined optimal cooler height given the ability of the base and pipes to move heat to the fin array.

Will be testing the Curve against the Cryorig C7 Alum Cryorig C7 Cu, L9i and LP53 + A9x14 combo.

Once I have encouraging results, I plan to file my provisional patents and once those are in place will be posting a comprehensive build log that goes from the concept stage, through engineering, simulation, manufacture, assembly and finally testing. Wish me luck guys!

Good luck!

 

[MultiDoc]

Hope you get excellent results KMPKT!

 

[Thehack]

After two or more months of trying to get someone to make the jig and spacing shims for assembly I finally have everything. Should be putting this thing together on June 30th and will hopefully have testing results for all four prototypes at that time. Rough specs right now are:

Width: 92mm
Length: 92mm
Height: 31mm
Height with 14mm thin fan: 45mm

Variants:
1.) Alu Base / Alu Fins / 2 Heat Pipe
2.) Alu Base / Alu Fins / 3 Heat Pipe
3.) Cu Base / Alu Fins / 3 Heat Pipe
4.) Cu Base / Cu Fins / 3 Heat Pipe

Will be testing all four models with the AF-9x14 fan to determine the most effective combination of heat pipes and materials. Once I've tested all prototypes, then I plan to progressively shave away fin material until I've determined optimal cooler height given the ability of the base and pipes to move heat to the fin array.

Will be testing the Curve against the Cryorig C7 Alum Cryorig C7 Cu, L9i and LP53 + A9x14 combo.

Once I have encouraging results, I plan to file my provisional patents and once those are in place will be posting a comprehensive build log that goes from the concept stage, through engineering, simulation, manufacture, assembly and finally testing. Wish me luck guys!

I have a couple comments:

1. I think 43mm total height would be better as it avoids the issue of turbulence inside the Dancase.

2. I think 105mm x 95mm may be a better design, with the extra 10mm offset in one direction. This allows you to use maximum amount of surface area, and while keeping compatibility similar. It would be very difficult to come up with a better performing cooler than what's on the market if we don't go a lil' bit bigger.

At the same time, if you go for 45mm, you encounter the issue of turbulance so 43mm will offer some relief there.

3. Perhaps experiment with keeping the heatpipes curve outside of the heatsink (vs C7, AR11). I'd like to see 3 to 4 pipes instead of 2-3.

4. Can we have it parallel to the RAM modules when used with AMD?

Regardless, good luck! Hope this project is successful.

 

[Kmpkt]

1. I think 43mm total height would be better as it avoids the issue of turbulence inside the Dancase.

This is quite likely. I started with the fins at the maximum desired height, but intend to grind them down 1mm at a time until we start to see performance loss. I am hoping the final unit will be shorter than 45mm.

I think 105mm x 95mm may be a better design, with the extra 10mm offset in one direction. This allows you to use maximum amount of surface area, and while keeping compatibility similar. It would be very difficult to come up with a better performing cooler than what's on the market if we don't go a lil' bit bigger.

My concern is fitment conflicts with boards like the Asus Strix that have absurdly oversized heatsinks on various different components.

Perhaps experiment with keeping the heatpipes curve outside of the heatsink (vs C7, AR11). I'd like to see 3 to 4 pipes instead of 2-3.

Because of bend radii and how heat pipes respond to them, this likely won't be possible. I also want to keep the overall dimensions of the heatsink down so that users don't encounter fitment issues like those using the LP53

Can we have it parallel to the RAM modules when used with AMD?

This is certainly a consideration in design.

 

[Thehack]

This is quite likely. I started with the fins at the maximum desired height, but intend to grind them down 1mm at a time until we start to see performance loss. I am hoping the final unit will be shorter than 45mm.



My concern is fitment conflicts with boards like the Asus Strix that have absurdly oversized heatsinks on various different components.



Because of bend radii and how heat pipes respond to them, this likely won't be possible. I also want to keep the overall dimensions of the heatsink down so that users don't encounter fitment issues like those using the LP53



This is certainly a consideration in design.

I'm a bit concerned that if you don't make it bigger or have better heatpipe contact itll perform too close to already available coolers. We would have to sacrifice some compatibility for the sake of performance.

I also hate that manufacturers put fancy ornamental "heatsink" that end up just getting in the way.

 

[Kmpkt]

To be honest, to do what you're suggesting would likely require me to lengthen the heat pipes and make a longer heatsink which in turn would only allow installation on Intel boards with the fins perpendicular to the RAM. This would likely negate any advantages provided by a slightly longer heatsink and, in the case of my design, would seriously inhibit the exit velocity of warm air (this would be significant).

Furthermore I tried my design with more than two heat pipes to the fin stack and it did't make a significant difference in performance. This is largely due to my novel approach to moving heat to the fins via the base. You can only move so much heat to a fin before it is saturated. It would appear we hit that point quite easily.

 

[Thehack]

To be honest, to do what you're suggesting would likely require me to lengthen the heat pipes and make a longer heatsink which in turn would only allow installation on Intel boards with the fins perpendicular to the RAM. This would likely negate any advantages provided by a slightly longer heatsink and, in the case of my design, would seriously inhibit the exit velocity of warm air (this would be significant).

Furthermore I tried my design with more than two heat pipes to the fin stack and it did't make a significant difference in performance. This is largely due to my novel approach to moving heat to the fins via the base. You can only move so much heat to a fin before it is saturated. It would appear we hit that point quite easily.

This is what I mean:

I found that most intel motherboards is pretty cramped, 95mm width is all you get. But quite a few of them allows beyond 95mm towards the PCIe.

AMD doesn't have this issue due to have less components around their socket.

Though really, it's just a beefed up LP53 at this point. I don't think we can get much better than the LP53 without doing something outside of the specs.

Unless, your approach is significantly different and provides more performance. There's certainly a lack of focus on LP coolers so I'm sure there is still progress to be made. Tower coolers on the other hand, we're fast approaching the threshold of current heatpipe tech.

I also wonder if a full vapor chamber block provides good performance if we use copper fins. I don't think ID-Cooling vapor chamber is particularly high quality, so there's little to make comparison to.

edit: I also just measured my LP53. The heatpipe length is 100mm, and the fin width is 95mm, so they got a similar idea. Except their's wouldn't work for motherboard that has exactly 95mm of width allowed, since the fins are now perpendicular to the RAM.

 

[Kmpkt]

But quite a few of them allows beyond 95mm towards the PCIe

This is exactly why I can't make a heatsink like this. I'm already serving a niche within a niche in that I am looking to sell to Small Form Factor enthusiasts that want to run high wattage CPUs. While I could be wrong, I am pretty much certain that the Z270/Z370 Strix boards are the most popular on the market right now. This means that by interacting negatively with their M.2 and Choke heatsinks I'm basically cutting a significant part of the market out of my client base. Considering I have somewhere north of 20K into the design and prototyping of this heatsink so far, the last thing I need to do is chop my market in half trying to eke out an extra 1-2C.

I don't think we can get much better than the LP53 without doing something outside of the specs.

This is where my heatsink should hopefully shine. Without giving too much away, my design should transmit considerably more heat through the base to the fins and have markedly more surface contact between the base and each individual fin than anything else out there. It should also allow for far less recycling and stagnation of warm air around the cooler which will hopefully keep the delta at the fins very high.

 

[Vlad502]

Furthermore I tried my design with more than two heat pipes to the fin stack and it did't make a significant difference in performance. This is largely due to my novel approach to moving heat to the fins via the base. You can only move so much heat to a fin before it is saturated. It would appear we hit that point quite easily.

Maybe something similar to Akasa Vortexx Neo heatsink (28mm height and 33 fins in 64mm width):

Spoiler: Photos

http://s.webry.info/sp/rarara.at.webry.info/200807/article_22.html

Performs good with 100W TDP: Vergelijkingstest: 6 videokaartkoelers (hardware.info)

 

[VegetableStu]

take note @Kmpkt isn't publicising the heatsink design at this point due to patent strategy, so having to answer physical thermal transfer suggestions is pretty disadvantageous for them at the moment.

besides the prototype's about to be assembled, so any pipe or transfer related suggestions would already be too late, unless a major redesign is required after testing

 

[Thehack]

take note @Kmpkt isn't publicising the heatsink design at this point due to patent strategy, so having to answer physical thermal transfer suggestions is pretty disadvantageous for them at the moment.

besides the prototype's about to be assembled, so any pipe or transfer related suggestions would already be too late, unless a major redesign is required after testing

Would it be US patent? If so, then us uses a first to file patent and if he submitted it already he would be safe. Regardless, I'm just in patient since I love lp coolers.

 

[Kmpkt]

Since the patent process is very expensive and I don't have conclusive real-world evidence that my design will be truly superior (engineering simulations are very encouraging), I've opted to hold off on filing a patent until I have results in the interest of not wasting thousands of dollars. I'm also not keen to divulge my design any earlier than necessary because I realize that someone in China could rip the design off and get a jump start on me in manufacture etc. quite easily were I to do so.

 

[Bonk]

Very interesting project, looking very much forward to seeing your results!

If/when you start selling this, do you know yet if you will bundle it with a fan?

 

[Kmpkt]

I haven't honestly gotten that far yet. Probably yes, but knowing that people will almost certainly opt to bundle with an aftermarket fan I may choose not to.

 

[ECHO]

Man, just wanted to stop by again to say how exciting this is. Can only imagine how exciting it must be for you! Best of luck with the assembly