Cooling KMPKT Heatsink and Fan Poll
- Thread starter Kmpkt
- Start date
Basically I have to fund the Dynamo, Dynamo Mini and Merge first. Once that's done, depending on how much money I have left, I'll try to move on the heatsink ASAP. If things go well, I'd hope to have these available by the end of the summer. Basically I have to make an 8000 dollar gamble on tooling to even get a sample. Once I have that, I'll need to test it extensively to make sure investing more money is worthwhile. If it outperforms the L9i and LP53 by a significant margin (I can't imagine it won't), I'll move straight to my first batch of 500 units.
I would, but I don't really want to give 7+% of my total revenue to a crowdfunding platform if I don't need to. It's just more about timing than cost. I have a couple other things going on the background as well, so I really want to make sure I stretch my available capital as much as I can so I don't have to hold anything up for a couple of month while I replenish funds.
What makes you think that your heatsink/fan combo out-performs the L9i? We know Noctua puts ridiculous amounts of R&D into their products, so if there was an easy way to improve performance significantly, you'd think they'd have already done it.
I wouldn't risk $8,000 of your money on something that you are unsure will sell enough to recover your investment.
I know that you are not keen on the crowdfunding platform adding 10% to the product price, but you could counter that increase by raising the goal quantity which should lower the per product manufacturing cost?
Crowdfunding reduces your financial risk to zero, the difficulty is you have to get more backers to achieve the same price, but this will mean more profit for you if you sell more units!
I know that you are not keen on the crowdfunding platform adding 10% to the product price, but you could counter that increase by raising the goal quantity which should lower the per product manufacturing cost?
Crowdfunding reduces your financial risk to zero, the difficulty is you have to get more backers to achieve the same price, but this will mean more profit for you if you sell more units!
A full copper heatsink will outperform any aluminium equivalent, and more surface area outperforms less surface area. We don't know by how much, but I think Kmpkt is right in thinking this heatsink would outperform the L9i.
Under otherwise equal conditions. Depending on how the additional surface area is achieved, a cooler with more surface area could actually perform worse than one with less. Just spacing the fins tighter together might not help depending on the fan that's used.
I know that copper is in general a better heat conductor than aluminium, but none of us know how much of a bottleneck the heat conductivity of the fin material actually is, and the heat radiation properties of copper and aluminium are very similar.
Don't get me wrong, I'd choose this heatsink over the L9i anytime if it really is better, it just doesn't seem realistic to me that one guys first ever heatsink could easily outperform the competing product of a company that has over a decade of experience in making nothing but heatsinks.
The way I understand it is that the thinner the fins, and the farther away they are from the heatsource, the more thermal conductivity is the bottleneck. This is because the heat will already have been mostly dissipated to the air before it got transfered to the end of the fins (which makes the end of the fins pretty inefficient).
Considering Kmpkt is planning on making longer fins than the L9i, they will be farther away from the heat source, thus the thermal conductivity becomes more important compared to the L9i.
As far as I can see on noctua's site, they don't have any copper heatsinks. Since the thermal conductivity of copper is different than aluminium, their heatsink designs are probably optimized for aluminium fins. When going copper, there are probably different optimums in terms of fin density, fin length, distance to heatpipes, fin thickness etc.
Last edited:
There are some online thermal-calculators for this. I think
direct-contact heatpipes (as many as possible, going in both directions, or even better a '3d vapor-chamber).
Why not make the fins as wide as possible. E.G. 140mm x 140mm, and put a big 140mm fan over it, like the 'Be Quiet'.
It would only be too big for the tiniest STX builds, and even then, the tiniest STX cases should still fit it
Spend your resources on the sink, as I feel like beating the super-quiet fans would be difficult.
direct-contact heatpipes (as many as possible, going in both directions, or even better a '3d vapor-chamber).
Why not make the fins as wide as possible. E.G. 140mm x 140mm, and put a big 140mm fan over it, like the 'Be Quiet'.
It would only be too big for the tiniest STX builds, and even then, the tiniest STX cases should still fit it
Spend your resources on the sink, as I feel like beating the super-quiet fans would be difficult.
I don't know it's going to outperform and that's the gamble. At this point I'm simply banking on the fact that what is a somewhat poorly designed and finished heatsink (LP53) crushes the L9i in every test I've seen by a very significant margin. Since they both have roughly the same fin area and heat pipe size, the determining factor pretty much has to be the copper versus aluminum heatsink fins. Furthermore, the LP53 utilizes a non-optimal bend radius of the heat pipes which reduces conduction to the fins. This is a pretty significant strike against its performance if the fin array has the capacity to dissipate more heat than the pipes can carry at that radius.
My logic at present is that improving the the L53's design by adding fins out over the heat pipes is free dissipation area when compared to the LP53. Additionally I am planning to add several mm of additional fin space above and beyond these two heatsinks since with a 15mm fan the LP53 would come out to 43mm and the L9i 38mm and my final goal for height is 45mm. Right now how much I'm trying to figure out exactly how much that would be since apparently a lot of these heatsinks measure height differently (ie. no standard!). Hypothetically I could end up adding as much as 7mm to the L9i's height which would increase surface area to about 30% more than the L9i.
As far as Noctua's legendary R&D, I don't doubt for a second that the L9i is optimized for the market it was designed for. That being said, I don't think it was developed with SFF gaming and 95W CPUs in mind, but instead for HTPC use. From what I can tell the heatsink was developed in 2012 at which point I don't think legitimate SFF gaming and workstation PCs like we are now building were even a thing. HTPCs however were becoming increasing popular and Steam Machines hadn't even been announced yet. I think the HTPC use case is further strengthened by the fact that the fan only outputs about 20 dBA in noise while most other 92 x 15mm fans I've come across run in the high 20s to low 30s. I think the reason the L9i is still the king of the hill is exactly because Noctua is so good at design and optimization.
Also to clarify, while the original design was not done with OEM consultation I am now working with the engineers at a major OEM to optimize the design for both fin count and pipe configuration. Hopefully once they've had a crack at the heatsink it will provide a clear and definitive upgrade over anything else on the market. Furthermore (and this should be obvious) my design is pretty heavily influenced at this time by the L9i as I'm not that keen to completely reinvent the wheel.
- Jul 19, 2015
- 818
- 1,359
I dont have the article link any more but i do remember a while back that noctua said (something to the effect of) the main reason their not doing full copper or anything crazy isnt so much because there isnt a cooling benefit or a market for it but because of the higher production cost and sticker price vs their target price and sales targets. I guess they know who they want to sell too and how much they can charge to sell to as many of those people as possible
I agree that based on the reading that is available there are no hard proofs showing major cooling differences in constructed heatsinks, however that actual heat transferring properties of metals (which are of course tested to a degree that is indescribable here and are not up for debate anymore sadly) seem to validate this urban myth. We have all been over this before:
I tried to confirm this based on what I have read about the properties of metal and drawing on my materials science books. I think it ultimately, at the few degrees differences folks are seeing, comes down to heatsink design and properties rather than properties alone.
So, although the focus here is on the materials, I think you will be fine (but I AM an idiot!). You have a good product, use the material you think is best. The few degrees you change from managing the properties differently wont make as much of an impact as changing the design itself.
If you would like to know, you can be the first to break it. Make the same design in both different ways and see which is best. Hopefully it breaks margin of error so it is worth it
Thanks to all of you, that made things much clearer! I didn't know the LP53 was performing so much better than the L9i, that's a very encouraging data point to have.
Also very good point about Noctua optimizing their heatsinks for HTPCs and thus noise rather than pure performance, I didn't consider that.
That's good to know! I tried to find out why noctua didn't use copper, but gave up after a while. Another thing about copper might be that it clashes with their already unusual colour scheme, so they'd have to use nickel-plated copper or something like that, which would make the heatsinks even more expensive.
Also very good point about Noctua optimizing their heatsinks for HTPCs and thus noise rather than pure performance, I didn't consider that.
That's good to know! I tried to find out why noctua didn't use copper, but gave up after a while. Another thing about copper might be that it clashes with their already unusual colour scheme, so they'd have to use nickel-plated copper or something like that, which would make the heatsinks even more expensive.
For those that aren't familiar with the difference between the L9i and LP53's performance, here is a great read:
https://hardforum.com/threads/thermolab-cooltek-lp53-review.1923635/
https://hardforum.com/threads/thermolab-cooltek-lp53-review.1923635/
Anyone think the nickle plated base has something to do with Noctua lagging behind so much with this specific cooler? Intel moved away from the all copper core included heatsinks on consumer CPUs too, didn't they?
I think the results focus too much on materials, it really should say: "The LP53 handily beats the reigning favorite of the SFF crowd, while using the Noctua cooler's own fan no less. Given the same fan was used for both it really demonstrates the difference full copper..." and the much heftier design attributes "...makes over the nickle plated copper base and aluminum fins on the noctua."
Cool to see this stuff being tested - I really would love to see the same design heatsink but different materials head to head. I mean, look at those beastly copper pipes! It is almost not fair!The L9i looks anemic next to the LP53.
The conclusion is worded problematically, thinking the only thing that needed to be standardized for real control was the fan. The heatsink is different, man! Hehe. Still, a really good set of data.
I think the results focus too much on materials, it really should say: "The LP53 handily beats the reigning favorite of the SFF crowd, while using the Noctua cooler's own fan no less. Given the same fan was used for both it really demonstrates the difference full copper..." and the much heftier design attributes "...makes over the nickle plated copper base and aluminum fins on the noctua."
Cool to see this stuff being tested - I really would love to see the same design heatsink but different materials head to head. I mean, look at those beastly copper pipes! It is almost not fair!The L9i looks anemic next to the LP53.
The conclusion is worded problematically, thinking the only thing that needed to be standardized for real control was the fan. The heatsink is different, man! Hehe. Still, a really good set of data.
Last edited:
Similar threads
