Maybe you could license it to them and get a % of the profit instead?
Cooling KMPKT Heatsink and Fan Poll
- Thread starter Kmpkt
- Start date
Maybe you could license it to them and get a % of the profit instead?
Btw the noctua actually isn't 14mm. The 14mm is the frame depth. It is actually 15-15.5mm due to the rubber inserts around the screw hole.
This would likely be the arrangement, but it still doesn't get me out of paying the 40K up front
This would likely be the arrangement, but it still doesn't get me out of paying the 40K up front
Maybe get an arrangement where they agree NOT to pay you a percentage until AFTER the initial development costs are recouped?
I think I'll probably go the route I did with my heatsink and just find an aerodynamics engineer with relevant background to do the design. In the meantime I'll probably just go with Yate Loon's 92x15 fan at launch as it seems to work quite well and noise doesn't seem to be all that big an issue from my preliminary testing.
I know this is a weird question, but how much better would an all silver heatsink be compared to copper or aluminum?
Hypothetically silver conducts heat approximately 6.5% better than copper while copper provides approximately double the performance of aluminum. The fun part of this equation is that aluminum costs six US cents per ounce, copper costs 19 US cents per ounce while silver costs 16.43 US dollars. This means that a 20 dollar aluminum heatsink would likely cost close to 50 USD if made from copper and approximately 4300 USD if made from silver - any chance you want to be the first to try out a silver heatsink?
Are you still thinking about making a threadripper version if it’s good enough?
If the AM4/115X version works well, then I will be looking at the applications of this heatsink design in both servers and higher end workstations. I think it will actually scale better with larger chips and dies than it does with standard desktop processors. I haven't looked at the particulars, but I am hoping I can make a variant that will accommodate both Threadripper and Socket 2066 after I have the 115X/AM4 into production.
One of the fundamental advantages of my design over existing top-down heatsinks is how exhaust air is handled. In the case of servers in particular, I am hoping enclosure fans could be run at a much lower rate as a result of using my cooler design. The obvious advantage of this would be considerable electricity savings for larger operations.
As a side note, if anyone here could sufficiently explain how chassis cooling in servers typically works and how chassis airflow is handled I would be pretty appreciative. I have read up on the fundamentals, but more specific knowledge is always welcome.
One of the fundamental advantages of my design over existing top-down heatsinks is how exhaust air is handled. In the case of servers in particular, I am hoping enclosure fans could be run at a much lower rate as a result of using my cooler design. The obvious advantage of this would be considerable electricity savings for larger operations.
As a side note, if anyone here could sufficiently explain how chassis cooling in servers typically works and how chassis airflow is handled I would be pretty appreciative. I have read up on the fundamentals, but more specific knowledge is always welcome.
Along the lines of TR and 2066, can anyone tell me if we could actually even fit a 92mm cooler on either? It looks like 2066 would manage one (about 100mm x 100mm area), but TR4 looks like it might be too narrow. If there is adequate room for say an L9i to fit, then I might just release an aftermarket baseplate for TR4/2066 since it will be swappable (and TIM swappable!)
Not sure if I can sufficiently explain but I would be happy to offer my experiences working in enterprise situations. Your typical server chassis runs cooling front to back, if there is active airflow to be had in the first place. In almost all server boxes I have personally deployed there is just a(n) exhaust fan(s) populated in the back to just get that hot stuff outta there - I have so rarely seen intake fans in these applications that I can't even remember...
That being said those CPUs of course need cooling - and I have not seen anything applied to these other than short, fat little towers with at maximum 80mm fans pushing air through them. Sometimes, very rarely, top down - still anemic looking little aluminum blocks. In this case I have always seen horizontal flow work better than downward for the CPU and the VRM, etc. These stock coolers often come with server boards and/or server grade CPU's, or the chassis - at least they did in the past when I was working on them. They are the stock cooling solution equivalent and they work, sure, but of course they are loud.
Something like this:
Something important to note, from my experience there is never really any other cooling solutions applied to server boxes in enterprise environments. The stock cooler that is included with a purchases from a major company, like HP, almost always is what the IT staff implement when building. Not only is it easier, but it is cheaper, and it is hard to justify to your CFO that you need the coolest all copper or innovative cooling solution for your server farm. From an economic perspective, nowhere I have worked would even consider buying an aftermarket solution for the cooler in the first place. That, and I have never rolled out racks full of server boxes to any room that wasn't air conditioned in the first place. So, maybe there would be some comparatively smaller electrical savings to be had by turning exhaust/enclosure fans down some hundred RPMs, but the places I worked at would laugh me out of the room if I mentioned it as they would just leave the AC on, and take all the fans out altogether.
For these reasons it seems like the aftermarket server cooling solution is best suited to really small server setups without climate control, hobbyists, or something like Google's warehouse monsters. With the big ones, however, would have to convince them to ditch all that previous capitalization.
Then again, I really have no idea about markets or anything - so I hope this has been even slightly useful!
Last edited:
Kmpkt - are you contemplating releasing the cooler on its own - and potentially develop the fan later on… I appreciate that you want to provide a complete 'package', (not a 'frankenstein'). However, developing a fan will take some months… and a considerable amount of money or horse-trading to get there. The cooler on its own - appears (from your feedback), performing thermally very well and conceivably quite close to coming to market (Q1/Q2 2018)? There is a clamour for a high performance LP cooler as you know… All prepared/willing to attach alternative fans (which there are lots to choose from). I am sure this has crossed your mind - but it might be interesting to conduct a poll and see if there is a consensus?
All prepared/willing to attach alternative fans (which there are lots to choose from). I am sure this has crossed your mind - but it might be interesting to conduct a poll and see if there is a consensus?
People have gold, diamond-studded iPhones, so...Hypothetically silver conducts heat approximately 6.5% better than copper while copper provides approximately double the performance of aluminum. The fun part of this equation is that aluminum costs six US cents per ounce, copper costs 19 US cents per ounce while silver costs 16.43 US dollars. This means that a 20 dollar aluminum heatsink would likely cost close to 50 USD if made from copper and approximately 4300 USD if made from silver - any chance you want to be the first to try out a silver heatsink?
Currently tendering prototyping contracts. 3D printed is out at 1600 USD per heatsink!
Similar threads
