DAN HSLP-48: A powerful sub 50mm heatsink

Uwe2

n00b
Joined
Feb 21, 2017
Messages
13
For the screw driver gap it would be sufficient to cut 3 fins and not 5 as planned.
 

StrawKite

Limp Gawd
Joined
Nov 17, 2016
Messages
164
There aren't any ways a double 92mm fan heatsink will fit. I thought about it many times, but I can't see a solution. If you have a idea let me know.
What do you think about a layout like this Dan? from my understanding it's still 9mm too long at 184mm

 

dondan

[H]ard|Gawd
Joined
Apr 15, 2013
Messages
1,743
The fan will block the PCIe Port so you can't mount the riser :(
 

okwchin

Limp Gawd
Joined
Aug 29, 2013
Messages
357
Heatsink design for heatpipes and fins also need to keep in mind how its put together. The fins of these heatsinks are pushed onto the heatpipes (or vice versa?), and where they meet a flange is punched into the fin to maximise contact area between the two, as well as maybe helping with controlling the fin spacing. Some heatsinks might also have the contact soldered to ensure good contact, and potentially further improve performance (with increased costs too)

Being able to physically push fins onto the heatpipes is best acheived by having heatpipes parallel for as much of the heatsink as possible. Any bends in the heatpipes within where the fins are would require a widened hole to be cut in the fins instead of the flanged hole. A widened hole cannot be flanged to fit a non-perpendicular heatpipe, and therefore significantly reducing contact area and contact force between the fins and heatpipe. If there is poor contact between the heatpipes and fins, there is no heat transferring to the fins, and therefore the cooling efficiency is significantly reduced for that area. What appears outwardly as additional cooling fins may not actually be contributing as much performance wise.

Having a look at the Cryorig C7, it has a moderate fin area for its size, but where the heatpipes are curving in the heatsink, a widened hole has been cut into the fins, The fins in that area might not be receiving heat from the heatpipes as efficiently. From a tooling point of view, additional punches are also required for this additional shape of fin too.

A simpler and cheaper heatsink to manufacture might be one like the Silverstoen Ar05, which has heatsink bends outside of the fins, and a single shape of fins. Unfortunately this means balancing additional dimension to accomodate the bends while still having enough fins that are blown by the fan.

The Noctua L9x65 is a compromise between the two, with the heatpipe curving into the fins, but has some of the heatpipe hanging out, and some fins slotted.
 
Last edited:

Veritas187

Weaksauce
Joined
Feb 9, 2017
Messages
84
We could for sure design such a fan but I don't think people would like to wait 1-2 years for it to be ready. :p
Seriously though, I don't see us going for such a fan anytime soon, as it is an extreme niche size where it is questionable if we would ever even earn the development costs. It was a bit easier with the A9x14, as it can be used easily on coolers that have to be inside the Intel and AMD keep-out zones, which is not the case with a 100mm fan.

Regarding a cooler, please send me a PM with your contact address and I'll get in touch with you.
I can't promise anything but we can at least take a look at the topic.
I didn't know you are from Germany but it might make things a bit easier. :)
Why doesn't noctua just release limited edition l9i that's 100% copper and charge accordingly for it? Would simple material change tame i7 7700k level of heat to reasonable levels again not looking for a miracle.
 

dondan

[H]ard|Gawd
Joined
Apr 15, 2013
Messages
1,743
To give you a idea how the HSLP-48 could perform if I make it like a flat Noctua NH-L12 with A9x14 bottom fan please take a look at this review on computerbase:

https://www.computerbase.de/2013-05/7-top-blow-kuehler-test/12/

In the diagramyou can see the NH-L12 with only bottom fan, the AXP100 and the L9i. They tested them on a 2500k with default and 4,2Ghz clock. The test was made with Prime95.
Keep in mind that the A9x14 can spin with 2500RPMS while the NF-B9 has only 1600RPM. But the NF-B9 is thicker as the A9x14. Keep in mind that the AXP100 is 8mm higher as the HSLP-48. The NF-L12 with 1200mm fan could be in the same class as a HSLP-48 with the Noctua unreleased 120x15mm fan with low profile ram.

default all FANs @ 1600RPM
  • L9i+A9x14 = 74,6°C
  • AXP100+TY-100 = 63,5°C
  • NF-L12+NF-B9 =61,2°C
  • NF-L12+NF-F12(120mm)=54°C- 1400RPM


default all FANs @ max possible RPM
  • L9i+A9x14 = 100 - 64,3°C - 2500RPM
  • AXP100+TY-100 = 58,4°C - 2400RPM
  • NF-L12+NF-B9 = 61,2°C - 1600RPM
  • NF-L12+NF-F12(120mm)=54°C- 1400RPM


4,2GHZ all FANs @ 1600RPM
  • L9i+A9x14 = 95°C+ (so no values)
  • AXP100+TY-100 = 84,8°C
  • NF-L12+NF-B9 =76,6°C
  • NF-L12+NF-F12(120mm)=68,7°C- 1400RPM


4,2GHZ all FANs @ max possible RPM
  • L9i+A9x14 = 100 - 84,4°C - 2500RPM
  • AXP100+TY-100 = 76,6°C - 2400RPM
  • NF-L12+NF-B9 = 76,6°C - 1600RPM
  • NF-L12+NF-F12(120mm)=68,7°C - 1400RPM
 
Last edited:

StrawKite

Limp Gawd
Joined
Nov 17, 2016
Messages
164
To give you a idea how the HSLP-48 could perform if I make it like a flat Noctua NH-L12 with A9x14 bottom fan please take a look at this review on computerbase:

https://www.computerbase.de/2013-05/7-top-blow-kuehler-test/12/

In the diagramyou can see the NH-L12 with only bottom fan, the AXP100 and the L9i. They tested them on a 2500k with default and 4,2Ghz clock. The test was made with Prime95.
Keep in mind that the A9x14 can spin with 2500RPMS while the NF-B9 has only 1600RPM. But the NF-B9 is thicker as the A9x14. Keep in mind that the AXP100 is 8mm higher as the HSLP-48. The NF-L12 with 1200mm fan could be in the same class as a HSLP-48 with the Noctua unreleased 120x15mm fan with low profile ram.

default all FANs @ 1600RPM
  • L9i+A9x14 = 74,6°C
  • AXP100+TY-100 = 63,5°C
  • NF-L12+NF-B9 =61,2°C
  • NF-L12+NF-F12(120mm)=54°C- 1400RPM


default all FANs @ max possible RPM
  • L9i+A9x14 = 100 - 64,3°C - 2500RPM
  • AXP100+TY-100 = 58,4°C - 2400RPM
  • NF-L12+NF-B9 = 61,2°C - 1600RPM
  • NF-L12+NF-F12(120mm)=54°C- 1400RPM


4,2GHZ all FANs @ 1600RPM
  • L9i+A9x14 = 95°C+ (so no values)
  • AXP100+TY-100 = 84,8°C
  • NF-L12+NF-B9 =76,6°C
  • NF-L12+NF-F12(120mm)=68,7°C- 1400RPM


4,2GHZ all FANs @ max possible RPM
  • L9i+A9x14 = 100 - 84,4°C - 2500RPM
  • AXP100+TY-100 = 76,6°C - 2400RPM
  • NF-L12+NF-B9 = 76,6°C - 1600RPM
  • NF-L12+NF-F12(120mm)=68,7°C - 1400RPM
Dan, my german is pretty bad. Did they do the test inside a case? And how mouch worse do you think performance would be inside the dan case? Not much I hope.
 

dondan

[H]ard|Gawd
Joined
Apr 15, 2013
Messages
1,743
The first step is done. I am in contact with Thermalright and Noctua.

Both of them are in the process of validating if they are interested in.


I updated the renderings in the 1. post of this thread. The new renderings includes the changes of my last update, but it does not include the new fin count.

Here is the old design vs the new one:



 
Last edited:

Ca11idus

Limp Gawd
Joined
Jan 9, 2012
Messages
493
That looks really cool. I'm interested to see how that will fit the narrow ilm slot on the asrock x99e-itx
 

Ca11idus

Limp Gawd
Joined
Jan 9, 2012
Messages
493
Will this be patented and sold by you or will thermalright or noctua sell these? If it is accepted by either.
 

okwchin

Limp Gawd
Joined
Aug 29, 2013
Messages
357
Great to see this is moving along the right way, I'm excited :)

I see there is a gap for the mounting screw through the fins, For those fins either side of the access hole, where are those fins attached? to the outer heat pipe only?

And whats the dark grey area on the fins near the other end of the heatsink, about 10 fins from the end of the heat pipes. is this a shadow of something behind? or nothing significant.
 
Last edited:

Shaav

Limp Gawd
Joined
Sep 28, 2011
Messages
163
My thoughts on this:

1. Our current problem is the small size of the cooler. Instead of designing a cooler that can use 25mm thick fans, I think it is better to use a slim fan and a large cooler.
2. In the current design the fan will push the hot air out of the pc case. This design prevents any turbolence noise caused by the side panel, but we will have turbulence issues with the mainboard instead. I don't think you will be able to run the fan on high speeds. I would optimize the cooler for the efficient NF-A9x14.
3. I would prefer the Noctua-FAN. It performs better at 1850rpm (which is super quiet!) and has less problems to push air through a very dense radiator.
4. If low-profile RAM would make it possible to mount a larger fan on an even larger cooler, I think this would increase its performence massively.
5. I don't think there is a fan on the market for this kind of scenario. We would need one with low noise and high prease. I assume it also had to be thicker than 15mm, probably 20mm.
6. I would put a thin ring of rubber between the cooler and the side panel to prevent the air of going elsewhere than directly out of the pc case.

How much space is under the fan?
24mm
On the pictures it looks more like ~12mm.

- Here is the new surface calculation:
normal fins = 18.5 (H)mm * 130 (L) mm * 60 count = 86394mm²
small fins = 18.5 (H)mm * 59 (L) mm * 5 count = 5015mm²
5% lost area for heatpipes hole
Total: 142269mm²

I don't think extending the cooler to 130mm or any much larger than 100mm will increase the performence by a lot, because there is just a tiny 100mm fan below. I would go for a high density, like the noctua and C7 did.


All in all, I don't believe this cooler would perform better than
- a copper version of the L9i
- a copper version of the L9x65 where the fan is mounted between the heatsink and the cooler
If you ignore the noise I might be wrong, but I will definitely throttle a 25mm fan to 1300rpm or a 15mm fan to 1850rpm.
 
Last edited:

dondan

[H]ard|Gawd
Joined
Apr 15, 2013
Messages
1,743
@Ca11idus: Yes it will fit on Narrow ILM and the X99e-ITX. It will be sold by DAN Cases and manufactured by Thermalright or Noctua

@okwchin: In the current design they are attached to nothing because the heatpipe spacing isn't final. It will be attached to the outer pipes. The dark line is a rendering error :D

@Shaav:

1.) The current design is for 15mm coolers

2.) The space between motherboard surface und buttom side of the fan is 25mm hopefully enough.
3.) If I cooperate with Noctua I will use the A9x14 and plan to also include the new 120x15mm fan that was publisehd on CES 2016
4.) Yes it will but this should only be an option. Because LP DDR4 is very expansive and hard to get
5.) The TY-100 has a very high pressure but it is only 100mm.
6.) I will think of that but, but I don't think this is neccessary because there will be only 2mm between pannel and heatsink.

The space between buttom CPU plate and fins is 20,5mm. But you will need 2mm for the rentention plate.

If you use a 120mm fan it will help also I plan to add bends so that more air will reach the side areas with a 100mm fan.


@ALL: Here is a picture how a copper version could look like:



 

Shaav

Limp Gawd
Joined
Sep 28, 2011
Messages
163
dondan:
2. I am certain 25mm would be way more than enough. Is there a problem with turbulences with the heatspreader and teh fan, or why is this distance between the fan and the mainboard so huge? I would prefer a thicker cooler and less free space. My thoughts are that a faster spinning fan will improve the cooling performence not much, but a thicker cooler will. Therefore I would opitimize the cooler for lower fan speeds. Lower fan speed means that there are less turbulances an we don't need to reserve that much space here.
3. The CES-fan ooks very much the same like the one I saw in my dreams. I believe in combination that with a 120mm-wide cooler and VLP-RAM this cooler could outperform alll the other coolers.
4. I would recomment to build a cooler that is optimized for that fan from CES, but can be used with the A9x14 if normal sized RAM is used.
5. I noticed that this fan performs worse than the A9x14, if you run it with fan speeds of 1850rpm on a cooler with very dense cooling fins (NH-L9i).
6. I think so too, but maybe interesting for modding the shit out of it :D

The space between buttom CPU plate and fins is 20,5mm. But you will need 2mm for the rentention plate.
Doesn't that mean that the fins could be almost 30mm high? Why do you go for 18,5mm?

How much space is under the fan?
24mm
I'm confued. Shouldn't it be 20,5mm + 2mm rententionplate - 15mm fan thickness?
 

dondan

[H]ard|Gawd
Joined
Apr 15, 2013
Messages
1,743
Here all meassures front the front view:

I have to correct myself distance between fan and topplate is 5mm instead of 6.5mm



Shaav:
I can not lower the fan because of rhe keep out areas on the board. It has to fit over the heatsinks and other components of the board.

With a 30mm thick fin area you can not use normal ram and it will not fit on the topside of the I/O area. On the drawing above you see that on the left and right side the fins are only 16,5mm height so it will fit over the I/O.
 
Last edited:

Petroj

n00b
Joined
Oct 17, 2016
Messages
33
Sure hope Noctua or Thermalright agree to cooperate with you and that a prototype will get made quite soon. A kickstarter in the early summer would be nice. I'm just worried that we are able to get our hands on it no earlier than this time next year. Either way, I'm interested :)
 

cowsgomoo2

Limp Gawd
Joined
Jan 8, 2016
Messages
144
If you chop off the ram side of the heatsink, won't it become compatible with tall ram? I know that you'll lose a bit of cooling and it will trigger OCDs with a non-symmetrical heatsink.
 

cowsgomoo2

Limp Gawd
Joined
Jan 8, 2016
Messages
144
Yes tall ram with high heatsinks will be incompatible. This is a compromise to have a better cooling performance.
Yes. I just don't think that the compromise is worth it. I have a feeling that many would-be buyers don't even know their ram height. Outside of hardforum that's even more so.
 

StrawKite

Limp Gawd
Joined
Nov 17, 2016
Messages
164
I have a new idea, how about a 80mm x 160mm heatsink with dual 80mm x 15mm fans (NoiseBlocker-blacksilent pro)



Dan, what program do you use for your sketches?
 
Last edited:

dondan

[H]ard|Gawd
Joined
Apr 15, 2013
Messages
1,743
And this will also block the PCIe port so you cant install the riser. The max. Length that could work is 140-150mm. Furthermore for a dedign like this it would be the best to make it like the gpu heatsink in an older bost with fans on top and a height of 29mm without fans.
 

dondan

[H]ard|Gawd
Joined
Apr 15, 2013
Messages
1,743
Nearly the complete height because i would not recomment to bend it that sharp.
 

jjsyht

Weaksauce
Joined
May 29, 2016
Messages
86
Nearly the complete height because i would not recomment to bend it that sharp.
And you don't want your fan to slowly chip the riser.... Or heatsink slowly melt the riser.

Basically, like Dan says, nothing over the pci slot
 

jjsyht

Weaksauce
Joined
May 29, 2016
Messages
86
About the fan direction, I see that it is pushing air from the motherboard to heatsink and outside.
I'm no expert, but
  1. Won't the PSU get warm air? (Applicable for Dan's case only)
  2. Being the only fan expelling air, won't the CPU area be the natural exhaust of the case, thus pushing warm air through the heatsink?
  3. Won't positive pressure concept of the case be ruined with different & independent fan speeds?
Just thinking...
 

kingtron

Limp Gawd
Joined
Feb 2, 2017
Messages
254
dondan what is the maximum Ram height ? if:

1- if you use 100mm fan and default height Ram
2- if you use 120mm fan and VLP Ram

I need to know if innodisk DDR4 2400 VLP UDIMM is compatible or not ? ( if used with 120mm fan of course), here is the diagram for it:
Capture.GIF
Memory-Form-Factor.png

It is the only option we have right now for a DDr4 VLP (nonECC and un-buffered)
 
Last edited:

Boil

[H]ard|Gawd
Joined
Sep 19, 2015
Messages
1,439

If I could get this in full copper, manufactured by Noctua, with their (forthcoming) NF-A12x15 fan (dunno if that is the actual model designation, but it seems to make sense), and giving maximum coverage of the MB, but allowing clearance for both the PCIe riser & a brace of Corsair Vengance LPX DDR4 RAM, AND have compatibility with AM4 sockets; I would easily pay $100 to $120 US dollars for this product…!!!

My usage would be in the Firewolfy MI-6 chassis, hopefully with an AM4 socket/X300 chipset ITX MB & Ryzen R7 1700 CPU & a GTX 1070 class (if not a GTX 1080 class) ITX-sized Vega GPU (or a MSI Aero ITX GTX 1070)…! So no worries about exhausted air feeding into the PSU…

Or if Nvidia would just make us a Titan ITX…!!!
 
Last edited:

octagoncow

Limp Gawd
Joined
Aug 31, 2014
Messages
130
Wow! Thanks for the render of it inside the A4. It does a great job of showing just how large the heatsink really is.
 
Top