DAN HSLP-48: A powerful sub 50mm heatsink

Even with the NF-A9x14 (2.5w 2500rpm version) at 50-70% fan speed the Black Ridge can keep my 9900K within the 70s-80s in intensive gaming loads like modded GTA V.

Undervolting helps a good amount — I have my 9900K at -30mV offset which translates to ~1.12v in the most extreme vDroop (e.g. Prime95 12K-12K fixed in-place FFT).

NF-A12x15 should have a pretty easy time.
 
I'm really hoping that I can make the Black Ridge work on an x570 itx board, but the manufacturers sure are making it tough! Thought I'd come to here to share what my research has dug up:


The gist of it is, ASUS and ASRock look like they are definitely not going to work. The chipset heatsinks are quite tall. As for Gigabyte, it appeared to be low enough to clear the fin stack, a Gigabyte rep confirmed this to be the case, however the 92mm fan looked like it would be a tight fit. Upon closer inspection, I'm a lot less confident that the 92mm fan will fit - possibly missing by just a couple mm. From here I'm going to do 2 things.

1) I will try to figure out a way to offset the 92mm fan a bit. If the heatpipes are mounted on the north side of the mobo, I'm wanting to move the 92mm fan slightly more north. I think this will require it to be dropped down a bit as well, because the heatpipes appear to almost immediately start to curve. I don't have my Black Ridge on hand to know this for sure, but I will have it next Tuesday evening. As for the "how", I'm really not sure, haha - maybe mount angle brackets (https://cdn.sparkfun.com//assets/parts/4/5/7/8/10228-01.jpg) and then use zip ties to fasten the angle brackets to the fan? Maybe use non conductive wire (nylon??) to wrap around the finstack/heatpipes and through the screwholes of the fan? Some concern about its ability to be moved after I do either of these... *shrug*

2) See if I can actually try to mount the CPU cooler without needing to pay a restocking fee in the event it doesn't fit. Being in Brisbane, I don't have a ton of options. So far I've asked UMart and Megabuy about this and both of them told me that I would have to purchase the motherboard to try it out. I do have a contact that is near a store that will let them try to mount it (hi Dazr) without needing to buy it, so hopefully that pans out if I am unable to personally give it a shot. By chance, is there anyone here in Brisbane that is planning on purchasing the Gigabyte x570 itx board on launch day that would be willing to let me try? Or knows of a local store that would let me try? :)

Please let me know your ideas of alternate mounting techniques!

The only other bastion of hope I have for x570 at the moment is that the M.2 heatsink on the ASUS board is not actually required and that the installed m.2 will sneak under the Black Ridge like the b450i's m.2.
 
Last edited:
Please let me know your ideas of alternate mounting techniques!


You should be able to remove the M.2 / chipset heatsink / cooling fan above the PCI-e port. I was waiting to see if that would be possible before getting this board myself but I've now decided that I don't need the 12C chip, either the 3700X or 3800X will do me, so settled on the Gigabyte B450 board in its place (it's also the only motherboard that'll fit the sentry 2.0 with this cooler).
 
You should be able to remove the M.2 / chipset heatsink / cooling fan above the PCI-e port. I was waiting to see if that would be possible before getting this board myself but I've now decided that I don't need the 12C chip, either the 3700X or 3800X will do me, so settled on the Gigabyte B450 board in its place (it's also the only motherboard that'll fit the sentry 2.0 with this cooler).

Yes, this heatsink can be removed. You have to do so to install the m.2. My assumption though is that this heatsink is going to be a requirement for the new chipset to not overheat. Afterall, every x570 board except for one has active cooling for the chipset, and that one is still passively cooled as part of a giant monoblock.

That said, maybe it will turn out to not be necessary - maybe the airflow from the 92mm fan will be more than enough to compensate?
 
So I started out with the heat sink. I knew the top of the delidded CPU sits about 2mm lower than the stock heatspreader (actually it's 2.5 mm after measuring it but this measurement wasn't documented anywhere online) and the mounting kit for the Black Ridge (BR) doesn't allow for any adjustment. So I had to shorten the embedded nuts on the mounting kit by those millimeters. I used my bench grinder to do it and then finished everything of with a small keyhole file.

7k6RFsW.jpg
XhOFWFS.jpg
zBfdZKW.jpg
9zKuD7b.jpg



Then I had a look at the board. The problem is that the BR doesn't work with the board to begin with (which I knew). It ever so slightly interferes with the VRM heat sink / IO cover and with the BR now sitting over 2 mm lower it's not slight anymore.
So I took of the heat sink. It's easy to by just unscrewing 4 screws from the bottom of the board. Interestingly enough I found that the thermal pad was squishing the CMOS battery cable between the heat sink and the board.

ezkYALI.jpg
I1xLg54.jpg


Now here begins the real fun part. I needed to cut the heat sink down to size so I marked it out with a pencil then took it to the basement to mutilate it. I first tried using a Dremel but the aluminium is quite thick an malleable which makes it hard to cut with a speed tool. So I went old school and used a vise and a small metal hand saw. Then I took it to the bench grinder to clean it up which took some time because it of course got very hot and I had to go slow. I then finished it of with a file and got a decent surface finish.

jQL3Slp.jpg


Remounting it was even easier than unmounting it since there now were only two screws left after all.

Now I tried the BR again but to my dismay I found that it now conflicted with the small caps next to the VRMs.

jQL3Slp.jpg


So now what do we do? Of course we bring out the Dremel to cut it to size. I cut of everything a close to the screws as possible with a milling bit and then finished it of with some sandpaper.

n7T3Sb4.jpg
NTDC54m.jpg
ATuOKLs.jpg


Trying it out that now it fits very tightly but it fits.

FUsd9IR.jpg


Since I didn't know for sure if the case of those caps is actually connected to one of the poles I put a small piece of tape on them just as a precaution.

Now there was still another problem to solve but this too could be taken care of by chopping up some hardware. The fan has almost no clearance to the VLP ram to begin with and now since the whole unit sits over 2mm lower... you know the drill. I marked it out then once again took to the basement.

SLRn3Tu.jpg


Fortunately plastic cuts pretty easily but it melts and stinks when using a speed tool. I took care to tape away the cable and fix the fan blades in place so that I wouldn't accidentally cut any of that. The corners where the hardest to cut so I took them out in chunks. One of the spokes that hold the motor in place had to go too. This time I used a knife and again a file to clean up the cut. This piece could be slightly cleaner I guess but it's going to be on the underside of the heat sink pressed up against the RAM anyway.

HSADUNK.jpg


Now everything was ready to begin assembly. Using the stock metal clip to mount the fan on the ram said wouldn't be possible for multiple reasons (one of them being that part of the fan wasn't there anymore) so I used some very thin wire to affix it.

lNIQ5Hu.jpg
dd1xllV.jpg


I then found out that you have to mount the board first before the PSU otherwise it isn't possible to get it in. Maybe on a board where the IO shield isn't fixed to the board it is possible in a different order and maybe reading the manual could have told me that sooner but ain't nobody got time for that. After reassembling everything I found out the hard way that it is basically impossible to connect the power button to the pin header with the CPU heat sink and fan in place so of course I disassemble and reassembled the whole thing again. Then I took to cable management which was going nicely thanks to the sleeved custom cables I bought from pslatecustoms.

Now came the magic moment, turning it on for the first time. And... nothing. The fans spun up but I wasn't getting any picture. After some head scratching I looked at the board only to find it showing a CPU error. So after double checking that nothing was shorting anywhere and everything was connected I faced the facts and disassembled everything again. I took off the heat sink but everything looked fine. So I thought maybe the OC frame wasn't holding the CPU down properly or maybe one of the tiny springs in the socket is bent. I removed the frame and CPU and checked the springs but they looked completely fine even after a detailed inspection. I then remounted the stock Intel retention module and put the BR back on as an interim solution taking care not to tighten it down completely as that would be too much with the shortened standoffs. Testing the board outside of the case (which is easy since there is an internal GPU) still showed the same error.

So that's where we are now. Something is broken. Either the board, the RAM or the CPU. The board is showing a CPU error and not RAM so the likelihood of it being that is lower. Now I might have killed the board but taking of and putting back on the VRM heat sink shouldn't have done that. Neither should changing out the retention module. So in all likelihood the CPU is actually broken.

I bought the CPU from Caseking already delidded and including the OC frame. I chose not to delid it myself since I've had issues with broken DIEs in the past and it seems like a risky operation on a soldered CPU to me. Now that might sound funny since I've chopped up half of my hardware here but if you think about it that all was basically just some metal working on components that didn't even contain any electronics (aside from the fan motor). They sent it over in a small carton with foam but they put the heatspreader and the CPU on the same piece of foam with them rattling around in there. They should have put the heatspreader under the bottom piece of foam and used some tape to hold the CPU in place or something I think. My guess is that the edges of the heat sink bumped up against the DIE and chipped the edge. It is incredibly hard to tell but one corner at least seems ever so slightly chipped (bottom left).

MvbfqJ8.jpg


The packaging seemed sketchy to me to begin with which is why I took a picture of how it arrived originally.

D1UXrS0.jpg


I've contacted them to get a replacement immediately and hope they will sort it out without blaming me. I would guess customers breaking delidded CPUs is pretty common so they might not be willing to replace it. However I have some experience with this stuff and took extra care not to damage anything so I am 99.9% sure I didn't. If they end up replacing it it could still be the board. In that case I will just buy another one and swap heat sinks. We will see how this story ends but for today it doesn't have a happy end.

If you try to do this: Don't. If you still want to do it you should either like a challenge, be a masochist or be prepared to spend a couple hundred or even thousands on dead hardware, preferably all three. Now I knew what I was getting myself into and at least 2 of those attributes fit me so I am not going to be discouraged this easily but again do not try this at home.
 
Last edited:
Those of you who compared the Black Ridge to the C7 and said the Black Ridge was quieter, were you using a Noctua fan on the C7?
 
The C7 is load because the gap betweetn the fan and the side panel is so small. Doesnt really matter which fan u use. Both Black Ridge and NH L9i are therefor much quieter
 
In case people weren't sure of what I was trying to describe with offsetting the 92mm fan to make it work with the Gigabyte x570 itx board, I made a paint drawing.

http://imgur.com/tZCl8hr

Any other suggestions for the heatpipe-side screws? Thought about glue, not sure I want to go that route.

Or any totally different ideas?
 
Just my thoughts: remove that heatsink completly. The cooling with the cpu cooler fan will probably be better anyway.
 
Just my thoughts: remove that heatsink completly. The cooling with the cpu cooler fan will probably be better anyway.

Agreed. You'll be blowing air directly over the M.2 and chipset anyways with this cooler.
 
Hi guys
I want to know if the vrm's heatsinks from the motherboard gigabyte z390 gigabyte z390 i aorus pro wifi are compatible with a 120mm slim fan+ABR and the adata vlp dd4 ram without mods nothing

Thanks!!
 
So I started out with the heat sink. I knew the top of the delidded CPU sits about 2mm lower than the stock heatspreader (actually it's 2.5 mm after measuring it but this measurement wasn't documented anywhere online) and the mounting kit for the Black Ridge (BR) doesn't allow for any adjustment. So I had to shorten the embedded nuts on the mounting kit by those millimeters. I used my bench grinder to do it and then finished everything of with a small keyhole file.

View attachment 167945 View attachment 167946 View attachment 167947 View attachment 167948


Then I had a look at the board. The problem is that the BR doesn't work with the board to begin with (which I knew). It ever so slightly interferes with the VRM heat sink / IO cover and with the BR now sitting over 2 mm lower it's not slight anymore.
So I took of the heat sink. It's easy to by just unscrewing 4 screws from the bottom of the board. Interestingly enough I found that the thermal pad was squishing the CMOS battery cable between the heat sink and the board.

View attachment 167949 View attachment 167950

Now here begins the real fun part. I needed to cut the heat sink down to size so I marked it out with a pencil then took it to the basement to mutilate it. I first tried using a Dremel but the aluminium is quite thick an malleable which makes it hard to cut with a speed tool. So I went old school and used a vise and a small metal hand saw. Then I took it to the bench grinder to clean it up which took some time because it of course got very hot and I had to go slow. I then finished it of with a file and got a decent surface finish.

View attachment 167951

Remounting it was even easier than unmounting it since there now were only two screws left after all.

Now I tried the BR again but to my dismay I found that it now conflicted with the small caps next to the VRMs.

View attachment 167952

So now what do we do? Of course we bring out the Dremel to cut it to size. I cut of everything a close to the screws as possible with a milling bit and then finished it of with some sandpaper.

View attachment 167953 View attachment 167954 View attachment 167955

Trying it out that now it fits very tightly but it fits.

View attachment 167956

Since I didn't know for sure if the case of those caps is actually connected to one of the poles I put a small piece of tape on them just as a precaution.

Now there was still another problem to solve but this too could be taken care of by chopping up some hardware. The fan has almost no clearance to the VLP ram to begin with and now since the whole unit sits over 2mm lower... you know the drill. I marked it out then once again took to the basement.

View attachment 167957

Fortunately plastic cuts pretty easily but it melts and stinks when using a speed tool. I took care to tape away the cable and fix the fan blades in place so that I wouldn't accidentally cut any of that. The corners where the hardest to cut so I took them out in chunks. One of the spokes that hold the motor in place had to go too. This time I used a knife and again a file to clean up the cut. This piece could be slightly cleaner I guess but it's going to be on the underside of the heat sink pressed up against the RAM anyway.

View attachment 167958

Now everything was ready to begin assembly. Using the stock metal clip to mount the fan on the ram said wouldn't be possible for multiple reasons (one of them being that part of the fan wasn't there anymore) so I used some very thin wire to affix it.

View attachment 167959 View attachment 167960

I then found out that you have to mount the board first before the PSU otherwise it isn't possible to get it in. Maybe on a board where the IO shield isn't fixed to the board it is possible in a different order and maybe reading the manual could have told me that sooner but ain't nobody got time for that. After reassembling everything I found out the hard way that it is basically impossible to connect the power button to the pin header with the CPU heat sink and fan in place so of course I disassemble and reassembled the whole thing again. Then I took to cable management which was going nicely thanks to the sleeved custom cables I bought from pslatecustoms.

Now came the magic moment, turning it on for the first time. And... nothing. The fans spun up but I wasn't getting any picture. After some head scratching I looked at the board only to find it showing a CPU error. So after double checking that nothing was shorting anywhere and everything was connected I faced the facts and disassembled everything again. I took off the heat sink but everything looked fine. So I thought maybe the OC frame wasn't holding the CPU down properly or maybe one of the tiny springs in the socket is bent. I removed the frame and CPU and checked the springs but they looked completely fine even after a detailed inspection. I then remounted the stock Intel retention module and put the BR back on as an interim solution taking care not to tighten it down completely as that would be too much with the shortened standoffs. Testing the board outside of the case (which is easy since there is an internal GPU) still showed the same error.

So that's where we are now. Something is broken. Either the board, the RAM or the CPU. The board is showing a CPU error and not RAM so the likelihood of it being that is lower. Now I might have killed the board but taking of and putting back on the VRM heat sink shouldn't have done that. Neither should changing out the retention module. So in all likelihood the CPU is actually broken.

I bought the CPU from Caseking already delidded and including the OC frame. I chose not to delid it myself since I've had issues with broken DIEs in the past and it seems like a risky operation on a soldered CPU to me. Now that might sound funny since I've chopped up half of my hardware here but if you think about it that all was basically just some metal working on components that didn't even contain any electronics (aside from the fan motor). They sent it over in a small carton with foam but they put the heatspreader and the CPU on the same piece of foam with them rattling around in there. They should have put the heatspreader under the bottom piece of foam and used some tape to hold the CPU in place or something I think. My guess is that the edges of the heat sink bumped up against the DIE and chipped the edge. It is incredibly hard to tell but one corner at least seems ever so slightly chipped (bottom left).

View attachment 167961

The packaging seemed sketchy to me to begin with which is why I took a picture of how it arrived originally.

View attachment 167962

I've contacted them to get a replacement immediately and hope they will sort it out without blaming me. I would guess customers breaking delidded CPUs is pretty common so they might not be willing to replace it. However I have some experience with this stuff and took extra care not to damage anything so I am 99.9% sure I didn't. If they end up replacing it it could still be the board. In that case I will just buy another one and swap heat sinks. We will see how this story ends but for today it doesn't have a happy end.

If you try to do this: Don't. If you still want to do it you should either like a challenge, be a masochist or be prepared to spend a couple hundred or even thousands on dead hardware, preferably all three. Now I knew what I was getting myself into and at least 2 of those attributes fit me so I am not going to be discouraged this easily but again do not try this at home.

So unfortunately the whole affair with Caseking took around a month. It was a shit show from beginning to end. First they couldn't manage to reply to my support ticket. Then they told me on the phone they sent out a return label which never arrived. Tried again never arrived. Sent it back by myself (as discussed with them, they promised the money back which of course I never got) marking it appropriately to be connected to the support case. No response. Asked them "it is being worked on". One week later "we don't know where it is" then "ok we found it it will need to be tested". One week later "still testing" (WTF can you test for one week on a broken CPU?). Well anyways I now have a working delidded CPU.

I had used the meantime to order 2 more Noctua A9x14 PWM fans. Placed one below the board and replaced the one in the PSU. The platinum version of these PSUs already have PWM fans so replacing it is easy but requires a little bit of soldering work.

HSmPvLw.jpg

SDwmQX4.jpg

JHs8ZVs.jpg

The Noctuas seem to have a very different fan curve than the stock fan or I fucked something up (judge for yourself by the pictures) because the fan basically never runs. Maybe I will need a new PSU at some point then but so far it does not seem to get very hot.

I also added some thick foam tape to direct the airflow. Basically both the CPU and the GPU can only draw air from the bottom (where the case fans are) or through the side panel.
EXAsHKu.jpg


And here is the final build:
aW2xljx.jpg

uoX3Kko.jpg

1SEcfNz.jpg

K4ij13Q.jpg


I tuned the thing as follows: The ram is only DDR4-2400 so I OC'ed it to 2600, 2666 seems to be unstable. The new Nvidia "super" cards just came out (fortunately the 2080 super looks to be not a big step up) so I had to keep up with that. Let the EVGA VF curve tuning tool do it's thing and ended up with 2050 MHz on the GPU, added 700 MHz for the ram manually (which comes out to 15.4 GHz effectively). Undervolted the CPU with an offset of -85mV in the BIOS after doing some preliminary voltage sweeping in XTU. It is stable in the XTU CPU and memory test and prime stable. Here is what the temps look like in prime small FFTs where it hovers between 3.9 and 4 GHz after the initial 5GHz turbo runs out.
NpN39X1.png
 
So unfortunately the whole affair with Caseking took around a month. It was a shit show from beginning to end. First they couldn't manage to reply to my support ticket. Then they told me on the phone they sent out a return label which never arrived. Tried again never arrived. Sent it back by myself (as discussed with them, they promised the money back which of course I never got) marking it appropriately to be connected to the support case. No response. Asked them "it is being worked on". One week later "we don't know where it is" then "ok we found it it will need to be tested". One week later "still testing" (WTF can you test for one week on a broken CPU?). Well anyways I now have a working delidded CPU.

I had used the meantime to order 2 more Noctua A9x14 PWM fans. Placed one below the board and replaced the one in the PSU. The platinum version of these PSUs already have PWM fans so replacing it is easy but requires a little bit of soldering work.

View attachment 174001
View attachment 174002
View attachment 174003
The Noctuas seem to have a very different fan curve than the stock fan or I fucked something up (judge for yourself by the pictures) because the fan basically never runs. Maybe I will need a new PSU at some point then but so far it does not seem to get very hot.

I also added some thick foam tape to direct the airflow. Basically both the CPU and the GPU can only draw air from the bottom (where the case fans are) or through the side panel.
View attachment 174004

And here is the final build:
View attachment 174005
View attachment 174006
View attachment 174007
View attachment 174008

I tuned the thing as follows: The ram is only DDR4-2400 so I OC'ed it to 2600, 2666 seems to be unstable. The new Nvidia "super" cards just came out (fortunately the 2080 super looks to be not a big step up) so I had to keep up with that. Let the EVGA VF curve tuning tool do it's thing and ended up with 2050 MHz on the GPU, added 700 MHz for the ram manually (which comes out to 15.4 GHz effectively). Undervolted the CPU with an offset of -85mV in the BIOS after doing some preliminary voltage sweeping in XTU. It is stable in the XTU CPU and memory test and prime stable. Here is what the temps look like in prime small FFTs where it hovers between 3.9 and 4 GHz after the initial 5GHz turbo runs out.
View attachment 174009

Obviously a huge pain in the ass but those are incredible temps for that CPU in a Dan A4 SFX.
 
New AM4 bracket available! Claims to fix the issue with at least the Gigabyte and ASRock b450 motherboards.

 
I didn't search, if this has already been addressed, so sorry, if that is unnecessary.
I tried how the cooler fits the ASUS ROG Strix X570-I gaming. Without the cooler for the SSD fits the pure heat sink. But there is very little room for the fan because the fan is mounted too far out of the middle of the socket. If you disassemble the sound card with the front NVMe mount the cooler would have fit with a 120 mm fan (with VLP-RAM, only measured). If the radiator was centered over the socket, it could be a good combination, imho. I'll do something different with the board, so I didn' t try anything else.
IMG_20191012_190348.jpgIMG_20191012_190319.jpg
 
Was there ever a consensus on X99 ITX 2011-3 narrow ILM support? Any unofficial brackets or files out there?

Worst comes to worst anyone have longterm success with zip ties?
 
Plan on getting this for a new processor and putting a new better fan to replace bottom fan and then a slim/skinny fan on top too.

I hope that makes the perfect low profile cpu cooler but I am worried it won't work well.
 
Plan on getting this for a new processor and putting a new better fan to replace bottom fan and then a slim/skinny fan on top too.

I hope that makes the perfect low profile cpu cooler but I am worried it won't work well.
Optimum Tech did a review a while back and tests with a 2 fan setup, cooling is very good.
 
Was there ever a consensus on X99 ITX 2011-3 narrow ILM support? Any unofficial brackets or files out there?

Worst comes to worst anyone have longterm success with zip ties?
Dan dropped support for 2011-3 narrow ILM pretty quickly. The problem was how close the RAM slots were in relation to the CPU socket. It was just easier to scrap it. It was also reaching it's EOL so there was no point in working it. I haven't tested it and I sold my x99e itx when he dropped the support so I haven't really checked to see if someone else has made custom brackets.
 
Back
Top