Cooling Eisbaer LT orientation on ryzen

paulesko

Master of Cramming
Original poster
Jul 31, 2019
415
322
Hello everyone.

I have been testing this week my new eisbaer and the ncase, and I was wondering if there is a review somewhere, where I can see the fin direction on the cold plate of the block. I'm searching this so I can mount it in the right direction according to the dies on the ryzen CPU.

Also, anybody knows how to remove the bracket that holds the block-pump combo to the motherboard? I have to make a really big amount of force and haven't succeeded in putting it apart.

Thank you!
 
  • Like
Reactions: Kyuss

Curiosity

Too busy figuring out if I can to think if I shoul
Platinum Supporter
Bronze Supporter
M...M...M...M...Multi-Tier...Subscriber...
Apr 30, 2016
708
827
The best image I've found so far, in this manual:PDF link.

The best way to find out would be to just remove the cold plate and look. (And take pics!)
 

paulesko

Master of Cramming
Original poster
Jul 31, 2019
415
322
The link doens´t work, but I guess you mean this pdf


theres an exploded view but it doesn´t show much. I guess I´ll have to remove the cold plate. I´ll post pics, because also, I think that the pump it has it´s the dc-lt 2600 that goes to 2600 rpm, and it´s identical to the dc-lt 3600 that goes up to 3600 rpm. So maybe it´s possible to interchange these two pumps and make a custom loop with this tiny unit, like Optimum tech has done, but with a little more water power.
 
  • Like
Reactions: Curiosity

Curiosity

Too busy figuring out if I can to think if I shoul
Platinum Supporter
Bronze Supporter
M...M...M...M...Multi-Tier...Subscriber...
Apr 30, 2016
708
827
Sorry about the broken link, but yes, that's the one.
I found this a few minutes ago, not sure if it's any help for you. I assume they use the same cold plate.

 
  • Like
Reactions: paulesko

paulesko

Master of Cramming
Original poster
Jul 31, 2019
415
322
I have looked inside the eisbaer this weekend.

The fins have the same direction than the fittings... which seems very logical once you see it. So the water comes into the block, and it´s pumped into the cold plate which has fins directed the same way as the water comes in and out.... I don´t know if I´m explaning myself...


maybe this pics help, the flow goes from the middle of the cold plate to the sides, and outside the block to the radiator.
Another thing I noticed was that my coldplate was not falt by any means, and it was noticeable, so I lapped it with water sand paper for like 1/2 hour (a pain in the ass, yes). You can see in the photo that I took in the middle of the process that I was doing more sanding in the middle of the coldplate than on the sides because of the curve it had.

Also lapped the 3700x, but it was almost perfect, I did this because I remember how I had to lap almost every athon 64 bits (old stuff) back in the day, but apparently AMD has learnt to do it right, at least in my sample.

After all the work, I also rotated the pump 90 degress th flow I previosly had was the orange arrow, and now I have the blue arrow type

After all this work, the improvement is ashtonishing, I wasn´t expecting this.
Before doing all this, I hitted 96 degrees in less than 5 minutes on intel burn (maximum heat aoutput config) and only did it once... Now the maximum dances between 83 and 82 and the fans instead of 2000 rpm stay at 1800 rpm a little more when it goes up to 83 and then less when it goes down to 82.

All this was at 1.26 - 1-25 volts and 4.150 mhz.

So now, I´ve ramped up the vcore a bt to see how it behaves, now I can do intel burn for 1h 30m at 4.300 mhz and 1.37v and temps stay at 93 maximum. In more realistic scenario, it does 10 lopps of cinebench with a maximum temperature of 81 and fans are at 1700 rpm, so it has some headroom still.

Very pleased with the results, and very surprised, I wasn´t expecting this... to be honest I was a bit disspointed with the eisbaer, but I thought it was how it worked and that was it, I thought that the chips were hot and difficult to cool because the die is so small and not in the center and so on. But you see that my coldplate was very curved and lapping it really helped.

Before lapping I did the turn 90 degrees and the temps improved by 3-4 degrees, not that much, it was the lapping what helped a lot. Also, I thought having the fins in my previous direction was the best idea, but it turned out the opposite, I think that this could be different with cpus that have two dies like 3900x and 3950x.

And to finish up, I pluged a black ice nemesis gtx to see if I can improve it further but no, temperatures dropped by 1 degree if any... so no improvement here. even with this thick rad I saw no difference between the noctua 25 and 15 mm thick fan, and also with the 25 mm fan in pull the temps were the same within 1 o 2 degrees which is very little difference. My guess is that the eisbaer handling a 3700x at 1.37 is at its limit, but its not the limit of the radiator by any means, so improving on that is going to yield minor improvement.

And this is all.

Next weekend I will try to change the OEM pump which is 2600 rpm with alphacool version of the same pump that can work at 3600 rpm to see if theres any difference. I like the 3600 pump better because it can undervolt to 5v safely while the 2600 model can only go down to 5v, so theres nothing too loose if I decide to undervolt it for idle loads, because it´s not going to be noisy.

If there´s any interest I can take some more pics or gather data in a more scientific way... anyway, this time the improvement has been so big that theres no doubt that something was wrong before.
 
Last edited:

paulesko

Master of Cramming
Original poster
Jul 31, 2019
415
322
I have recieved the new pump today, and have installed it... temps have gone down a bit more. From 93 degrees at intel burn to just 88 but it´s bit early to say because I´ve had it there only for 15 minutes or so.

The thing I see is that the pump is terribly noisy even at the same rpm than the previous one, this thing whines like crazy. I´m thinking that maybe I have put too much pressure on the mounting screws of the pump and vibrations translate into the chasis better, I don´t know.

The way you mount the pump is: You put a ring aorund it and push that metal ring with the screws against the block... the ring itself is curved so that it wasn´t necesary to put too much pressure, bt I didn´t pay attention to this when disassembled it, so.. maybe I put too much pressure. I will try again tomorrow.

Here you can see some photos, and a video so you can hear the noise. At the begining of the video, everything is at full load (3.800 for the pump and 2000 for the noctuas) an then at the end of the video the pump goes down to 2.000 rpm and the noctuas to 800.


So, I wrote this message yesterday and didn´t send it by mistake. I took the pc home at night and looked into the pump noise a bit. You cannot put less pressure on the pump because it will leak, so it´s properly tightened... and yet it makes that noise... I will try to get all the water outside of the pump to see if that´s the problem, and put the old pump again to see if it´s just the way this more powerful pump sounds.

Also, as you can see in the pics, you don´t acces the pump via coldplate but the other way round, which is easier, but impossible not to break the filmsy aluminium cover on the block... kudos to the double sided tape manufacturer that alphacool uses there, my god, is strong AF.

More info soon.
 

paulesko

Master of Cramming
Original poster
Jul 31, 2019
415
322
Just write to say that I´ve found no way to reduce the noise coming from this pump, so I´ve read everything I´ve found on the internet and discovered a thread where a guy had also a problem with this 3600 unit and the alphacool 5.25 bay dual pump + reservoir. He seemed to have the same problem as I do, and alphaccol told him to buy this O-ring (which I don´t have): This ring helps the impeller not to hit it´s housing in the block, I will try it and will post results here to see if it makes any difference or not.


I´ve tried to put both pumps at the same rpm ant hold it in my hand without mounting them on the block and I can´t tell one from the other, the problem is when they are mounted, so my guess is that the impeller is hitting or grubbing slightly somewhere. I hope the problem will dissapear with this new ring. I´ll keep you posted!
 
  • Like
Reactions: Sashby

stepnyVLK

SFF Lingo Aficionado
Sep 23, 2019
91
29
Just write to say that I´ve found no way to reduce the noise coming from this pump, so I´ve read everything I´ve found on the internet and discovered a thread where a guy had also a problem with this 3600 unit and the alphacool 5.25 bay dual pump + reservoir. He seemed to have the same problem as I do, and alphaccol told him to buy this O-ring (which I don´t have): This ring helps the impeller not to hit it´s housing in the block, I will try it and will post results here to see if it makes any difference or not.


I´ve tried to put both pumps at the same rpm ant hold it in my hand without mounting them on the block and I can´t tell one from the other, the problem is when they are mounted, so my guess is that the impeller is hitting or grubbing slightly somewhere. I hope the problem will dissapear with this new ring. I´ll keep you posted!
Did you tryed to swap the impellers? After all the should be identical you should be able to just pull it out like a ddc
Like here
 

paulesko

Master of Cramming
Original poster
Jul 31, 2019
415
322
I will try, but it looks like it is not possible to remove it from the dc-lt 2600... I didin´t put too much effort into it tbh, but will try this weekend.
 

stepnyVLK

SFF Lingo Aficionado
Sep 23, 2019
91
29
I will try, but it looks like it is not possible to remove it from the dc-lt 2600... I didin´t put too much effort into it tbh, but will try this weekend.
if you mess it up ill send you a new one :) i have like 3 of them now :))) im currently trying to squieeze the ddc into the ghost under the psu but it seems impossible with the standard alphacool top V4 it just has the in/out ports on the wrong side... will have to wrap my head around that
 

paulesko

Master of Cramming
Original poster
Jul 31, 2019
415
322
Hahahha ok, I take your word.

Have you ever tried this EK pump? It loks to have the same size as a ddc but without the heatsink as it consumes less energy (thus a bit smaller than a ddc). I found it looking for very quiet pumps, it seems to be one of the quietest.


Here a member of this forum has done a cpu block+pump with this one, I think you´ll like to see this thread

 

stepnyVLK

SFF Lingo Aficionado
Sep 23, 2019
91
29
Hahahha ok, I take your word.

Have you ever tried this EK pump? It loks to have the same size as a ddc but without the heatsink as it consumes less energy (thus a bit smaller than a ddc). I found it looking for very quiet pumps, it seems to be one of the quietest.


Here a member of this forum has done a cpu block+pump with this one, I think you´ll like to see this thread

no i just re-rediscovered it today while researching the build for KronDestroyer13
its in this build but the person states its only one top available for it and a top port is a nogo in the ghost :)
plus if its as big as an ddc and has no obviouse bennefits than just go for a ddc there is no reason not to you can regulate the 3.25 down
Im actually loking for the stock top
that would fit in the ghost but they only sell it with the standard ddc and i dont want to use the standard because of the 12v rail i dont want to use just for that :)
the DDC 3.25 ended up being the quietest of the real pumps if you believe the testing (witch makes snese to me)
 

stepnyVLK

SFF Lingo Aficionado
Sep 23, 2019
91
29
Here a member of this forum has done a cpu block+pump with this one, I think you´ll like to see this thread


yes its amazing what you can doo with computers these days :) but i honestly dont know if the pump is any good there is only one review where it destroyes everything and the ddc is the loudest and i know from experience that the ddc can be quiet so im not shure about that and that test is quoted on EK website so dunno
But i dont need the quietest pump on the planet or the most powerfull the dc-lt is enough for me if it doesnt break down every 2 years
I think i will go with the 2 dc-lt 2600 combo that should be ok or try the 3600 if its not enough but i dont see the point in that since it only runns higher rpm and is louder ( i dont want it to be any louder)
I want to use as many off the shelve parts as possible and i think you cant get anny bether packaging and thats the DC-LT that also fits in the ghost
You should be able to put 2x240 rads and 2 blocks in there and be reasonably quiett.
I want to have the performance i had 6 years ago whisper quiet no matter what you did with the pc just have one settings for cooling all fans went 800-900 rpm and the temperature did what ever it wanted between 30-60 degrees :)
since we are in a 13 liter package compared to corsair 700D and its 84 liters it could be 35-70 deg maximum ;)
 

paulesko

Master of Cramming
Original poster
Jul 31, 2019
415
322
Well I finally did some more testing. I´ve been very busy that´s why it´s been so long, sorry for that.

First these new alphacool fittings. the HF model are vey nice, they look good. Just beware that there are nylon and metal models. make sore you buy the metal ones because the nylon are a bit... to plastiky for my liking, it´s not like they are going to fali, but I don´t know, I don´t feel comfortable tighting them because I was afraid to brake them.. .and it´s just not as nice. that´s all.

Some pics will help.



Here you can see the difference in height between the 90 degree HF fitting (this is the metal one) and a ekwb torque. The alphacool ones are very nice for us who don´t have much room. The EKWB 90degrees adaptor has the nylon Alphacool HF fitting attached to it.

One thing that doesn´t give me much confidence about this alphacool 90 degrees is how it moves around... both of my samples moved the same way, I don´t know wether this is normal or not, but it doesn´t leak so... it works I guess

Then some pics of the fittings themeshelves






Well, and now the dc-lt 2600 and 3600. I did some testing... the improved flow over the 2600 doesn´t give me much improvement, maybe a degree or two, I tried several times and one of them I had a bit worse temperatures than with the 2600 si it depends greatly on how you place everything, but it´s going to be a matter of one degree or two. Maybe if you add a graphics card to the mix this can change, I don´t know.
One thing I can tell you is that the 3600 model is way louder than the 2600. I really don´t understand why but you put them to the same rpm and the 3600 model is just louder, and way louder, it´s not a subtle thing. For me this is a deal breaker, I wouldn´t be using this pump even if it made a difference in temperatures ust because of the worse acoustics.

Lets see how they are different




the dc-lt 2600 is on the left, and you can see that the design of the impeller is not completely different, but they are not interchangeable. Notice the o-ring on the dc-lt 3600 that I bought trying to silence it.. didn´t work out completely but it mutes a bit the vibrations on the 3600. I used it on the 2600 just to test it, and I think the noise has improved maaaaybe a liiiitle bit, but I don´t know if this is myself trying to justificate why I´ve bought all this stuff hahaha



Again 2600 on the left.


And a closer look at the fancy o-ring.

Also, I dissassemble the coldplate again to lap it, and found out that the coldplate bends when you mount it to the rest of the block (captain obvious here lol) so you have to lap it screwed to the rest of the block, wich is what I did at first, but I donñt know why... I thought... "lets try with the coldplate alone it would be less combersome"
the result


And after attaching more work I got something like this


just to mount it -> get worse tmps -> realize I have to lap with everything together -> begin again... It was a happy day.

And to end up. I mounted the Hardware Labs gtx rad into my ncase m1. In order to do this, I have modded (just a prototype, I will draw a case to print) my psu so it doesn´t have a fan, because it´s getting air from the rad anyway... and I´ve measured the temps in there via laser and it doesn´t go much above 45 so it´s fine for now.

The radiator provided a bit better temps, but not much. The big improvement has been that I can have those good temps with the fans at 1200 rpm and not at 1800 like with the original radiator, so I´m very happy noise wise.


The fitting is clooooose to that heatsink... lets hope they don´t leak

 

stepnyVLK

SFF Lingo Aficionado
Sep 23, 2019
91
29
Thats an awfull lot of R&D ;) Thats the beauty of the black ice its cool at low rpm with the noctuas you get by with 12-1300rpm
 
Last edited:
  • Like
Reactions: Curiosity

ExplodingWaffle101

SFF Lingo Aficionado
May 11, 2018
129
82
that radiator picture... that kinda cramming is what sff is about.

knew this kinda stuff would happen the moment i saw those new itty bitty fittings.
 

paulesko

Master of Cramming
Original poster
Jul 31, 2019
415
322
that radiator picture... that kinda cramming is what sff is about.

knew this kinda stuff would happen the moment i saw those new itty bitty fittings.

Because of this forum, I´m always looking somewhere tu put things inside a case. Now everything just seems too big jajjaa

Thats an awfull lot of R&D ;) Thats the beauty of the black ice its cool at low rpm with the noctuas you get by with 12-1300rpm

Yep, I´m very happy with the rad, even at 800 rpm it´s able to maintain the cpu under control. that´s really awesome
 
  • Like
Reactions: stepnyVLK

stepnyVLK

SFF Lingo Aficionado
Sep 23, 2019
91
29
Yep, I´m very happy with the rad, even at 800 rpm it´s able to maintain the cpu under control. that´s really awesome
Yes i just got the second one will test the vga setup next weel :)
Im curiouse how the dc lt will perform :)
 

SSJ4

SFF Lingo Aficionado
Feb 10, 2020
131
210
www.ospreygaming.com.au
Well I finally did some more testing. I´ve been very busy that´s why it´s been so long, sorry for that.

First these new alphacool fittings. the HF model are vey nice, they look good. Just beware that there are nylon and metal models. make sore you buy the metal ones because the nylon are a bit... to plastiky for my liking, it´s not like they are going to fali, but I don´t know, I don´t feel comfortable tighting them because I was afraid to brake them.. .and it´s just not as nice. that´s all.

Some pics will help.



Here you can see the difference in height between the 90 degree HF fitting (this is the metal one) and a ekwb torque. The alphacool ones are very nice for us who don´t have much room. The EKWB 90degrees adaptor has the nylon Alphacool HF fitting attached to it.

One thing that doesn´t give me much confidence about this alphacool 90 degrees is how it moves around... both of my samples moved the same way, I don´t know wether this is normal or not, but it doesn´t leak so... it works I guess

Then some pics of the fittings themeshelves






Well, and now the dc-lt 2600 and 3600. I did some testing... the improved flow over the 2600 doesn´t give me much improvement, maybe a degree or two, I tried several times and one of them I had a bit worse temperatures than with the 2600 si it depends greatly on how you place everything, but it´s going to be a matter of one degree or two. Maybe if you add a graphics card to the mix this can change, I don´t know.
One thing I can tell you is that the 3600 model is way louder than the 2600. I really don´t understand why but you put them to the same rpm and the 3600 model is just louder, and way louder, it´s not a subtle thing. For me this is a deal breaker, I wouldn´t be using this pump even if it made a difference in temperatures ust because of the worse acoustics.

Lets see how they are different




the dc-lt 2600 is on the left, and you can see that the design of the impeller is not completely different, but they are not interchangeable. Notice the o-ring on the dc-lt 3600 that I bought trying to silence it.. didn´t work out completely but it mutes a bit the vibrations on the 3600. I used it on the 2600 just to test it, and I think the noise has improved maaaaybe a liiiitle bit, but I don´t know if this is myself trying to justificate why I´ve bought all this stuff hahaha



Again 2600 on the left.


And a closer look at the fancy o-ring.

Also, I dissassemble the coldplate again to lap it, and found out that the coldplate bends when you mount it to the rest of the block (captain obvious here lol) so you have to lap it screwed to the rest of the block, wich is what I did at first, but I donñt know why... I thought... "lets try with the coldplate alone it would be less combersome"
the result


And after attaching more work I got something like this


just to mount it -> get worse tmps -> realize I have to lap with everything together -> begin again... It was a happy day.

And to end up. I mounted the Hardware Labs gtx rad into my ncase m1. In order to do this, I have modded (just a prototype, I will draw a case to print) my psu so it doesn´t have a fan, because it´s getting air from the rad anyway... and I´ve measured the temps in there via laser and it doesn´t go much above 45 so it´s fine for now.

The radiator provided a bit better temps, but not much. The big improvement has been that I can have those good temps with the fans at 1200 rpm and not at 1800 like with the original radiator, so I´m very happy noise wise.


The fitting is clooooose to that heatsink... lets hope they don´t leak

Hey mate did you see a big jump in performance after lapping the cold plate? My cold plate is absolutely terrible. Full of grooves and isnt flat lol.

Thinking of doing this myself.
 

paulesko

Master of Cramming
Original poster
Jul 31, 2019
415
322
yes, I did see an improvement, but also lapped the ihs of the processor so everything fit together properly