Eisbear 280 Extreme, getting it running in a case that is too small.

Thomas_S

... the nice marketing guy next door
Staff member
So I did get a chat request on the Alphacool system, I think from someone claiming to be a young woman selling nude pics of herself. I assume that related to the 2 in 9 out? Why did I not dare to check the details of that communication?
Do you have the name of the user? They are actually completely removed from the system!
 

Anneve

New member
There are two such messages so far.
The first as a conversation on 12.9.2021, Adronis452.
The second as a conversation on 10.10.2021, adozel828, which is no longer listed.
I assume it is highly unlikely that either of these are the sort who browse water cooling forums? They are just victims, too, lured into this???
 

Anneve

New member
Late February 2022, cooling characteristics are still around the same, a bit less as I no longer use push fans on the Extreme radiator as they block access to the top of the mobo. Delta T is 51C at 100%, 45.5C at typical high load. That is with a Ryzen 9 3900X 12-Core in an X470 Crosshair Wifi.

Trying it with my old 1700 8 core CPU, how did this compare? For the high load, Handbrake dvd conversion test, the typical CPU temperature was 37C, an amazing delta T of 13C, but taking a very noticeable three times longer, 220 minutes rather than the 63 minutes with the 3900X. The delta T with the 100% soak test is only 11C, I find it difficult to understand how it can be lower.

I interpret that as it is looking as if things are moving towards the latest (lower power and cheaper) 65W 8 core becoming almost as fast as my 105W 12 core. Those ultra low delta T could eventually become practical reality.

*

I waited until the new year for the latest upkeep work and tests, so that it would be around the time of an annual check and clean-up.

The biggest change, apart from removing the top push fans, was cleaning and redoing the Conductonaut thermal compound.

When I used thermal paste I cleaned the surfaces with turps or white spirit until they were visibly clean, then added fresh compound, with turps to soften it and help spread it around. With Conductonaut I get the same performance as with thermal paste plus turps, suggesting that there might be some flaw in my watercooling to CPU interface. I might look into that a year from now, things are okay enough as is.

With liquid metal conductonaut I am discovering that I need to clean it with Isopropyl alcohol and that visually clean is not good enough. To get the thermal compound to spread easily I needed the surfaces to be visibly clean, but then clean it further with a rubber ‘ink grade’ eraser, with a further rinse with isopropyl. The test as if the compound spreads easily. I had first found that it could take a few minutes to get the metal compound to start to spread, metal compound seemed difficult to apply, like this it went on easier, but there is still no improvement in thermal performance.

As a system tidy up I changed my GPU from a heavy Strix GTX 960 to a choice of two low profile, low weight, low power GPU, a Radeon Pro WX 3100 and a Nvidia T400. In a linux system I prefer the T400, simpler to adjust, but if I accept a GPU temp in the mid 40’s then no problem with either.

I also changed the motherboard to an X470 Crosshair Wifi, from a Asus TUF Gaming X570-Plus Wifi. Better BIOS and heavier heatsinks, at the cost of the GPU operating as x8 rather than x16 and only 6 SATA sockets.

I have still been occasionally looking for a pump/reservoir that is a plug in replacement for my Extreme, though my original reservoir is running okay, just not pretty. With forthcoming price rises in England, I do not expect to be spending on cooling and computing for a while. If an Extreme reservoir became available just now I would hold back.
 

Anneve

New member
At September 2022 the delta T for 100% Load and Normal High Load are now:
33.3C and 30.5C delta T. These used to be approx 50C and 45C.

This is with the same Ryzen 9 3900X 12-Core + X470 Crosshair Wifi + Extreme 280 + extra 280.

The time taken for the Normal High Load test is no change, approx 41min. It used to be 63min but I am now using a different test.

I am now also using the utility: Linux / System Profiler / Benchmarks / GPU Drawing, as a fast indicator of system effectiveness.

The big improvements come from using undervoltage. The method used was first an internet search on undervoltage, particularly for Linux and Ryzen 9 3900X. Then a check in my Bios to see how the settings compare to the articles. It then needed a second read through of the articles. Finding the appropriate variables was not so easy.

I also changed the Bios fan settings to get a quieter run, at high load the fans are now only running at around 50%, 100% only at 70C, a higher temperature than is expected to occur. There is a big improvement with quietness for low and high load. Note that means the cooler effectiveness has been reduced in order to reduce noise, but there is still a massive cooling improvement due to undervoltage.

I had considered that any CPU upgrade would need to be to a modern 65W CPU, but undervoltage allows more than enough improvement with my current 105W.

Update: This is what I hoped CPU cooling would be like.

Enclosed a pic of the undervoltage test runs.
 

Attachments

  • alphacool forum updates, 2 rad - 7.9.2022.png
    alphacool forum updates, 2 rad - 7.9.2022.png
    345.4 KB · Views: 0
Last edited:

nostriluu

New member
Hello all from Montreal, Canada. Where it is very expensive to order parts because of shipping, and two dealers that I contacted checked out on me.

But I ended up with a Alphacool 11389 Eisbaer 140 kit, and the 280 Extreme Solo (no head unit). I think the 280 is an older version, it has blue at the "cold" output.
I will be using this with a 7950X3D, for machine learning, which involves long running full-cores execution. I may also try to loop a 3090 in. I think the open case will benefit ventliation?

I actually mismeasured for the 280, it doesn't fit my XTIA ATX open case, but I realized after searching that the parts it contains make it much better value than ordering them independently.

This is a long thread, but I gather that I cannot re-use the reservoir, and that I must get a separate D5 reservoir? How would I detach the current one from the radiator?

I would use the head unit from the 140 kit, the LT Solo. I guess I would need two GPX tube kits. Would it be worth it looping in that kit's 140 as well?

Thank you very much for consideration to get these fine parts working together.
 
Top