Fractal Design Celsius+ S28 Prisma AIO

[crazyeyesreaper]

Fractal Design has revamped its venerable line of Celsius liquid coolers with the all-new Celsius+ S28 Prisma. This latest release delivers the topnotch performance you would expect along with ARGB lighting and a solid warranty, which gives it a leg up in the crowded liquid cooler market.

Show full review

 

[FreedomEclipse]

That price compared to the Deepcool Assassin III though...

Fun fact - The Noctua NH-D15 is currently selling cheaper than the deepcool on amazonUK.... I really dont know why but maybe Covid-19 bumped up prices. Deepcool has to export from china while Noctua is in Europe.

 

[Caring1]

@crazyeyesreaper
In the mounting position it was tested in, you had no problems, but if the Radiator was inverted so the hoses were at the bottom, would the hose length have been an issue?
Would the Graphics card be in the way?
Also, for aesthetic reasons, if the hose outlets on the pump were facing the VRM instead of the Ram, does the Fractal facia rotate to keep the name horizontal and the correct way up?

 

[crazyeyesreaper]

@crazyeyesreaper
In the mounting position it was tested in, you had no problems, but if the Radiator was inverted so the hoses were at the bottom, would the hose length have been an issue?
Would the Graphics card be in the way?
Also, for aesthetic reasons, if the hose outlets on the pump were facing the VRM instead of the Ram, does the Fractal facia rotate to keep the name horizontal and the correct way up?

No the hoses arent long enough to orient the radiator at the bottom same can be said of pretty much every AIO on the market as it would put too much stress on the tubing near the fittings on the rad and vice versa, with a vertical GPU mount it would clear the GPU its just not long enough to work here. Your mileage may vary depending on the chassis. Example putting the AIO in the front of a case it would likely be okay less stress and direct path to the pump limits potential problems but its all comes down to case / components /etc. That said Fractals AIO is 400 mm tubing length.

The facia does not rotate to my knowledge as it has a built in function (you give it a slight twist to select between Auto or PWM functionality since the logo is backlit if it rotated the LEDs would be in the wrong spot so to speak and if mounted to the logo rotating would likely destroy the wiring so yeah what you see is what you get.

 

[Colonel Sanders]

For a long time now I have questioned the validity of the earlier temps recorded on the Fractal Design S24 and this review losing out to it by a significant amount, just makes me question those earlier temps even more.

 

[crazyeyesreaper]

For a long time now I have questioned the validity of the earlier temps recorded on the Fractal Design S24 and this review losing out to it by a significant amount, just makes me question those earlier temps even more.

The older design had a better build used proper fittings and focused more on performance than looks. If you look at the charts coolers with RGB typically perform worse across the board. I have gone back and retested many coolers they all perform as they did previously. The older Celsius unit is in a friend's system but the other aios are the same including the Reveen which also performs well. Aseteks changes over the years have also affected performance. Different pump designs focused on noise different impellers. Othr fans etc. All comes into play. That's why the Celsius line lasted for so long. It was good all around. As such they didn't need to revamp it every six months.

 

[AsetekDennis]

The older design had a better build used proper fittings and focused more on performance than looks. If you look at the charts coolers with RGB typically perform worse across the board. I have gone back and retested many coolers they all perform as they did previously. The older Celsius unit is in a friend's system but the other aios are the same including the Reveen which also performs well. Aseteks changes over the years have also affected performance. Different pump designs focused on noise different impellers. Othr fans etc. All comes into play. That's why the Celsius line lasted for so long. It was good all around. As such they didn't need to revamp it every six months.

Just want to chime in here with a little information

We (Asetek) have made changes over the years. None of which should have affected performance negatively (moreso the other way around). That being said, over the past few generations a lot of emphasis has been put on acoustic performance which, just like thermal performance, is where we always strive to lead the market and of course continiously improve upon ourselves.
Thermal performance is very difficult to measure and many things can influence how a cooler performs. Whether or not DeltaT is measure matters a fair bit, as well as case consistency, airflow consistency, fixed voltages and so much more.
What should also be taken into account is the natural variance from run to run, which isn't typically accounted for. Also margin of error

 

[crazyeyesreaper]

Just want to chime in here with a little information

We (Asetek) have made changes over the years. None of which should have affected performance negatively (moreso the other way around). That being said, over the past few generations a lot of emphasis has been put on acoustic performance which, just like thermal performance, is where we always strive to lead the market and of course continiously improve upon ourselves.
Thermal performance is very difficult to measure and many things can influence how a cooler performs. Whether or not DeltaT is measure matters a fair bit, as well as case consistency, airflow consistency, fixed voltages and so much more.
What should also be taken into account is the natural variance from run to run, which isn't typically accounted for. Also margin of error

All testing is done in a climate controlled within reason space (temperature and humidity) DeltaT is not used these are the temps reported by the CPU in an again controlled environment with all tests run 3 times each at 15 mins with the peak temps taken from those runs. Testing follows an Idle / Wprime / Aida64 (CPU, FPU, Cache, System Memory) / Aida64 FPU in succession thus making sure to reach thermal saturation. this is repeated 3 times in order. Typically variance between runs is usually 1'C on the hottest core which determines CPU throttling if that point is reached. So margin of error from around 40-50 coolers is +/- 1C.

Just figured I would toss that out there.

As for the comment which was answering another user asking why an older Asetek based unit is beating a new one, I noted in that comment that pump design, fan design, etc all come into play which you then just reiterated. Fact remains older Celsius units were in my opinion as a reviewer some of the best Asetek based AIOs on the market. Now since everyone has Asetek based units and just slaps less than optimized RGB fans on them the performance doesn't look as good. I'll give Fractal Design credit here they didn't castrate performance for looks in fact performance is good and they kept some nice features. Looking at many other RGB AIOs on the market I can't say they offer impressive performance regardless of OEM. Many offer middling performance at high noise levels. Obviously some manufacturers step up there game and delivery good units (Celsius etc) others not so much. Proof is in the results.

 

[AsetekDennis]

All testing is done in a climate controlled within reason space (temperature and humidity) DeltaT is not used these are the temps reported by the CPU in an again controlled environment with all tests run 3 times each at 15 mins with the peak temps taken from those runs. Testing follows an Idle / Wprime / Aida64 (CPU, FPU, Cache, System Memory) / Aida64 FPU in succession thus making sure to reach thermal saturation. this is repeated 3 times in order. Typically variance between runs is usually 1'C on the hottest core which determines CPU throttling if that point is reached. So margin of error from around 40-50 coolers is +/- 1C.

Just figured I would toss that out there.

As for the comment which was answering another user asking why an older Asetek based unit is beating a new one, I noted in that comment that pump design, fan design, etc all come into play which you then just reiterated. Fact remains older Celsius units were in my opinion as a reviewer some of the best Asetek based AIOs on the market. Now since everyone has Asetek based units and just slaps less than optimized RGB fans on them the performance doesn't look as good. I'll give Fractal Design credit here they didn't castrate performance for looks in fact performance is good and they kept some nice features. Looking at many other RGB AIOs on the market I can't say they offer impressive performance regardless of OEM. Many offer middling performance at high noise levels. Obviously some manufacturers step up there game and delivery good units (Celsius etc) others not so much. Proof is in the results.

Just to clarify. I did not mean to suggest our testing was inaccurate. Rather I tried to say that any testing done is hard and requires a lot of effort. Which you obviously have given a lot of in your CPU cooler reviews.

As for Celsius vs. Celsius+ I can say for sure that the older pump generation did not perform better specifically, but it is entirely plausible that the older fans performed better with the radiator. That is something which would actually be interesting to cross-test (Celsius+ fans on Celsius AIO and vice-versa)

 

[crazyeyesreaper]

Just to clarify. I did not mean to suggest our testing was inaccurate. Rather I tried to say that any testing done is hard and requires a lot of effort. Which you obviously have given a lot of in your CPU cooler reviews.

As for Celsius vs. Celsius+ I can say for sure that the older pump generation did not perform better specifically, but it is entirely plausible that the older fans performed better with the radiator. That is something which would actually be interesting to cross-test (Celsius+ fans on Celsius AIO and vice-versa)

Depending on the AIO and the fans the inclusion of ARGB lighting can cut overall blade surface area by up to 20%. As you might imagine that causes a serious hit to performance vs high quality standard fans as it creates larger dead spots. Then add in larger gaps around the outside of the fan frames and you get blow back through those gaps between the fans resulting in lower performance. Stuff like this gets overlooked but its as easy to test by just taking some tape and masking that spots off at the radiator shroud level. This usually results in lower perceived noise (dBA reads the same but tonal change is typically easier on the ears).

This is very simple stuff that when paired with optimizations of the radiators FPI to better relate to the selected fans you can usually shave off 2C and drop noise levels by around 1-2 dBA. I did some minor work testing various designs for a couple manufacturers previously and it was interesting to see how going from say 21 FPI to 18 FPI with the same fans could show the above changes in performance. Many will just slap some fans on an AIO and call it a day. But optimize fans and FPI and you can get a much better product that is not only quieter but performs better. Wish more design teams experimented in that regard.

 

[AsetekDennis]

Depending on the AIO and the fans the inclusion of ARGB lighting can cut overall blade surface area by up to 20%. As you might imagine that causes a serious hit to performance vs high quality standard fans as it creates larger dead spots. Then add in larger gaps around the outside of the fan frames and you get blow back through those gaps between the fans resulting in lower performance. Stuff like this gets overlooked but its as easy to test by just taking some tape and masking that spots off at the radiator shroud level. This usually results in lower perceived noise (dBA reads the same but tonal change is typically easier on the ears).

This is very simple stuff that when paired with optimizations of the radiators FPI to better relate to the selected fans you can usually shave off 2C and drop noise levels by around 1-2 dBA. I did some minor work testing various designs for a couple manufacturers previously and it was interesting to see how going from say 21 FPI to 18 FPI with the same fans could show the above changes in performance. Many will just slap some fans on an AIO and call it a day. But optimize fans and FPI and you can get a much better product that is not only quieter but performs better. Wish more design teams experimented in that regard.

I absolutely agree that fan optimization vs. the FPI and additional rad characteristics is something that is not being considered enough. Ironically this has lead to the perception that air = water, when it is absolutely not the case (if people need a credible, non-vested, source have a look at Gamers Nexus' comparison). However, I personally believe that what air coolers are doing (in most cases) excellently at this point in time is matching their heatsinks with the fans used. This leads to optimal performance for their product. As you mention, it would be cool to see more emphasis on this part of product development on AIO's as well, to truly show the power of liquid cooling AIOs.

 

[crazyeyesreaper]

I absolutely agree that fan optimization vs. the FPI and additional rad characteristics is something that is not being considered enough. Ironically this has lead to the perception that air = water, when it is absolutely not the case (if people need a credible, non-vested, source have a look at Gamers Nexus' comparison). However, I personally believe that what air coolers are doing (in most cases) excellently at this point in time is matching their heatsinks with the fans used. This leads to optimal performance for their product. As you mention, it would be cool to see more emphasis on this part of product development on AIO's as well, to truly show the power of liquid cooling AIOs.

Pretty much nail on the head. Its why a cooler like the Scythe Fuma 2 for its price makes $150 AIOs seem quite silly, 40 dBA vs 48 dBA with similar performance. Its honestly a bit sad to see. Yet I see it day in day out. Then add in the biggest drawback of an AIO water cooler that being pump failure or leaks, Fractals previous Celsius units with the G1/4 fittings was a breath of fresh air, good performance acceptable noise output and it can be refilled and properly maintained. Back then it was a huge leap forward especially for the price, Now AIOs with RGB lighting are pushing prices into the same realm as all copper RGB equipped pre-filled custom water solutions. While these may be more nerve wracking for some, It shows a huge weakness. Makes it hard to recommend a $150-200 AIO when a $150 gets you custom water. More focus on improved designs and optimization is essentially necessary at this point. Hopefully i will see these changes on my test bench sooner rather than later but I still have my doubts.

 

[AsetekDennis]

Pretty much nail on the head. Its why a cooler like the Scythe Fuma 2 for its price makes $150 AIOs seem quite silly, 40 dBA vs 48 dBA with similar performance. Its honestly a bit sad to see. Yet I see it day in day out. Then add in the biggest drawback of an AIO water cooler that being pump failure or leaks, Fractals previous Celsius units with the G1/4 fittings was a breath of fresh air, good performance acceptable noise output and it can be refilled and properly maintained. Back then it was a huge leap forward especially for the price, Now AIOs with RGB lighting are pushing prices into the same realm as all copper RGB equipped pre-filled custom water solutions. While these may be more nerve wracking for some, It shows a huge weakness. Makes it hard to recommend a $150-200 AIO when a $150 gets you custom water. More focus on improved designs and optimization is essentially necessary at this point. Hopefully i will see these changes on my test bench sooner rather than later but I still have my doubts.

Hopefully you will see those changes ;-)

Now I do think there's still innovation going on in AIO's, however maybe not as fast as everyone would like to see it.
As for longevity and leaks, I can naturally only attest to Asetek SKUs. Leaks is a non-issue on Asetek SKUs - we have industry leading test regimes to counter leaks as well as many, many years of experience in liquid cooling. Pump failures are rare or very rare, and also non-catastrophic. However, I will agree that there is a life-span on units, absolutely. I mean, that would go for any moving part or anything that has power pumped into it (pun intended! )

 

[Toast5254]

Does anyone know if the pump speed on this can be set to the full 3,600rpm permanently if plugged into a motherboard header or direct to the PSU?

 

[Caring1]

Does anyone know if the pump speed on this can be set to the full 3,600rpm permanently if plugged into a motherboard header or direct to the PSU?

Usually you can change those settings through the Motherboards Bios.

 

[Toast5254]

Usually you can change those settings through the Motherboards Bios.

Hello. Thanks for your reply. I can adjust rpm through the header on my motherboard, but this Celsius+ range is listed as having a maximum pump speed of 2,800rpm, with 3,600rpm only unlocked in "thermal protection mode" (60 degrees + fluid temperature), so I'm wondering if I would be able to set it at the full 3,600rpm pump speed constantly if using my motherboard as opposed to fractal software.