Hmm, well it's nice to see some test results for this. It's used all over the M-ITX scene since it actually fits in the cases. The pump noise might be to much for me though.

By the way, does the orientation of the radiator make any difference? Changing the maximum height the pump has to move the water might change results. For that matter, I wonder if it would do better orientated like it would be in a tower, so it's pumping sideways rather then up.

Interesting article - thanks for the review Lawrence & MikeC.

Since the pump is run off a standard 3 pin fan header, would it be possible to do some tests with undervolting the pump? I vaguely recall one review site on the web that did this with a H50 when it first came out, but I can't find the review again unfortunately . From memory, the performance was a few degrees C worse with the pump at 7V.

Given that the pump was the dominant noise in these tests, would it be possible to do a re-test with undervolting the pump? The performance isn't fantastic to begin with, and I suspect it would drop further. However, like Mr Perfect said, the H50 strikes me as useful for mini ITX systems to maintain reasonable cooling power while being more flexible than the 160mm tower coolers.

I have to say that I (subjectively) disagree with the assessment - I have the H50 on a Quad Q9300 overclocked to 3GHz. Temps were reasonable with the stock fan, but when I combined it with 2 Noctua NF-P12's in a push-pull exhaust configuration, they dropped to 24C degrees at idle and 40C at full load, with the cores about 25-30C away from TJmax.

There was virtually no noise coming from it except the wind.

I since replaced the noctua's with 2 Gentle Typhoons 1850rpm, which I typically run at 60% speed. This gives me the same performance but I can push them to 100% and drop the temperatures even more in summer when my house temp can hit 33C!

As for the pump, I cannot hear it through my Antec P182 at all.

I have to say you have a different opinion about the same model based on your experience with a different sample. There's absolutely no getting away from the sound of our sample's pump. Listen to our test recordings.

It may be that the pump....
1) became better in later production runs
2) became worse in later production runs
3) represents an unusually bad sample

Without knowing more about you and your gear and your environment, it might even be that you hear exactly what we hear -- but perceive it very differently. We don't really know at this point... though your simple presence here suggests otherwise, that you have good noise sensitivity.

Judging from forum discussions on this product elsewhere on the web, it appears there is definitely some variations in noise. We'll touch base w/ Corsair and see what they have to say.

Mike,

You're quite right. I qualified my disagreement with the word "subjectively", but what I should have said to be clearer is that my experience was very different.

Indeed I have an strange sensitivity to noise because while I'm used to a quiet environment, I can tune out certain low pitched sounds. Once I started experimenting with quiet pc's I realized that since I like performance computing, I would find impossible to achieve perfect quiet so now I concentrate on getting the sound to a level and quality that I can forget about. Others may perceive it differently.

And of course the fact that my case is closed and sound dampened, means that I have a different result from your testbed.

It's worth noting however that other H50's sometimes have noisy pumps which is usually due to a fault and its worth pursuing an exchange. It may also be a good idea because a faulty pump could eventually leak too!

Thanks a lot for the great articles and reviews!

dkazaz -- right.

One other thing I want to point out is that your Quad Q9300 overclocked to 3GHz is a much cooler load than our i7-965. TDP of the Q9300 is 95W (stock) which we know is higher than can be measured at the AUX12V socket w/ any type of CPU load. In contrast, our test setup pulls 180W DC w/ only 3 sticks of RAM, an SSD, 11W graphics card and the motherboard. Our guesstimate on CPU power is 130W~140W.

Indeed the loads are very different.

However do note that pump speed should not be varied to avoid potential problems with overheating. Ideally the H50 pump should be relatively quiet and running at its nominal 1400rpm while tha fan speeds should vary...

I've been reading Corsair forums and it seems that the "noisy pump syndrome" is not that uncommon. It seems some pumps are defective.

Could it be possible your unit was defective too?

We have no way of knowing. There's nothing to compare it to -- tho no question it's way noisier than the pumps in the passive Zalman tower WC coolers from days past. Coolingwise, it isn't bad, just not great for the noise.

Is there any possibility to RMA your unit and get a new one?

I'm really surprised that everybody talks so very good things of this cooler and in your review it seems to have been such a frustrating experience (noise-wise speaking)

I wouldn't call it frustrating, just disappointing. We run often across products that are noisier than advertised or hyped. This is normal and expected at SPCR. But Corsair has had some nicely quiet products in the past, so this is a bit of a surprise, too.

Company reps are aware of our review & experience but have not yet got back to us.

Thanks for the review.

The mp3 recording links don't seem to work :

* Corsair Hydro H50 pump and stock fan mounted at 1m
— 9V (28~29 dBA@1m)
— 10V (24 dBA@1m)
— 12V (21~22 dBA@1m)

* Corsair Hydro H50 pump and Nexus fan mounted at 1m
— 7V (21~22 dBA@1m)
— 9V (21~22 dBA@1m)
— 12V (23 dBA@1m)

Whether the pump was defective or not doesn't make a difference here. This solution is not a good value when you look at how it performs. Even if you throw a high speed fan like the stock fan on it, it still performs worse than top HSFs and is loud as hell. And costs more.

fixed

There's a chance that a noisier pump is also a less efficient pump. I mean, if our sample is noisier than usual, it's possible that the thing also doesn't pump as well as usual. Until we get a another sample to examine, we really can't tell. At this point, it's not clear whether Corsair thinks we have a bad sample.

They're probably wondering if it's good PR to admit there's a QC issue.

I'd tend to chalk it up to most computer enthusiasts definition of "quiet" or "silent" being a lot different than ours. It's all very subjective, so whenever I read a claim like that on other sites, I take it with a grain of salt.

But who knows, could be sample variance, maybe SPCR just got a really bad unit.

I'm a big fan of Corsair, but I just don't like this product as it's kind of like the worst of both worlds. You don't get the performance of a water cooling setup, and you get additional noise from a pump compared to just the fan on a tower heatsink.

The only real use I see for it is small setups like htpc's, but since you're not going for performance in most cases, you could get away with just a passive heatsink.

One thing I don't like about WC setups in general is the completely lack of VRM cooling. Even with a tower heatsink you have airflow over the VRM heatsink (wouldn't buy a board that didn't have one) almost like a push/pull setup but when you have the CPU watercooled, there no fan to blow over it in conjuction with the exhaust case fan.

That might true if you had no additional sources of air flow in your case (i.e., no case fans at all), but given the efficiency of motherboards today, I seriously doubt it'll be much of an issue. As such, an H50 won't provide any vrm cooling either, and could potentially worsen the air flow you have in your case.

I tried 3 different H50, all the pumps were loud at 12V. I don't think Mike's H50 was defective.

Still, if you compromise, they can be very quiet.

In my setup, the pump runs full speed during boot which is absolutely necessary during a cold boot. When speedfan loads up, the pump is severely undervolted so that speedfan gives really jacked up readings, but it is inaudible for all practical purposes.* The Lynnfield are at the low 40s during light load.

If i run prime, my Lynnfield reaches low 60s with 2 fans on the Radiator running ~6v. The pump becomes audible at 1100rpm but it’s not too bad. Further tweaking could fix it, but lack of incentive. I only run prime for stress testing.

In my setup, completely undervolting pump raises the temps a few degrees. I would like to say 5C.

*if you have the side panels off and listen to the pump from a a feet a way, you will hear the pump.

@Mike: on page 7, "Corsair" links to Prolimatech's website

But the cooling results you get are simply not worth the price of the H50; you can get as good or better with a $30 Scythe HSF.

I too had a noisy pump in my H50, but I found that just enabling the fan control on the motherboard for it was enough for me to not hear the noise, and I struggled to see any difference in cpu temp under load with my overclocked i5-750. I've also heard that two fans makes a large difference in cooling especially and low fan speeds, and would love to see the H50 retested with the pump undervolted to an acceptable noise level and two undervolted nexus fans mounted push+pull.. This would certainly put an end to most of the objections I can see.

You (SPCR) seem to try a lot of different methods to quieten cases and record results during reviews, why not do the same for this? It has applications where most normal coolers do not, and I'd imagine some people would want to know just how far it is possible to silence the H50. Other people like myself can post our subjective thoughts and experiences, but you're the ones with all the test equipment that can actually provide scientific comparison.

As for Corsair not recommending undervolting the pump, i'm sure some of the fan manufacturers would say the same thing about their fans, but that doesn't seem to stop most people.. It's not like the motors work much differently when it comes down to it.


Edit: Damnit the review was posted almost a month ago, the H50 probably isn't even in the labs anymore

Actually we never move that fast...

Not only is the original still here, but a second sample is also in hand. Alas, the 2nd is, if anything, perhaps even noisier than the first. We might try a few things, then put up a postscript. The pump noise we reported appears to be typical, though.

Sorting through all watercooling communities, those that are looking for quiet computing will say that all brands of water pumps exhibit electrical noise at stock voltage. I would not expect a budget pump to be any different than the norm. The typical noise is worse than a quality fan and is usually suppressed by decoupling from case with tubing/elastic/silicon padding.

The gurgling is air still present in the loop which gets caught in the waterblock and pump during operation until a large enough air accumulates to pass to the next impass as a larger bubble. Bleeding out the air is critical for quiet operation which typically takes twenty four hours of operation.

My case for using the H50 instead of air coolers is I sell the workstations I build. Shipping a system via courier with a very heavy air cooler has proven troublesome on more than a few occasions. I have to do very careful inside packing so the motherboard won't suffer from that load. Also, fans must be attached to the heatsinks very carefully or they'll fall out of place during transportation (those clips on Megahalems or Thermalright are really flimsy).

Now I can safely ship my worsktations with a working and quiet -for technical workstation standards- cooling solution. I always sell my machines "sensibly overclocked" (Core i7 920 @ 3.4GHz is pretty usual) and the H50 works wonderfully on that sort of setups.

My customers are mostly computer graphics artists and while this setup is not bedroom-quiet, it's definitely office-studio-quiet. Only during heavy rendering sessions the full throttle of all the fans inside the case can be heard, but it's hardly annoying.

I am wondering the answer to the first question in this thread. Is changing the orientation of the radiator effect anything? Apparently in manual it says any orientation is fine, but well, can we trust that without testing?

- You do not get the setup and maintenance problems normally associated with water cooling.
- You do not get the weight and stress on the motherboard related to similarly performing heatpipe coolers.

As a result you do get a robust cooling solution that won't break the motherboard or CPU when the case is handled roughly. (Like sent by mail or generally dropped on the floor.)
This is great for building a mobile gaming computer that is to be hauled around a lot!
It's by no means designed with total noise minimisation as a top priority.

Cheers
Olle