Zalman's HD135 HTPC case: Gasping for air

Viewing page 6 of 7 pages. Previous 1 2 3 4 5 6 7 Next

TEST RESULTS

Ambient conditions were 22°C and 20 dBA.

Testing was conducted without the intake 80mm fan or the airflow duct installed, as there was not enough room above the heatsink to fit it in over top. The heatsink fan was close enough to the top of the case that an additional fan would have made little difference to cooling and degraded noise significantly.

Noise Profile: Zalman HD135
Fan Speed	System Noise
0%	28 dBA@1m
40%	32 dBA@1m
50%	35 dBA@1m
70%	37 dBA@1m
100%	40 dBA@1m

Our initial impression of the system was pretty bad. With the exhaust fan plugged into the fan controller, the baseline noise level was 35 dBA@1m — a long way from quiet. After investigating with Zalman's M-Play software, we discovered that the fan speed defaulted to "50%", and could be manually set to 0%, 40%, 50%, 70%, or 100%. Our ears told us that, even at 40%, the exhaust fan was still the main source of noise in the system, yet we could barely feel any air movement around the fan.

With the exhaust fan disabled entirely, the system could no longer be called noisy, but neither could it be considered quiet. 28 dBA@1m might be acceptable during a movie when the sound track will drown out most of the background, but it's not quite good enough for a system that will be permanently on in the living room. Besides — with no exhaust fan active, the HD135 would certainly overheat before long, as the only airflow generated in the main chamber would come from the CPU fan.

Eventually, it was decided that very little could be done to improve things, so the exhaust fan was reset to the default 50%, and thermal tests were run with all fans set at constant speed. The system noise level during all of these tests was 35 dBA@1m.

The major noise sources, in order of descending importance, were as follows:

• Stock 80mm exhaust fan, spinning at 50% (as determined by the M-Play software).
• Western Digital 500 GB hard drive, hard mounted in idle.
• SilverStone Element Plus 500W Power supply (fan speed never increased from minimum).
• Nexus 92mm fan at 12V, mounted on the CPU heatsink.
• Zalman VF900 VGA cooler at 5V, mounted on the GeForce 6800XT
• Seagate Momentus 7200.1 hard drive, soft mounted in idle.

Thermal Test Results: Zalman HD135
System State	CPU	GPU	HDD	System Power	Noise (SPL)
Idle (Cool'n'Quiet)	36°C	50°C	44°C	95W	35 dBA@1m
Idle (No Cool'n'Quiet)	45°C	50°C	44°C	112W	35 dBA@1m
2 x CPUBurn	75°C*	52°C	44°C	187W	35 dBA@1m
2 x CPUBurn, with top cover removed	59°C	39°C	44°C	181W	35 dBA@1m
*Test halted to prevent permanent CPU damage from thermal runaway.

The test results speak loudly for themselves. As configured, the HD135 was incapable of preventing thermal runaway, and we had to halt the test to prevent damage shortly after putting the system under load. Should we have left the intake fan in place? It's hard to believe that we should have. With the CPU fan about a centimeter from the top vent, it should not have been starved for fresh air.

Besides, thermal performance was marginal even with the top cover removed. Obviously, our modified IceTank cooler was not quite up to the task of cooling a hot X2 5000+. It's clear that the thermal conditions inside the HD135 didn't help — a 15°C drop is nothing to sniff at — but it's unfair to blame the HD135 entirely for its failure in this first test.

ADDING A NEXUS 120

However, before swapping out the heatsink (a chore because it's difficult to find a high end, top-down heatsink that is also quiet), we decided to try one more thing: We installed a new fan in the left drive bay, hoping to generate more system airflow than could be provided by the puny exhaust fan whirring away at 50%. A quiet Nexus 120mm fan running at full speed was jury-rigged with some foam and duct tape blowing into the main chamber using the front two vents as intakes. The result is illustrated below:

The orange fan front and center was added, to spectacular effect.

Thermal Test Results: Zalman HD135 with Nexus 120 Fan Mod
System State	CPU	GPU	System Power	Noise (SPL)
Idle (Cool'n'Quiet)	23°C	38°C	95W	35 dBA@1m
2 x CPUBurn, with top cover removed	50°C	45°C	179W	35 dBA@1m
2 x CPUBurn	53°C	50°C	179W	35 dBA@1m
2 x CPUBurn with exhaust fan disabled	54°C	51°C	179W	29 dBA@1m

The results surprised us — so much so that we went back and reran every test a second time to confirm our results. With the addition system airflow generated by the Nexus 120, the CPU temperature dropped a full 9°C with the cover removed — and only went up by 3°C when the cover was replaced. We were now seeing acceptable results, achieved with zero acoustic cost. Even at full speed, the Nexus 120 didn't affect the noise character of the system, as the whiny exhaust fan drowned it out completely.

Of course, the next step was to disable the exhaust fan entirely — which dropped the noise level to 29 dBA@1m and had precisely no effect on temperatures. If you were looking for proof that the stock exhaust fan is inadequate, this is it.

With the system running smoothly, there was little reason to replace the heatsink now that the system airflow had been improved. However, the question remained: Was the terrible performance in the first test caused by a poor choice of heatsink, or lack of airflow in the HD135? The fact that the problem was fixed by improving system airflow points to the case, but why then did the system perform so poorly with the top cover removed?

The question caused a lot of head scratching until Mike Chin, SPCR's editor-in-chief, took a look at the system and pointed to the motherboard as the culprit. The motherboard??!? It just so happens that the M2N32-SLI Deluxe that we used in our test system consumes a substantial amount of power on its own. It also happens that it's passively cooled by a collection of copper heatpipes and heatsinks that surround the CPU socket. These heatsinks rely on system airflow to move heat away from them, and without adequate airflow across the motherboard the heat didn't go anywhere, heating up the motherboard and the CPU in the process. The result: Thermal runaway, even with the cover removed.

At this point, we could have finished the job by tinkering with the fan speeds (and drive mounting) to drop the noise down further, say, to below 25 dBA@1m. With appropriate modifications, this would not have been difficult, but it's not clear we would have learned anything new about the HD135, so we let things lie. Besides, a $300 case should not need modifying to get an acceptable level of noise or performance. We leave it to those who enjoy tinkering to discover the best way of silencing the HD135.

Previous 1 2 3 4 5 6 7 Next

Cases & Damping - Article Index

Help support this site, buy the Zalman Home Theatre PC Enclosure Blk from one of our affiliate retailers!