Modding the Zalman 9500 Heatsink

Table of Contents

Seagate joins Maxtor and Hitachi in offering a 500 GB hard drive in this era of ever increasing storage capacity. The Barracuda 7200.9 is a 4-platter drive with 16 mb of cache and SATA 2.5 compliance. We take a sample for a spin around our lab….

Nov 22, 2005 by Mike Chin

Zalman’s CNPS9500 made a very strong showing on our test bench. In the conclusion of our review posted in September, we wrote:

“The Zalman 9500 provides excellent cooling performance, close to the best
that we’ve tested, the Scythe
Ninja
, and matches the Thermalright
XP-120
. It provides this performance even
when airflow is reduced by undervolting the fan with the included Fanmate, a feature for which we’ve long applauded Zalman.

“However, similar performance could be achieved at a lower noise level with
the two heatsinks mentioned above by using a quiet Nexus 120mm fan. By contrast,
the fan in the 9500 is a step backwards even from its predecessor, the Zalman
7000, which sounded nicer (if only slightly quieter) at the same voltage.

“The performance of the heatsink, especially under low-airflow conditions, makes
it a prime candidate for a fan swap. This is a more
difficult procedure than a standard fan swap, but the results are likely to
be well worth the effort. It’s been done with both the Zalman 7000 and the 7700 series.”

This article is a followup to the last paragraph above. It documents the procedures used to remove the stock Zalman fan and replace it with the guts of a Nexus 92mm “Real Silent Fan”, which is the quietest 92mm fan we’ve yet encountered. The objective was to have a heatsink as quiet as the Nexus-fan modded 7000 and 7700 heatsinks described above — but with even better cooling performance and all the benefits of the 9500’s superior overall design.

THE BEGINNING

We begin, of course, with the stock 9500 that was used for our testing. In case you’ve forgotten what it looks like, here’s a photo:


Stock Zalman CNPS9500.

As you can see in the above photo, two machine screws affix the metal fan bracket to the base of the heatsink. These screws were removed, using a small Phillips head screwdriver with a long shaft. Care had to be taken not to apply pressure against the fan blades, to avoid damage.


Fan and bracket removed.

The photo below shows two smaller screws that attach the fan to the bracket. These were removed to separate the fan from the metal bracket.


The fan attaches to the bracket with two small screws.

The distance between the mounting holes on the fan was measured carefully. It was exactly 4 cm, center to center. Shown below is a Nexus 92mm fan, already cut away from its frame. The Nexus fan came from our long time supporter, End PC Noise. A pair of sharp metal cutting pliers was used carefully to avoid applying torque on the bearing.


Nexus fan cut from frame.

The next photo shows the tiny holes drilled into two of the arms that extend from the center of the fan. They had to be very precisely done. I was off ever so lightly; the distance between the holes to be just under 4 cm, perhaps 3.95 cm. I knew this might be enough to cause a break around one of the holes as the screw was tightened and tension applied. The plastic “arms” are very narrow. I should probably have drilled the hole before cutting the fan from its frame and used a sharp pin of some kind to score the starting point. It was a non-slip, self centering drill bit I used.


Holes drilled for mounting the fan to the 9500 fan bracket.

The remaining arms were cut and trimmed away carefully, again using the cutting pliers and a large file. A layer of 1/16″ thick double-sided foam tape was applied between the bracket and the back of the Nexus fan to stop the screws from going all the way into the holes that I suspected might break. The double side tape also added a bit of soft cushioning.

I managed to attach the Nexus fan to the metal frame with only a small crack along one of the holes. With the tape and the other screw, it was secure, and there seemed little risk of the hole breaking open. The photos below show the final result, mounted on the nearest motherboard.


No LED lights in this fan, but it’s much quieter.

The work took perhaps an hour. You might be able to do it in less time. There are probably many other ways you could accomplish the same task; using epoxy glue, for example, instead of trying to duplicate the mounting holes. You could also try something more extreme, like using a 120mm fan instead. This would mean the blades have to be outside the recess in the heatsink fins.

TEST SETUP

Test Platform

  • Intel
    P4-2.8A
    The Thermal Design Power of this P4-2.8 (533
    MHz bus) is 68.4 or 69.7W depending on the version. As the CPU is a demo model
    without normal markings, it’s not clear which version it is, so we’ll round
    the number off to ~69W. The Maximum Power, as calculated by
    CPUHeat
    & CPUMSR
    , is 79W.
  • AOpen
    AX4GE Max
    motherboard – Intel 845GE Chipset; built-in VGA.
  • OCZ
    DDRAM
    PC-3200, 512 MB
  • Seagate Barracuda IV 40G 1-platter drive (in Smart Drive)
  • Seasonic
    Super Tornado 300
    (Rev. A1)
  • Arctic Silver
    Ceramique
    Thermal Compound
  • Nexus 120 fan
  • Two-level plywood platform with foam damping feet. Motherboard on
    top; most other components below. Eases heatsink changes and setup.

Measurement & Analysis Tools

  • CPUBurn
    processor stress software
  • Speed Fan
    software to show CPU temperature
  • A custom-built fan controller that allows us to dial in the precise voltage
    to the fan
  • Electronic Anenometer (to measure fan air flow)

Noise measurements were made with the fan powered from the fan controller while
the rest of the system was off to ensure that system noise did not skew the
measurements.

The ambient conditions during testing were 17 dBA and 21°C.

TEST RESULTS

Acoustically, the salient fact of the comparison below is that even at the lowest setting of the Fanmate (5V), the stock fan is 2 dBA noisier than the Nexus fan at 12V. The Nexus 92 modded Zalman at 12V is probably quiet enough for most users. Other noise-producing components will likely prevent this fan from becoming an identifiable, audible source of noise.

At 9V, the Nexus fan becomes difficult to hear even from just a foot away. The smoothness of the sound is a huge improvement over the buzzy quality that dominates the stock Zalman fan when it is undervolted. Overall, there seems to be less turbulence noise with the Nexus 92 stripped of its frame and mounted in the Zalman 9500. The measured noise actually dropped a dBA or two.

Zalman 9500: Nexus 92mm Vs. Stock
Fan
Fan Voltage
°C Rise
°C/W MP
Noise (dBA@1m)


Nexus92

Stock


Nexus92


Stock


Nexus92

Stock
12V

21

14 

0.27

0.18

20

37
9V

25

15 

0.32

0.19

18

32
7V

29

16

0.37

0.20

<17

27
5V

N.A.

18

N.A.

0.23

N.A.

22

Load Temp: CPUBurn for ~20 mins.
°C Rise: Temperature rise above ambient at load.
°C/W MP: Temperature rise per Watt, based on CPU’s Maximum Power (79W)
Noise: SPL measured in dBA@1m with high accuracy B &
K 2203 Sound Level Meter

The 21°C Rise (above ambient room temperature) of the Nexus-equipped 9500 at 12V is probably good enough to keep most AMD Athlon 64 and Sempron processors below 60°C at full load in a reasonably well designed system, even in 30°C room temperature. This probably holds true for Intel Prescott cores up to ~3.2 GHz. The cooling performance is still OK at 9V, but borderline. (These comments assume that the baseline temperature inside a PC typically runs ~10°C hotter than ambient room temperature. A better configured system could run cooler; others could run hotter.)

At lower voltages, the Nexus fan does not create enough airflow for useful cooling. Performance was not tested at 5V, considering the predictably poor cooling at 7V. There is little need to slow the Nexus fan below ~9V anyway. It’s plenty quiet enough at that level.

CONCLUSIONS

The question is whether this Nexus 92mm fan swap mod is worth doing. If you’re interested in the lowest possible noise and want to use the Zalman CNPS9500, then the answer is yes. The cost is not low, as the 9500 plus a Nexus 92 will probably set you back ~US$70 or more. The main advantages are:

  • Extremely low noise, typically 20 dBA@1m or lower at any drive voltage up to 12V.
  • Very good cooling at >9V.
  • Easy to direct the airflow towards the back exhaust fan.
  • Better cooling airflow across the voltage regulators on the motherboard than with most tall blow-across-the-motherboard heatsinks.
  • Sheer geek satisfaction.

As a parting shot, an alternative 120mm fan mod is shown in the photos below. It’s the work of YHL, a friend who heard what I was doing and promptly sent over photos of his mod. Obviously, the goal was higher airflow with less noise. YHL says it is quieter than the stock 92mm fan; he refused to say which 120mm fan was used. With the fan blade area doubled, more effective cooling is achieved as well.

Link to our original Zalman CNPS9500 review

* * *

Comment on this article in our Forums.

Silent PC Review is reader-supported. When you buy through links on our site, we may earn an affiliate commission. Learn More

Leave a Comment

Your email address will not be published. Required fields are marked *