Home

Nexus NX3500 120mm PSU

Submitted by Mike Chin on Wed, 2003-12-17 08:54. Power
Viewing page 4 of 4 pages. Previous 1 2 3 4

TEST RESULTS

Measurements were made at 5 output power levels: 65W, 90W, 150W, and 350W. The PSU was allowed to run for 10~15 minutes at each power level before measurements were recorded. The room temperature was 21C.

A. Load on the PSU

+12V
24
36
60
156
+5V
20
20
40
110
+3.3V
16.5
26.4
42.3
78
-12V
2.4
3.6
3.6
3.6
-5V
1
2
2
1
+5VSR
1
2
2
1
Total DC Output
65W
90W
150W
350W

B. On test bench, in 21C ambient temperature

AC Power Draw
103W
135
215
516
Efficiency
63%
66.7%
69.8%
68%
Power Factor
0.62
0.67
0.68
0.71
Fan Voltage
4.1 V
4.2 V
4.3 V
10 V
Noise (1 meter)
<20 dBA
<20 dBA
22 dBA
38 dBA

C. In thermal simulation case, over light bulb

Total DC Output
65W
90W
150W
350W*
AC Power Draw
103W
135
215
525
Efficiency
63%
66.7%
69.8%
66.7%
Light bulb
60W
60W
100W
100W
Fan Voltage
4.1 V
4.4 V
7.2 V
10.8 V
Noise
<20 dBA
22 dBA
30 dBA
43 dBA
Case Temp
29
30
33
38
Exhaust °C
33
33
36
40

* Only after the tests were conducted did I realize that 350W total DC output represents a 20W overload for this model. It's one explanation of why in this thermal simuation high power test, the AC power draw went up: We may be seeing a tiny bit of thermal overload in the circuits.

ANALYSIS

1. VOLTAGE REGULATION was good, within -/+2% on all lines in any combination of loads. The low and high voltage seen on each of the main lines is shown:

  • +12V: 11.87 to 12.34
  • +5V: 4.90 to 5.31
  • +3.3V: 3.29 to 3.42

We have no way of testing line regulation, so AC conditions are steady-state, not dynamic as it would be (potentially) in a real PC. The AC line in the test lab as measured by Kill-a-Watt is usually very stable, within a couple of volts of 120V.

2. EFFICIENCY was modest throughout the power load range and never quite reached 70% even at high loads. This is a bit below average performance compared to other PSUs tested by SPCR.

3. POWER FACTOR was as expected, ranging from a low of 0.64 at low loads to a high of 0.72 at maximum power load. This was expected, given the use of passive PFC.

4. FAN VOLTAGE: The fan receives full voltage (10-11V) for a couple of seconds upon startup to ensure that it always starts. The default voltage just after a cool start is ~4V. In the thermal simulation case, the first significant change comes at the 150W output level with the 100W bulb. The fan voltage jumps from 4.2V (without external heat) to ~7V. At full power without additional outside heat, the fan speed stabilizes at 10V.

5. NOISE was measured at 1 meter from the exhaust grill. The test environment is live, so readings are higher than would be obtained in an anechoic chamber readings. (See explanation in Test Methodology section above.)

Subjectively, the Nexus NX3500 is very quiet. After the startup burst, the 120mm fan has virtually no bearing noise, and sounds much smoother than the fan in the Fortron-Source Aurora. As the speed increases, the fan retains its smooth sound. Even up past 6V, the primary noise is the relatively benign whooshing sound of air turbulence. No bearing chatter, whistling or whining. It is substantially smoother sounding than the SuperRed 120mm fan used in the Seasonic Super Tornado as well.

When the fan was deliberately disconnected or halted, some coil hum could be heard with the output above ~150W. Normally this noise would be masked by the fan noise. At the highest fan speed, wind turbulence noise dominates completely, accompanied by a bit of humming.

The measured noise at minimum is a couple dBA better than the Seasonic Super Silencer 400 or the Nexus NX3000. Only the Seasonic Super Tornado and the ExoticPC-SilentX 14 dBA PSUs manage to best it for residual noise. However, we all know this is only where the noise begins. The real question is where it ends up under load in a real system.

6. THERMAL IN-CASE SIMULATION

The temperature of the case was monitored with a thermal probe positioned about 1" below the PSU intake vent and about 1" away from the center. The temperature of the exhaust air from the PSU was measured with a thermal probe positioned about 1/2" away from the center of the PSU exhaust grill panel.

Judging from the upturn in fan speed at the 150W output level with the 100W bulb, in order to minimize noise from a system using this PSU, it is probably best to keep total system power draw to under 150W. This is not difficult to do even with a very capable system -- one with a CPU rated to >2.5 GHz, a couple of hard drives and a mid-line gaming video card. It is probably possible to run such a system without a case fan, with just the fan in the PSU and with a quiet fan on the CPU heatsink. In this room ambient temperature, once the load reaches 150W, however, the fan speed jumps to ~7V and the fan noise reaches 30 dBA at 1 meter, which is not so quiet any more.

7. WHAT ABOUT WITH A CASE FAN?

At the 150W and 350W loads, measurements were repeated with a Panaflo 80mm low speed fan (FBA08A12L1A -- our reference) mounted on the back panel of the case, fed 7V via a Zalman Fanmate1 controller. There is some question about how this fan interacts with the 120mm fan in the PSU.

Power
150W
350W
Case Fan
Off
Exhaust
Intake
Off
Exhaust
Intake
Fan V*
7.2 V
8.1 V
7.1 V
10.8 V
11.1 V
10.8 V
Noise
30 dBA
33 dBA
30 dBA
42 dBA
43 dBA
42 dBA
Case Temp
32C
33C
29C
33C
33C
32C
Exhaust °C
37C
38C
33C
39C
40C
34C

At the 150W power load, turning the exhaust case fan on had the effect of increasing the fan voltage from 7V to over 8V, with a concomitant 2-3 dBA rise in noise. This implies that the temperature seen by the internal PSU thermistor increased. It is the same result found with another 120mm PSU.

Flipping the case fan around to blow outside air into the case had an immediate effect. Both case and exhaust temperatures dropped slightly with increase in the PSU's minimum fan speed. Note that the presence of a CPU heatsink and cooling fan directly below the PSU may have an impact on airflow; experimentation is recommended.

CONCLUSION

The Nexus NX3500 is a good contestant in the 120mm fan PSU sweepstakes. The smoothness of its fan is a very endearing quality and provides unobtrusive low noise PC operation with a modicum of care in system assembly. This brightly colored Yate Loon is probably the smoothest, least obtrusive 120mm fan I've heard yet. (And no I have not hear them all!) The Nexus NX3500's fan noise characteristic puts it in at the top of the 120mm PSU class. As with other 120mm fan PSUs, it may be possible to run a mid-range power system without any other case fans due to the large air-moving capability of that big fan.

The Nexus NX3500's strengths include:

  • Very smooth, quiet fan
  • Good stability and voltage regulation
  • Good self-cooling
  • Good directed airflow design

There are a couple of weaknesses:

  • Linear-line fan control may produce more noise than absolutely necessary when pushed hard.
  • AC/DC conversion efficiency could be higher.

Our thanks to our sponsor EndPCNoise for the Nexus NX3500 review sample and for their continued support.

* * *

POSTSCRIPT: Unraveling the Mystery of PSU Noise Vs. Watts Graphs

Consider the noise / power output chart provided by Nexus below. It compares their earlier NX3000 PSU, a highly successful and popular model, with the NX3500. Just looking at the graph at face value, it's clear the noise advantage of the NX3500 comes only below 150W or above ~260W. In that middle range, the NX3000 stays 2-3 dBA quieter than its bigger sibling. That's at face value.

A Key Fact: We do not know at what temperature the information on the graph above (or a similar one supplied by Seasonic and other PSU makers) was collated.

This is a most important point: There are NO ATX power supplies that adjust fan cooling speed in accordance with power load. Rather, they adjust their fan speed in accordance with the temperature of a thermistor within the PSU housing, usually clamped to a heatsink. When graphs like the one above are shown, the power points for the changes in fan speed are NOT absolute, they are quite dependent on environmental or ambient temperature. The fan speed-up point is plainly and simply preselected by the designers to a specific thermistor temperature, and any combination of factors which causes the thermistor to reach this temperature will have the effect of turning the fan speed up.

Consider what this means: After subjecting the new model to a battery of torture tests, the PSU design engineers decide that the thermistor must stay below 50°C in order to keep key components in the PSU within their safe operating range. Just because the thermistor is at 50°C doesn't mean those other components are at the same temp; they could be running usually 30°C hotter. We don't know. In any case, they have chosen 50°C to be the maximum safe allowable temperature of the thermistor.

So. Let's say tests have shown that at the rated output of 300W, this new PSU dissipates 100W of heat (which means an AC/DC efficiency of 75%). The engineers test further and determine that given the characteristics of the PSU heatsinks and casing, and the anticipated temperature in which the PSU will be tested for safety certifications, a directed airflow of 30 CFM is adequate to keep the thermistor at just below 50°C. They prefer a healthy margin of error, so they choose a fan of 40 CFM airflow capability.

They want to minimize noise, and so design a fan control circuit that begins speeding the fan up from its minimum consistent start voltage of 5V at around 35C, going up linearly to 50°C where the fan will get the full 12V. Above that thermistor temperature, the engineers know that the efficiency of key components will become less linear so that power output actually begins to decline. But that's OK, they know that with 30 CFM or more at the test temperature, the PSU can keep working indefinitely at full power output and not exceed 50°C.

Remember, all this assumes a certain predefined environmental temperature! If the ambient temperature is at a very cool 20°C, then the PSU may be perfectly capable of 350 W output before overheating begins to take place. On the other hand if the ambient temperature is a scorching 40°C, that 50°C temperature may be reached at a low 250W, beyond which thermal overload behavior will ensue.

The final point is simple: Fan speed (and thus noise) is NOT a simple function of power output, and noise vs. power graphs as presented by many PSU makers is simply an idealized behavior curve they would like to see and to present. Real operating conditions are subject to far greater variations -- such as actual environment temperature dictated by climate and season, air conditioning or the lack thereof, amount and effectiveness of system case ventilation, the amount of heat dissipated by other components in the PC, the efficiency of the PSU, the noise/airflow curve of the fan used, and the workload on the system as a whole.

The system builder's challenge is to keep any of these noise-reduced PSUs operating at or below the speed-up turning point of the fan curve. The rules for any prudent silent PC system builder to follow are:

1. Choose the highest efficiency PSU with the quietest fan.
2. Provide this PSU with the lowest temperature adequate intake airflow.

This second rule brings up the idea of a PSU fresh air duct, which I have mentioned many times in the forums, and which forum member Lilla took and ran with in the thread: Building a PSU Channel.

It's obvious that one day I have to witness a PSU to getting tested for certification by a safety agency like CSA or UL. One day.

* * *

Discuss this article in the SPCR Forums.



Previous 1 2 3 4

Power - Article Index
Help support this site, buy from one of our affiliate retailers!
Search:         
» 55669 reads