You are here

Coolmax Taurus CF-300 Fanless ATX PSU

May 26, 2004 by Mike Chin

Product
Coolmax Taurus CF-300 Fanless PSU
Manufacturer
Coolmax
MSP
US$120

Fanless PC components are becoming more common. As fans comprise the major sources of noise in computers, fanless operation is generally regarded as the ultimate in silent computing. This goal is not without challenges, however, as forced air cooling by fans is integral to PC design: Fans are assumed for cooling the CPU, the case and the power supply in all the prevalent standards and guides.

The PSU generates a fair amount of heat, depending on...

  • how efficiently it converts AC into DC voltage, and
  • the power requirements of the system it is running.

Making the PSU operate safely without a fan in a typical ATX tower case is a serious challenge. The PSU is located at the top where the hot air rises and pools, and one of the designed exhaust paths for heat in the tower case is through the PSU. Anyone who has experimented with electronics and fan cooling can attest to the huge cooling advantage even a small amount of airflow (say <10 CFM) has over passive air convection.

Still, fanless power supplies are appearing in greater numbers in the marketplace. Where there is a demand...

The Coolmax Taurus CF-300 is the second fanless ATX power supply to be reviewed by SPCR. It is the lowest power model in a range of fanless PSUs offered by Coolmax. The others are rated for 350W, 400W and 480W. Like virtually every computer power supply we've seen thus far, it is manufactured in China.



Unlike ordinary PSUs, the fanless Coolmax Taurus features a casing made from extruded aluminum pieces.

Coolmax appears to have started in 1992 as a "thermal solutions provider" making AC and DC fans, then gradually expanded its product range to include heat sinks, external enclosures for hard drives, optical drives, cables, case mods and power supplies. It is not clear whether they manufacture the PSUs themselves or subcontract it out to one of the big OEMs.

COOLMAX TAURUS FEATURE HIGHLIGHTS

FEATURE & BRIEF COMMENTS

Real No noise Design.

No fans, so if there's no humming or bussing...

High performance components are designed for extra long life and provide maximum system performance

Doesn't everyone say this?

Heavy duty transformers designed to support high performance system several drives

OK.

Cable-tube on main power cables for better cable routing and neatness(optional)

More useful for simplified inventory control than for users.
Aluminum Chassis(optional)

Yes.

Built in EMI filter, low ripple noise

OK.

Over voltage, Over power and Short circuit protection

Not unusual.

Designed refer to UL, CSA, TUV regulations. Meet Intel ATX12V/ATX 2.03

Safety certificates are less plentiful than others. Does not conform to ATX12V/ATX 2.0 in at least regard; see comments in SPECIFICATIONS.

Compatible with all leading processors including Pentium III, 4 and AMD Athon XP

Hard to sell one that isn't.

SPECIFICATIONS

There are some discrepancies between the label on the PSU and the data provided on the Coolmax web site. The following was compiled from on the web site, as it is more complete.

Coolmax Taurus Model FL-550ATX (CF300)
AC Input
115 VAC/ 4A or 230 VAC/ 8A, 50-60 Hz
DC Voltage lines
+3.3V
+5V
+12V
-12V
-5V
+5VSB
Max Output
25A
30A
15A
0.8A
0.3A
2A
Maximum Output
180W
180W
1.5W
9.8W
10W
279W
* Efficiency is specified as >60% at full load.
* Designed refer to UL, CSA, TUV regulations. Meets Intel ATX12V/ATX 2.03

The numbers look quite substantial and promising. They are what you would expect from a legitimately spec'ed 300W PSU. However there are oddities.

1. A temperature is specified on the web site specs: The number given for the CF300 is 75°C. The numbers provided for the higher power models are 83, 83, and 89°C, respectively. The problem is that there is no indication of what these numbers represent. Certainly they cannot be the maximum ambient operating temperatures; they are much too high. The operating ambient temperature cited by the most reputable PSU companies is 50°C. Perhaps they refer to the maximum safe temperature of core components such as the MOSFET devices? Without more clarification, these numbers are essentially meaningless.

This calls into question the temperature at which the maximum rated power can be achieved. As many readers may be aware, the ambient temperature around the PSU in a typical modern PC is likely to be 35~45°C, and PSU power output capacity drops rapidly when the thermal limit is reached. We do not know the maximum temperature at which the power output of the Coolmax can be assured. This is of particular concern in a PSU without a cooling fan.

2. Despite the last statement in the specifications table above, the efficiency of 60% minimum at full load does not meet the Intel ATX12V/ATX 2.0 requirements, which calls for 70% under “Full” load, 70% under “typical” load, and 60% in a “light” load or idle condition. For a fanless PSU, the 60% efficiency figure is not promising. It means 40% of the AC power is lost as heat inside the PSU. So in delivering 150W in DC voltage, according to this spec, the PSU would generate as much as 100W within itself, which is a lot of heat to evacuate without a built in fan. We hope this is a typo... or that Coolmax is being very conservative with their efficiency spec.

BASICS

The Coolmax Taurus CF-300 comes in a standard size box with two somewhat unusual accouterment:

* A 20-pin ATX plug to 24-pin plug adapter: The recently released ATX12V V2.0 PSU Design Guide calls for 2 x 12 pin output to support 75 watt PCI Express requirements. Presumably, the adapter is supplied for use with a PCI Express motherboard.
Or with a dual-CPU server board (which is less likely given the modest 300W rating).

* A dual-12V to quad-12V plug adapter. Used for dual-CPU server boards or any boards that call for EPS12V PSUs.

Surprisingly, there is no manual or user's guide. Given the special attention that I think a fanless PSU needs for proper use, this strikes me as a lack of customer care.

You've seen the angled perspective view on the previous page; here's a square on view from the back.



The power switch, voltage selector switch and AC input socket are on the slot-vented back panel. It has vents on the back so air can flow through it. The top panel has no vents...



...but all the other panels have slotted vents.

COOLMAX APPROACH to FANLESS COOLING

The vent slots on five surfaces indicate the approach to cooling taken here:

We don't have an integral fan, but there is bound to be some airflow in the PC case, so we'll take every opportunity to make sure that that airflow can go through the PSU.

It is a reasonable approach. Hot air in a typical tower case will rise to the top, and it will pass through the ventilation slots on the bottom, back and sides of the CF-300. How easily it flows OUT the back vent slots will depend on how much air is flowing into the case. A positive pressure case, one in where there is more incoming forced air than outgoing, is probably be best for this PSU.

This approach is quite different from that used by the proSilence 350, which has a sealed back and a set of large heatsink fins on the outside. The approach there is to allow the PSU heat to be transferred out by conduction through the heatsinks, and then air-cooled by convention via the external fins. It provides no way for rising heat in the case to escape, however, which means hot air tends to collect at the top of the case.

In contrast, the Coolmax CF-300 allows an evacuation path for not only the PSU heat but also the rising hot air in the system. At least in theory, it has an advantage over the proSilence cooling approach, as it attempts to take advantage of natural convection.

Of course, bear in mind that the air that will pass through the Coolmax CF-300 will always be warm, which will certainly limit its cooling potential. Depending on just how hot the components get and how high the ambient temperature is, it might not even be much of a benefit.

CONNECTORS - There are 7 wire sets:

  • 2 - 32" long, each with two 4-pin IDE drive connectors and 1 floppy drive connector
  • 1 - 26" long, with two 4-pin IDE drive connectors
  • 1 - 20" long, with main 20-pin ATX connector
  • 1 - 20" long, for 12V (P4) connector
  • 1 - 20" long, for 3.3V connector
  • 1 - 20" long, for single SATA drive connector

The ~5 lb. weight of the CF-300 is high but not as heavy as the proSilence 350 with its large external heatsink and heavy steel casing.

INSIDE THE CF-300

Unlike the typical ATX PSU, which is formed of two U-shaped pieces of sheet metal, the CF-300 is made of four pieces of extruded aluminum:

  • Two U-shaped pieces for the top and bottom
  • Two flat pieces for front and back

Screws on the front and back panels need to be removed to take the cover off.



Bottom cover removed.



This is the bottom of the PSU; it is upside down



The back panel screws have also been removed.



From the other side.



MOSFETs visible on flat side of large internal heatsink. This heatsink is bolted to both front and back panels for structural stability and conduction.



Smaller heatsink vertical, also connected to the back panel.

The sheer size of the internal heatsink made it difficult to poke around much. I did not attempt to move the big heatsink for fear of accidental damage.

The basic layout is not vastly different from a standard ATX PSU:

  • The PCB is set up so that the components hang upside down when the PSU is installed.
  • The large internal heatsink is perforated with many small holes for some of the rising air to get through to the other side, to the rest of the components.

TEST SETUP

For a complete rundown of testing equipment and procedures, please refer to the article SPCR's Revised PSU Testing System. It is a close simulation of a moderate airflow mid-tower PC.

In the test rig, the ambient temperature of the PSU varies proportionately with its actual output load, which is exactly the way it is in a real PC environment. But there is the added benefit of a precise high power load tester which allows incremental load testing all the way to full power for any non-industrial PC power supply. Both fan noise and voltage are measured at various loads. It is, in general, a very demanding test, as the operating temperature of the PSU reaches 40°C or more at full power. This is impossible to achieve with an open test bench setup.

The testing was conducted in the "sound lab", a 20' x 10' x 8'(ceiling) carpeted den with heavy drapes on windows across one of the short walls. Acoustics are fairly well damped. Ambient conditions during testing were 21°C and 15 dBA, with input of 120VAC at 60 Hz.

TEST CUSTOMIZATION FOR CF-300

Initially, the standard test setup was used. In SPCR's PSU Testing System, there are four fans at 5V that blow the heat from the loaded resistors in the PSU load tester into the thermal simulation box. There is a slight positive airflow pressure in the box which actually helps with PSU cooling. As shown in the photo above, an exhaust Panaflo 80M fan at 5V serves to reduce the positive pressure somewhat and to better model a typical low noise system.

  • Because side of the CF-300 is open to the outside in the test rig, the vent slots on that side would act as an exhaust rather than an intake. I decided to close off those vents with a piece of cardboard. In no PC case can hot air exit the PSU to the outside in this way, while internal PSU side vents could easily be blocked.

    At the same time, the exposure of the top PSU panel to the open air also seemed unrealistic. This large radiant surface would never be exposed in a normal system setup. So the top was blocked off as well. It's true that close proximity would allow some transfer of heat to the top panel of a typical case, but I wanted to err on the stringent or conservative side in testing.

    The end result is shown in the photo on the right. Note the tiny blue thermal sensor held down by masking tape near the top of the PSU. It was positioned to be 1~2 mm from the PSU back panel.

  • During initial testing, it seemed worthwhile to ask the question,
  • What impact does the exhaust fan have on the PSU exhaust and test box temperature?
Because the PSU has no fan, the amount of air that flows through it depends on convection and the positive pressure in the case. The exhaust fan was removing some of the heat, on the one hand, but also perhaps reducing the amount of positive pressure in the case. In an attempt to answer the question, the exhaust fan was removed, thus:

After an initial minute or two of temperatures dropping and rising, once things stabilized, there was no change in temperature with the fan blowing out at 5V versus the fan completely removed. This held true for all four test loads from 65W to 200W. I believe this means that the airflow produced by the fan is effectively no greater than that created by convection alone, although it does not feel like that -- when I place my hand in front of the hole. Perhaps some cool outside air enters while some hot air exits.

TEST RESULTS

The testing procedure utilized was considerably more complex than usual. Please read the notes carefully and do not simply skim the data in the table.

Coolmax Taurus CF-300
DC Output (W)
65
90
150
200
250
300
AC Input (W)
92
121
190
246
310
370
Efficiency
71%
74%
79%
81%
81%
81%
Exhaust Fan
None - see text at bottom of previous page
*See Box Text: Higher Power Testing*
Case Temp (°C)
27
30
33
38
39 / 43
- / 45
PSU Exhaust (°C)
31
37
41
50
55 / 47
- / 49
Power Factor
0.62 ~ 0.66 (higher with higher load)

NOTE: The ambient temperature during testing was 21°C. It has a direct impact on all measured temperatures. Please take this into account when comparing test data from other SPCR PSU reviews.

* Higher Power Testing and Case Airflow*

I wanted to avoid burning out the PSU while testing it at 250W and 300W. These are loads I've put on other PSUs, but the reality is that the most power hungry P4-3.2 system I've tested could only draw ~180W maximum peak in DC voltage from the PSU. So these are artificial benchmarks that are highly unlikely to be duplicated in real use. After 15 minutes at the 200W load, the temperature seen by the external probe at the exhaust side of the PSU had risen to 50°C. I was not certain how much more heat the PSU could withstand before suffering damage.

At the 250W load, without the case exhaust fan, the exhaust side temperature reached 55°C within 15 minutes. It's not clear whether the temperature had stabilized yet, but the smell of "electronic burning" (most likely caused by evaporation of thermal interface material [TIM] or the epoxy used in coils and transformers) in the lab made me very uncomfortable. So I reinstalled the "case" fan, but blowing in at 5V instead of blowing out.

My thinking was simple: I wanted to save the PSU from potential heat damage, and I knew the back fan in exhaust mode had no effect on PSU temperature. By making the fan blow in, cooler outside air would be added to the mix, and the overall air pressure in the box would be increased. This is probably something an end user might try with this PSU.

The effect was immediate. Within a couple of minutes, the internal case temperature went up by several degrees, and within 10 minutes, the PSU exhaust temperature had dropped by 8°C, down to 47°C, where it stabilized.

The data in the table for Case and PSU temps at 250W and 300W shows the result with no case fan first, followed by the temp with case fan blowing in at 5V.

The 300W load test was done for 10 minutes, only with the Panaflo 80M fan blowing in. PSU exhaust temp only went up by two degrees over the 250W load.

A NOTE on Voltage Regulation

VR is a relatively easy test for any good PSU to pass as long as it is operating within its power output limits. Many are accurate within 1%. The variances commonly found in reports from motherboard voltage sensors (such as provided by Motherboard Monitor) are not at all useful to test power supplies. The problem is that those readings are the sum of interactions between the motherboard, its power circuitry, the connectors and cables between the board and the PSU, and the PSU itself. Hence, it is difficult to identify the source of any anomalies.

The output voltage of the PSU must be monitored in isolation from external influences while it is doing work (delivering current). The only practical way to do this is to use a voltmeter (multimeter) to check the voltage across the terminals to which the power is being delivered.

In SPCR's VR testing, a multimeter is connected to each of the voltage lines for several minutes. The voltage reading is monitored continuously while the loads on each and all the lines is varied, and the peaks and valleys recorded manually.

1. VOLTAGE REGULATION is very good, well within the required -/+5% on all lines in any nominal combination of loads; it is closer to -/+2%. The low and high voltage seen on each of the main lines is shown:

  • +12V: 11.78 to 12.33
  • +5V: 4.78 to 5.22
  • +3.3V: 3.19 to 3.39

2. EFFICIENCY is exceptional at >80% through the top half of the output range. In spite of the >60% efficiency spec, the CF-300 matches the current efficiency champs among SPCR reviewed PSUs, the Enermax Noisetaker 475 and the Seasonic Super Series, Rev.A3. For a fanless PSU, especially, such high efficiency is extremely important. It minimizes the amount of self-generated heat that must be dealt with.

3. POWER OUTPUT: The unit ran with good stability all the way to the full 300W output. However, within the test rig, the CF-300 reached 50°C and higher at outputs of 200W or more. With adequate positive pressure case cooling, the unit will be as stable with high loads as any 300W PSU. However, in a low airflow, quiet-optimized PC, it's prudent to limit long-term output to below 200W, especially if the ambient room temperature is much higher than the 21°C in the lab during testing.

4. POWER FACTOR is low, as expected for what is probably a non-PFC model. It is not specified; this usually means there is no power factor correction, and the low numbers confirm it.

5. OTHER NOTES

HOT - At all loads above 150W, the PSU was HOT to the touch. It became hot enough that if run in a system that routinely draws >150W, you might have concerns about using it in a PC that small children or pets could touch the back of accidentally. This was in a cool room, just 21°C. Higher ambient temperature will cause hotter operation.

No buzzing or humming could be discerned even at high power loads, but it must be pointed out that rare is the PSU that buzzes with the pure resistive load of the power tester. This is no guarantee that the unit will not humm or buzz when faced with the more complex (though less power-hungry) loads of PC components. Coil buzz is often the result of interactions between components.

IN-SYSTEM TRIAL

In-system testing has not been a part of SPCR's PSU reviews for some time, due to time constraints. The lab testing takes enough time and effort as it is. But an exception was made for the unusual CF-300.

The test system was the one assembled for the ARM System Stealth PC Foundation Kit. This system is in use still as a "temporary" backup for the main machine. The details are essentially unchanged except for the increase of RAM to 1GB.

  • Evercase 4252 case modified with top PSU intake duct and channel
  • Zalman 400B PSU (ATX12V v1.3) modified with Panaflo 80L fan
  • AOpen AK89 Max (nVidia3 Athlon 64 board)
  • AMD Athlon 64 3200+ (spec'd at 89W)
  • Zalman ZM7000A-AlCu heatsink w/ fanmate1 set to 5V
  • ATI 9800 Pro VGA with Arctic Cooling VGA Silencer (fan set to low)
  • Crucial PC3200 memory, 1MB
  • Samsung SP0802N hard drive, mounted with the supplied EAR soft grommets
  • Samsung SM-252B CD-RW Drive
  • 120mm exhaust & intake fans (as supplied in the kit) set to 5V

With identical 120mm fans running at the same speed on intake and exhaust, the slow out-blowing 80mm Panaflo fan in the Zalman PSU is matched by the air flowing in through the side duct over the slow-spinning CPU cooling fan. Overall, there is neutral air pressure in the case: About the same amount of air is actively blown in as out.

Some basic temperature measurements were taken on the above machine. Then it was shut down and the modified Zalman PSU swapped for the Coolmax Taurus CF-300. The duct was removed, and the top case vent blocked. No other changes were made to this machine.

With the fanless Coolmax in place, we have a slightly positive air pressure case: It's the CPU fan that makes the difference, which is very small, as it does not pull much air. A thermistor was placed on the back exhaust grills of the Coomax (in the same spot as during the lab testing). The following temp measurements were taken after 30 minutes in each state. Note that the ambient temperature for the in-system testing in this room (my office) was 24°C, 3°C higher than in the lab.

PSU
State
Temperature (°C)
CPU
Board
PSU
AC Power
Coolmax CF-300
Idle
50
41
43
136W
CPUBurn
61
45
49
158W
Zalman 400B modified
Idle
50
40
42
140W
CPUBurn
61
42
47
168W
Ambient room temperature: 24°C

Some observations about this PSU swap:

1) There is less forced air with the fanless Coolmax. Even though the airflow of the Panaflo 80L in the modified Zalman PSU is very low, it is enough to affect the board temperature by 3°C at maximum load.

2) The reduction in airflow is not enough to affect CPU temperature at all.

3) The Coolmax is more efficient than the Zalman, especially at higher power output. Here is a simple table showing in-system power and efficiency, based on data from SPCR tests:

PSU
Idle
CPUBurn
AC
Efficiency
DC
AC
Efficiency
DC
Coolmax CF-300
136
75%
102
158
77%
122
Zalman 400B
140
73%
102
168
73%
122

Two facts jump out:

  • The maximum power that this fairly high end system draws is just 122W, which is only about 40% of the maximum rated output of the Coolmax.
  • The temperature reached by the Coolmax in this system is far higher than on the test rig, which suggests that the positive pressure of the test rig is very advantageous for this PSU. The 49°C reached in the system with CPUBurn (at 122W DC output) was not reached in the test rig till 200W output.

BOTH 120mm FANS BLOWING IN

There was one obvious thing to try: Reverse the direction of the 120mm exhaust fan so that both front and rear 120mm fans are blowing in, causing positive pressure in the case, which should force more air out through the PSU, cooling it better in the process. The side duct for the CPU may become an exhaust port, but we will see.

Uh-oh! A Glitch!

The test system had been running CPUBurn with the Coolmax PSU for about 45 minutes when it was turned off. The PSU was not hot enough enough to burn, but definitely quite hot to the touch. The back case fan was flipped over, the PSU plugged back into AC, and the power switch hit.

No response.

The fans spun up for a few seconds and then died. The motherboard LED indicating idle current was on. With the next power button press, the fans remained motionless. I repeated this about 10 times before realizing it was not going to boot. I checked all the cables; none had been dislodged. I unplugged the PSU from AC, reset the CMOS, plugged the AC cord back in. No luck.

The only thing that could explain this was some kind of thermal circuit breaker in the PSU. Had it tripped just as power was turned off? Did the PSU get hotter after the power was turned off ? Only way to find out was to leave the PSU to cool off for a while.

Two hours later, the system powered up and ran without a hitch. I don't know what caused the shutdown; I do believe a thermal fuse may have been tripped.

Later, I discovered that every time I powered the system down, this Coolmax Taurus CF-300 sample refused to power back up until it was completely cooled off for at least an hour. It's not the kind of behavior anyone wants to see in a power supply.

Finally, I tried turning the system off within a minute after turn on, just after Windows stabilized. It still refused to start again. I had to unplug the PSU from the AC and leave it for 10 minutes before trying again -- at which point it started OK.

Temperatures were measured again.

Coolmax CF-300
State
Temperature (°C)
CPU
Board
PSU
Both fans blow in
Idle
50
43
48
CPUBurn
64
49
52
Original test results*
Idle
50
41
43
CPUBurn
61
45
49

*Front fan blowing in, rear fan blowing out.

Ambient room temperature: 24°C

The results were worse in almost every regard. Perhaps the back fan blowing in does not really create positive pressure because of the fresh air side duct for the CPU cooler. There are too many factors to sort out exactly what is happening.

For the Record: All Fans at Full Power

Just to see what this PSU will do in a high airflow system. We know the 120mm fans are capable of ~60 CFM each. They are in push/pull mode, so let's just call it 60 CFM. The 92mm CPU cooling fan is probably capable of min 35 CFM, maybe 40, and it is drawing in cool air through the side duct. There could be as much as 100 CFM of air flowing in and out of the case. Suffice it to say there was a lot of air flowing around and through in the case with all the fans at 12V. (An Aside: My bare sandaled feet were a lot cooler due to the extra air circulating under the desk.) Results?

Coolmax CF-300
State
Temperature (°C)
CPU
Board
PSU
All fans at 12V
Idle
40
33
34
CPUBurn
49
37
49
Original test results
Idle
50
41
43
CPUBurn
61
45
49
Ambient room temperature 24°C

The CPU and board temperatures are no surprise. The PSU temp at idle is very nice, but the huge jump of 15°C
for just 122W DC output even with all that airflow is rather disappointing. (Admittedly, it was difficult to feel much in the way of airflow at the back of the PSU.)

CONCLUSIONS

This review took — nay, prised! — an inordinate amount of time and effort. An epic week after beginning the review process, I am left with a frustrating feeling of incompleteness. A second sample as requested of all review sample providers, could have made this a more conclusive review. As it stands, I have no definitive way to gauge whether the characteristics of this sample are typical for this model or not. Perhaps near 50°C temperature at the back of the PSU is simply too high for stable operation. I don't know because there is no clear manual or cautionary notice about the proper way to use this power supply, and no other sample to compare it to.

It could well be that the Coolmax Taurus CF-300 is designed as a drop-in substitute only for system that have conventional forced air cooling — i.e., high speed case fans all whirring and whining and blowing at the full 12V voltage. I simply don't know.

The best I can do in terms of a comprehensive summary is a pros and cons list. You can make up your own mind whether I received a bum sample or damaged it during the review process, whether it's a waste of money or a great product that only needs the love and care of an inspired silencing modder.

PRO

  • No fan; silent operation
  • No buzzing noted during in-system trial
  • Very high efficiency
  • Five-sided venting take advantage of convection
  • Actually meets output claims in test rig
  • Very good voltage regulation

CON

  • No restart a sample-specific glitch?
  • If cause of above is overheating or overloading, it's bad news.
  • No manual is a bad oversight with this unusual product.

Finally, if the >80% efficiency of this sample is common to Coolmax's fan-cooled PSUs, I would certainly be most interested in reviewing them as well.

Postscript on the Sample's Glitch, June 30, 2004

A few days after my first email anout this problem, Elaine at Coolmax responded saying this was obviously a defect of some kind, one they had not seen before. They would do their best to get a replacement sample to me ASAP and would please I hold off on publishing the review till after I had examined the second sample.

As you can see from the May 26 publish date on this article, that was over a month ago. I won't venture to say why no replacement has yet turned up, but if one does, I will be open to taking another look.

* * *

Discuss this article in our Forums.

Sections: 

Google

www SPCR