Test Systems

To find out how Seasonic S-12 500 performs, my plan is to test it in real world application. First compared to OCZ Powerstream 420, a residence of a P4 Prescott S478 system, then it will be moved to another system, A64 Venice currently occupied by Thermaltake 420 watts.

P4 Prescott Test Setup
CPU	P4 Prescott S478 2.8E @ 3.50 GHz 1.525 Vcore
Mainboard	Asus P4P800 Deluxe
Memory	2*256Mb PC3200
Other	R9000 Pro 128Mb Two Optical drives 120 GB SATA Hard Drive Leadtek TV tuner 120 mm exhaust; 90 mm intake

Athlon 64 Test Setup
CPU	A64 Venice 3000+ @2.7 GHz 1.6 Vcore
Mainboard	DFI NF4 Ultra D
Memory	2*512Mb PC3200
Other	Apollo 6200PCX Single Optical Drive PATA 80GB Hard Drive 2x80 mm intake; 2x80 mm exhaust, 1x80 top-blower

Testing Methodology

Before I start moving things around, I recorded the three voltage rails under minimum load using a digital volt meter.

• 12 Volt = 12.02 V
• 5 Volt = 5.01 V
• 3.3 Volt = 3.34

Although some review sites prefer load tester to simulate maximum wattage load; oscillator to measure jittered and AC to DC noise, I simply use a primitive method by using what I could get my hands on. A few two-pin old speaker wires from an ancient Intel 286 board. After all, the power supply will end up in your "expensive rig" and not with the tester. This way, you might know if the motherboard circuit/sensor or software is the culprit. Granted, a load tester will be able to exert accurate maximum load ensuring the maximum rated wattage is attainable.


Put some solder on the leads.

Pressed it down to the main connector - orange for 3.3 volt; red for 5 and yellow for 12 volt (black for ground), check continuity and secure the wirings

Making sure the wires are long enough so that I have enough "room" to measure the voltage outside the case with the system running.

The measurements will be used to reconcile with MBM5 and Smart guardian readings should there be any discrepancy amount the three different means of measurement. Although the P4 will be using MBM for comparison and Smart guardian will be in the A64, the comparison will remain "apple to apple" in terms of voltage reading disregarding if OCZ 420 versus S12-500 versus ThermalTake 420 are fair deal.


Prime95 Running in background + 3DMark2001se


Seasonic S12 500 Prime95 load screen shot - A64 3000+ Venice @ 2.7G. 1.6Vcore.


Seasonic S-12 500 Prime95 load screen shot - P4 Prescott 2.8E @3.5G 1.525Vcore.


OCZ Powerstream Prime95 load screen shot - P4 Prescott 2.8E @3.5G 1.525Vcore.

Thermaltake PurePower Prime95 load screen shot - A64 3000+ Venice @ 2.7G. 1.6Vcore.




Interestingly, both OCZ and Thermaltake are holding up better than Seasonic S-12 using onboard sensors and read from MBM5 and Smartguardian software. Please note that OCZ provides voltage trimpots for all 3 voltage rails while the Thermaltake Purepower provides a single trim pot for 5 and 12 volt rails. Although I did not adjust the trim pots pre-set with the OCZ Powerstream, I did adjust the Thermaltake to arrive a higher voltage about six months ago. Without knowing how accurate the motherboard sensors are, I use the Craftsman Autoranging Multimeter 82028 to measure the voltage rails below.



Seasonic S-12 500 holds a surprisingly tight tolerance in all 3 rails using Multi-meter measurement.


Sound level without load – (ambient 37dBA)

Using CEM DT-8850 digital sound meter
• ~ 10 cm (4 inches) away 37 dBA
• < 1cm 39.8 dBA

The 12cm fan turns at a lazy ~650 RPM at idle and reaches a high of ~1,050 RPM after 30 min of Prime95, any water cooling radiator fan or CPU fan or case fan will overcome the noise from this power supply.