I have a server that operates 24.7. It occurred to me that I may be able to save some money on electricity bills if I could reduce its power consumption. It already has a low power AMD Sempron LE-1100 CPU so the power draw of the system should be quite low, but the PSU is quite old and not especially efficient. This got me thinking about Power supplies designed for low power systems. There are plenty of 80+ certified PSU’s out there with 500+ watt ratings, but are these actually worth the investment for a low power system?
|Base system (e.g. atom or Core i5)||18W|
|4x 2Tb (e.g. Samsung EcoGreen F3)||19W|
|RAID Card (e.g. Areca 1220)||5W|
|1x 120mm Case / HDD Fan (e.g. Nexus Real Silent 120mm)||1W|
Using the above figures, I estimate that a new system with a similar role and capability to what I have will consume just under 45watts at idle.
For a typical low spec (450w) PSU, this load is only 10%. At this low a load a typical PSU can be quite inefficient. The ATX specification requires 70% efficiency at 20% of rated power, but a 43W draw is below 20% rated power for all PSU’s rated for 250W or above (in effect all PSU’s). Below is a comparison between a bare minimum ATX PSU and a few highly efficient PSU’s.
The figure for the generic PSU is based on the minimum requirement and is therefore a worst case. Even with a very poor efficiency PSU the AC draw from the wall won’t increase as the DC draw decreases. I have calculated a best-case yearly saving, based on a kWh charge of 10p and assuming 24.7 usage.
|Make / Model||Efficiency at 43W||Est. Total power draw (idle)||Saving|
|ATX v2.2 250w ATX||70% (at 50W)||61W|
|Seasonic X-400||80.4% (at 42.6W)||53W||£6.97|
|Corsair AX850||77.3% (at 43W)||55W||£5.09|
|Enermax Modu87+ 500w||77.5% (at 43.4W)||56W||£5.21|
|Corsair CX400W||73.2% (at 43.2W)||59W||£2.35|
|Sparkle Power SPI220LE FlexATX 220W||80.5% (at 42.4w)||53W||£7.02|
|HuntKey Jumper R90 300W 80+ Gold||85% (at 30w)||51W||£9.50|
Conclusion: For a low power system it is cheaper, and probably more environmentally sound to keep (or re-use) an existing power supply than to replace it with a high power model, even if the new model is very efficient.
Corrected a calculation error and added the HuntKey Jumper R90. Even a very good value, highly efficient PSU can’t change the conclusion much in this scenario — the savings are just far too small to justify the cost (and environmental cost) of a new PSU. The HuntKey would take in excess of 3 years to pay for itself, and this is in an ideal scenario, compared to a a worth case scenario PSU! For an even lower power system (e.g. a AMD E350) the savings would be lower still. Even a “perfect” PSU would only net savings of around £16 a year!