Recently I migrated my home server (a Linksys NSLU2 a.k.a. ‘Slug’) to a more powerful device (Mac Mini). The Mini has more capabilities and hence will run more services. Time to invest in a UPS that will keep the computer running when the electricity grid goes down.
Since the power consumption of my home server is a major concern to me, I’m also interested in the extra power consumption generated by the UPS.
I was able to measure the power consumption of a MGE Ellipse 750. Power was measured with a Voltcraft Energy Check 3000 and the test load was a normal PC (note: not the Mac Mini!).
The result was: the setup consumes 76 Watt without UPS and 80.5 Watt with the UPS. Note that the batteries of the UPS were fully loaded when the measurement was done.
3 responses to “UPS power consumption”
Ouch, 4.5 Watts lost. Was that power usage continues?
You should check if part of the extra power usage is not a result of a “correction effect” of the power usage of the PC by the UPS though.
Depending on the design the UPS might in fact improve the power factor (cosinus-phi) of the PC’s power supply. So you might actually finally be measuring the PC’s power usage better then before.
I guess it also depends on the accuracy of the Voltcraft device you are using.
Try with a purely restive load like a light bulb. That is if you have such outdated tech handy 😉
Hey Bart,
yes, the power usage was continuous.
It is clear that the measurement I did was not completely accurate. The measurement device is a (very) simple one, and there could indeed be influence of the power factor. I just wanted to get an idea on the impact of adding a UPS to my server setup.
Currently I cannot rerun the test with another load, as the UPS is not mine (I quickly borrowed it for this test). However, I’m convinced that adding the UPS is not ‘too bad’ so I will buy one and then I’ll run the resistive-load test.
Update: I ran the resistive-load test with a incandescent lamp of 50 W. This time with an Ellipse 600 ASR.
Without UPS: 49 W
With UPS: 54 W
So apparently the original measurement was already rather accurate.