Testing (continued) 

Just to show off the basics of the Windows version of the software, I plugged the test system into the UPS but ran the USB cable to another computer running Windows XP Professional.  The screenshot below shows the test system's data as monitored from this other computer.

The Windows software has a rather basic (read old school) interface, but it is functional and provides alot of information and other features that are accessible from the various tabs.  It provides live monitoring, data logging, diagnostics, and a few ways to send alerts regarding power events that you may want to react to.  It may not be pretty, but it works, which can't be said for the Linux version.

Now it was time to test the UPS.  The first step I took was to leave it plugged in for a total of 48 hours, just to make sure the batteries were fully charged.  I then connected the computer and monitor power cords to two of the battery backed up outlets on the UPS and turned everything on.  I then let the system idle (no applications running, no screensaver) for another 12 hours before getting down to business. With a stop watch in hand I pulled the plug on the UPS and timed how long it took for the system to power off.  I then repeated the test a few days later after the battery had been allowed to charge for another 48 hours.  The results are inconsistent, and as follows:

• Time to Shut Down - Test #1: 14 minutes 55 seconds
• Time to Shut Down - Test #2: 11 minutes 49 seconds

Both tests prove that I would have plenty of time to react to the problem, and that I should be able to shut down safely without having to scramble.  Based on data collected from various online UPS sizing calculators, even the longest time of 14:55 seems well less than expected.  Although Ultra Products does not offer such an online calculator, other manufacturers indicate that this sort of performance is more indicative of a 400-625 VA unit, and not a 1025 VA unit.  The significant drop in uptime between tests 1 and 2 is also of concern, as we are talking about over 3 minutes less until the system powers off.  The table below is taken from a reference document at Tiger Direct, and shows that the load of the test system should have been sustained for 75 minutes or so!

With a Windows based computer you could setup the software to safely shut down the system before the battery died completely, but that was not an option here.  Linux users may want to conduct this type of test just to see how long they have before their system will power off, so that when it really matters they will not lose any unsaved data, etc.