Give Your UPS Equipment the Acid Test

Recently I had the great fortune (and misfortune) of replacing my UPS (uninterruptable power supply) batteries. My two UPS units are familiar brands. One is used for a server and the other is for my video editing equipment. One is rated for 30A at 120V and the other is 20A at 120V. The rated operating time is where the dirty details come into play. My 20A unit has a rated run-time of 15 minutes with a load drawing 1000 watts. I was getting nowhere near that amount of time. I wanted to test the old batteries and then check out the new, replacement ones. I could not test the old batteries in my 30A unit, because its batteries were open circuit.

There are two standard types of battery tests, a performance or State of Charge (SoC) and the State of Health (SoH). The SoC compares published data in terms of A-h (ampere-hour) of capacity to the actual capacity. This test starts with isolating a single cell or a group of cells (i.e., a battery). You then connect the cell or battery to a load-box (a bank of calibrated resistors) that can dissipate the expected power. The intent here is to discharge the cell or battery to a specific level and measure the elapsed time. Specifically, this discharge test drops the charge to a Depth of Discharge (DoD) level of greater than 90%. This provides a reasonable and reliable check of performance, but only at the time of test.

An alternative to the full DoD is a partial discharge using the system as a load, not an external load resistor. This system load is a more accurate run-time test. This allows the batteries to support the entire actual load, so we'll call this a real-life test. Performance is measured by the published run-time chart or graph. Capacity can be estimated by extrapolation if you know the size of the actual load. The highest accuracy of the test is when the DoD discharge is close to 100%.

A second type of test is the State of Health (SoH). Fault modes such as sulfation, chemistry dry-out, operating temperatures outside of permissible range, and incorrect float charge will affect batteries negatively. Electrical capacity of SoH will have a baseline from ohmic measurements. We will go over ohmic testing in the next article. For now, you just need to know that an ohmic test is an impedance (or conductance) derived from an AC signal driven into the battery; or the internal resistance derived from a DC technique.

This capacity test will detect when the SoH has declined to the point of near-end of battery life. It does not show how much life is remaining until enough testing is done to show a trend. What we need is a test that changes over the life of a battery. Ohmic might be the answer. The problem is that a battery is a non-linear device that is elegant in one manner and clumsy in others. Mixing chemistry with electricity can be most difficult to predict.

Resolution of the testing meters and impedance of test load (if used) will all affect test outcomes. A method of testing may require several readings. Even with this care in the techniques used, test readings may not match up with one another; or from battery to battery; or with just one battery but on different days. Your baseline results may vary.

What I have learned about batteries is that testing for SoC, SoH, and DoD are the present methods for determining battery condition. All tests need to be performed and a lot of time dedicated to meter reading, both before and after discharges. You must pay careful attention to the loads and to equilibrium of battery cells.

Of course, another approach is to look at the manufacturer's specifications. Print out the run-time graph or chart. See if the manufacturer provides a half-life spec. Test your batteries based on this information. My UPS has a hot-swap battery compartment so I can test individual batteries or replace the entire battery compartment without powering down. If I have less than half of the run-time available, it is time to change out. Batteries are cheap compared to test equipment, servers, and the expense of having a long-duration test crash before you get your results.

What testing methods have you used to measure your batteries? Do you have any horror stories related to battery tests?

Related posts:

4 comments on “Give Your UPS Equipment the Acid Test

  1. eafpres
    December 29, 2013

    Hi Peter–back when The Connecting Edge was still up and running, we talked about some aspects of this.  In one example I did some empirical observations and some rough calculatons regarding jumper cables.  For large-capacity lead-acid batteries, the ohmic resistance of the load can have a significant contribution from the cables used to connect said load.  This can distort your measurements, as well as causing other unpleasantness such as melting, arcing, fire, burns, and other so-called “neagative outcomes”.  

    I agree with you that, say, testing over some time frame with a high resistance load may not tell you how well the things work pulling full load for 10 minutes of desperate shut down window.  On the other hand, drawing full load for any length of time into a test circuit can be tricky for the above reasons among others.

    In the mobile phone world, and before that in the mobile radio world, smart circuits and continuous monitoring of batteries has been standard for a long time.  I have not looked into it but is it common for UPS systems not to provide real-time battery status?  I would think at least high-end systems would operate with at least two banks of cells and be optimally discharging one then recharging and vice-versa, to maintain cell optimal life.  Such as cycle would provide ample data on health.

  2. Victor Lorenzo
    December 30, 2013

    Once in a road my father's car engine just stopped, it was a '52 Crestline Ford, a rainy night and no one else in the road.

    While my father was going for his lamp a frind of his that was coming with us declared himself in charge of testing the battery, took a wire and shortcircuited the battery terminals, “to see if it was charged, he said”. The fact is that part of the Pb battery terminal melted and the guy was not able to remove the shortcircuit, the wire got red, really hot and really red, and the plastic cover from the wire ignited. Fortunately the wire was thin enough to melt in a few, but almost eternal, seconds.

    The failure was simply due to the ignition coil.

    Lesson learned!

  3. vbiancomano
    December 30, 2013

    One path I do not recommend is relying on the UPS' “self test” mode to ascertain how much sock the battery has left, especially as the battery reaches its “last quartile.” Over many years in one of those environments where the AC-mains infrastructure seems to look for any excuse to fail, I've seen that battery-test algorithms are frequently not sufficiently sensitive to determine that the UPS can suffer an AC-mains glitch, let alone a total loss of AC mains, without taking the computer system down with it—-thus defeating the purpose of why the UPS is there in the first place. While I haven't replaced a UPS with a newer model in five years or used any other systems having a UPS that's less than 5 years old,  I believe this is one forgotten area where “self-test” algorithms need to be much better.

  4. David Ashton
    January 9, 2014

    Some years ago my employer had quite a few small (<1KVA) UPSs around the place and I got the job of maintaining them.  I built a UPS Tester with a counter and a small circuit that reduced the 50Hz mains frequency to 1 pulse per 6 seconds and sent it to a counter.  You reset it when you removed the mains feed to the UPS and the counter would count (in tenths of minutes) until the UPS stopped.  I had 6 100W incandescent light bulbs with switches as a load - yes they have heavy inrush current but so do PCs and other real-world loads.  It worked a treat.  I still use it occasionally - the other day I tested a UPS that I had last seen a few years ago and - from the label I put on it then - the unit still had about 2/3 of its previous capacity.

    Some APC 1KVA units had a steel frame and the batteries used to expand a bit when they got old, and were very difficult to extract to replace,  That good old all-purpose electronic fault-finding tool, the hammer, came into play a few times.

    The APC units were very fussy about the batteries used,  Seemingly identical batteries would gas and leak if used in APC units, I had to order the batteries from APC to do a successful replacement.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.