Why would a standby UPS fail to power devices when there's no power outage?
I have an ordinary consumer-grade standby UPS with, I believe, ample power for the devices connected to it.[1] It's five years old. For the second time in about a month, I've found my computer powered off and the UPS apparently without power (indicator light was off). Both times, I pressed the power button until the UPS's light came on, then restarted the computer and carried on without apparent problems. In neither case was there a power outage or discernible power disruption (nothing else in the house, including another UPS, had problems).
My first thought was that the UPS's battery had failed (I know they are not immortal), but then I read up on the difference between standby and line-interactive UPSs and found that my mental model was wrong. I thought the house current fed the UPS which then fed the computer, and if the battery in between was bad that would cause problems downstream. According to several articles I read (example), that's actually how a line-interactive UPS works and a standby UPS, in contrast, switches to the battery during a power outage but otherwise just passes house current through to the devices plugged into it. If that is the case, then even if the UPS's battery is failing, I don't understand why that would matter when the house power isn't out.
In the end I want to ensure that I have reliable power to my computer, which might mean replacing this UPS, and I almost asked this question on Power Users, but I'm asking on Electrical Engineering because I'd like to understand how a UPS works and why I'm seeing this behavior. What could cause a UPS to shut itself down and stop passing power through? For example, if it can't keep the battery charged, does a UPS shut down entirely rather than lead you to believe you have backup power? (If so, is that universal, or is it something I can screen for in my next UPS?) Are there diagnostics I could do as an ordinary user without any special electrical-testing tools to understand what's happening?
-
The UPS spec says 800VA 450W. It's powering a Mac Mini, a monitor (that sleeps when not in use), one external hard drive (Time Machine), a USB hub (keyboard, headset), a network hub, and an infrequently-used printer. This was the biggest UPS I found on the ordinary consumer market at the time I bought it (2018). I don't understand enough about voltage and wattage to evaluate this collection of stuff against this UPS, so maybe I am in fact overloading it? ↩︎
4 answers
Started writing a comprehensive answer, interrupted by some phone calls, finally finished it...and then accidentally clicked to another page and lost it. Arghh!!! So this answer may not be quite as complete/well-written.
TL;DR Normal for a UPS with a Low Battery - Replace the Battery or the UPS
A typical uninterruptible power supply (UPS), unless it is a true dual conversion UPS often has this problem as the battery ages. Note that age is primarily "charge cycles" and not calendar time.
A typical UPS works in two modes:
- Good AC Power
UPS filters (basically the same as a good-quality surge protector) incoming power and passes it on the powered equipment. At the same time, the UPS monitors the battery (meaning: checks voltage) and charges it as needed.
- Bad AC Power
This includes:
- Low voltage (a.k.a., brownout)
- High voltage (unusual in most places except if you have a serious utility or electrical system problem such as a lost neutral)
- Power outage
When any of these problems is detected, the UPS disconnects the powered equipment from incoming AC power and turns on an inverter and other circuitry to draw power from the battery (typically 12V DC) and convert it to 120V AC power for the equipment. This switchover happens very fast - typically 4 ms to 12 ms. Compare that to a single AC cycle in the US (60 Hz) of 16 ms. The powered equipment doesn't know anything has happened.
The problem is when the battery starts to wear out. Despite the metaphors used when describing battery level, it is not a simple level like checking the amount of water (or gas or oil or whatever) in a tank. Battery charge is checked by reading the voltage. It isn't 12V = 100%, 6V = 50%, 0V = 0%. In fact, the range tends to be rather small.
For example, this article and chart shows the numbers for lead-acid batteries, with a range of 11.63V to 12.89V. Lead-acid batteries are quite common in consumer-grade UPSes. Lithium-ion batteries have a different scale, but the same general concept applies.
A good UPS will tell you when the battery can't hold enough of a charge any more, with some sort of indicator light or other display or beeping. However, it is actually hard to get this right, so sometimes the problem isn't found until it is too late. People also tend to not like beeping, so many UPSes deliberately do not beep to tell you there is a low battery.
If the battery is low, whether you (or the UPS control circuitry) "knows" about that or not, when it tries to switch and draw full power from the battery to convert to AC to power your equipment, it will fail very quickly. The result is exactly the problem you have found. It is actually worse than no UPS at all! Why?
If you have no UPS then the computer (or router or whatever) will rely on its own power supply to ride through very short outages. A typical example is when a thunderstorm comes through and the lights blink for a fraction of a second. The UPS will attempt to switch as fast as possible - less than one AC cycle, but then fail due to insufficient battery power. On the other hand, a computer power supply is typically designed to handle at least one cycle, often more, before causing a reboot or other problems. At least in my area a storm will often cause power outages that are so short that many (but usually not all) digital clocks (microwave, clock radio, etc.) will not even need to be reset, even though there was a visible outage for a fraction of a second.
Brownouts are another issue, in two ways. Many computer power supplies can now handle a wide voltage range - e.g., 100V to 250V for a single design to work worldwide. A UPS will typically treat anything outside a relatively narrow range - e.g., 105V - 125V - as a problem, so a dip to 102V might cause a problem for the UPS but not cause a problem for a computer connected directly to AC power. In addition, large loads such as a laser printer printing or a refrigerator compressor startup can cause enough of a voltage drop that a UPS will detect a brownout, and these repeated UPS cycles can wear out a battery very quickly. I had problems of that sort until I finally had my electrical service upgraded a year ago, which made it very easy for my electrician to split the main basement circuit - previously there just weren't any spare circuits.One circuit was simply running so much equipment that the critical loads on a UPS were subject (every couple of years, now hopefully longer) to the UPS battery wearing out due to frequent cycling.
A note about UPS/battery replacement. Unlike cell phones, laptops and many other devices, most UPSes have easily replaceable batteries. There is a good reason for that. Cell phones, laptops and similar devices are as likely to have other problems (physical damage, other components wearing out) or simply become functionally obsolete before the batteries wear out. Manufacturers like non-user-replaceable batteries because it encourages device replacement rather than battery replacement, but much of the time the user replaces the device anyway, even if the battery is perfectly good.
But with UPSes, it is quite common for a good UPS to last through 2 or 3 or even more battery replacements. The rest of the UPS is basically a surge protector, charge controller and an inverter - relatively simple components with no operating system upgrades or cell phone network changes to worry about. You don't have to get the battery from the UPS manufacturer. I generally go to Batteries Plus, formerly Batteries and Bulbs but there are other retailers as well. Be careful with mail-order for batteries, as returning the old batteries for recycling is much more of a hassle, though at least in my area Batteries Plus will take batteries for recycling even if you don't buy new batteries there, and I often bring them batteries from old UPSes if, for example, I replaced the UPS rather than just replaced the batteries.
1 comment thread
This actually started out as a comment, but I think it deserves an answer of its own even though it's more about whole-device behavior than electrical engineering per se.
I don't know if your particular UPS does it (it probably does), but a good UPS will occasionally do a self-test. Typically this will involve switching to battery power for a while even though mains power is considered to be of adequate quality, and in one way or another monitoring the discharge characteristics of the battery to make sure the battery's response is within acceptable limits.
You mention that you have occasional mains power failures, on the order of several times per year for more than a few minutes. A very rough estimate of the power draw on your UPS would be on the order of maybe 100-200 W typical, and you mention that your UPS is specified to be able to handle a load of 450 W. A typical consumer UPS might be able to run for 10-15 minutes at load capacity. The upshot of all this is that in your situation the battery would be discharged at a reasonably high rate while there is a mains power outage.
When the UPS switches to battery power and starts drawing power from the battery, the battery terminal voltage will drop. This in itself is entirely normal; every battery has some non-zero internal resistance, and this internal resistance will manifest itself as a voltage drop while current is being drawn from the battery, increasing as more current is being drawn. Therefore any given battery will have some maximum current that can be drawn from it while keeping the terminal voltage above some given value.
However, if the UPS switches to battery power but the battery, being marginal, can't keep up with the discharge rate, then the battery voltage will drop below the requirements of the UPS' inverter, which is the part which takes the DC power from the battery and turns it into AC power for the outlets. That in turn will cause the inverter to shut off, either because of built-in protection circuitry or simply because it cannot work with the too-low voltage. (In a good UPS, you'd expect it to be the former.)
The result: apparently random power failures on the nominally power failure-protected outlets of the UPS in the presence of normal AC mains power.
Therefore, apparently random power loss on equipment protected by a battery-powered standby UPS is a good indication that the battery is overdue for replacement.
0 comment threads
Now that you've supplied a link to the product, I can provide a different take on this affair. Your "UPS" is an Amazon Basics model described at https://www.amazon.com/gp/product/B073Q3BSPG/?th=1.
First, it's clear that unit is optimized for one parameter only, which is minimum possible price while still getting enough ordinary unsophisticated consumers to buy it.
The sales blurb does mention "battery backup", but it seems impossible to nail that down to any quantifiable specs. They claim (at this point I wouldn't be too surprised if some of these claim are flat-out fabrications) to be able to provide 450 W. That's a reasonable number and sounds nice, but the all-important spec of how long it can provide that when the power goes away is completely missing. It might only be able to hold up the output for 2 seconds, and you wouldn't be able to complain it wasn't working.
If you scroll down that page past a bunch of other stuff they're trying to sell you, you eventually get to a section "Compare with similar items". Note that they won't tell you the runtime, even though they do for other models!
In other words, the run time is so bad that they don't want to admit what it is. This thing has the cheapest possible battery they could find that holds together just long enough to arrive intact and look like it's working for a while.
I got curious, so tried to dig into this more. Scrolling even further down the page you get to "Product guides and documents". The "product documention" just describes the warranty. Clicking on "User guide" lets you download a ZIP file. However, the ZIP archive contains only one file, and that's an EXE, not a PDF despite what the link says. No, I'm not going to run a mislabeled EXE from a questionable company, so I deleted it.
The "User manual" did turn out to be a PDF. I can see why it's hidden way far from where they try to sell you the unit. There are things in there they don't want you to see before buying it. Point 1 under "Basic Operation" has an interesting disclaimer in bold:
There are a few more gems in the warnings section:
I didn't look up those standards, but they clearly know this unit doesn't provide the performance required by those who have looked into such things.
In other words, "We cut as many corners as possible to make this thing cheap, because we're aiming for a market where customers buy on price. As a result, it's so flimsy you can't even take it for a car ride without something breaking. We figure enough will arrive working anyway after shipment to not look too bad. Besides, we plan on being outta here and on to the next scam by the time these things all blow up in the field".
There there is this gem:
They warn you to replace the battery only with one of a suitable recommended type. OK so far, but nothing says what the recommended type is, let alone any spec at all about the battery! Clearly nobody was ever intended to actually read this manual. It's just there to make it look like there's a manual and allow someone to check off a box.
The Technical Specifications section provides some interesting information (to the extent you can believe any of it):
Note the brownout transfer of 96 V, and the transfer time of 8 ms. This corroborates what manassehkatz speculated. The unit will switch to battery operation when the line power drops to 96 V, and it will switch over in 8 ms once it decides to. 8 ms is half a line cycle at 60 Hz. Most appliances can ride out 2-3 line cycles of drop out without incident.
It's reasonable for a UPS to be more twitchy in deciding when to switch over. However, that does make you more vulnerable to a bad battery because the unit will switch over in many cases where the bare appliance would have been OK.
All in all, this unit is a piece of crap, with a battery to match. It was deliberately designed to be as cheap as possible and be sold to people that can't tell the difference. Amazon should be ashamed of themselves, but they long ago gave up selling only reputable products.
I'm no longer surprised the battery only lasted five years. The best you can do at this point is to carefully open it up (unplugged and off first!), see what battery is in there, and rig up a way to connect a roughly equivalent off the shelf sealed lead acid battery.
Sorry for the bad news, but it is what it is.
I see that manassehkatz has already given you a good answer and plausible explanation for what you observed, so I'll fill in a few other points.
Your battery is 5 years old, so is getting to the point where it could start to fail. I wouldn't expect a 5 year old battery to be bad, but it's to the point where it's plausible. If the battery were 10 years old, I'd just replace it immediately, then see what the symptoms are.
Don't compare UPS batteries with car batteries, even though both are probably lead-acid (in a consumer grade UPS). Cars are very harsh environments for batteries. Car batteries have a lot of robustness built in. This results in a much heavier and larger battery for the same amount of energy storage, but you get reliability in return. They are also over-specified so that they can still do their job with some degradation.
Your UPS battery is probably a sealed lead-acid type, probably 12 V, but I've seen 24 V types used too. These are designed for storage efficiency, deep-cycle ability, and no chemical leakage. They are not designed to handle much vibration or wide ambient temperature swings. As a result, they are physically more delicate.
Individual lead-acid cells produce about 2.1 V when fully charged. A "12 V" battery actually has six cells in series. These cells are packaged together into this unit we call a "battery", but are really independent electrically. A plausible failure mode is that one cell went short inside the battery. This can happen simply over time due to stuff migrating, dendrites growing, or a physical failure. Again, car batteries are physically beefier, so you don't usually see this kind of failure before other effects make the battery no good anymore.
If one cell out of six is shorted, the battery voltage will only be 10.5 V instead of 12.6 like the UPS expects. That's enough of a difference that the UPS would either consider the battery "bad" or too discharged to use, especially when the UPS knows the battery should be fully charged.
Your UPS manufacturer will be more than happy to sell you a replacement battery, at a hefty markup for the convenience and for people that don't feel comfortable doing this themselves. I just looked, and Mouser lists 836 different varieties of sealed lead-acid batteries (https://www.mouser.com/c/power/batteries/sealed-lead-acid-battery). Of course, there it's up to you to know what you need. You need to know the voltage, roughly the capacity in amp-hours, maximum size, and physical connection type.
If you're not comfortable with all that (or can't find someone to help you who is), just get the replacement from the UPS manufacturer, and consider it the cost of having a UPS every 5 years.
If you're more adventuresome, consider that the new battery doesn't have to go in the same physical place the old one did. You can extend the cables to the battery outside the UPS box and have the new battery sitting next to the UPS instead of being nicely tucked inside. That's fine electrically, as long as the cables are thick enough and not too long. I'm envisioning a 1-2 foot extension, just enough to place the battery next to the UPS.
Now you only need to know the voltage and capacity. If you're lucky, the old battery has its spec printed on it, or there is a plaque on the UPS or you can find it in the UPS documentation. The voltage needs to match exactly. Fortunately, it's probably either "12 V" or "24 V". There is some leeway in capacity. I'd say ±50% is OK. Capacity directly relates to run time.
You can estimate capacity from other specs. Let's say you found the battery needs to be 12 V. You have already said the UPS is rated for 450 W, but you also need to know the run time at that 450 W. Let's say for sake of example that the run time is rated for 15 minutes at 450 W.
Ultimately, the battery provides a fixed amount of energy, which the inverter in the UPS converts from 12 VDC to 115 VAC. Energy is power times time. If the inverter is putting out 450 W for 15 minutes, then the total energy delivered is 113 Wh (Watt-hours). There will be some loss in the inverter. For a consumer-grade UPS, let's figure the inverter is 85% efficient. That means the battery has to cough up 132 Wh.
The battery has to be able to hold 132 Wh in this example. However, that's not how batteries are sold. You need to convert this to capacity, which will be listed in Ah (Amp-hours). You do this by dividing the Wh by the battery voltage. (132 Wh)/(12 V) = 11 Ah. So in this example, you're looking for a 12 V sealed lead-acid battery that has 11 Ah capacity. Anything in the 10-15 Ah range should be OK with your UPS. 10 Ah will reduce the run time a bit. Higher capacity will give you more run time, but if you go too high the UPS may think something is wrong when it can't charge the battery as quickly as expected.
4 comment threads