CyberPower UPS - bad?

All other things being equal, I'd spend a bit more and get a LiFePO4-based device due to much longer anticipated life time since they tolerate being deeply discharged much better than NMC cells. Wirecutter's reviewers usually only review on a very superficial level - I'd consider underlying battery chemistry to be the most critical factor when reviewing battery power stations, but they don't even mention battery chemistry in their reviews.

Jackery uses LI NMC cells.
https://www.jackery.com/pages/why-choose-us
"Equipped with lithium Nickel Manganese Cobalt Oxide (NMC) batteries,the same type of battery that Tesla uses in their Powerwall units. It provides high energy density and long cycle life."

Anker has a mix of NMC and LiFePO4, and they warranty them 18mos or 5y, respectively, depending on battery type. In some cases different capacities within the same model series can use different chemistries, which is not confusing at all.
https://support.anker.com/s/article/Anker-Portable-Power-Station-Battery-Types
"Anker portable power stations have two battery types: LifePO4 batteries and MCN (Nickel Manganese Cobalt) batteries.
Models built with LifePO4 batteries come with a 5-year warranty, while models built with MCN (Nickel Manganese Cobalt) batteries come with an 18-month warranty."


Ecoflow Delta power stations also mostly use LiFePO4, but some some use NiCad.
https://blog.ecoflow.com/us/upgrade-ecoflow-battery-capacity
"The EcoFlow DELTA 2 and DELTA Pro have a respective Smart Extra Battery that utilizes LiFePO4 (LFP) technology. The RIVER 2 portable power station series also runs on LiFePO4 (LFP) batteries. The DELTA Max and Smart Extra Battery use Nickel Cadmium (Ni-Cad) battery chemistry."

EDIT: edited for clarity

Meanwhile at Walmart: EF EcoFlow DELTA 1300 Portable Power Station 2 Units,1260Wh Capacity,Solar Generator,1800W AC Output for Outdoor Camping,Home Backup,Emergency,RV,off-Grid
$1,500.00 and this "spec":



Is 30ms considered continuous?

Auguste_Fivaz said:

Is 30ms considered continuous?

Click to expand...

No.

ATX spec hold-up time is 16.66ms (17ms), so 30ms hand-off time would be seen as a power interruption.

I did some poking around on this because I was curious.

continuum said:

No.

ATX spec hold-up time is 16.66ms (17ms), so 30ms hand-off time would be seen as a power interruption.

How We Test Power Supply Units

Our PSU reviews explore performance, noise and temperature ratings, along with the build quality of every power supply we test. Today we introduce you to our methodology, testing equipment and benchmarks.

www.tomshardware.com

The Importance of Hold Up Time

Many do not know the meaning of hold up time in the electrical specifications of power supplies.

www.fsp-group.com

Click to expand...

That 16.66ms number comes from the 60Hz US power grid frequency - 1 cycle at 6HZ is 16.666...ms, so the idea is that an ATX power supply must maintain at least 90% of it's rated voltage at 100% power draw for 1 AC cycle.
... implies that 16ms is a guideline accepted globally for common power supplies (not just PC ATX), even though most of the rest of the world uses 50Hz (20ms period) grid frequency.

I spot-checked a few UPS devices, and I'm seeing transfer times of 6ms-12ms for various APC models and <10ms for a couple of smaller Eatons I looked at. Looks like the intent is to stay well under that 16.6ms hold-up time.

Looks like 30ms is really insufficient unless you are using it with portable devices anyway.

Some actual testing, lower-end PSU’s often don’t even make the ATX spec of 16.6ms, many fall quite short at around 11ms:

Higher-end units do better but still mostly short of 25ms:

so confirms what @Xelas concluded.

I'm not sure the Jackery units are designed for that kind of duty cycle. I'd ask their support dep't if I were you.
They're fairly dependable.

continuum said:

Some actual testing, lower-end PSU’s often don’t even make the ATX spec of 16.6ms, many fall quite short at around 11ms:

Asus TUF Gaming 550W Bronze Power Supply Review

The Asus TUF Gaming 550W is a good-performance budget power supply.

www.tomshardware.com

Higher-end units do better but still mostly short of 25ms:

Corsair HX1000i Power Supply Review

The Corsair HX1000i has silent operation and top-notch build quality.

www.tomshardware.com

so confirms what @Xelas concluded.

Click to expand...

Right, some equipment can't even handle that 10-15ms transfer time. That's why (more expensive) online double-conversion UPSes are thing (which have no transfer time), so that more "sensitive" equipment doesn't power cycle during interruptions (plus the power is always conditioned).

It's always appealing to look at much cheaper (for a given battery capacity) consumer gear and wonder if you can get away with it. There's a reason why proper UPSes are a thing, however.

If you don't actually mind if the attached equipment power cycles during a power interruption, though, there's no reason you could use one of those EcoFlows or equivalents.3

I'll also second what Xelas said about making sure you're getting LiFePO... but add that anyone managing UPSes in general should be looking at that rather than continuing to rely on VRLA-based systems. The battery should last as long (or longer) than the electronics in the unit itself, so labor is greatly reduced over having to swap out heavy lead-acids 2 to 3 times over the life of the UPS. Not to mention much less chance of the battery failing or not living up to its promised run-time at inopportune moments. The up front cost is, of course, higher, but the TCO can be equivalent or even better. The only issue I've run into is that APC's lithium-ion offerings are overpriced and lackluster compared to the competition. Hence why I'm moving my org to Vertiv Liebert.

In the last 2 weeks, 2 Cyberpower UPS devices experienced utility power failures. Because of bad batteries (both over 3 years, not totally sure of the age), the devices did not restart when power was restored. And I didn't check if these would have restarted when power waa restored if the batteries had been good and simply ran out.

This is AWFUL design! No matter what the state of the batteries, at the very least, when a unit's power is restored, the device should restart AND give an audible indication that the batteries need replacement (if needed).

In an ideal world, you would also be notified in advance that the batteries are bad. Alas, this is more difficult to do. So the words to the wise are: Stress test your UPS batteries regularly, no matter what brand UPS, and find a device that will auto-restart in the event that batteries don't hold a charge.

Kyuu said:

I am migrating my organization to Vertiv Liebert models for more critical infrastructure (specifically their lithium-ion offerings). While obviously I don't yet have good historical data for their performance, things are going well so far

Click to expand...

How are those units holding up for you? We’ve picked up a few as well, and have run into a few firmware issues where the unit goes online, fan kicks on and stays on, even at no load and good, stable input voltage. Leaving affected unit turned off and unplugged for 12 plus hours two or three times seems to help. Vertiv Liebert support is aware but hasn’t been able to do anything about it yet, and the problem is sufficiently intermittent thar troubleshooting it has been painful if not impossible.

(and just to be clear other UPS’s off the same breaker panel behave just fine, it’s just some of our Vertiv Liebert ones)

As for those talking about using the "power stations" as a UPS solution: by design, these are not made for continuous use. They are also by design, made to charge slower on input than output (taking in say 200w, while giving the ability to go up to 2000w out) and do not simply straight pass power. Thus, you may drain these dead before the recharge and force them into a strained cycling. Such use case situations (using the units as attached powered UPS's instead of temporary off-grid power) are often excluded from warranties, so buyer/user beware.

@continuum Which model number are you using exactly? My fleet is almost entirely the PSI5-1500RT120LI. I also have a couple GTX5LI-1500LVRT2UXL where we need external battery cabinets for additional runtime.

I haven't encountered the issue you describe with either unit, at least not that I've noticed. I did have issues with the first GTX5LI we tried to install where it wouldn't detect any voltage from the external battery cabinet and had to get multiple replacement cabinets and a replacement UPS to sort it out. That turned out to be an issue of mismatched firmwares between the battery cabinet and the UPS.

We have a mix of the PSI5-1500RT120LI and PSI5-1500MT120LI. I haven't personally handled the support calls on the RT units but I have on the MT units. We've, like you, also had a few defective ones. Can't tell for sure if DOA rate is higher than APC, Eaton, etc. or if just a victim of small sample size so far, but the units were all drop-shipped direct from Vertiv Liebert...

Just to clarify, I haven't had any defective units. The issue was only with the GTX model, and only with attaching the external battery cabinet, and as I said I eventually determined that was due to mismatched firmware between the UPS and the battery cabinet.

Granted, we don't have a huge fleet of them yet as they only get installed as our existing APC units age out. Next month we'll be ordering another 20 or so for EOL replacements, so my sample size will get a bit larger.

Let us know how it goes, a fleet size of 20+ is getting decent. Since I don't have full visibility into our deployment, you might have the best data out there.

Do any of you monitor the status of a large number of UPSs? If so, how? Is there some sort of standard protocol? Syslog? Proprietary shit?

I inherited monitoring of the APC fleet with using software called StruxureWare. For the Vertiv models, I'm using the same monitoring solution we use for all our other devices: Zabbix. Zabbix just uses SNMP and ICMP.

I think we currently have about 15 of the Vertivs, so I'll be in the mid-30s once we're done deploying the next shipment. Every fiscal year we're going to order right around 20 for EOL replacement, and I believe our total number of UPSes in service is right around 100. You can see why not having to replace batteries every few years and the reduced weight is a big plus for me.

Just SNMP here.

Thanks, guys. How well does SNMP actually work?
Reason I'm asking is that we have ~6k locations. The locations are all totally unique in their layouts and requirements. We have access to conditioned and redundant power in a fabulously clean IDF at some locations, or we may have a worn-out outlet with a bad ground at a site that loses power for a couple hours twice a week and the "IDF" is a half-rotten piece of plywood on a wall behind a leaky boiler (thankfully rarely, but I do have a few of those). We do NOT have any influence over the locations. We're slowly working on figuring out an approach to get a better handle on how we monitor deployed devices. The network equipment itself is more-or-less a solved problem, but now we're getting into other stuff like UPS devices.
A site may have have anywhere from zero to 2-3 UPS devices. My best guess (and I would probably know better than anyone else in our corp) is we have a few hundred UPS devices at a few hundred sites. The sites don't have a VLAN back to our NOC, so securely managing SNMP is not currently possible, but it sounds like we'll need to get this set up at some point. We do NOT have any servers at any sites, so there is no good place to run a local SNTM agent.
Managing thousands of VPN tunnels is a total PITA, though. We have some network equipment that has IPSEC tunnels back to an MSP, and even the limited ~300-400 devices that are set up that way have 10-20 devices that are offline at any given point for one reason or another and it's a full-time job to manage that. I can't begin to imagine how we would manage thousands.