I worked a battery backup Y2K project back in '99. Covering the server room and close to 50 bank branches, and never learned as much about UPS's as I did in the last 20 minutes. Love going to your school Professor Dave.
My dad volunteered to be on call that New Year's Eve. He was convinced nothing bad would happen, and he said he could charge as much as he wanted. I don't know how much he ended up charging, but as expected, nothing did happen, at least not on his watch.
@@aaronperl some things did happen but it was companies that completely ignored patching the issue beforehand and then acted surprised the thing they’d been warned to patch happened. Personally i never got why in a lot of cases it would even cause any long them issues as its not like most servers had data under 1901. Though anything that referenced back to the previous years may or may not work depending if it would buffer overflow the other direction. And honestly if they worried they could have always run a test server that they’ve manually clocked forward to just before 2000.
Dave, I'm a generator guy among other things. It seems like you once mentioned that you were running a 70kw backup genset. If you haven't already, do a "lights out" test. Kill your utility and let the generator run the house for a few hours or half a day. The UPS's are kicking off some really wild power factor. The old rule of thumb was that the genset needed to be 5x the total UPS load to handle the pf. I've had it bite me a few times. Maybe your generator guy knew how much UPS load you have, maybe not, maybe he sized things ok, maybe not. Either way, test run everything before you have to rely on it.
Interesting. Inrush currents can be insane. I recall hearing when switching power supplies started to become the norm it created massive issues during brownouts. In the old days when your house mains dipped things just shutoff; when it returned they fired back up. But with switching supplies which have very wide operating voltages when you dip due to a brown out the supply compensates by pulling more current. So the brownout goes into avalanche mode as thousands of houses just pull more and more from the grid rather than backing off nicely.
@@dennisfahey2379 Most high-draw appliances nowadays will have a delayed startup time after an outage. Those brownouts will wreck an AC or refrigerator compressor. For older appliances you can buy wall plugs that wait for things to stabilize before powering the appliance and they also shut off if there is a brownout.
Great rundown Dave! In the telecommunications industry we never loaded a UPS or a generator more that 80% of its nameplate kVA. So your 1200W (1500VA) workstation would require a UPS rated at1875 kVA. VERY important to know your loads. Super important point!
Shame... With computer engineering any computer not running at 100% all the time is considered wasted resources... Should try a real man's field sometime
@@craigpeacock1903... I imagine that the official computer spec normally already has some built in overhead protection. Meaning they can do more than they say they can do ... So the 80% thing is already in there from the manufacturer. With the power systems, when you hit the written capacity, they immediately flash overload and go off ... Meaning they are honest, and you have to do the 80% thing yourself.
This video is amazingly timely for me. I've been thinking that I should get a UPS for my little home server, just so that I can have it shutdown cleanly (which takes a few minutes, to shutdown all the VMs). My dad worked his whole career in this area, designing large UPS systems like you showed with the rooms full of batteries and diesel generators. He retired about 15 years ago, but got sick with cancer during COVID. I got a chance to ask him about a month ago how I should go about selecting a UPS, and he did seem to perk up as he tried to explain it to me, but he didn't have the energy to really go into it. He passed a week ago, so I only had and incomplete explanation to go on .. until now. He had told me that I first need to find out the power factor, so I've been meaning to pick up a Kill-a-Watt, but haven't done so yet...
Damn that sucks... Sorry for your loss... This video did do a really good job explaining things though without going into marketing crap and other nonsense...
Sorry to hear that about your dad. I lost my mother to COVID. Those remembered conversations and thinking of things you would ask if they were here really lingers
To say that lithium batteries have an environmental advantage because they don't have lead might be a bit off. Lead acid batteries are one of, if not the most recycled consumer good in the world. 99% recycle rate. You can drop off lead acid batteries at virtually any auto parts store. (Some even give a store credit if you're not bringing one back as core for one you bought from them!) Lithium battery recycling, while improving, has a way to go.
The lack of Lithium battery recycling is due to the insane amount of disposable devices you can find them in. People just don't bother. In a UPS with lithium battery packs, when those go bad and get e-wasted, they are VERY likely to end up being recycled.
Lithium battery production is far more polluting than anyone who recommends them seems willing to admit as well. Matter of fact they always seem to completely gloss over (not even mention it at all in in 99% of cases I've seen/read) how dirty actually getting usable lithium is while at the same time criticizing the same aspects in other power sources.
@@JimmyRussle Agreed on the mobile / small devices being the biggest source of "loss" in the lithium recycle circle of life. The thing would be to incentivize turning them in for recycle instead of tossing in the trash. That's what happened with lead acid batteries. They started adding a core charge to new unit purchases, and as I said before, some places offering credit / $$ even if it wasn't a core return for a purchase made at that establishment. Note that I am in favor of lithium batteries as a whole, and do plan on converting my UPSs over to LiFePo4 the next time the lead acid units need replacing. Would have done it the last cycle, but the projected cost per year of service was still too high in comparison to lead acid.
This is probably the best explanation of the various factors affecting UPS selection that I have ever come across, especially concerning Watts and VA, off-line, interactive, etc. The only thing to add (probably a full video on its own) would be a feature on the software options for UPS management (Windows, Linux, environments like Unraid, etc.). For what it's worth, I use my Unraid server for my UPS with the desktop PC as an apcupsd slave.
And this video actually came at the right time as well since i was in the middle of trying to figure out if the issues i have been having with a new switch is a result of it being sensitive to power fluctuations and what type of UPS i would need to get to hopefully stop a switch from resetting every time the ups returns back to main line power...
@Michael Hanson Does the switch have an internal or external power supply? I have replaced a couple of supplies on small switches that simply failed due to age / heat / poor component quality. Since the mains supply can be missing for half a cycle or so when your UPS switches to and from the mains (depending on UPS type), it's possibly the switch's power supply that is at the root of the reset issue you are seeing. I would try an alternative supply first if that's an option.
I have a garbage picked APC 1400 XL. It's been backing me up for 23 years now. I hard wired it into my main panel with permits on my critical circuits. I got rid of the internal 4x7 Ah batteries and have two 100 Ah external batteries in series. I can't believe it's still working perfectly after 25 years.
Depends on the outages here when someone sneezes the grid power goes down and often long enough to kill the batteries in my APC700. But saying that I am just about on my second set or 2x7ah replacements and had them since 2016 so about 3yrs a set. I have thought about putting larger AH batteries in place but in the end bought for fun an MPP Solar PIP5048 and 8 300AH batteries and got the place on a manual transfer switch so when the grid goes out don't have to sit in the dark or go to bed early - with latest outage I realized I don't have a simple method to tell when power is back on as all that's a bit remote so looking to run a mix of critical and non-critical loads. The whole place is transferred to the PIP on manual cut over so you cant use heavy appliances or will trip it but its mostly to keep lights and internet up to ride out the outage.
The following is more info than you need in your video but a few may be interested: Your example of what manufactures call modified sine wave (more like modified square wave) is very generous. Most products with modified sinewave use a square wave with 3 levels: max positive, short time at 0, max negative, then back to 0 for a short time in a way that averages to the usual AC rated voltage. Some may use 5 levels so they have a half step positive & negative added (each used twice per cycle). For a given resistance, the AC RMS voltage is equivalent to the same amount of DC voltage through the same resistance giving the same power flow, but sinewaves make it complicated. AC voltage sinewave means it goes higher than that RMS average for some of the time but the non-RMS average (if you divided cycle time into 12 or more evenly spaced samples) is slightly different to the RMS average.
I live mobile and have a few UPSs to backup and filter power here. Super good information and helped me verify I am doing it correctly. Since my place is essentially a faraday cage I tend not to worry but with Summer coming up I think it best I establish a full grounding of the cage. Thank you, Dave for another solid lesson!
Been working around the APC UPSs for over 20 years now. All that they guarantee is every 3 years your server will lose power when the battery goes dead.
Yeah. I have a low opinion of UPSs. I've only used APC ones, but they seem to cause power problems, and since we almost never lose power they have almost never saved anything. Overall we have had much more problems having UPSs than if we hadn't bothered.
Amen to this comment. Having also worked in it for 23 years I too have experienced more power losses caused by APC UPS failure rather than the grid. YMMV on other brands but APC are not alone in what seems like poor design.
They seriously caused more problems than they solved at my hospital. We abandoned ever replacing the SLA batteries because the electronics failed so often. Not even going to mention the medical grade "smart" power strips we had to use that disconnected as soon as the USP fired up...🙃
80% power factor and 80% max long-term load are good rules of thumb. Plug your dual power supply servers and switches into a UPS-backed PDU and a grid-backed PDU. Make sure the PDUs run consistently ≤ 40% so that if the grid fails, the UPS will handle the full load and the PDU will be running ≤ 80% rated load. Just remember that the load and runtime listed on the UPS will be halved if it becomes active
Used online UPS units are usually pretty cheap because companies switch them for new units whenever they install new servers. They will need new batteries but those can be cheap if you buy from a wholesaler in bulk. One thing I can’t recommend enough is setting up a NUT server to shut down all your computers and servers even if they’re not connected to the USB port of the UPS.
This is what I did. Got a Tripp Lite 3000VA online UPS rack mount for under $200 and found batteries for $150. It is showing a 10% load for the whole server rack!! (Only minor whining point is the WEBCARDLX managment card costs about as much as I paid for the whole UPS.)
@@DavesGarage Yup, that's where all of my UPSes came from including the rather large 6kVA online unit with the 240V DC bus (and about a quarter tonne of batteries after I was done with it) which I ran basically my whole house off through a number of 2+ hour power outages when I was living in a place with really bad power (mostly in summer because of all the A/C's, reason I went so big was so I could run MY A/C off it ;) ). Currently running an 3kVA online SmartUPS (oldschool, commercial-grade, pre Schneider) and a 3kVA dual-conversion Eaton unit (each on a separate, dedicated circuit) in my rack that feeds out to outlets in my study.
@Dave's Garage - I've always relied on an UPS to protect my desktop PC. But to save myself some money, I bought a used small capacity UPS unit and connected it instead, to two high capacity 12v leisure batteries wired in series, thus replacing the two original internal series wired lead acid cells. The UPS easily keeps my desktop PC, two monitors and router running for much longer than normal as a result.
on my servers at work a have two ups, each running from a different phase, and each feeding one of two redundant PSU in the servers. This allows to loose a phase, or an entire UPS and still have the system up and running.
Dave this is a fantastically complete review of all UPS types and technologies. I suffered through learning all this the hard way with older technology UPS units in the 90s and 00s. Back then it seemed that the manufacturers all set the battery charging voltage too high so that the battery would prematurely fail. Your video mentioned but didn't dive into the "costs" of power issues. That'd be another great video topic. Please keep up the great videos... always very informative, and your production quality makes watching them very enjoyable!
Thanks for the excellent overview of UPS, I use them at work and at home. At work we run a dual 600 KVA Online UPS system into separate switchboards and then dual power all the equipment from each UPS, because UPS do fail sometimes. We also run some dual power systems with UPS and Raw, Again although the Raw power fails more often, we have had the UPS fail and Raw power keep going so critical services stay up. A Minor correction at 19:42 when you were talking about your fathers IBM UPS having huge banks of 1.5 V Lead acid cells, I would observe that Lead Acid cells have a nominal voltage of 2 Volts per cell and a low to high range of 1.85-2.35 Volts. This is why a 12V battery has 6 series Cells.
I upgraded our UPS at work, and am so happy with it. We have power outages all the time downtown, and it was sooo annoying to get calls randomly 2-3 times a month. Especially during thunderstorm season. I got one that is 3000VA to run our 2 servers, the switches and the router. It's overkill, but it wasn't that expensive compared to one at half the price. My rule of thumb for anything electronic, is always get a power supply/UPS that is double what is needed. That way, they'll tend to run more efficiently.
Great video. If you do a follow up episode it would be interesting to cover: - the difference between LiPo cs LiFePO4 - interactions between UPS and whole-home battery like Powerwalls
Great overview. Loved the simple explanation of power factor. Try the conversion of transmitter power to ERP (Effective Radiated Power. Hox you can put a bunch of RF into a conduit and get four times that power out makes the "free energy" crows drool all over thenseves. We were feed 440 three phase at 60Hz. KCP&L (our local utility) would drop a phase on us without warning and smoked our input transformer. Those were over a grand apiece in the 80's. I built a "crowbar" curcuit. Three transfomers (440 volt to 120 volts), three relays with contacts in series, and a heavy duty three phase contactor. If one phase drops the contactor opened in way under a second.👌
re: " Try the conversion of transmitter power to ERP (Effective Radiated Power. ... a bunch of RF into a conduit and get four times that power out makes the "free energy" crows drool all over thenseves." Prime failure here is know that it is in ONLY the direction of desired 'coverage' where this increase occurs. The other area loses EIRP ...
@@crankshaft3612 Well, it is an increase, but only in a limited, predetermined area and at the 'cost' of another 'area' of coverage (speaking geographically now) ... but you get the concept. In the case of a vertical omni antenna, it is useful, as coverage at high elevation angle is useless for persons 'on the ground' or surface of the earth. Antennas designed to talk to aircraft, though, need the high elevation angle ...
I got my hands on a free APC 750VA unit because the batteries were knackered .... It's now modified with better heatsinking / fan cooling and running from 3 x 12v car batteries charged from my 2 240W solar panels. it runs my old 4770K workshop pc for ....well its never run out yet despite using it for well over 4 hours at a time also powers the network switches and router but they're loads are very small in comparison.
Great video as always Dave. When i upgraded my whole computer system 2 years ago, first thing after buying my gear I looked at was a ups system... Peace of mind was my overriding factor for power outages whilst I am using the system and came with a guarantee if the shit hits the fan and I do get an actual surge I am covered for up to $20,000 of damages for ten years. Now that's peace of mind 😉.
Thank you thank you for explaining the difference between "Online" and the other two types. I've spoken to so many IT and facilities professionals who have no idea what the difference is and it's a very important distinction to make especially if you have the possibility of dirty power from the grid or a questionable genset.
People often forget cooling power when designing a whole room UPS. One place I worked had around 50-60 servers that were used for development and QA. The UPS took one whole rack, with 3 racks for the batteries. One day the power went out and we discovered that the room was simply too hot to leave the servers on. Turned out someone forgot to pay the utility bill and it would be over a day to sort out. The batteries would have run the servers that long. or so it was said, but with the AC would have been good for only 4 hours or so .
Excellent explanation of what they do and how they work! I personally use a unit big enough to run my entire work space (4 computers, 12 monitors, 2 printer, all of my led work lights and other miscellaneous devices) for between 3.5 at full load and almost 7 hours with minimal load. Even for basic home users I highly recommend having one for any desktop computer.
yes, new information, I did not know existed or was a thing UPS I thing is more completed than small computer, them small UPS, large computer, the larger UPS, ?
I used to work for a UPS manufacturing company called Galatrek Int'l, they manufactured the 6min backup for desktops all the way up to full cabinet versions supporting Hospitals, Water treatment plants to name but a few...
After looking at UPS's, portable backup power and replacing UPS batteries I found the most cost efficient setup however slightly manual is a cheap UPS (most devices are fine with a modified sine wave for a short period) and a large lipo battery+inverter. You can get huge lifepo4 batteries (300Ah+) these days for relatively cheap and a charger for around $50. Paired with a big pure sinewave inverter for another couple hundred and if the power goes out you can just hook up the UPS to the battery/inverter and get many hours of runtime without any downtime. The UPS will take the load for a few minutes while you hook it up and then will simply passthrough the clean inverter power as required. You will need to manually hook up the inverter and charge the batteries but this setup has the added benefit of being a portable power source when out camping or travelling. It works out to be around 1/5th the price of a premade solution.
Another alternative is a low voltage ups, i build one for my rack at 12 V, the microtic switches working at 12 V and my Microserver has a MicroPSU, the POE Injectors are even using 12V, the benefit of a lower power System is that the efficancy is much better, i have only one longlive power supply with 12V, which works constant at a point where it is at > 89% efficancy, although the runtime at power lost is mich longer, the dc supplys are all very efficant compared to the dc to ac and dc to ac conversion from normal UPS. My UPS has even a usb/serial connection for monitoring it is recognized by the os as normal UPS. I even build one for work at production testing systems with an intel NUC it just works great and has a runtime of 12 hours, equipt with LiFePo4 batterys. I can just recommend it, there are even 12/24V UPS ATX supply’s for this.
As someone that did environmental systems for computers, I have been trained on computer power. One of the things that I was taught is to NEVER feed a UPS with a UPS. One UPS will fight the other. You can see this if you look at the current and voltage between the two UPS. Almost every time I've seen this done, it has caused problems. The exceptions were the computer room sized Liebert UPS that can handle 100kW and more. But there you really didn't need the second UPS. Still, while I know that specific APC UPS, I know nothing about the Ecoflow. Maybe it was designed to handle a UPS. Most standard UPS can't handle powering a UPS or being powered by a UPS. But I am willing to update my recommendations. So here it is: Unless you have Dave's budget and expertise, never feed a conventional USP with a conventional UPS.
This is why I have my Rackmount 1540W UPS on a dedicated circuit. It only has to cover for a few seconds while the Gas Generator kicks on automatically. I put all my Network gear, POE switches and Rpi Servers on UPSs as well. Not dropping my main Machine amd network during a power outage. Same reason I currently have both FiOS and Xfinity services with Failover and VLAN routing for streaming.
Q1. How does "VLAN routing" improve the availability of your setup? Q2. VLANs are a Layer 2 construct while routing happens at Layer 3 so what exactly is "VLAN routing?"
Great video sir! One thing to add: be sure to test your UPS setup every so often. Actually pull the plug out of the wall and see if the battery is really in good shape, and that comms link to your PC is still functioning. Nothing worse than doing the good work to buy, configure, connect a UPS and then find out you disconnected the USB cable when you tested that new ... And test your backups for the same reason. On the subject of UPS batteries... I think one well-known UPS mfg, or a large online retailer, or both, are shipping undersized replacement batteries. I've never seen a 9Ah replacement test out at >6.5Ah. Something you might want to look into? Cheers!
SLA batteries the amp hour rating is determined at a specific set of conditions. They charge it and see the capacity, say 9Ah. Then discharge it at a rate that it will take 10 hours to discharge, so around 1A for that battery, a C/10 rate of discharge. However discharging at a higher current the amount of energy you get out before the low voltage cut off is lower, so it will likely, at around C current draw, only show 6.5Ah capacity, and conversely, at C/100 it might run for giving an apparent capacity of 11Ah capacity. I took a 20Ah SLA pack, known to be poor, but still delivering 13.8V at 100% SOC, and loaded it down with a 3s white LED string, that drew 100mA initially, and left it to go. After 5 days it was now down to 11V, theoretically fully discharged, but still was able to light the LED string with 5mA, and after a further 2 days it was at 9V, and the LED string was still glowing, albeit dimly, and you could see the individual dies in the yellow phosphor coat. Another 2 weeks and it was still glowing enough to cast a shadow at night.
If you pull out the plug, the equipment will lose its earth protection. I've run into situations where this is not ok! All UPSs instead have a button for quick test...
@@emilalmberg1096 Noted. So just pull the fuse/flip the breaker off at the main panel to keep the protective earth. But I would also say if your setup has some weird earthing issues, best to deal with them before someone trips over the power cord and pulls it out of the socket. And I don't trust any sort of self-test to tell me what's really going to happen when the power really fails. Cheers!
@@grottyboots Simpler UPS and devices have ground via the cable only. The test function turns off the rectifiers for a while, that's enough for most times.
I just bought a 1500 VA APC lithium unit due to time. Just had to power a Synology and a few switches. When I get the time, I am going to adapt small 12v Victron inverters coupled with 4 cell LiFePo4 batteries.
VERY informative, Dave. I will send a link to everyone I have talked to about UPSs and their absolute need. Hell, I have at least 8 in my house alone, guarding every sensitive device. Thanks again.
excelllent discussion of power factor but do not forget inrush to capacitive or inductive loads. inductive loads with large transformers can have enough inrush to trip breakers do to resetting the core polarity
Good low-down - but.... The story starts with surge protection. Having correctly installed surge protection devices (SPDs) is an essential element. They should be installed in the consumer unit - or 'panel' and be connected to all supplies. In Australia - we connect them to all three in-coming phases (3 of them) and bond then to a main bond. We configure for 64A of surge current for domestic supplies - or 200A commercially. from there - we have bonded secondary SPDs in comms racks, computer rooms, building services rooms, pumps, etc.. They are also protected by an equipotential bonding arrangement. This way, the UPS shouldn't see much over 150V US or 300V/phase in the rest of the world.
Thanks Dave! You're videos are always well-researched, and you rarely present any bias unless your own personal experience bends things that way. I never knew that UPS systems could be this intriguing!
Thanks Dave. That is the most impressive talk about UPS that I have ever heard in 70 yrs of ac & dc systtems working. Yes, VI Cos theta / thi is the way to calculate for the "imaginary" element which is as you said the A vs V differential phase. But you did slip in a short comment about 3 phase. Are there really requirements & solutions that are 3 phase mains? Servers or data banks I assume. Well done for being so pedantic.. Others could learn a lot from your great talks.. About time you polished gramps old truck.
Plugging your home/office router into a UPS means that you can have internet/phone when the power is down. Even a small UPS can keep your router running for a long time.
In a town full of furniture manufacturing, we found the voltage protection against sags, especially in plants that already had low voltage (105 a lot of times) and equipment turning on randomly, kept data from corrupting overtime. Boss swore he had to reload Windows at least yearly till he started running with a UPS.
I have a standby gen with an ATO, so the vast majority of the UPS units I have around the house are very small. Generally, they are under 600VA with a couple being only 425va. Those small units are more than enough to power AV equipment and networking equipment for a minute or two until the standby gen set starts providing power. For my desktop, I have a 1500va/1000w model. The bad news is that if you need to power a 13900ks paired with a 4090, you may need a larger unit than a 1000w UPS and the prices go up dramatically once you get past 1500va/1000w.
I have a 12900K w/3090 and 10980Xe w/2080Ti protected by a big UPS. Running on a dedicated 20Amp circuit. Luckily I have a backup generator (nat gas) that picks up within a few seconds. Lots of smaller units on POE switches and network.
UPS's have gotten me through several power outages and surges. I also have my entire home theater protected by a UPS since if any parts of that fail they can't be repaired like a PC and would be a pain to swap out.
We learned some basics about VA and Reactive Power in DC/AC circuit analysis class. Trying to get an associates degree in electrical engineering. Its cool to finally have an idea of what you are talking about for a change :)
In my 20+ years of using UPS I had noting but disappointment from consumer grade standby devices. My go to for a long time was the Eaton Powerware line. My only complaint is the price. They have been discontinued for a bit now, but replacement batteries have continued to be available and fairly cheap. I will have to give Extreme a look for future purchases.
Well, from the late 1990's I had good luck with a consumer/office grade 500VA APC unit ... I was doing daily recording of voice material, so this solved power glitching issues which caused missed recordings!
I found out the hard way that line-interactive has another drawback, you have to get much bigger than you think you need. I had one loaded to about 80% but when the power failed the surge from the capacitors in the PSU draining during the switchover caused it to overload when they tried to charge back up again.
As a person who's also on the spectrum, of all things, the martial art aikido really helped me socialize. It also helped that it allowed me to nerd the heck out but the idea of Aikido is that it's a martial art that has had it's will to cause harm removed from it. It a way, it was conceived as a form of therapy for people lost to psychological horrors of warfare and combat. It creates a vocabulary of martial stances and techniques but uses them to, "say," controlled, stable and calming things. It's like you're training to become a combat therapist. Or a martial crisis negotiator. To be more explicit, you can't perform any Aikido techniques without a partner. So you first train to be a good partner. To me this alone is good for people on the autistic spectrum because A: it's very organized, contrived and controlled social behavior. B: it teaches by showing you rathe than telling you what to do. When you're playing the part of the attacker, you want to honor your partner with an attack that's enthusiastic but not more than what they can handle. You're learning in real time to watch, understand and empathize with your partner. In a way it teaches you how to be a, "power bottom," as they say. Which gets me to my final point why I think Aikido is good for people on the spectrum. C: If autism really is a form of hyper masculinity, by teaching you the value of human interaction is such a manner that if very passive, socialize, forthright and sympathetic, I believe that Aikido works almost as a form of controlled exposure therapy for people with Autism. If gives them space to both leave and return to your own comfort zone and to experience and value the perspectives of the people in the world around you. You start to see how we reflect the world and each other with our minds. You start to value people because of how their opinion is different than your own. You start to realize, in practical terms, the value of being part of society and interacting with it harmoniously. If autism is the cooption of the social part of your brain to perform other analytical tasks. Then I see Aikido as a way to coopt the analytical part of your brain, to learn of to better process social interactions. Big fan of your channel, especially your breakdowns of caching. This video's great as well, all info, no bs. Thanks! =D
I have the same APC smart-ups 1500 that you showed in my house for my server rack. I doubled its battery capacity by adding 2 more batteries. Keeps my rack of switches online and internet for around 6 hours if I want or server for about 2 hours. I have a whole home generator but it's manual transfer.
@@wickedcode007 I'll be more panicked about how my computer's PSU withstand against the UPS with modified sinewave or UPS that failed. Usually at power outage some of them behave badly against active PFC supplies (especially those with low hold time and shitty filtering design (15-20ms is ok, can't remember what ATX specs say)
@@GbpsGbps-vn3jy ATX spec is 17ms, at 47Hz and full load and minimum input voltage. Translated from specs to real data could mean somewhere in the region of 15ms at 60Hz mains to 20ms at 60Hz (a bit shorter than full mains cycle to a bit longer). If specs were to take PFC into account, than those could be following: "17ms, at 47Hz and full load and any input voltage".
I've been really afraid to go with lithium UPS for fire risk. I want to wait another couple of years and see how it goes. The lead-acid batteries are SO heavy. Got a nice 208V Tripp-Lite sinusoidal full time conversion system over here, never going back to line-interactive.
Dave said the the batteries in the unit were LiFEPO, Lithium Iron Phosphate. They don't have the thermal runaway problems that more energy dense chemistries have.
Electrical engineer here. Very good explanation on real/reactive/apparent power and power factor, congrats. It gets even funnier when you consider the harmonic distortions that typical IT loads impose to the grid. Good UPS systems have filters to mitigate that.
I use the APC 1500VA Smart UPS like you on my workstation. An AMD 7900X and 4080 FE. I took my 3960X and tossed it into a server chassis running XCP-ng. My rack has a V7 (3000VA)
Just like the book, the explanation I wish I had before buying my UPS. Mine is amazing and all, but I think I could've spent less, even while this is an old model. Thanks, Dave!
To be honest most of devices using power electronics should withstand one skipped AC cycle under full load. 1ms switchover time is just nothing for them (unless device is severely derated due to aging).
Here is a 'fun' one from 1997. We had a RAID tower that started eating drives in an accelerating curve. It ended when all the spare drives on the shelf had been swapped in and then we lost 3 out of 4 active drives in a 7 hour time period. 😱 (It only cost us $90K in a mail order company 2 weeks before Christmas.) - The culprit? The switching unit in the rather expen$ive rackmount 3000VA APC UPS was sending power spikes into the equipment. After losing $90K the only fix we could afford was a TrippLite ISObar surge suppressor. That worked for the rest of the company's life... (Tripp Lite ISObar surge suppressors are the best there is. Get some!)
5:47 "It sounded like imaginary numbers to me." I had quite a chuckle at this since VA consists of a real part (W) and an imaginary part (VAr) in the complex number space.
An important feature to look out for is the hold-up time at the intended load, I've just seen a cheap UPS offering 1 minute hold up time at rated VA, which is probably not long enough for an instant orderly shut down. As ever the issue is trying to better the usual mains MTBF while having the facility to switch out batteries or a whole UPS. A really good UPS will be tracking the mains waveform to cut in with little noise and sync to an incoming AC waveform for a clean zero voltage point change back to mains or generator.
Dave, you missed one type of UPS. I had never run across this until we acquired a newer hospital that had it, and it’s a flywheel ups. Yup a motor spins a flywheel really fast, and that flywheel drives a generator. Not that I’m a huge fan of it, we inherited it.
I recently learned a good visual example of Power Factor. You pour a glass of beer to full. The bottom 3/4 of the glass is beer and the upper 1/4 is foam. So the PF would be .75 as the good stuff is at the lower 3/4
I modified an APC smart 3000VA rack unit with custom firmware to use deep cycle marine batteries. I have 8 giant batteries for a total of 400ah at 48volts. I can survive a decent power outage with my whole network powered... I do have to keep them topped up with distilled water about once a month. They are also heavy AF.
Buying UPSs were the biggest peace of mind investment I did, 1800va Sine Wave, 24v 7A Lead Acid battery, manufacturer sells replacement batteries and extra ones for extended uptime, hold my setup, PC, PS5 (never both at load at the same time), router and 2 screens, it holds on load for about 7 or 8 minutes, plenty for a safe shutdown of everything if the power doesnt come back quickly
Online UPS, one more drawback if the juice always flows through the battery: If the battery fails, it's worse than having no UPS at all. Your connected equipment goes down while everything else in your house continues to work normally. I imagine the best of them can bypass the battery when it isn't needed. I lost my confidence in the APC brand when a glitch in the power too short to affect anything else caused my PC to shut off suddenly. The battery had failed, but the UPS switched over to it upon the power glitch anyway, and didn't go back to AC for several seconds. My current UPS (not APC) will switch back immediately in that case. When I heard the relay double-click a couple of years back, I knew it was time to check the battery. Sure enough, it had to be replaced. But it didn't screw me during one of the many quick power glitches we get around here.
Wrong. Typical online UPS include: PFC stage, bi-directional DC-DC and output invertor stages. PFC output, DC-DC high voltage side and invertor input are connected together. DC-DC low voltage side is connected to battery. Whenever battery fails, unless internally shorted, whole UPS is powered in the same way as normal operation.
Actually, power factor is between -1 and 1, though below 0 is probably not relevant to the matter in mind. I felt the need for this statement. Otherwise great video so far!
Two other notes: There's more lithium battery chemistries such as LiFePo4, which lasts even more cycles. An online UPS with a Li-ion or LiFePo4 can be more efficient than an offline or in-line UPS with a lead acid battery, due to the charge algorithms and other differencen electronics as there isn't much innovation in the low end anymore.
I have an UPS. Last year, PG&E threatened to shut down electricity. I think I'm too close to the Apple HQ. The UPS started to go click-bzzzz-click-bzzzz. I thought the UPS was dying, but it turns out that the electrical circuit was dipping down to low volts, and the UPS was kicking in. The shutdown never happened, but wow, I'm glad I had the UPS. I also got a cheap UPS for my modem due to power clicks for a while. My Internet connection kept having momentary drop-outs for several minutes, and the UPS keeps the modem running. I realized that my modem and router together costed more than $400, so got the UPS.
i am running the Wroung one on 2 Lawn and garden batteries. APC Back - UPS RS 1200 it needs 24V so i wired in 2 Lawn and garden batteries it's just on my truenas. so it can shut down in a safe maner. then i have a APS 500 on my Cable modem / PFseance Box Keep in mind the Power Supplys in like 99% of every thing made today is a Switching power supply they just converting every thing to DC any way so a square wave is fine and more efficient.
at work we have a 380v DC UPS for our servers, our server 240v power supplies just use rectifiers on their input so they can handle DC voltage just fine, this removes the conversion stage and increases efficiency. Afterall you are in the end converting AC to DC to AC to DC im sure when 12VO power supplies take off someone will make a power supply with an external battery to skip even that final DC to DC step and initiate a soft shutdown or just kick the device to sleep, just accepting that your monitor would not work
You said they have surge protection - well - that's a "maybe". In my last job, we had three GE UPSes - 15kVA or so - and a longer offline job was planned. Since we didn't have a generator, we hired one from the power company. They came and put it outside, laid down the cables and switched off the main power - the UPSes took over. Then they powered up the generator and for some reason, it sent a 4kV pulse into the UPSes, which paniced and went into passthrough and sent it all into the datacentre. Popcorn time! Quite a few PSUs were fried among with some disks. IIRC no motherboards were damaged. Interestingly, the stack of SuperMicro machines just buzzed on through this mess - nothing even in the logs. Good PSUs on those.
Regarding battery longevity: li-ion has a longer average longevity than lead acid, but its longevity is significantly shortened if (a) maintained at full charge instead of around 80%, and (b) maintained at warm temperatures. You can increase the longevity of li-ion by de-rating them down to their sweet spot of around 80% and by keeping them cool. You can sort of consider the top 20%-ish of capacity as "over-charge", because it kind of is. And for temperature, if it feels cool to you then it's probably good, but if it feels warm to you then it's probably too hot for the battery too.
I wish there was a UPS that skipped over the thing with double conversion and charged the 24V batteries and feed that DC directly into the computer after stepping it down. Just like servers PSU do. In fact, you only need a single conversion AC to DC to battery then 24V to 12V DC/DC PSU to the rest of the computer (plus 5V and whatever, most motherboards only use 12V either way and regulate themselves what goes to the devices.
I have a smaller UPS that my ISP provides (and maintains) for my network gear. I run line conditioners on all my electronics. Even have one on my refrigerator. These line conditioners are not just cheap MOVs that only deal with spikes. These clean up the AC and correct under and over voltage situations. Tripp Lite LC1800 or something similar. You can get 120V and 230V versions. When power goes out my network runs on UPS and I use my laptops.
Some online UPSs do not completely shut down above their rated load when they get the AC input from the line, they instead switch to a weird offline mode (unregulated output!!) with battery charging continued.
One of the more insidious characteristics of lead acid batteries is something called the Peukert effect, in which the overall capacity drops as you increase the load. Lead acid battery capacity is usually specified at a 20 hour discharge rate. So a 100 amp hour battery can be discharged at 5 A for 20 hours. But try to discharge that same battery in 30 minutes as might be typical in a UPS situation, and you will find the overall capacity has been reduced by around 2/3, so your 100 amp hour battery is now a 30 Ah battery capable of delivering 60 amps for 30 minutes. Lithium ion batteries, especially LFP batteries, do not behave this way. A 100 amp hour battery is pretty much always a 100 amp hour battery regardless of discharge rate (within reason).
I hope you'll include efficiency testing in your next videos. I had a second-hand double-conversion UPS that looked like an ancestor of the P91. I resold it quickly when I measured how much energy it was eating! I'd be interested if you could measure how much energy both the P91 and the Delta Pro consume under different scenarios, for instance: no-load, 10% load and 80% load. In my case, my unit (4.5kVA) was consuming ~200W on top of what was plugged in it. This number seemed to decrease with load however! Which is strange because at first my explanation was that there seemed to be a constant ~0.1A going through the batteries (20 batteries in series, charged with ~270V DC since I live in a 240V AC country), so the amount of energy would be correct but not decrease with load... I may have missed a change in power factor though... Maybe this consumption was because the batteries were old and recent units wouldn't need that much constant energy going through the batteries? Anyway, looking forward to your test of the Delta Pro!
Lightning always best to have external lightning protection in the breaker panel, then add in surge arrestors in the outlets that feed the UPS. For lightning you need multiple stages of clamping, high power but high standoff at the power input, that can handle the high energy incoming surge, and attenuate it down to a lower, but still very high level, surge on the power to the breaker panel. Then the breaker panel unit will reduce it further, using the impedance of the wire from the meter to the panel, and ensure a lower level to the outlets. Then the outlet units can handle the lower level surge, and clamp to a safe voltage for the equipment. Then the UPS can handle this higher voltage without problem. Remember the lightning protection starts at the power line, with power companies using lots of surge arrestors and arc gaps along the power lines, because they all run outdoors, and are often the highest point on the landscape, and all the pylons are well grounded, so all of them will receive multiple hits in a storm. Thus there is a lot of protection, to limit the spikes that go to the customer side, where local hits cause a lot more damage, simply because the surge clamp ability is smaller there, as the power levels are lower, so the utility sizes them smaller. But also local lines tend to be hit less, as many are not that high, and are often buried as well.
The VA value is related to average current (amps) which IF it's higher when voltage is lower (than if it was a resistive load) means there's more heat but less average power. For part of each cycle (Hz), the current is similar to a much larger load then briefly lower than expected but not enough to make up for the time it had excessive current. Power is more efficiently supplied when the highest current is when the highest voltage happens each cycle. A 1500VA UPS is often around maximum of 1050W RMS continuous, the VA value assumes a 0.7PF (eg. old PSU with transformer), this would be ~1167VA@0.9PF (eg. modern PSU with some PF correction/mitigation) but the power (W RMS) is still limited to the same 1050W in this example.
Car audio amplifiers and subwoofers are great example of a/c reactive load. Lots of RU-vid channels test the amplifiers on amp dyno that’s a resistive load giving you s best possible power output of the amp at specific ohm loads also a good stress test on the amps to see if they are quality built or go up on smoke😂
A typical PC power supply does not care if it is fed a modified sine wave input. It is going to chop it up anyway. On the other hand electric motors *DO* care. So a pure sine wave inverter is more important for power tools than for a PC. One thing becoming more common is whole home battery backup, usually in conjunction with solar panels. These systems often have an unfortunate interaction with the UPS. When the grid connection is lost and the home is running off battery backup (e.g. Tesla Powerwall) the frequency is increased above the usual 60Hz as part of the safety protocol. This can have the consequence that your UPS thinks the power is bad. So, while the rest of your home may have power, your UPS will activate, run out of battery backup and then your PC will shutdown. Some systems allow the frequency tolerances to be adjusted to avoid this issue.
Minor nit: The float batteries at 19:46 are lead-acid, so the voltage per cell is 2V. Decades ago I worked in a computer and telecommunications center where a whole floor of the building was an array of about 40 rows of batteries as shown in Dave's example. In the basement there were three 1 Mega Watt diesel generators, for reliability. One day during a storm the building was hit by lightning and the master breakers disconnected the building from the grid power. The backup system perfectly switched over to the battery array, which was sized for about 3 hours operation, the expectation being that the generators would start up within minutes of loss of grid power availability. Tragically all three generators failed to start up, and after a few hours the batteries had been depleted and the building and all computers and telecom equipment lost power. The system had never had a real "live test" of losing grid power. The problem was that the starter motors for the diesel engines were wired to grid power via the master breakers.
Living in the Santa Cruz Mountains, UPS is a way of life. As far as brownout protection, when a power line fell on a road and cars drove over them for an hours, the 'protected outlet' took out the caps on a Genelec Speaker. So I do not trust the 'protected no battery' outlets anymore.
that output is protected with MOV against high voltage transient thunder spikes, not low voltage brownouts. Use the "battery output" to protect brownouts perfectly with voltage boost ! 😊
The cool thing about being an electrician (or just a person who has tools and knows how to crimp wires properly) is that you can make your own UPS. If you hook everything in parallel as 12V, you can use any mix of used car batteries instead of throwing them in the ocean. Just remember that car batteries are not intended for deep cycling. If you run them down in a UPS, you permanently damage the battery.
When we ran an ISP back in the 90ies, we had an online UPS. We never ever lost power, but the USP did break and left our server room dead in the water.
Typically a data center of any magnitude or risk of lost work should include two completely separate power feeds from main services. That is IF you can get that at your location.
@@wickedcode007 doesn’t help when the UPS dies, though. I guess if you had all devices with redundant power supplies and use 2 ups, ok, but we didn’t swim in money then.
I once made a "poor man's" online-ups out of two line interactive ones. My work was replacing some old APC XL1400s so they let me take them home. One of them was a 240v model the other two were 120v. I needed all three to provide power for my computers, switches and servers. I put the 240v on my servers since all that equipment was rated for it. Well the issue was my generator only output 120v. So the server would lose power during an extended outage. So what I did was connect the battery banks of the 240v and one of the 120v UPSes together. So when the generator was running the 120v UPS was charging the battery bank so the 240v ups could keep running. Technically with the set up if I never plugged the 240v ups into an outlet it would have been running "online" all the time. However in my case it only ever ran that way on generator. The "on battery" beeping would have gotten old fast too. Thankfully I've since gotten a generator that can output 240v so I don't have to worry about that any more.