[HN Gopher] Is it okay to daisy chain a UPS?
___________________________________________________________________
Is it okay to daisy chain a UPS?
Author : nuker
Score : 108 points
Date : 2024-04-28 08:17 UTC (14 hours ago)
(HTM) web link (www.eaton.com)
(TXT) w3m dump (www.eaton.com)
| delichon wrote:
| I've never found a straight answer on whether it's ok to daisy
| chain surge protectors. Anyone know? I've been trying to avoid it
| based on a warning I don't understand, but it can be tough to
| find power bars without surge protection.
| 4gotunameagain wrote:
| yes, as they are passive device unless in the case of a surge,
| which would be shorted to ground by the first one in the row
| (technically, the first one whose limits are lower than the
| surge)
| kosma wrote:
| The nature of surges is not simple like that - a lightning
| strike can easily blow MOVs and inrush limiting resistors in
| multiple devices. I come from a rural area and coming to
| someone's house with a bag of fresh MOVs and resistors is not
| an uncommon thing after a big storm.
| 4gotunameagain wrote:
| I believe that it highly depends on the type of the surge
| protector, their ratings and the cable network involved. I
| do not think that it would cause issues in two surge
| protectors from power outlets connected in series, and
| depending on their rating and switching characteristics
| maybe both could trigger, I agree.
| numpad0 wrote:
| I believe the question is if it's safe at all to do so
| rather than whether the protection works or not, though I'd
| have to be all ears about interactions between multiple
| parallel surge arrestors.
| kosma wrote:
| It's not very different from daisy chaining normal extension
| cords - safe if you know what you're doing (not exceeding the
| current rating on any of them). Most surge protectors are
| fused, making them safer to daisy-chain than normal extension
| cords.
| sgarland wrote:
| As an aside, increasing the length of extension cords can
| cause premature failure of some devices (mostly motorized
| tools, especially cheaply-made ones) if the wire gauge is
| inadequate, due to voltage drop.
|
| As a general rule, I wouldn't run tools past 50 feet on
| anything smaller than 12 AWG (and really, 14 AWG is the
| smallest I'd go for any length; anything smaller isn't safe
| for most loads).
| quickthrowman wrote:
| > It's not very different from daisy chaining normal
| extension cords - safe if you know what you're doing
|
| It's not safe, and it's expressly forbidden by the NEC, see
| 11.1.5 below:
|
| _> 11.1.5 Extension Cords
|
| > 11.1.5.1
|
| > Extension cords shall be plugged directly into an approved
| receptacle, power tap, or multiplug adapter and shall, except
| for approved multiplug extension cords, serve only one
| portable appliance._
|
| Daisy chaining extension cords is unsafe and not recommended.
| Only use extension cords that you've inspected and are
| properly rated for the environment (don't use indoor cords
| outside, don't use an outdoor extension cord outdoors unless
| it's GFCI protected) and power usage of the device you are
| powering.
|
| Any time electricity has to flow through a splice or
| mechanical connection, the possibility of a loose connection
| causing an arc and subsequent fire exists.
|
| It's unlikely to happen to you specifically, but it does
| happen and avoiding electrical fires is a good thing if it
| can be avoided.
|
| Daisy chaining power strips is also forbidden by the NEC:
|
| > 11.1.4.2
|
| > The relocatable power taps shall be directly connected to a
| _permanently installed receptacle._
| dvdkon wrote:
| Yes, there is a risk of failure involved with _anything_
| electrical, but I don 't see why anyone would consider
| chaining extension cords inherently dangerous enough to
| ban. It increases the number of connections, but that's a
| miniscule risk compared to the 5+ connections an extension
| cord might have on its own. The only significant risk I
| know is people disregarding the max amperage rating of
| everything in that chain.
|
| For anecdotal experience, I've had both extension cords and
| wall plugs fail (nothing serious thankfully, but they did
| get a bit melted), but in those cases it had nothing to do
| with my extension cord chains, but rather an internal
| connection failure.
| fbdab103 wrote:
| It is probably one of those little process changes to
| minimize chance of catastrophic failure. Sure, the risk
| of the daisy chained system going poof is low, but not
| zero. Instead, you should try to re-work your plans so
| you do not need to daisy chain.
| marcosdumay wrote:
| AFAIk, the rationale for extension chords is that they
| are sized to cause a controlled amount of voltage drop
| within their lenght. If you keep adding them, you will
| increase the drop, and many devices will react by
| increasing the current.
| quickthrowman wrote:
| This is correct. For a 120v 12FLA load at a distance of
| 50', you're fine using a #14 cord. If you double that to
| 100', you need to use #10 cord to account for the voltage
| drop. As voltage drop increases, the amount of current
| flowing through the cord increases, which can potentially
| heat up the insulation beyond its rated temperature.
| AstroJetson wrote:
| You need to know what the cable size is of the power strip and
| the current you are going to draw. If you have bar A with 14 ga
| cable plugged into the wall, and bar C With 18 gauge plugged
| into that, then with devices that don't exceed C current rating
| (12 amps) and devices in A ( including the devices in C ) that
| don't exceed A's rating, you will be fine. The other way wall -
| C - A is the problem, C isn't able to manage the full load.
|
| Only buy known surge suppressor, there have been tear downs
| where the surge components were missing / fake.
|
| Since surge comments are passive, chaining the surge components
| is not a problem.
| xprn wrote:
| Today I learned. I always thought it was some "electricity
| magic thing" like additional heat generated within the power
| strips causing issues between the connected devices, but this
| makes a lot more sense than whatever I was thinking of.
|
| Although I do think I might have mixed some things up between
| regular power strips and those outdoors/industrial ones with
| a long (double/triple digit meter) rollable cable which my
| dad was a big user of back when he used to work in
| construction. Basically back when I was little he used to
| tell me never to plug power tools into a rolled-up "power
| wheel", and I think that when I was later heard you shouldn't
| daisy chain power strips I must have made that (wrong)
| connection.
| vbezhenar wrote:
| It's about heat. Cable has some resistance and emits heat.
| This heat has to dissipate somewhere. If cable is rolled
| out, it'll dissipate heat to the air. If "power wheel" is
| not rolled out, cable will heat cables around. Outer cables
| will dissipate heat to the air, but inner cables will not.
| So with enough current and enough time, this thing will
| melt.
|
| You probably won't have issues charging iPhone from this
| thing or powering something for few seconds, so no need to
| go crazy about it, just something to keep in mind.
| willis936 wrote:
| Surge protectors do have one magical electrical thingy in
| them: metal oxide varistors (MOVs). They're what shunt
| current in an overvoltage transient and they do age with
| usage.
| aceazzameen wrote:
| I've always known surge protectors age and eventually
| stop protecting from surges. But what I always wanted to
| know is how can you tell when it expires? I'm assumings
| it's based on how good/bad/stable the electricity is in
| your area. But still, is there any way to know when it's
| time to replace?
| malfist wrote:
| Typically surge protectors have a little light to let you
| know if the surge circuitry is still good. Others will
| fail safe, meaning the won't power on if the circuitry is
| bad. Cheap ones may do neither.
|
| Either way, if your house has had a surge and other
| equipment has died that wasn't surge protected, probably
| a good time to replace all surge protectors in the house,
| they're not really meant to survive multiple large
| surges. They shunt the power destructively, just
| somewhere you don't care.
| neilv wrote:
| > _they 're not really meant to survive multiple large
| surges. They shunt the power destructively, just
| somewhere you don't care._
|
| Is this a concern when buying _used_ rackmount power
| conditioners (like used for live music setups?), to
| protect home IT gear? Can they be worn out without a sign
| that they are?
| cesarb wrote:
| What I've read is that their main failure mode is that,
| as they age, their trigger voltage gets lower and lower,
| and at some point the normal line voltage is enough to
| trigger them all the time. And when they overheat, either
| due to being triggered all the time or due to diverting a
| large surge, they fail open and no longer have any
| protective effect on the circuit. High quality surge
| suppressors would have fuses physically touching the
| MOVs, so that when a MOV overheats and fails, the fuse
| opens and cuts power to the now unprotected output.
| willis936 wrote:
| You are not supposed to. The upstream device is sized for a
| maximum current and daisy chaining can lead to a scenario where
| downstream devices are sized for larger current than upstream
| devices, which is avoided everywhere else in electrical
| distribution.
|
| So it's fine as long as you control the strip and keep track of
| loads (e.g. you know your spouse will never plug a vacuum into
| that handy receptacle you have there), but at work your EHS
| team will mark you down for it.
| vasco wrote:
| In your example there's a first undersized UPS A, then
| another UPS B, then whatever combination of electrical
| devices which power usage exceeds UPS A. You say this is
| problem.
|
| If you have said combination of electrical devices, and if
| you're assuming we're using an undersized UPS A + the combo
| of devices, why does the UPS B matter?
|
| If you're going to overload the UPS A you're going to
| overload the UPS A regardless of UPS B, no? Daisy chaining or
| not, that doesn't seem like the actual problem to a knee-jerk
| thinking.
| 4gotunameagain wrote:
| This is not specific to surge protection, but extension cords
| in general.
| xtqv wrote:
| Daisy chaining is irrelevant to the problem that you can buy
| 18ga "lighting-only" extensions that bear a 15A rated NEMA
| 15R but are limited to 8A.
|
| Daisy chaining a power bar with it's own circuit breaker can
| be ideal if it prevents someone from making the mistake of
| using a circuit in a way that trips a panel breaker, ie
| preventing your spouse from plugging a vacuum into a circuit
| shared by several rooms.
| ta1243 wrote:
| In the UK, if I have a 6 way strip sized for 10A, which thus
| has a 10A plug in, I could then plug in a 4 way strip
| downstream with a 5A fuse, then a lamp with a 3A fuse, and
| that's fine. I could even connect it the other way.
|
| If I plug in a heater pulling 10A then sure, the 5A fuse will
| blow.
|
| Daisy chaining multiways will increase the resistance in the
| earth wire which could mean you end up with a class 1 device
| with a fault connecting live to earth which would only
| punting say 8A to earth due to a high resistance (but then
| your circuit's RCD would trip with that), but is it a major
| problem?
|
| With the US system, do you not have wires capable of 3A (say
| 24 AWG) which you can connect to a normal socket which also
| takes a 10A vacuum?
|
| If that lamp has a fault where it pulls 6A, what protects the
| 3A wire -- i.e. there's a fault with your lamp which is
| plugged into a 15A circuit breaker, and the lamp draws 10A,
| it wouldn't trip the breaker, and that nice thin 3A lamp cord
| would melt.
| t0mas88 wrote:
| That lamp example happens nearly everywhere except the UK,
| due to the UK having fuses in plugs. We don't have that in
| the rest of Europe for example.
| vbezhenar wrote:
| As long as you keep load below threshold, I don't see any
| potential issues.
| quickthrowman wrote:
| It's not OK, it's a violation of the NEC:
|
| > 11.1.4.2
|
| > The relocatable power taps shall be directly connected to a
| _permanently installed receptacle._
|
| A surge protector is a 'relocatable power tap' and _must_ be
| plugged into a permanent receptacle.
| immibis wrote:
| "surge protector" is what Americans call a passive device that
| splits one outlet into several, yeah?
|
| The danger is overloading. Back in the days when the main
| things you plugged in were incandescent lights and space
| heaters, this was probably a big issue. With computer equipment
| and LED lights you have to have a lot more stuff - many
| outlets' worth - to reach the circuit's maximum capacity.
|
| If the circuit and "surge protectors" are rated for 1800W (15
| amps x 120V), officially you should limit yourself to 80% of
| that for continuous loads which is 1440W, so you can supply 14
| laptops or small small desktops that use 100W each, or over 200
| raspberry pis on USB chargers that use 5W each, and either way
| you're going to need a lot of outlets before you come anywhere
| close to that limit.
|
| At least that's a rough estimate. Power factor could decrease
| that number by up to 50% and you can use the full rating for
| intermittent loads; I'm not certified to know the fine print.
| Point is that 10 computers can easily use less power than a
| single space heater.
| jcalvinowens wrote:
| > or over 200 raspberry pis on USB chargers that use 5W each
|
| Those little switcher bricks are horribly inefficient: the
| 15W one I just pulled out of a drawer draws 0.8A on the
| primary. Realistically you're going to max out a 15A circuit
| around 20-30 of those, not 96 (1440/15).
| wnoise wrote:
| > "surge protector" is what Americans call a passive device
| that splits one outlet into several, yeah?
|
| Technically different, but often combined functions. The
| splitting bit is a "power strip", or sometimes a "power bar".
| The surge protection is switching off when there's a short or
| overvoltage in the supply, or other larger than expected
| power draw.
| callalex wrote:
| >surge protector" is what Americans call a passive device
| that splits one outlet into several, yeah?
|
| Not necessarily. There are "power strips" which turn one
| receptacle into several. Then there are sure protectors which
| are typically built into power strips. So not all power
| strips are surge protectors but almost all surge protectors
| are also power strips.
| cesarb wrote:
| > So not all power strips are surge protectors but almost
| all surge protectors are also power strips.
|
| There are panel-mounted surge suppressors (which can
| protect all circuits coming from the panel), and also
| inline surge suppressors with a single output like this
| one: https://www.lojaclamper.com.br/dps-iclamper-
| pocket-2pinos-10...
| GuB-42 wrote:
| I don't see the problem, provided that you are within the
| current draw limits.
|
| Surge protectors are usually made of components that cause a
| short when a surge happens, protecting the equipment
| downstream. It usually pairs with some kind of overcurrent
| protection (breaker, fuse, sometimes GFCI) to protect against
| the short the surge protector itself caused.
|
| Having chained surge protectors it actually quite common. You
| may have a surge protector in your breaker panel, then in your
| powerstrip, then in the power supply of the device you have
| plugged in. Most good quality ATX power supplies have built-in
| surge protection for instance. They also all tend to have
| overcurrent protection too. The breaker panel has breakers
| (duh), the power strip may have a simple breaker too, and the
| device may have a fuse. In the UK, the plug itself may have a
| fuse, plus the breaker from the utility company.
|
| The risk from chaining surge protectors is that it increases
| the risk of false triggers if one of them is defective. But it
| may also provide better protection. All in all, I wouldn't
| worry too much about it. Just don't overload that power bar and
| whatever it is plugged in.
| WarOnPrivacy wrote:
| > it may also provide better protection.
|
| This has been my experience. There was a corridor between 2
| FL counties known for heavy lightning strikes. I serviced
| small sites that had their IT equip fried (exploded, melted)
| once or twice a year.
|
| Putting it behind 4-6 decent, consumer-grade surge protectors
| turned out to be really effective. I was a bit surprised
| given how lightning can jump over protection during a strike.
|
| To illustrate the area: An XO's home was hit. Char marks
| lined the walls wherever wiring ran. Pipes burst all over.
| Nothing plugged in or wired survived. The front door was blow
| into the street.
|
| His grade school kids were home at the time; they were
| physically fine.
| LorenPechtel wrote:
| There's also the fact that consumer surge protectors are
| incapable of determining if they're any good or not.
| There's a small component in there that eats the surge--a
| severe surge and it's destroyed, it's obvious. However,
| there's a range in which it no longer functions but is not
| destroyed. The next surge goes on down the wire.
|
| I really wish someone would come up with some surge
| suppressors that have a string of field-replaceable
| suppressors. Periodic maintenance, replace the suppressors.
| orblivion wrote:
| This is funny, I actually am wondering about plugging a surge
| protector into a UPS. I have some equipment (probably not very
| high power draw but I can confirm) conveniently zip tied to a
| peg board along with a power strip that I assume has a basic
| surge protector on it. One power cord leaves the whole thing,
| which is very clean, but power goes out a surprising amount
| here. Can I plug this whole thing into a UPS or should I find a
| replacement strip without a surge protector?
| Eisenstein wrote:
| You can plug it in. The problems with the UPS being plugged
| into another UPS are related to the internal batteries and
| the inverters, not to the surge protection. Just don't plug
| in too many things and overload the UPS.
| radicality wrote:
| From what I remember reading, you should not do that. Only
| plug in PDUs or directly connect equipment to a UPS. You're
| meant to put the surge protection before the UPS. I think
| it's because the circuitry in surge protectors can mess with
| what the UPS thinks the actual load is.
|
| I believe some UPS brands might also void parts of your
| warranty if you use them with a surge protector plugged in.
|
| https://www.apc.com/us/en/faqs/FA158852/
| rinron wrote:
| probably because it Depends, If you dont know its safer not to.
| long answer (keep in mind i live in canada, im not a
| professional, this is just what i learned and trust, keep in
| mind above) - As long as no wire/bar exceeds its max Amp your
| safe no matter how many things are plugged in or how long of a
| daisy chain, the max amp is usually listed. - all devices
| should list the max amp they use, add up the amp's of all the
| devices connected to that wire and if its below your safe. - if
| your breaker is 15 amp, virtually all normal extensions/power
| bars are rated for at least 15 amps which means doesnt matter
| what you do the breaker should* flip before any damage is done
| - if breaker is above 15 amp most surge protectors have
| protection that will trigger if they exceed their max, most
| basic "splitters" and extension cords dont, eg if they dont
| have a switch or reset button they are pretty much guaranteed
| to not have this protection, be extra carful how much you load
| on those. some good rules of thumb to keep in mind, look for
| imprints or labels that list amps, smaller wires can handle
| less, damaged wires can handle less then they did before and
| shouldn't be used, if a wire or connector warms when its in use
| its overloaded reduce the load on it, especially if it warms
| quickly. here is a video that i liked that talks about it as
| well https://www.youtube.com/watch?v=K_q-xnYRugQ
| dreamcompiler wrote:
| The reason it's a "bad idea" is that most surge protectors are
| power strips and you must be cautious when you daisy chain
| power strips. Likewise extension cords.
|
| Why? Two reasons: You have to ensure the wire gauge on every
| link can handle the current, and at every junction (plug) the
| resistance is higher than in the wire itself. When electrical
| fires start they usually start at these plug junctions because
| they overheat.
|
| The surge protectors themselves don't mind being daisy chained.
| madaxe_again wrote:
| I mean, it _is_ ok, as long as everything is appropriately sized
| - I live entirely off grid, with several battery banks and
| inverters, which are essentially giant UPSs - and have regular
| UPSs for various equipment so that when I shut the power down for
| whatever reason the network and servers stay up.
| lambdaone wrote:
| There are a vast number of problems that all can be summarized
| as "this is a really stupid thing to do, unless you really know
| what you are doing". This is one of them.
| dgacmu wrote:
| Not really. In your case, you probably have a very high quality
| pure sine wave inverter that's much larger than the individual
| UPSes, which will make the downstream UPSes mostly happy as
| long as you don't approach max load.
|
| Many cheap inverters are not pure sine, and a UPS seeing this
| waveform may decide it needs to go to battery also.
|
| Practically, a UPS also adds to the current draw, and many
| people may accidentally exceed the circuit limit because they
| only look at the useful load but not also the UPS charging load
| after a power failure ends.
| K0balt wrote:
| Same here, I run a lot of different equipment on grden variety
| UPS's on our solar grid - but that is fed by redundant 18KW
| three-phase inverters and 72KWH of batteries... so that
| probably looks a lot like utility power from the perspective of
| the UPS's.
|
| Overall, the discussion tangentially reminds me of a common
| theme in aviation - twin vs single engine aircraft. With two
| engines, the chance of having an engine failure at a critical
| time is doubled.
| immibis wrote:
| But you also have a good chance to get through the failure
| with one working engine. A two engine plane with one working
| engine climbs slowly, and a one engine plane with zero
| working engines doesn't climb at all.
| geor9e wrote:
| This article is about daisy chaining UPSs into each other's AC
| outputs. I hope you're not doing that off grid. Daisy chained
| UPS means the energy takes the path of DC battery > AC inverter
| > DC rectifier > different DC battery > AC inverter > DC
| rectifier to whatever device you're powering. You could be
| losing up to 20% of the energy (server uptime) at each step of
| the daisy chain. And since there's only one path for power to
| take, if anything fails, it all goes down. The article's
| solution is super simple, just use them in parallel instead.
| madaxe_again wrote:
| UPSs typically operate in passthrough mode unless a deviation
| in supply occurs - so no efficiency loss day to day - and no,
| I don't chain UPSs, but as I said, the house batteries are
| essentially big damn UPSs.
|
| Also, not particularly caring about efficiency currently as
| we always have more energy than we can use, and I'm currently
| looking at filling a shipping container with sand as a dump.
| geor9e wrote:
| >UPSs typically operate in passthrough mode
|
| Oh, that makes a lot of sense. In that case, daisy chaining
| two sounds like no big deal.
|
| >filling a shipping container with sand as a dump
|
| That's pretty cool. I didn't know private folks were doing
| that.
| chx wrote:
| Note the last image of the article is not correct, if you want
| full redundancy you need two independent power feeds. How
| independent depends on your needs of course. You might just want
| to run it from a different circuit but the same utility power. If
| you've got insane needs and megabucks you can talk to the utility
| about being fed from two substations or at the extreme end you
| can get one feed from the utility and you can make your own
| second feed. Traditionally we did this with water turbine working
| from a river but today I might look into solar and perhaps molten
| salt.
|
| As an example, way back then when this was a very lucrative
| business, we placed the servers for a premium number erotic call
| in an industrial park on the border of two districts in Budapest
| because that's where we could get two independent power feeds
| without running our own lines. Internet connection wise, one was
| a simple leased line the other was a microwave connection to very
| far away. Short of bombing the entire site it was fair impossible
| for the installation to go offline and -- for the six years I
| knew about it, it never did. Note the site served German callers,
| that's where the big bucks came from.
| mcfedr wrote:
| Presumably that's what they are showing with two wires with
| different shape connectors
| adrianmonk wrote:
| And/or the box labeled "Utility Power" is not intended to
| convey any specific notion of single circuits or multiple
| circuits nor of single substations or multiple substations.
| It just depicts power that in some way comes from a utility
| and no more.
|
| In that sense, it is not incorrect about the configuration of
| utility power because it doesn't say anything about that
| subject.
| willis936 wrote:
| I'm working on a facility with redundant power and my
| impression is that it's not insanely expensive if you have
| expensive machinery to protect and that diesel generators are
| far and away the most common and inexpensive second feed.
| michaelt wrote:
| Diesel generators are great if you need a few hours of backup
| (assuming the generator actually starts when you call on it).
|
| But if you need enough backup capacity to survive something a
| multi-state, multi-day blackout [1] that probably gets
| expensive.
|
| You wouldn't need that for a premium erotic call processor,
| but a 911 call exchange might, for the portion of their
| workload they can't pass off to another exchange.
|
| [1] https://en.wikipedia.org/wiki/Northeast_blackout_of_2003
| segmondy wrote:
| Natural gas generators.
| j45 wrote:
| Looking at multi fuel generators is worth it too depending
| on your setup.
|
| For example some can run gas/propane/natural gas
| willis936 wrote:
| You can store about a day's worth of diesel on-site and
| have agreements in place to have daily refills in cases of
| emergencies.
|
| If you can't get gasoline within a day's drive then there's
| bigger problems in the world.
| progbits wrote:
| In many places going over what you can store might put
| you over emission quotas and you would have to shut down
| anyway. I'm familiar with one incident at large DC which
| had fuel left and could easily get more, but only had few
| hours before they were required to shut down by EPA.
| s0rce wrote:
| I worked at a site where we powered a bunch of stuff
| consistently from diesel generators until the grid hookup
| was finished, much longer than a few days. Probably was
| expensive.
| xen2xen1 wrote:
| Worked at a place with a natural gas genny as the second
| source. You don't have to feed it, it just keeps working.
| ocdtrekkie wrote:
| Technically you are just feeding it power from a different
| utility company than the electric company.
| hedora wrote:
| I wonder if natural gas is actually redundant at this
| point. Around here, phone company decided to take a
| dependency on the power grid, so phone + internet go down
| if the power is down.
|
| It wouldn't surprise me at all if some utilities have
| started installing smart meter upgrades or inline
| compressors or computer-controlled valves that don't have
| generators attached to them.
| wongarsu wrote:
| > phone company decided to take a dependency on the power
| grid, so phone + internet go down if the power is down
|
| I hope it goes down a couple hours later?
|
| Around here (Germany) your phone and internet is also
| dependent on a junction box with routers somewhere within
| half a mile or so of your home having power. But they
| have four hours of battery backup, and on normal-sized
| outages they send people out with diesel generators when
| the batteries start running low (prioritizing business
| customers). Having it go down from power loss is a
| decision made in triage, not something that just happens
| tharkun__ wrote:
| How common are power outages? Historically and in the
| coming years given nuclear power I hear is dead in
| Germany and France's reactors are old or need to shut
| down because of rivers being too hot in summer. Is the 4
| hours a 1980s decision that needs revising?
|
| Asking, coz Canada here. Power outages aren't uncommon
| (when I say that I mean: expect at least one that makes
| you get out the generator in 'shoulder season' per year),
| coz power lines (not talking transmission lines) are
| mostly above ground, except in large cities of course.
| But as soon as you get out of the "center" (which
| depending on city is larger or smaller too) it's good old
| wooden poles that carry power on the top and cable /
| phone on the lower level.
|
| Bucket transformer[1] on a pole near you blows up is a
| favourite but the lines are actually fine, including your
| cable / DSL. Last time we also lost internet it took
| about 24 hours after power went out that internet went
| down as well. Cell phone service from the same company
| was still fine. The entire metro area was out of power
| for days and I guess they prioritized topping up the
| diesel/LP for those.
|
| [1] These guys
| https://en.wikipedia.org/wiki/Distribution_transformer
| 20after4 wrote:
| In small town USA, my cable internet goes offline
| immediately even if the power outage is just a 1 second
| flicker. Then takes some equipment to reboot at the cable
| office before it comes back online. Very annoying. Phone
| companies typically have a good battery backup though.
| wongarsu wrote:
| Large-scale power outages are exceedingly uncommon in
| Germany, or really all of Europe. The last big one was in
| 2006 because of some mistakes when taking down a major
| transmission line [1]. I don't remember reactors being
| offline ever causing a power outage. When France has to
| shut down its nuclear reactors Germany just fires up more
| coal plants, that's the beauty of a large interconnected
| grid.
|
| What does happen are smaller scale outages. Power lines
| are mostly buried along streets and under the sidewalk,
| just like telephone lines. That doesn't stop the
| occasional excavator digging too deep and taking a street
| off the grid. At an individual level it's extremely
| uncommon, maybe once a decade. But deploy thousands of
| boxes with networking equipment all around the country
| and it happens to your equipment all the time.
|
| 1: https://en.wikipedia.org/wiki/2006_European_blackout
| tharkun__ wrote:
| Yeah historically that makes a lot of sense to me.
| Reactors going offline directly would usually be planned
| and thus not cause instability.
|
| The 2006 one I had read about before. I love reading
| timelines of such disasters. Shows how hard this actually
| is and how much work it is to keep it all running.
|
| Here's another one: https://en.wikipedia.org/wiki/Northea
| st_blackout_of_2003#Tim... And speaking of Canada and
| power lines (this time it does include transmission
| lines) https://en.wikipedia.org/wiki/January_1998_North_A
| merican_ic... While not so severe this is basically the
| kind of thing I was referring to us happening in
| "shoulder season". There's usually at least one ice storm
| or very wet snow event at the start and/or end of winter
| now and it's very likely that in our wooded area we get
| trees into power lines and boom the buckets go. When
| we're lucky it's localized and crews are available to
| come out and fix it in a few hours. If it's all over the
| place then it's gonna take a while and they'll have crews
| from other provinces and the US come in to help as well.
|
| I'd be interested in your outlook on the future of the
| grid in Germany and Europe though. Of course when France
| takes a nuke offline, that's usually planned, even when
| done for a "river water temperature emergency" it's gonna
| take a while and you can bring that coal plant online
| like you mention. But doesn't Germany want to reach the
| climate goals it set itself? How does coal make sense
| there? And how is shutting down their own nukes a thing
| when it's OK to use French nuke power?
|
| Or some natural gas, which is quicker. If you have the
| gas. Re: Russia.
|
| In all of the European (NATO) countries together, is
| there enough generation capacity if you assume zero
| Russian inputs (save for say untraceable third party
| transit or resources) and half of France's nukes going
| offline? Especially when the sun doesn't shine because
| bad weather and thus the winds are so high that you have
| to shut down your wind turbines?
| bobthepanda wrote:
| At least during the Texas grid failures during their
| freeze the other year, gas ended up not being a backup
| because the lines froze over.
|
| Something is only really a backup if the actual fuel is
| on site, at this point
| znpy wrote:
| > for the six years I knew about it, it never did.
|
| my dude i cannot tell you how much i love these stories of non-
| faang, real world engineering for extreme reliability.
|
| thank you for posting that.
| chx wrote:
| my fav real world engineering story is of the first
| commercial ISP in Hungary
|
| back then running leased lines would've been _way_ too
| expensive. So what did the kids running the show did? They
| got wind of a central office in an older part of Budapest
| have excess capacity so they rented an apartment in the next
| building and _drilled the wall_ :D no expensive trenching, no
| expensive equipment to demultiplex landlines, nothing, just a
| bunch of wires running straight from the CO equipment into
| retail modems... We had no idea what we were doing, mind you.
| I was already doing Linux at the time so among the few
| installers I was the lucky guy who got to install the
| Internet at a small business who wanted it to be done on a
| Solaris workstation. That was a fun challenge... Other
| installs were Trumpet Winsock. The ISP itself ran a custom
| linux app, you dialed in and landed on a text app or maybe it
| was Lynx? can 't quite remember, it's been 30 years...
| linsomniac wrote:
| >run it from a different circuit but the same utility power.
|
| If you're going to come from a different circuit, see if you
| can at least pull it from a different phase, if we're talking a
| 120v circuit on 240v service (typical US home service). It's a
| small improvement, but I'd say 10%+ of the power outages I've
| seen have been just a single phase going out.
| DrPhish wrote:
| I have also seen a single phase go out (bird poop into a
| transformer) and having redundant PSUs on different phases
| saved our bacon
| dylan604 wrote:
| The house I grew up in lost transformers multiple times a
| year from squirrels.
| hunter2_ wrote:
| The center tapped transformer that provides residential split
| phase 120/240 on its secondary winding connects to just 1 of
| the 3 grid phases on its primary winding. If that one phase
| goes out on the grid, both sides of the split phase service
| go out together. There would need to be a fault on just one
| side of the split, downstream of that transformer, for your
| suggestion to hold up. Certainly not impossible, but far less
| common than a "losing one phase" scenario which would
| typically originate upstream on the high voltage side.
|
| On a commercial 3 phase service, yes, connect redundant PSUs
| to separate phases, since each phase on the panel actually
| corresponds to each phase of the grid.
| kurthr wrote:
| But if the issue is a local breaker flip, then being on a
| different phase is very effective!
|
| I don't have a power outage more than once a year, but we
| manage to blow a breaker more than a few times a decade
| (vacuum + water boiler was one).
| anamexis wrote:
| If the issue is a breaker flip, then you only need to be
| on different breakers, right? Phase is irrelevant in that
| case
| hunter2_ wrote:
| Right, because the main breaker would be 2 pole (or at
| the very least handle tied).
| cesarb wrote:
| > The center tapped transformer that provides residential
| split phase 120/240 on its secondary winding connects to
| just 1 of the 3 grid phases on its primary winding.
|
| I believe that's a USA peculiarity. Where I live, the usual
| residential and commercial wiring is from 13.8 kV to
| 127V/220V through a three-phase delta-wye transformer, in
| which the primary connects between each pair of phases, and
| the secondary connects between one phase and the neutral
| (the high-voltage primary side does not have a neutral).
| When one phase of the high voltage side is lost (very
| common, since each high-voltage phase is a separate wire
| and has an independent fuse upstream of the transformer),
| what happens is that one phase of the low voltage side
| stays normal (the one between the two intact high voltage
| phases), and the other two have a lower voltage which
| varies depending on their relative load.
| kelnos wrote:
| > _I believe that 's a USA peculiarity._
|
| Yep. Sounds like you have true 3-phase service, whereas
| most places in the US just get split-phase.
| Maskawanian wrote:
| You can definitely loose one phase of a split phase
| transformer. I've seen it 3 times in my life. Sometimes it
| is completely out, other times I've seen only getting 80v
| rather than the full 120.
|
| All depends on how it failed.
| j45 wrote:
| I read the graphic differently and it reads ok to me.
|
| The two different power sources are different shapes.
|
| To someone who's setup things in datacentres, it seems pretty
| reasonable to see that could be 2 different circuits.
|
| Of course it should be 2 separate plugs with the 2 different
| shapes in them.
|
| As for batteries you can just get a UPS system that supports
| adding extra batteries to it.
| nuker wrote:
| The main point against seems to be that downstream UPS will not
| like "simulated or modified sine wave" if upstream UPS is not
| producing "Pure sine wave". To be tested :)
| dgacmu wrote:
| I've tested it before. TFA is correct, at least with the APC
| UPS I tried it with once. The apc ups went to battery and then
| turned off when out of battery.
| nuker wrote:
| Thanks! But why they don't like it? UPS, as a black box,
| should harvest whatever energy is in the source, regardless
| is it sine wave or freaking square mess. Even DC or RF is
| present :) Why so picky?
| boricj wrote:
| That's true to an extent for online UPSes, which converts
| from AC to DC and then from DC to AC, completely isolating
| the input current from the output, but these are the most
| expensive types of UPSes.
|
| Your run-of-the-mill UPS is likely to either be offline,
| which forwards the input to the output, or line-
| interactive, which can compensate to an extent under or
| over voltage conditions with a regulator. If the input
| current characteristics are outside allowable tolerances,
| they can't compensate and must switch the load to battery
| to continue powering it.
| nuker wrote:
| > If the input current characteristics are outside
| allowable tolerances
|
| This is the key to this whole discussion. I guess it
| boils down to existing line-interactive designs, why they
| can't work with simulated sine as a source.
| dgacmu wrote:
| A transfer-switch based UPS still needs to protect
| against brownouts, so it too has line behavior parameters
| outside of which it will switch to battery. It's an
| interesting question if/why a modified sine wave input
| would trigger that or not. Probably one of those "it
| depends on the design" things. (modified sine ->
| approximately an oscillating square wave.)
|
| A dual-conversion (online) UPS is almost certainly more
| robust as far as what kinds of inputs it can accept
| (though as GP noted, they're more expensive, and they're
| also less efficient due to the additional
| rectification->inverter).
| nuker wrote:
| > Your run-of-the-mill UPS is likely to either be
| offline, which forwards the input to the output,
|
| This one surely should "like" simulated sine wave as a
| source and don't drop to battery?
| boricj wrote:
| An offline UPS will switch to battery if it doesn't like
| the input. What qualifies as acceptable input depends on
| the design and specifications of the UPS. If it expects a
| real grid-like sine wave and doesn't see one it will
| reject it, regardless if the load would like it or not.
| nuker wrote:
| > If it expects a real grid-like sine wave and doesn't
| see one it will reject it, regardless if the load would
| like it or not.
|
| But why would it require grid-like sine wave and not go
| along with whatever is the source, provided source can
| still be used to charge its batteries? I saw no answers
| yet, and this "why" is the very key to the discussion.
| soneil wrote:
| There's a lot of compromise in UPS design.
|
| For example, a common topology in offline UPS is that the
| inverter and the charger are the same circuit driven
| differently[0] - so you can't charge the battery while
| the inverter is carrying load. This is popular at the
| low-end because you have literally half as much UPS, but
| makes what you're describing impossible.
|
| Another common issue at the low end is that the inverter
| isn't thermally sized to run non-stop, they know they can
| cut a corner because your battery presents a finite and
| known duty cycle.
|
| There are ways around this, but at some point you end up
| fixing the wrong problem - eg, it's cheaper, safer, and
| more resilient to buy a transfer switch instead of
| uprating two UPS to be capable of daisy-chaining.
|
| [0] https://patents.google.com/patent/US5302858A/en
| numpad0 wrote:
| I think you're being gatekept. It's inconceivable that
| just no one knows why square and triangular waves don't
| pass as AC.
| nuker wrote:
| In other words. I asked if A should lead to B. You
| answered "A leads to Not B" :)
| boricj wrote:
| Another important point is that an UPS can produce huge surge
| currents, for example when switching from battery to utility.
| The upstream load goes from zero to the sum of both the
| protected load _and_ the UPS charging its battery.
|
| When daisy-chaining an UPS to another, the downstream UPS can
| easily overload and trip the upstream UPS because of that.
| lucumo wrote:
| Wow. Your post is much more informative and clearer than the
| article. And it's only three sentences.
| nuker wrote:
| > Another important point is that an UPS can produce huge
| surge currents, for example when switching from battery to
| utility.
|
| It has not happened yet. If downstream UPS likes the
| source, it is not going to switch to battery, and battery
| charging will Not be happening because downstream UPS had
| it charged already long time ago.
| effluvium wrote:
| An easy fix for that would be to put a surge protector
| between the UPSs. (Joke)
|
| ------------
|
| I bought three UPSs, all them used independently; but I had
| to stop using all of them. They all had a capacitor whine
| that was driving me crazy.
| myself248 wrote:
| Another thing they don't mention is that when utility AC returns,
| each UPS in the chain tries to recharge, which can be a
| significant chunk of power. Any given UPS is sized so that its
| downstream load plus its own recharging power doesn't overdraw
| the circuit, but add two recharging powers and it's much more
| likely.
|
| I don't think this is an issue with double-conversion UPSs, since
| their input power is fixed by their rectifier size and they'll
| simply charge more slowly instead, but with standby type, it's
| very much a concern.
| lucumo wrote:
| What a stupid article. It doesn't tell you anything beyond "no,
| buy our products instead". WHY is it bad? HOW does it fail?
|
| I mean, it's possible that I don't know enough about UPSes to
| understand the finer details and all this would be obvious to
| someone who does. But presumably they already know why it's a bad
| idea, and don't need this particular article to explain them.
|
| The amount of scare words and the lack of detail honestly make me
| believe it probably IS okay to do, and Eaton just wants to scam
| some extra money out of people. It's just the posts in this
| thread that make me take it slightly more seriously.
| tssge wrote:
| >Eaton just wants to scam some extra money out of people
|
| Seems quite improbable that would be the purpose of the
| article. The article is basically saying "buy a single UPS
| instead of buying two to daisy chain them expecting to get
| better results". If anything, such advice would lead to people
| buying less devices.
|
| Of course if one had two UPSes laying around for nothing, then
| I guess daisy chaining might come to mind to get better
| capacity or something for no extra price. However, generally
| people don't randomly have UPSs laying around for no reason, so
| it would make sense to buy higher capacity UPS than two lower
| capacity UPSs to daisy chain them anyways (and for probably
| quite similar total cost of ownership).
| nixpulvis wrote:
| Let's say I do have two UPSs lying around though, and a
| soldering iron.
| pixl97 wrote:
| >make me believe it probably IS okay to do
|
| Yea, as someone that ran their own micro ISP for a while, and
| worked for larger ISPs, no it's not ok to do at all. APC brand
| was the absolute worst about immediately tripping when plugged
| into another UPS. Eaton online UPS' actually handled it ok in
| comparison, but they are typically pretty expensive units.
|
| This is definitely a case of the vendor attempting to save you
| money. Get an ATS on anything that doesn't accept dual power.
| tssge wrote:
| >APC brand was the absolute worst about immediately tripping
| when plugged into another UPS.
|
| At least from my experience of owning multiple APC UPS
| devices, they have a customizable acceptable power quality
| setting. In such cases setting them to accept the absolute
| worst quality of power could probably stop them from tripping
| on bad power input.
|
| No idea if this affects the end devices, however there's
| probably a reason other than simply extra profit for the
| default power quality tolerance setting on those devices.
| Generally they are set to rather low tolerance threshold as
| the expected usage scenario is servers and other relatively
| sensitive equipment.
|
| The models I own aren't really the most expensive either,
| some of the lower end tower models and they still have
| configurable acceptable power input settings available.
| Regarding them tripping on bad input and being "absolute
| worst", I consider this tripping a feature more than an anti-
| feature, especially as it is user configurable.
|
| EDIT: Also wanted to add that it is actually _preferable_ for
| UPS to trip as immediately as possible on bad power input.
| That is the only purpose of such product after all: to
| protect the devices attached to it from bad or otherwise
| inadequate power input.
| Waterluvian wrote:
| I'm guessing the sine waves wouldn't line up?
|
| Edit: derp. Wouldn't matter. I was thinking putting them in
| parallel.
| CapitalistCartr wrote:
| The places I've worked on used multiple generators, more than was
| required to keep running. Home Shopping Network had 8, 5 of which
| would keep them on air. That's an extremely reliable backup.
| toxic72 wrote:
| Can you share more about the Home Shopping Network's broadcast
| infrastructure - that's such an explored little nugget of the
| world I've never given any thought to before
| CapitalistCartr wrote:
| This is 30+ years ago, they had a halon system. They had
| calculated the volume of the rooms, including the contents so
| as to not kill people while supressing fire. So making a hole
| in a wall, or adding/subtracting _anything_ was a huge
| production.
| NelsonMinar wrote:
| This page is SEO crap. A thin article with little technical
| content peppered with links to the company's own products.
| rietta wrote:
| I know from experience that plugging an old school digital alarm
| clock into a UPS does not work. The clock drifts badly. Turns out
| those use the AC cycle rate as their clock.
| kadoban wrote:
| It should depend on the UPS, some I believe are
| indistinguishable from wall power, especially without actually
| trying.
|
| Even for the cheaper UPSes I wonder if the issue isn't the
| cycle rate, but the cycle shape? My understanding is that they
| tend to be able to hit 60Hz pretty easily, but the cheaper ones
| are a ~square wave instead of a ~sine wave. Maybe the digital
| clock just glitches on that more.
| yafosuda wrote:
| To clarify this comment: there are two primary types of UPS,
| double conversion and line interactive.
|
| Double conversion takes power from AC, converts it to DC to
| charge batteries, takes battery output, and inverts it back
| to AC. All power drawn from the UPS goes through the battery
| and inverter stack, and there is no transient/power loss when
| AC mains are lost. They tend to be more expensive, louder,
| run hotter, etc.
|
| Line interactive UPSs, on the other hand, tend to be cheaper
| and are in most cheap consumer products. They take AC mains,
| convert it to DC, and charge batteries. But AC mains is also
| connected directly to the output device through switch
| circuitry that will quickly switch the power source from AC
| mains to batteries/inverter if power loss is detected.
|
| Reputable UPSs will use pure sine wave inverters for
| converting DC battery back to AC. Modified sine waves are
| indeed a lot cheaper but are not suitable for some sensitive
| equipment.
| LorenPechtel wrote:
| Yup, made a variation of this mistake once long ago. Ordinary
| clock, plugged into a transformer connected to Chinese (220/50)
| power. It ran 20% slow.
| digitalsushi wrote:
| I had an underpowered battery backup for my desk, and was using
| it for my gaming PC, two laptops, and two external monitors, and
| all my peripherals. (It didnt occur to me I was overrunning it,
| otherwise my rant here is nonsensical)
|
| The monitors both had the same general issue - they would fail to
| find a signal every few weeks, and I'd find that waiting a few
| hours with them turned off would help.
|
| I have pages of notes here. Another way to get them to work would
| be booting up the windows computer, which would seem to 'trick'
| the monitor into getting a signal on hdmi, and then I could
| switch to display port for the mac laptops to be used.
|
| Anyways it's all crazy rambling notes, with copious timestamps,
| looking for patterns. I have an IR heat thermometer from the
| kitchen and my monitor vents would regularly have air over 160F
| coming out the tops when the monitors would not even post the
| vendor logo after a hard power reset.
|
| I removed the battery backup and it's been months now with zero
| blips. So the only obvious takeaway I have is that overrunning a
| battery is a completely worthless endeavor.
| lmpdev wrote:
| What I find insane is the peak current discharge those little
| 12V 7-9ah backup batteries can push out
|
| Well over 100A for a few seconds
|
| Lithium can't even do that in a single cell (but in turn is
| infinitely better for continuous current)
|
| It sort of baffles me intuitively that an extremely simple lead
| battery can for a short while compete with the _grid_
| jpgvm wrote:
| Yeah this is why lead acid was hard to replace in cars with
| lithium chemistries.
|
| The Cold Cranking Amps (pretty much the peak startup current)
| that they can provide is rather insane. Even my small
| motorcycle battery provides over 250 CCA.
| elintknower wrote:
| Super capacitors / just large capacitors in general might
| eventually save the day here. The only issue is super
| capacitors can't generally bake in the heat all day and
| still work for 3-5 years like lead acid batteries can.
|
| Supercaps is how "fast" charging in most smartphones works
| as well since they can soak current faster than the battery
| itself and also mitigate cycling batteries too hard.
| jnsaff2 wrote:
| It won't be anywhere near 12V while supplying those 100A.
| p1mrx wrote:
| 12V x 100A = 1.2 kW. A typical grid connection for a house is
| 240V x 200A = 48 kW.
| somehnguy wrote:
| Lithium polymer pouch cells can easily push out well over
| 100A, even relatively small cells hit this mark no problem.
| russdill wrote:
| Are they Asus monitors by any chance?
| j45 wrote:
| It might also be helpful to ensure you have a battery backup
| that provides pure-sine wave electricity when plugged in, and
| ideally as close to it as possible when on battery backup and
| the power is out.
| jjeaff wrote:
| most electronics don't need pure sine waves as they are
| converting the ac to DC anyway. but perhaps a square wave
| could throw off some PSUs.
| danjl wrote:
| Do kids these days actually make Daisy chains anymore? You know,
| with real Daisy flowers.
| AzzyHN wrote:
| I was unaware that was the origin of the term, so probably not
| elintknower wrote:
| Is anyone else here using Ecoflow batteries as a UPS backup? I
| really wish Eeaton would produce a similar consumer facing
| function that is intended to be used long-term 24/7 not just on
| weekend trips etc.
| necovek wrote:
| This is the type of article I find useless.
|
| It would be great if Eaton went to the trouble to explain what
| exact conditions the two UPSes need to fulfill to be successfully
| daisy chained, which would probably put people away from doing it
| anyway.
|
| But it would also be a much more informative article, and also
| positively framed, which is always a much better read.
| deadbunny wrote:
| That just leads to "I followed your instructions on your
| website and my house burned down". No company is going to open
| themselves up to that.
| zootboy wrote:
| They don't explain it because it basically never makes sense to
| daisy-chain a UPS. If you want redundancy in your power supply
| chain, use a dual-PSU computer. If you want longer backup
| times, get a UPS with a bigger battery.
| BuildTheRobots wrote:
| The only time I think it makes sense if if you're doing a
| belt+braces with colo'd stuff in a datacentre.
|
| The DC will have its own UPS system, but in the event it all
| goes wrong (which happens more often than I'd like) you
| probably want something in your cab that can give your
| equipment notification and time to shut down safely, and
| provide surge suppression and maybe some level of isolation
| from the inevitable back-emf caused by the rest of the hall
| going dark.
___________________________________________________________________
(page generated 2024-04-28 23:01 UTC)