Online UPS - what am i missing

[hacklordsniper]

As im renovating my lab im hunting a 3000VA OnLine ups.

after reading some brochures i cant understand the following. Most of them have a battery of 72 VDC(12 V SLA in series) and stated the following " max charger curent 1-4 A (manufacturer dependent)". But how will then this charger feed the battery which is feeding the 3000VA inverter all the time?

or they feed the inverter from separate power supply and battery kicks in when the line fails?

[alm]

or they feed the inverter from separate power supply and battery kicks in when the line fails?

I think this is the case. The batteries are not usually rated for the full 3kVa continuously, they are only designed to survive this for the ~10min or so it takes to discharge after power failure.

[Kiriakos-GR]

I am hunting a 3000VA OnLine ups.

Search for a used APC.
It would had an less salty price, and it would be a true 3000VA .

[NiHaoMike]

A true online UPS does not switch. The rectifier is rated for however many VA the UPS is, but there's a current sensor (shunt or Hall effect) that limits the battery charging current.

More likely, though (especially for a newer one), is that the UPS is a "high rail" design where the batteries are separated from the main DC bus with a DC/DC converter or two. That's actually a superior design - it offers the same power quality as an old style online UPS but with a greater efficiency and lower weight and cost. The large DC bus capacitors provide a buffer while the DC/DC converter(s) switches. The most advanced ones can even support alternative energy inputs. (Beware that while every old style online UPS can support alternative energy just by adding a charge controller of the proper voltage, the "high rail" ones generally cannot unless they're designed for it.)

[hacklordsniper]

Thank you for clarifying the situation to me.

Some cheap Chinese UPS are using even a trick. They are tagged "on-line" but they use new "innovative" approach working in bypass mode and switch to on line only if the line is not within the adjusted window. They cant be forced to work in "real" on-line mode or they would catch fire 

Finding an used APC of this size is not possible whatever i do. New 3000 VA on-line APC costs close to 3000 $ here which i think is too much. Regarding other manufacturers there is Centralion Titan series for 1200 $ and OPTI-UPS 1140 $. I like the OPTI-UPS and all seems fine except the UPS in default mode work also in bypass mode but can be switched to on-line without any problems.

There are only two more things bothering me.

1. The APC vendor says that the APC fan spins only when the device is working full load and heated while it is off in all other scenarios (however the APC vendor will not allow to test the unit so i think he is lying). The other mentioned manufacturers UPS-es sound like a jet plane taking off even when there is basically no load on them.

2. The APC in desired flavor has a weight of ~55 kg, while other manufacturers claim from 24 kg to 31 kg with same amount of batteries. Where did they implant the diet process in heatsinks or possibly in the transformer using a SMPS?

Did anybody open APC SURTD3000XLI or have it and say something about it? IS the fan working all the time? I want to hear my thoughts in the room 

[amspire]

In a previous life, I designed UPS's.

In terms of a UPS as a battery backup device, online ups's and all the varieties on non-online UPS's are about the same.  Sure there is a 10ms to 20ms loss of power while the relays are switching, but all online UPS's have a bypass relay for surges (like you are near full load and something is turned on). So even with an online UPS, the PC's have to be able to survive a temporary loss of power. All PC's on UPS's should have good quality power supplies that can cope.  The switchmode power supplies in all the computer gear don't really care how pure the sinewave is - they just don't like spikes.

The problem with all switch-over type UPS's is that if a true 3000A lightening strike hits them, they can prevent most of the surge reaching the computer, but you can still easily get 600V to 1000V - enough to kill most PC power supplies. That first lightening strike will kill the MOV's on the UPS, so if there is a second strike, the UPS will not offer much protection at all, and that's when the PC can get rather fried inside.

It doesn't help that to save money, the filter components in almost all UPS's are just nowhere near heavy enough.

A well designed online UPS will have the same problems, but even if the input inverter of the UPS is fried, the load is isolated and so your computers survive.

Fan noise?  Again it is all cost saving. We designed our UPS's with 100% fan-less cooling, but it means that we used very big custom heatsinks as part of the case. Our thinking was that fans are one of the most unreliable parts, but if you cut out the $20 fan, you probably have to replace it with $300 of extra heatsink and extra case and construction costs.  So they put in the fan.  Most UPS's are run at a fraction of full load, and I would believe that the APC fans are quiet at partial load.

Another joker in the pack is that again due to cost savings, UPS's are frequently less reliable then the computers they are meant to protect.  It has always been very hard to get customers to pay $2000 for a quality built UPS to power their $1000 computer.   The result of this is that even though using a UPS will mean much less disastrous damage to computer equipment in the case of lightening surges, it often does mean more computer outages.  In many situations, people make the choice of not using UPS's and allow for the fact that after a lightening storm, they make have to get new hardware.

There is one massive mistake people make with UPS's. In the case of a lightening strike, all the hardware directly connected to the computer must be connected to the same UPS, because in the short term, the ground and neutral can easily get several hundred volts above the  ground and neutral of other power points. The main culprit is the USB laser printer. It usually cannot be on the UPS due to the high surge current needed by the fusing roller, but if it is connected to a UPS protected computer, then the computer may still be destroyed via the USB connection.  Ethernet connections provide some better isolation usually to the extent that at worse, only the actual device on the cable heading off to computers gets fried.  So the switch dies, the computer survives.

About the weight, if you are going to put 22kgs of batteries into a 24kg UPS, is there any substance to the rest of the UPS?  Are you actually going to trust cut price lightweight components to be reliable and to give any kind of protection? The whole point is that the power supplies in a UPS are meant to be much much tougher then the supplies in the computer gear, and that almost always means heavy.  Huge inductors, huge capacitors, heavy wiring, and a proper construction designed to isolate the output inverters from anything happening on the input.

The last thing is before you buy a UPS, have a look at the connection to the PC and the software. In many cases you can download the software for free and try it out. The reason I am mentioning this is that if you want to see the worst quality software ever written, take a look at UPS's. You need the software as you only have a few minutes of battery life. Something has to shut the computer down before the batteries die, otherwise the UPS is almost pointless. You may have several computers which means you need agents to shut the other computers down too.


Richard

[hacklordsniper]

Thanks for your nice concerns. To power my lab i used solar power + sine inverter + grid charger (when there want enough sun and battery capacity and that was mostly always). I decided to ditch all of that setup because I'm not planning to use solar anymore and simply want a compact and nice unit to power all instruments and devices in my lab. The usual power draw is around 1000 VA MAX so this is why i decided to buy the 3000 VA ups to have a reserve and prolong the unit life by derating its specs.

All of that is because the line in my village is catastrophically and two years ago there was an incident when they reversed some wiring on the distribution transformer and burned all line connected devices on the street. Luckily i was just moving in and haven't installed all my toys

Today i have managed to see some cheap Italian mad UPS-es and its just crap 

I think i will go for the APC, would like any users with experience with mentioned model to write the fan noise.

[Kiriakos-GR]

I own the APC SUA1000XLI   and I had buy it new at about 800$ , the XLI writing translates to " Huge transformer inside " capable to charge four extra battery boxes that extends the on battery time  up to 24 hours ...

The  fan  in my unit it gets active immediately, if the unit start working as backup power source ,
and the fan it self it is noisy. ( High speed SUNON inside )
( Its advisable the UPS to be in a different room "storage room" and not next to your ears. )

In those XLI series that targets large Data centers and Servers , what boost the price is also the software interface that comes with them.
They can be remotely controlled and and remotely inspected , you could  add three or more 1000XLI  to act as a single 3000 unit.

Speaking of batteries its the most costly items today.
Inside my unit I found  some no-name (Made in India), that lasted for three years.

Panasonic offers three identical models with different aging  characteristics = lasting in time.
a)  2-4 years
b) 4-6 years
c) 8-10 years.

I got the 10 years version, even if its battery costs 150$ its one and I need two.  ( 40 min run-time)   
If I needed to add all the extra four battery boxes, plus batteries in the UPS unit,
I would need 10 batteries = 1500$ 

All those battery packs needs special "Anderson power connectors" SB50A to be connected in series with the master unit.
I got two of those connectors ( from UK) so to be able to extend the battery capacity by my own " home made battery packs " that I use on them used batteries who still have some life in them, even if they are not in top shape.
 
The UPS systems its a serious investment, and needs to be seen as such.
And it is very costly too.

What I did even if my needs was not to protect a data center, was to buy capable equipment for this task,
so to get in my hands an UPS that is designed to have a true long life.

The single UPS units, that do not accept external battery packs, cost 1/3 of the price,
everything in them as circuitry and charger, is made to serve a single battery pack,
and they are poor made speaking about longevity, because hey are designed to run for just few minutes.
And yes they are very prone to damage.

( I was ready to post pictures too, but there is a issue with the server !! )

[amspire]

A look at APC's site seems to indicate that the SUA models that Kiriakos-GR talks about have two speed fans, with the fans going to high speed if the UPS is on battery, if the batteries are charging or if the load exceeds 75%.  I guess that means that as long as the battery is charged and mains is on and you are running at 1000W, then the fans will be running in low speed mode and it would be fairly quiet.

It looks like for the SURT ups's, the fan is always on high speed and they claim a noise of 55dB at 1 meter.  As the fans get older, they get louder. Probably a bit too noisy if you are sitting next to it for hours.

Also the 3000VA UPS's need a 15A power socket which is usually more then the standard sockets.

It is a great pity that Constant Voltage transformers have problems : super heavy, noisy - noticeable 50Hz hum , distorted output waveform and expensive. They do probably exactly the thing you need - with good fuses, they are just about indestructible and they offer near complete protection to the loads under any kind of mains problems. You never need to buy batteries. If you found a second hand one, you definitely would need to put it in a separate room away from people.

Richard

[hacklordsniper]

Is there possibly a recomendation for a fan-less ups in desired flavor ~3000 VA?

I really dont have the possibility to put it somwhere else and sometimes i really need silence. The model im interested originally is http://www.apc.com/products/resource/include/techspec_index.cfm?base_sku=SURTD3000XLI&total_watts=200 but too loud

[A Hellene]

This is a very interesting video clip about the operational details of a True Sine Wave UPS (an APS NS3000RMT3U 3KVA UPS) in both its operating phases: The high voltage power inversion and the battery charging, using an ordinary low frequency (20KHz) power transformer (or two 1.5KVA ones in parallel, in this specific model) and a PWM/PDM drive approach.



Though this is not an on-line UPS, to protect the load better by isolating it from the mains, its response is almost instant because it does not use any relays to switch phases since it constantly monitors the mains waveform and takes over in a small fraction of a cycle after the detection of some input power anomaly.


-George

[NiHaoMike]

For extra surge protection, what about plug the UPS into a good surge protector? That'll also protect the UPS itself. Also, use a heavy wire connecting the surge protector ground to an additional ground (cable TV connector, water pipe, etc.) and loosely coil extra power cord to increase the inductance.

[jahonen]

I certainly wouldn't trust a surge protection what is solely on the mains socket, whether it is UPS or generic SPD, there is just too much energy to absorb if that is only protection.

Good protection can only be achieved if there is a multi-level protection. That can be accomplished by installing heavy duty SPD on the distribution board where good low-impedance connection to the equipotential strip (which is connected to the main grounding wire buried to earth around the house at least here in Finland), is available. I have done that (or, actually had an electrician to do that, I have no license to do mains stuff) in our house here, after lightning went too close for my taste in past summer. I have one suspected VDR failure due that. Well, it did probably what it was for but still I think I need some more insurance against conducted lightning damage.

The SPD I had installed was Phoenix Contact FLT-CP-3C-350, SPD for TN-C system.

https://eshop.phoenixcontact.co.uk/phoenix/treeViewClick.do?UID=2859725&parentUID=852612006

After that is done, the UPS surge protection (or other SPD in general) should be able to absorb rest of the surge pulse without blowing to smithereens if there is a lightning induced surge.

Another issue is that if one has also telephone line or other thing coming in, one must make sure that when surge pulse happens, the potential difference between mains and phone line should not be let to bounce wildly. If that happens, the DSL modem or whatever is connected between these two domains, is in great danger. In practice, that means that telephone line surge protection should work against this same equipotential strip to prevent that. It might cause to relatively thin copper wires of telephone line disappear if there is sufficiently large surge coming in.

Regards,
Janne

[amspire]

Janne,

I hope people read what you have written, because it is absolutely correct.

Whatever surge protection devices you get, most of the protection is done by a few relatively cheap devices such as MOV's or gas surge arrestors, and the place for the primary protection devices is definitely in your switchboard. Properly installed, the surge protector should be accompanied by some heavy duty fuses to get the best protection.

Now that earth leakgage breakers are common in switchboard to all the general purpose outputs, there is a major limit in the size of MOV's that you can put in devices such as UPS's that plug into standard power outlets. A large MOV from mains to ground  has enough capacitance to trip the earth leakage breakers, so you can only use small $0.20 versions of the MOVs.  And remember these are the primary protection devices in your $500+ standby UPS or the $60 power protection boards with the unclaimable 50 gazillion dollar damage protection guarantee. These cheap MOV's probably form 90% or more of the surge protection of standby UPS's or protection board. They might chuck in a few capacitors,and perhaps some undersized inductors, but these are very secondary in their effect.

The good switchboard surge protection devices have surge protection fault signal outputs, and it is important that these go to some kind of indicator that someone will  check every now and then, because if you are hit by a really big lightening strike, then the switchboard device has to be replaced. The way MOV's absorb big strikes is destructive - they are fine during the strike, but afterwards their capacity is diminished by some degree that depends on the amount of energy absorbed.  If you have to replace it, it has probably saved a lot  of the electrical and electronic devices in the building though, so it is worth it.

Richard

[NiHaoMike]

If it's acceptable to have a little more downtime in return for protecting the equipment, what about a SCR crowbar to blow a fuse if there's a 400V or so spike? Normally, the MOVs would clamp and prevent the crowbar from operating, but if that fails, the crowbar takes over. (A fast automatic reset circuit breaker can be added in series with the fuse if necessary.)

[amspire]

Fuses are slow, and even if they blow in time , they will conduct the complete energy of the spike, due to arcing, until the spike completes. It is only after the surge completes that the fuse will be an open circuit.

So if an SCR/Triac shorts the mains out, it will have to absorb a huge amount of energy for a short time, and you would need a massive device to do this without blowing. If the SCR/Triac  physically blows up, there will suddenly be a massive extra voltage transient generated by the inductance of the wiring and the sudden break in the current. Could even be worse then the initial problem. If you have a high power SCR, I doubt it will turn on in time to prevent a dangerous spike getting to the building wiring anyway. Which will probably mean that the SCR would blow from over voltage before it has time to turn on.

MOV's are brilliant at absorbing huge amount of short term energy and that really is the whole point - if a massive amount of energy is dumped into your switchboard, then something has to absorb the energy - the energy has to go somewhere.

The crowbar solution is one that as part of a solution along with MOV's, gas arrestors and filter inductor/capacitor modules could probably work, but I doubt if it a solution on its own.

Richard

[hacklordsniper]

Can somebody comment AEG ups, VFI, online double conversion models? The 3000AV model is priced 60 % of APC and as far i know AEG is a reputable company

http://www.aegps.com/en/data-it/ups/online_ups/protect_c/index.html

[amspire]

I had a look at the AEG site.

I stopped work designing UPS's about 1991.  When I look at the AEG products I am back in 1991 - nothing has changed!

20 years of development and nothing has got any better!

The UPS you mentioned looks like it was designed maybe 2005.  It has a RS232 port and the instructions say "plug this into your computer's RS232 port".

Has anyone told them that computers don't have RS232 ports any more?

Even though adding a RS232 to USB converter costs nothing, they sell it as an extra option! I bet it is not cheap either.  This is on UPS's that costs thousands of dollars. It costs much more to have it as an option, then to build it in.  It is just that is the way they have done it for the last 20 years - why change now?

I said earlier about starting off with the software, and this is very true with AEG.  Living back in the 90's, their software is licensed you get one license for the computer connected to the UPS, and 5 network licenses, so if you ever run more then 6 computers off the UPS, you need to pay more.  I mean, are they kidding.  Who else is going to use their software other then an owner of an AEG UPS, and shouldn't they be entitled to free licenses?   You of course are not allowed to know anything about the RS232 protocol from the UPS - that is top secret, even though as a peripheral device, you can't get much simpler then a UPS.

This may not sound all that important right now but you will have this UPS for many years during which time many computers, or instruments with embedded computers will pass through your lab. You will eventually regret having a UPS built by companies that don't believe in these modern computer thingy's, and who believe that all UPS owners are thieves who just want to steal the software that basically checks for mains fail, and if so, tells the computer to shut down. 

Here is what I recommend.

There is an open source UPS initiative led by Eaton Powerware.  This is code available totally free so you never have to worry whether you have to pay for a new license, or if AEG still have a program for your UPS in 10 years.  Open source is hugely more powerful in potential then the UPS software from ANY vendor.

Here is the list of compatible companies and hardware:

http://www.networkupstools.org/stable-hcl.html

What you are looking for is green and blue entries since these are the only companies openly cooperating with the open source initative. It includes Eaton Powerware, Dell, Compaq, MGE, MicroDowell, Microsol, and Powercom.  I would start with these companies. Even though APC has drivers for many of their UPS's, their latest initative is to use a new protocol that is not available to open source.  Go with the companies looking towards the future, not the guys living in the 1990's. Not the companies who think their very low grade software should be worth a lot on money, even though it is crippled in functionality.

That AEG by the way still has fan noise up to 50db, so it would still be noticeable.  I thought the distortion figure of less then 4% into a resistive load sounded pretty crappy - don't know why it is that bad.

There are definitely some good UPS's in the group above.  Definitely follow Janne's (jahonen) suggestion - it will do more to protect everything in your house including your lab then anything else. It will definitely protect your UPS big time. Then given that extra protection, think about just getting an Offline/Standby/Interactive (or whatever you want to call it) UPS as they will be totally silent under normal conditions, and the fan only has to switch on if there is a mains problem and it has to go to the battery. Paired with the switchboard surge suppressor, it is not perfect but will do a decent job.

If you are getting bigger then a 2000VA unit, consider adding a special power cable from the switchboard to the UPS without earth leakage protection, and add earth leakage protection after the UPS in your lab. You will have to check your local regulations, but it is the preferred way to connect large UPS's, as the protection circuits of a decent UPS can easily trip Earth Leakage protection circuits. Regulations may say you have to hard wire the UPS to the mains, but you can always sneak an inline plug/socket in there. It is really annoying if your power keeps tripping out at random for no good reason.

I think all 3000VA online UPS's will be noisy. Probably more chance of a quiet online 2000VA UPS as they are more consumer oriented.

Once you have a UPS, you will need to get batteries every 2 to 4 years, so don't get a UPS with any more batteries then you really need. Some of the UPS's may need $1000 plus every time you get new batteries.  If you can live with 10 minutes backup time under half load, then that is what you want to buy.

These are my suggestions anyway.

Richard.

[hacklordsniper]

I allready gave it up on the AEG and ordered an  2x APC (( http://www.apc.com/products/resource/include/techspec_index.cfm?base_sku=SURT1000XLI&total_watts=200 )

This UPS comes in 1000 V/A flavor and i decided to put it at my first bench equipment. The second will be used at second bench and i covered all my needs. If i will need  more in future i will buy one more.

Why such decision? Because 3000 costs same as 3x1000 VA, and in normal logic buying three 1000 VA UPS-es should be more expensive than one of 3000 VA.

I will also check is there an official representative here for mentioned protection.

Also i dont care about RS232 since 80%  of my equipment uses it and i have alot of RS232 ports ready and will surely use it for much years that will come