Motherboard Forums


Reply
Thread Tools Display Modes

UPS waveform question for SUN servers

 
 
JJ
Guest
Posts: n/a
 
      12-16-2008, 07:06 PM

I've been somewhat ignorant about UPSs. It was just recently that I
discovered that the main difference between the two UPSs we have is
that one produces a true sine wave (APC smart-UPS), while the other
(which was cheaper but with a higher VA rating) produces a stepped
approximation (APC smart-UPS SC).

These UPSs sit between the wall outlet and two aging sun blade
workstations (a blade 1k - over 5 yrs old, and a blade 2k about 4 yrs
old).

The more expensive UPS with the true sine-wave output just had its
battery die. During its life, this UPS has caused more trouble than
it has alleviated - with the battery sometimes discharging for no
apparent reason and the computer being unceremoniously shut off - why
in this eventuality the UPS would not just run the computers directly
off the incoming power is beyond me, but as I said, I'm pretty
ignorant about these things. I even had the UPS unit and the battery
replaced, but I've seen this type of anomaly occur about once a year.
I thought a UPS was supposed to prevent unwanted shutdowns!
Afterwards, the UPS battery would recharge and all would seem fine
again.

Anyway, because the battery is now truly dead (or so says the UPS
unit), I decided that now would be a good time to upgrade the unit to
one that is more dependable and with a higher VA rating. So I figured
I would get another of the SC units - haven't had any trouble with
that one since I got it about a year ago. Which is when I learned
about the sine-wave Vs. stepped approximation thing. Maybe the SC
unit has just never really been needed since I got it and has always
just passed the power directly through from the wall. At least it
hasn't shut off my system without warning.

My real question is, do these computers (Blade 1k and Blade 2k) really
need a true sine wave of current from the UPS, or is a stepped
approximation good enough (especially for the short periods of time
they would be running off of the UPS)? Would these machines just shut
down if they started seeing a stepped waveform? Would the stepped
waveform cause damage to any of the components if the power was out
for a long enough time? Would I be better off with no UPS for these
computers than ones with a stepped waveform?

Thanks for any info regarding this.

-J




 
Reply With Quote
 
 
 
 
Cydrome Leader
Guest
Posts: n/a
 
      12-16-2008, 09:43 PM
JJ <(E-Mail Removed)> wrote:
>
> I've been somewhat ignorant about UPSs. It was just recently that I
> discovered that the main difference between the two UPSs we have is
> that one produces a true sine wave (APC smart-UPS), while the other
> (which was cheaper but with a higher VA rating) produces a stepped
> approximation (APC smart-UPS SC).
>
> These UPSs sit between the wall outlet and two aging sun blade
> workstations (a blade 1k - over 5 yrs old, and a blade 2k about 4 yrs
> old).
>
> The more expensive UPS with the true sine-wave output just had its
> battery die. During its life, this UPS has caused more trouble than
> it has alleviated - with the battery sometimes discharging for no
> apparent reason and the computer being unceremoniously shut off - why
> in this eventuality the UPS would not just run the computers directly
> off the incoming power is beyond me, but as I said, I'm pretty
> ignorant about these things. I even had the UPS unit and the battery
> replaced, but I've seen this type of anomaly occur about once a year.
> I thought a UPS was supposed to prevent unwanted shutdowns!
> Afterwards, the UPS battery would recharge and all would seem fine
> again.
>
> Anyway, because the battery is now truly dead (or so says the UPS
> unit), I decided that now would be a good time to upgrade the unit to
> one that is more dependable and with a higher VA rating. So I figured
> I would get another of the SC units - haven't had any trouble with
> that one since I got it about a year ago. Which is when I learned
> about the sine-wave Vs. stepped approximation thing. Maybe the SC
> unit has just never really been needed since I got it and has always
> just passed the power directly through from the wall. At least it
> hasn't shut off my system without warning.
>
> My real question is, do these computers (Blade 1k and Blade 2k) really
> need a true sine wave of current from the UPS, or is a stepped
> approximation good enough (especially for the short periods of time
> they would be running off of the UPS)? Would these machines just shut


I haven't check the spec sheets on these products, but since they may be
OK.

The real issue is good vs. bad UPSes. Square wave UPSes are cheaper to
make and will be inherently closer to bottom of the barrel than one that
produces the correct sine wave.
 
Reply With Quote
 
 
 
 
DoN. Nichols
Guest
Posts: n/a
 
      12-17-2008, 03:10 AM
On 2008-12-16, JJ <(E-Mail Removed)> wrote:
>
> I've been somewhat ignorant about UPSs. It was just recently that I
> discovered that the main difference between the two UPSs we have is
> that one produces a true sine wave (APC smart-UPS), while the other
> (which was cheaper but with a higher VA rating) produces a stepped
> approximation (APC smart-UPS SC).


O.K. The pure sine wave is better for systems with input
transformers (typically linear regulated supplies), which are more
likely to be found in audio equipment these days than in computers.

> These UPSs sit between the wall outlet and two aging sun blade
> workstations (a blade 1k - over 5 yrs old, and a blade 2k about 4 yrs
> old).


The SB-1K and SB-2K use the same power supply. I've got a spare
(from a damaged chassis which I got from eBay -- but what I needed was
fine, so the damage did not matter to me. :-)

I not too long ago happened to take the power supply apart
(looking for a C-ring from one of the CPU modules which had gone into
hiding because someone had apparently tried unscrewing one jackscrew
completely before going to the other. :-(

The interior of the power supply is purely a work of art. And
It is a switching power supply with auto input voltage switching. I
doubt that it even knows that the waveform is only a stepped
approximation.

> The more expensive UPS with the true sine-wave output just had its
> battery die. During its life, this UPS has caused more trouble than
> it has alleviated - with the battery sometimes discharging for no
> apparent reason and the computer being unceremoniously shut off - why
> in this eventuality the UPS would not just run the computers directly
> off the incoming power is beyond me, but as I said, I'm pretty
> ignorant about these things. I even had the UPS unit and the battery
> replaced, but I've seen this type of anomaly occur about once a year.


Hmm ... circuit breaker tripped because too much load was on all
of the outlets serviced by that breaker?

Cat or dog knocked the plug from the wall? (This happened to my
wife's computer recently. I was hearing the alarm, but we have so
many things running that I could not track it down (I thought that it was
the keyboard beep). It communicated via the serial port to her SB-1K
and shut it down cleanly as intended, instead of suddenly dropping the
power from under the machine. If she had looked at the LEDs on the USP
front panel, she would have seen that it was running from battery power
and we could have prevented the shutdown entirely.

Automatic battery test with the VA load too close to the maximum
rating of the UPS and aging batteries? I've encountered this causing
the UPS to shut down for a short time when it shifts to powering the
system from the batteries to see how they handle the load. Normally,
all you may notice is a beep from the UPS when it switches to battery
power, and then silence. It typically gives beeps in Morse Code to tell
you what it is alarmed about. :-)

> I thought a UPS was supposed to prevent unwanted shutdowns!


Not if it loses AC power long enough.

You need to have the audible alarm turned on, and to notice it.
(I failed the latter in that one case. :-)

> Afterwards, the UPS battery would recharge and all would seem fine
> again.


It sounds like the power cord connector loose in the outlet. Or
perhaps a house made during the period when aluminum house wiring was
legal. The aluminum crushes under the screws over time and starts to
make poor contact, which could leave your UPS without power for long
enough to discharge the batteries and to force a shutdown.

> Anyway, because the battery is now truly dead (or so says the UPS
> unit), I decided that now would be a good time to upgrade the unit to
> one that is more dependable and with a higher VA rating. So I figured
> I would get another of the SC units - haven't had any trouble with
> that one since I got it about a year ago. Which is when I learned
> about the sine-wave Vs. stepped approximation thing. Maybe the SC
> unit has just never really been needed since I got it and has always
> just passed the power directly through from the wall. At least it
> hasn't shut off my system without warning.


I don't think that the SB-1K or SB-2k care whether it is true
sine wave, stepped approximation, or even just a square wave. :-

> My real question is, do these computers (Blade 1k and Blade 2k) really
> need a true sine wave of current from the UPS, or is a stepped
> approximation good enough (especially for the short periods of time
> they would be running off of the UPS)?


They should not care.

> Would these machines just shut
> down if they started seeing a stepped waveform?


Nope!

> Would the stepped
> waveform cause damage to any of the components if the power was out
> for a long enough time?


Nope -- except perhaps for transformer-powered audio equipment.

> Would I be better off with no UPS for these
> computers than ones with a stepped waveform?


Nope!

My own preference is an UPS no longer made -- but available on
eBay (usually quite a few). Look for "Best power systems" and
"FERRUPS". These have a Sola type of constant voltage transformer, so
low power is just boosted to the proper voltage with no need for the UPS
to actually run from the battery, and surges are automatically reduced
to reasonable values. When the power actually goes away, the UPS starts
driving other windings on the transformer from the battery, and
continues to produce a clean sine wave.

There is a serial port on the UPS which an be connected to ttya
or ttyb and monitored by a dameon called "checkups" which can be
downloaded from the people who took over Best Power Systems if you don't
get the CD-ROM which includes it as part of the package.

It comes pre-compiled for several versions of unix, including
Solaris (and even SunOs 4.x.x IIRC), and with source code to allow you
to compile it for other systems.

You start it from an entry in /etc/init.d linked to one in
/etc/rc2.d (or wherever you want it to start). You can pass to it how
long before the UPS runs out of battery charge you want the system to
start shutdown, and how long you want to give the system to shut down
before you actually drop the power to preserve the batteries.
(Discharging them fully does terrible things to their life.) The UPS
shuts off the power when it is told to, and then returns power once AC
power returns. The system can be set up to boot on power restoration.

I've got a 700 VA BEST UPS powering my wife's SB-1K, a 1.4 KVA
one powering my SB-2K with a tray of seven 72 GB FC drives in a zfs
system, and a 2KV one with four 12V batteries in series powering the
server (Sun Fire 280R and two trays of seven 72 GB FC drives in zfs
filesystems, plus an Exabyte 430 tape library, and two Sun Ultra 10s,
two SPARCstaton 5s, a Sun Ultra 5 (acting as a firewall), a
router/T1-(CSU/DSU), several twisted-pair ethernet hubs, and
occasionally an Intel Mac Mini.

If you get a BEST UPS -- ask them to ship it without the battery
or batteries, and get a new one to fit it once it is at your home. The
UPS itself is quite heavy, and add a gel lead-acid battery and you will
have something too heavy to move. I did bring the 2KVA one up through a
trapdoor from the garage/workshop with a hoist and winch. The others
were carried upstairs as the main chassis, the case, and the battery,
and all were joined only when in the final location. (You will likely
need some wrenches to attach the cables to the batteries, and be careful
to not short across the two terminals with the tools. :-) Typically, I
needed a 1/4" drive ratchet set with sockets for 1/4-20 bolts and nuts
on the larger ones, and a 3/16" Allen wrench for the smaller one for the
700 VA unit. I got that battery (34 AH) at a hamfest (A vendor cycling
his stock which had been kept properly charged to sell newer stock to
his regular customers), and Batteries Plus for the larger 80 AH ones for
the 1.4 KVA unit and four of them for the 2.0 KVA unit.

As long as the batteries are replaced every four years or so I
expect no problems -- other than from my wife or the cats kicking the
plug out of the outlet -- and then I expect a clean and graceful
shutdown.

Good Luck,
DoN.

--
Email: <(E-Mail Removed)> | Voice (all times): (703) 938-4564
(too) near Washington D.C. | http://www.d-and-d.com/dnichols/DoN.html
--- Black Holes are where God is dividing by zero ---
 
Reply With Quote
 
westom1@gmail.com
Guest
Posts: n/a
 
      12-17-2008, 05:46 PM
On Dec 16, 2:06*pm, JJ <(E-Mail Removed)> wrote:
> I've been somewhat ignorant about UPSs. *It was just recently that I
> discovered that the main difference between the two UPSs we have is
> that one produces a true sine wave (APC smart-UPS), while the other
> (which was cheaper but with a higher VA rating) produces a stepped
> approximation (APC smart-UPS SC).
> ...
> My real question is, do these computers (Blade 1k and Blade 2k) really
> need a true sine wave of current from the UPS, or is a stepped
> approximation good enough (especially for the short periods of time
> they would be running off of the UPS)? *


Output from a computer grade UPS (your cheapest ones) is also called
stepped sine wave or modified sine wave. For example, the output from
a 120 volt UPS in battery backup mode is two 200 volts square waves
with a spike of up to 270 volts between those square waves. That
output (even the spike) must not be destructive to any minimally
acceptable computer. Industrry standards for computer power supplies
make that apparent. So why spend maybe five or ten times more money
for the 'so called' pure sine wave?

When a UPS is installed for serious purposes, its battery typically
lasts something less than 20 years. But plug-in UPSes are made as
cheap as possible. Often its battery lasts only three years. A
replacement battery may cost as much as an entire new UPS.

A cheap UPS has only one function: to maintain power so that data is
not lost. The typical cheap UPS can output power so 'dirty' as to be
harmful to small electric motors and power strip protectors. But this
power must never harm a computer. Also called a computer grade UPS.
 
Reply With Quote
 
JJ
Guest
Posts: n/a
 
      12-17-2008, 06:36 PM

Thanks for the detailed thoughts. It sounds like the stepped wave
function should be safe for these computers.

As for the failure of the old one, I guess the point is moot now, but
none of your theories sound likely for the setup we have. All this
stuff is sitting in a small but dedicated server room on a university
campus that had all the electrical wiring updated about 5 years ago -
I'm sure it well surpasses code for domestic wiring. No dogs or
anything around to knock the plug out. Loads were all reasonable.
Other computers and UPSs in the room suffered no problems during these
anomalies.

This one might be the closest culprit: "Automatic battery test with
the VA load too close to the maximum rating of the UPS and aging
batteries? I've encountered this causing the UPS to shut down for a
short time when it shifts to powering the system from the batteries to
see how they handle the load." Best I can tell is that the batteries
discharged for some unknown reason - then the UPS tried one if its bi-
weekly tests and *bang* - sorry not enough battery power for the 1/2-
second (or however long it is) test.

Whatever the problem - I've given up on that particular model and it
looks like the one with the stepped approximation to a sine wave
should be fine for these computers.

Thanks.

-J



 
Reply With Quote
 
Cydrome Leader
Guest
Posts: n/a
 
      12-17-2008, 06:38 PM
(E-Mail Removed) wrote:
> On Dec 16, 2:06?pm, JJ <(E-Mail Removed)> wrote:
>> I've been somewhat ignorant about UPSs. ?It was just recently that I
>> discovered that the main difference between the two UPSs we have is
>> that one produces a true sine wave (APC smart-UPS), while the other
>> (which was cheaper but with a higher VA rating) produces a stepped
>> approximation (APC smart-UPS SC).
>> ...
>> My real question is, do these computers (Blade 1k and Blade 2k) really
>> need a true sine wave of current from the UPS, or is a stepped
>> approximation good enough (especially for the short periods of time
>> they would be running off of the UPS)? ?

>
> Output from a computer grade UPS (your cheapest ones) is also called
> stepped sine wave or modified sine wave. For example, the output from
> a 120 volt UPS in battery backup mode is two 200 volts square waves
> with a spike of up to 270 volts between those square waves. That
> output (even the spike) must not be destructive to any minimally
> acceptable computer. Industrry standards for computer power supplies
> make that apparent. So why spend maybe five or ten times more money
> for the 'so called' pure sine wave?


The reason to spend more money is so the UPS actually works. It's not the
sine wave you're after, it's that the UPS is designed and made to actually
work correctly. Good output is a side effect of this.

you'd be surprised at how many people drag garbage UPSes into datacenters
with real power conditioning equipment and then cry when their UPS fails,
even though it never saw a drop, brownout or surge.




 
Reply With Quote
 
JJ
Guest
Posts: n/a
 
      12-17-2008, 07:41 PM
This is a tangential question - and bear in mind that it was the
"expensive" UPS that gave me all the troubles.

Why would a UPS ever cause loss of power to the equipment connected to
it (even if its battery dies), if the power coming from the wall is
clean and steady? I mean, you're supposed to be able to pull the
battery out and replace it with a fresh one without losing power to
your hardware (as long as there is no power anomaly from the wall
outlet during that process) - so why would a UPS that is not seeing a
loss or fluctuation of power from the wall cut off power to the
hardware? Surely if the thing is going to do a self test and cut off
power from the wall for a bit, it should be smart enough to check
before it does whether or not its battery is low (it knows that) or if
the load is too high (which it also knows).

-J

> you'd be surprised at how many people drag garbage UPSes into datacenters
> with real power conditioning equipment and then cry when their UPS fails,
> even though it never saw a drop, brownout or surge.


 
Reply With Quote
 
Cydrome Leader
Guest
Posts: n/a
 
      12-18-2008, 01:58 AM
JJ <(E-Mail Removed)> wrote:
> This is a tangential question - and bear in mind that it was the
> "expensive" UPS that gave me all the troubles.
>
> Why would a UPS ever cause loss of power to the equipment connected to
> it (even if its battery dies), if the power coming from the wall is


That's a really good question.

No real UPS should fail that way, but there's plenty of junk out there
that does. Tiny UPSes lack bypasses, so when they fail, they just shut
off.

> clean and steady? I mean, you're supposed to be able to pull the
> battery out and replace it with a fresh one without losing power to
> your hardware (as long as there is no power anomaly from the wall


It depends on the size. Some let you swap batteries, some can't be opened.
Some have shitty connectors to the internal batteries, were some treat the
batteries as a module with real (Anderson) connectors.

> outlet during that process) - so why would a UPS that is not seeing a
> loss or fluctuation of power from the wall cut off power to the
> hardware? Surely if the thing is going to do a self test and cut off
> power from the wall for a bit, it should be smart enough to check
> before it does whether or not its battery is low (it knows that) or if
> the load is too high (which it also knows).


Most deskside UPSes are pretty stupid. If (more like when) the battery is
weak/dead and it tries to switch to the inverter, it will just die and
shut off. Maybe as you said, some unit try to test themselves and fail in
the process.

Giant UPSes are smarter. They have multiple bypasses where if they fail
or are being serviced, they will continue to let line power though, even
with the inverters dead.

The logic is any power is better than no power.

Also, proper maintenance is done in larger installations where somebody
tests each battery yearly. If a battery is tested as weak, it gets
replaced before it fails.

You can't even get to the batteries is a small UPS, so there's no way to
even test it without losing power if it's bad.


 
Reply With Quote
 
Benjamin Gawert
Guest
Posts: n/a
 
      12-18-2008, 06:37 AM
* JJ:

> My real question is, do these computers (Blade 1k and Blade 2k) really
> need a true sine wave of current from the UPS, or is a stepped
> approximation good enough (especially for the short periods of time
> they would be running off of the UPS)?


No. Computer PSUs are usually switched PSUs, and they don't care about
if the output is true sinus or stepped sinus. Besides that, inverters
(the components inside a UPS that convert DC to AC) always output a
stepped sinus. The only difference is the resolution (number of steps).
They never put out a true sinus.

UPSes exist in three classes:

- Standby UPS: simplest and cheapest variant, usually for low
capacities. These UPSes just pass through mains power, and if mains
power fails, they switch to battery power. The switching times (times
between switch from mains to battery or vice versa in which power output
is interrupted) are relative long, and they don't do power conditioning
(filtering spikes and overvoltages in mains power to protect the
connected equipment). I'd avoid them.

- Line-Interactive UPS: more sophisticated than the Standby variant. The
basic principle is the same but the Line Interactive variant has shorter
switching times and offers power conditioning (filtering of surges,
spikes and overvoltages, voltage levelling etc). They switch to battery
power not only when mains fails but also when mains power is out of
range. Line Interactive UPSes are usually used in capacities up to
around 2.5kVA and are usually the best and most economic option
(inverter which has energy "loss" is only used when run from battery)
for equipment that falls into that load rating.

- Online UPS: usually used for capacities from ~2kVA and above. An
Online UPS operates its inverters constantly. In normal operation it's
feeded by mains power that is transformed into a low voltage (usually
the same voltage as the collective battery voltage) DC current, and when
mains fails, the inverter gets fed from battery. These UPSes always
output the same "quality" of power (offer the best power conditioning)
but also they have the inverter constantly running which means they
"consume" a fair amount of power for themselves, and since the inverter
has a limited operating life it means higher service costs.

> Would these machines just shut
> down if they started seeing a stepped waveform?


No, they don't care.

> Would the stepped
> waveform cause damage to any of the components if the power was out
> for a long enough time?


No.

> Would I be better off with no UPS for these
> computers than ones with a stepped waveform?


No. There is no need to worry about the output waveform, but you should
worry about the quality level of the UPS as such that you buy. As to UPS
brands, my experience is that most APC UPSes are utterly crap and act up
especially after the first batch of batteries has died.

Better brands are Powerware/Best Power and (especially for capacities
from 2kvA) also Sola/Hevy Duty.

Benjamin
 
Reply With Quote
 
Cydrome Leader
Guest
Posts: n/a
 
      12-18-2008, 06:07 PM
Benjamin Gawert <(E-Mail Removed)> wrote:
> * JJ:
>
>> My real question is, do these computers (Blade 1k and Blade 2k) really
>> need a true sine wave of current from the UPS, or is a stepped
>> approximation good enough (especially for the short periods of time
>> they would be running off of the UPS)?

>
> No. Computer PSUs are usually switched PSUs, and they don't care about
> if the output is true sinus or stepped sinus. Besides that, inverters
> (the components inside a UPS that convert DC to AC) always output a
> stepped sinus. The only difference is the resolution (number of steps).
> They never put out a true sinus.
>
> UPSes exist in three classes:
>
> - Standby UPS: simplest and cheapest variant, usually for low
> capacities. These UPSes just pass through mains power, and if mains
> power fails, they switch to battery power. The switching times (times
> between switch from mains to battery or vice versa in which power output
> is interrupted) are relative long, and they don't do power conditioning
> (filtering spikes and overvoltages in mains power to protect the
> connected equipment). I'd avoid them.
>
> - Line-Interactive UPS: more sophisticated than the Standby variant. The
> basic principle is the same but the Line Interactive variant has shorter
> switching times and offers power conditioning (filtering of surges,
> spikes and overvoltages, voltage levelling etc). They switch to battery
> power not only when mains fails but also when mains power is out of
> range. Line Interactive UPSes are usually used in capacities up to
> around 2.5kVA and are usually the best and most economic option
> (inverter which has energy "loss" is only used when run from battery)
> for equipment that falls into that load rating.
>
> - Online UPS: usually used for capacities from ~2kVA and above. An
> Online UPS operates its inverters constantly. In normal operation it's
> feeded by mains power that is transformed into a low voltage (usually
> the same voltage as the collective battery voltage) DC current, and when
> mains fails, the inverter gets fed from battery. These UPSes always
> output the same "quality" of power (offer the best power conditioning)
> but also they have the inverter constantly running which means they
> "consume" a fair amount of power for themselves, and since the inverter
> has a limited operating life it means higher service costs.
>
>> Would these machines just shut
>> down if they started seeing a stepped waveform?

>
> No, they don't care.
>
>> Would the stepped
>> waveform cause damage to any of the components if the power was out
>> for a long enough time?

>
> No.
>
>> Would I be better off with no UPS for these
>> computers than ones with a stepped waveform?

>
> No. There is no need to worry about the output waveform, but you should
> worry about the quality level of the UPS as such that you buy. As to UPS
> brands, my experience is that most APC UPSes are utterly crap and act up
> especially after the first batch of batteries has died.
>
> Better brands are Powerware/Best Power and (especially for capacities
> from 2kvA) also Sola/Hevy Duty.


I'd second the Eaton/Powerware units. They offer good driver support for
weird systems, and are very efficient and small. The extenal rackmount
batteries scale nicely as well, and can be replaced on a live system.


 
Reply With Quote
 
 
 
Reply

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are On
Pingbacks are On
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
Refurbished Sun Enterprise Servers from Sun Microsystems EOL Batteries Embedded 0 03-11-2011 06:30 AM
waveform edition Stephane Embedded 5 11-07-2005 06:52 PM
Nov 9th Sun Expert Exchange: "Inside the New Sun Fire UltraSPARC IV+ Servers" John D Groenveld Sun Hardware 0 10-22-2005 02:16 AM
Waveform Brett Foster Embedded 5 03-05-2005 03:33 AM
Waveform storage & var. freq. playback Bob Martin Embedded 6 11-12-2003 04:52 AM


All times are GMT. The time now is 12:30 PM.


Welcome!
Welcome to Motherboard Point
 

Advertisment