[sci.electronics] UPS's that aren't

mason@tmsoft.UUCP (08/08/87)

Sorry if this gets discussed every week here, but I don't usually read the
newsgroup.  I will read it for a while in case this is an interesting topic
for you EEs, and you want to post replies.  If it's a boring topic, please
reply by mail.  I'll summarize if asked.

I was going away on 3 weeks vacation & wanted tmsoft to stay running & receiving
news & mail.  So, I thought, I'll just go out & buy an Uninterruptible Power
System.  So I did: a Sendon UPS-400W (400 watts, 5 millisecond max switchover).

Problem: the wave-form that comes out is positively ugly, and while switching
power supplies don't seem to care, tmsoft has a linear supply.  It will run on
the battery and when the power comes back, it seems to switch OK; but when the
power goes away power drops long enough to be a problem (i.e. system crash 1/2
way through vacation)

I've seen switchers work (although they were probably pulling less current)
fine with it.

The wave form looks like a bad cross between a sin wave & a square wave,
switching between the two about the top of the crest & bottom of the trough.
There are also some ugly transients when it's switching between mains power and
the battery.  I added a Radio Shack "Voltage Spike Protector with Broadband
Noise Filter" to the output with no discernable difference.

What I would like is something I can add on after the UPS that will clean
things up so the power supply works and I get rid of the nasty transients
(which mess up a terminal I use).  (My electrical theory is pretty shaky, but
it seems to me a huge 60 Hz pass filter would do the trick).  (I can also add
some more capacitance to my power supply, but I'm a little concerned about
inrush current).
	Thanks
	../Dave Mason,	TM Software Associates	(Compilers & System Consulting)
	..!{utzoo seismo!mnetor utcsri utgpu lsuc}!tmsoft!mason

larry@kitty.UUCP (Larry Lippman) (08/10/87)

In article <161@tmsoft.UUCP>, mason@tmsoft.UUCP (Dave Mason) writes:
> So, I thought, I'll just go out & buy an Uninterruptible Power System.
> So I did: a Sendon UPS-400W (400 watts, 5 millisecond max switchover).
> 
> Problem: the wave-form that comes out is positively ugly, and while switching
> power supplies don't seem to care, tmsoft has a linear supply.  It will run on
> the battery and when the power comes back, it seems to switch OK; but when the
> power goes away power drops long enough to be a problem (i.e. system crash 1/2
> way through vacation)
> 
> The wave form looks like a bad cross between a sin wave & a square wave,
> switching between the two about the top of the crest & bottom of the trough.
> There are also ugly transients when it's switching between mains power and
> the battery.  I added a Radio Shack "Voltage Spike Protector with Broadband
> Noise Filter" to the output with no discernable difference.
> 
> What I would like is something I can add on after the UPS that will clean
> things up so the power supply works and I get rid of the nasty transients
> (which mess up a terminal I use).

	Unfortunately, one gets what one pays for, and the plethora of
low-cost UPS devices flooding the market generally have poor quality standby
power characteristics.
	The only altermative which _may_ help is to place a ferroresonant
voltage regulating transformer on the output, such as those made by Sola
Electric.  I can't answer for certain if a Sola will help since I don't
know the exact nature of your power waveform.  Unfortunately, the cost of
a 400-watt Sola regulator ($ 350.00) is probably close to the same money
that you paid for the UPS in the first place.
	Good quality sine-wave UPS systems are not cheap. A 400-watt Sola
UPS (p/n 26-00-50400-3001) is around $ 1,400.00.
	Vendors with good-quality UPS systems include Sola, Superior Electric,
Elgar and Lorain.  We have used Elgar for our industrial process control
systems because they offer rack-mounted units.  As an example, we have an
Elgar p/n 172-1 for our site `kitty'; this is a 1.75 kva sine-wave UPS
with solid-state reverse-transfer option.  It gives a picture-perfect
sine-wave with less than 5% THD; however, this nice waveform cost us
close to $ 4K with batteries. :-(

<>  Larry Lippman @ Recognition Research Corp., Clarence, New York
<>  UUCP:  {allegra|ames|boulder|decvax|rutgers|watmath}!sunybcs!kitty!larry
<>  VOICE: 716/688-1231       {hplabs|ihnp4|mtune|seismo|utzoo}!/
<>  FAX:   716/741-9635 {G1,G2,G3 modes}   "Have you hugged your cat today?" 

wtm@neoucom.UUCP (Bill Mayhew) (08/11/87)

Somebody around here had some bucks and bought a 50 KVA UPS system
that runs our 3 Vaxes, Phone System, and Fire Alarm.  With
batteries, it is about the size of three refrigerators!  I think it
cost in the neighborhood of 40-60K dollars (ouch!!).

In our case, the administrator opted for a motor/generator type
UPS.  The motor/generator takes advantage of the intertia of the
mot-gen armature to keep power going while the switch over from
mains power to batteries takes place.  There is also never any
glitch on the load since it is always connected to the generator
(well, really three phase alternator) output and not switched.  The
disadvantage of the mot-gen type of UPS is that eventually the
bearings will have to go out, then somebody will have to descend
into the inner sanctum of steam pipes in the basement where the UPS
is and  patch around it.  The mean time before failure (MTBF) on
mot-gen units is usually quoted to be around five years.  This
seems reasonable, since the mechanical part is a sealed unit.

An alternative UPS system is a static sine-wave unit.  In a static
unit, the incomming AC is fed through diodes to the batteries.  An
inverter circuit then powers the load continuously from the
batteries, even when normal line voltage is present.  Static UPSs
work well, but don't provide 100% isolation from incomming
transients.  The switching transistors or SCRs in the inverter
circuit are a sensitive area that can be damaged by transients.
MTBF ratings are quoted at 1 to 5 years.  Power delivered to the
load is usually pretty good since there is no switch-over when the
mains power is lost.

The least desirable type of UPS usues a transfer relay or SCR bank
to cut in the inverter only when mains power is lost.  The
advantage of such a unit is that the inverter is only operated when
required, thus potentially lengthening its life.  The disadvantage
is that nasty switching glitches can result.  Also, the inverter in
this type of UPS might be designed less robustly, since it is not
expected to be in use continuously.  MTBF ratings vary widely, but
usually fall into the 1 to 5 year category.

Maintenance interval can vary too.  With transfer type and mot-gen
UPSs, the batteries can be serviced without taking the UPS offline.
Static UPSs must be shut down to service the batteries (at least,
as a general rule).  Types of batteries vary.  Wet cell car-battery
type units usually require attention every 6 months, and
replacement of the batteries every 5 to 10 years.  Gel electrolyte
batteries are available too.  Usually gel cells require only
periodic inspection.  Some manufacturers quite optimisticly claim
that their gel batteries last up to 20 years.  I'm not so sure I
believe that claim.

The rule of thumb is that you get what you pay for in a UPS.  I've
been using a little 200 watt Questa UPS on my XT compatible at home
for about the last 6 months and it kept going through several black
outs.  Occasionally, the noise pulse from turning on a lamp will
make the UPS kick in for a few seconds, even when the AC is good,
but that seems to be no big deal.  I bought my UPS used, and was
surprised to find out the list price was $400 when I looked it up
in a catalog.  Seems like a lot of $$$ for a seemingly simple
device.  One nice feature of the Questa UPS is that it has
terminals in the back so that a large storage battery can be
attached for extended back up power.  The little interal battery is
only good for about 5 to 15 min. depending on the load applied.


  --Bill
(wtm@neoucom.UUCP)

larry@kitty.UUCP (Larry Lippman) (08/12/87)

In article <654@neoucom.UUCP>, wtm@neoucom.UUCP (Bill Mayhew) writes:
> In our case, the administrator opted for a motor/generator type
> UPS.  The motor/generator takes advantage of the intertia of the
> mot-gen armature to keep power going while the switch over from
> mains power to batteries takes place.  There is also never any
> glitch on the load since it is always connected to the generator

	I once saw an interesting variety of a UPS that was installed
in a remote pipeline pumping station to maintain control power and
emergency lighting.
	This UPS consisted of an AC motor operated from the line, an
AC generator which powered the load, a large enclosed flywheel, all
of which was connected to a 4-cylinder propane-fueled engine.  The
engine turned over continuously with the AC motor and generator, but
was not running since the ignition and fuel were turned off.
	When power failed, the ignition and fuel were turned on, and
the engine would start and be up to load within a cycle time since
it was already running at governor speed.  The flywheel provided
sufficient inertia that only negligible degredation of output power
occured during the power loss transition.
	This UPS was only rated at about 7 kva.  It was an OLD system
dating from the 1950's, and I believe it was made by Bosch (but I
could be mistaken).  I have never seen another UPS of this type anywhere
else, and I rather suspect that few were sold due to the continual
mechanical wear on the engine.

<>  Larry Lippman @ Recognition Research Corp., Clarence, New York
<>  UUCP:  {allegra|ames|boulder|decvax|rutgers|watmath}!sunybcs!kitty!larry
<>  VOICE: 716/688-1231       {hplabs|ihnp4|mtune|seismo|utzoo}!/
<>  FAX:   716/741-9635 {G1,G2,G3 modes}   "Have you hugged your cat today?" 

ssm@munsell.UUCP (Steve McLafferty) (08/13/87)

In article <1932@kitty.UUCP> larry@kitty.UUCP (Larry Lippman) writes:
>In article <654@neoucom.UUCP>, wtm@neoucom.UUCP (Bill Mayhew) writes:
>> In our case, the administrator opted for a motor/generator type
>> UPS.  The motor/generator takes advantage of the intertia of the
>> mot-gen armature to keep power going while the switch over from
>> mains power to batteries takes place.  There is also never any
>> glitch on the load since it is always connected to the generator
>
>	I once saw an interesting variety of a UPS that was installed
>in a remote pipeline pumping station to maintain control power and
>emergency lighting.
>	This UPS consisted of an AC motor operated from the line, an
>AC generator which powered the load, a large enclosed flywheel, all
>of which was connected to a 4-cylinder propane-fueled engine.  The
>engine turned over continuously with the AC motor and generator, but
>was not running since the ignition and fuel were turned off.
>	When power failed, the ignition and fuel were turned on, and
>the engine would start and be up to load within a cycle time since
>it was already running at governor speed.  The flywheel provided
>sufficient inertia that only negligible degredation of output power
>occured during the power loss transition.
>
The UPSs described in the above two articles seem to make a clean
transition from main power to backup power.  My question is:
How do you go back from backup power to main power once the main power
has been restored?  Presumably there is a relay at the main power
input to the UPS that opens when main power is lost, so that the power
from your UPS goes only to the desired load, not back into the main
power grid.  (Otherwise your UPS might be supplying backup power to
your neighbor's computer :) .)

Once the main power is restored, you will want to close that relay to
let the main power back in.  The problem is that your UPS frequency
may have drifted slightly off from the 60Hz frequency of the main power 
grid, so that when you close the relay the two power sources might be 180
degrees out of phase.  The resulting current surge could damage the relay 
and the generator in the UPS, and possibly the equipment the UPS was 
supposed to protect.

Did the UPSs described above make the switch back to main power
automatically, or was operator intervention involved?  (Such as the
system involving lights between the two sources of power to show the
phase difference, where the relay is closed when the lights are off.)


Steven McLafferty                      "If I lose my mind,
Eikonix Corp                            I hope some one
Bedford, Mass                           will turn it in"
(617) 663-2115 x468                        -George Carlin

{adelie,encore}!munsell!ssm

chen@gt-stratus.UUCP (Ray Chen) (08/19/87)

I have a question for you folks.

A combination of old age and flakey Georgia Power finally killed my
old computer.  In an attempt to prevent a repeat of this sitation, a
friend of mine found me what seems to be a 2500 watt isolation
transformer.

It's black, *heavy*, and made by Freed Transformers of Brooklyn, NY.
Dimensions are 10" long X 6" wide X 7" high (or so).  The thing weighs
a ton.  We managed to put it inside an industrial strength milk crate.
So now 2 people can lift it and not get hernias if they're careful.
There are 2 sides with terminals.  One side says "Input 115 V/50 60 Hz".
The other side says "Output 25KVA".  We think that's a misprint that
should read "Output 2.5KVA" as the concept of a 1 cubic foot
25,000 watt transformer boggles the mind.

There are some numbers on the top that read "5950-856-1800" and
"37076".

First, anyone know anything about this beastie?  We think it's
your standard lots-of-coils transformer just because the thing weighs
so much.

Second questions -- how to safely connect it?  It's got +/- input
and output terminal.  I'm assuming that you connect the + and -
input terminals to the wall socket (via a plug and cable) and
the + and - output transformers to a power strip.  The real
question is what to do about grounding.  I'm assuming that you
don't ground either side and you just let the ground float.  So
the hookup looks like --

	Output	    Xformer	Wall
	+ ---------- +|+ -------- +
	GND --        |        -- GND
	- ---------- -|- -------- -

Is this the right way to do it?

Third, anyone got any idea on how much E&M (RFI, whatever) this
thing will put out?  I wasn't planning on keeping diskettes or
tapes near it, but if screws with the microwave oven (which will
be within 6 feet of the thing) or the neighbor's TV, I'm going
to have problems.

Fourth, any ideas on heat output?

Fifth (just to make sure), from my vague memories of my (one)
analog circuits course, even though the transformer is rated
at 2.5 KVA, it shouldn't draw the full load.  The load it draws
should depend on the load on the transformer.  But I'd hate to hook
the thing up and blow every circuit breaker in the apartment to
find out that I was wrong.

Sixth (just for fun), how bad do you think the noise might be?

As you may have gathered, I an *not* an EE-type and any help you
can give me would be much appreciated.

Thanks.

	Ray Chen
	chen@gatech

larry@kitty.UUCP (Larry Lippman) (08/20/87)

In article <845@gt-stratus.UUCP>, chen@gt-stratus.UUCP (Ray Chen) writes:
> A friend of mine found me what seems to be a 2500 watt isolation
> transformer.
> 
> It's black, *heavy*, and made by Freed Transformers of Brooklyn, NY.
> Dimensions are 10" long X 6" wide X 7" high (or so).  The thing weighs
> a ton.  We managed to put it inside an industrial strength milk crate.
> So now 2 people can lift it and not get hernias if they're careful.
> There are 2 sides with terminals.  One side says "Input 115 V/50 60 Hz".
> The other side says "Output 25KVA".  We think that's a misprint that
> should read "Output 2.5KVA" as the concept of a 1 cubic foot
> 25,000 watt transformer boggles the mind.
> 
> There are some numbers on the top that read "5950-856-1800" and
> "37076".

	The "5950-856-1800" is a universal military stock number, indicating
that this was built for the government.  I don't have any NSN directories,
so I can't offer any help. 
	The "37076" is probably a date code of sorts, since normal Freed
part numbers for transformers are all alphanumeric.  An example Freed
p/n for an isolation transformer is IT4, and for a ferroresonant transformer
is MCV61000LC.
	According to a Freed catalog, they don't manufacture isolation
transformers greater than 500 va, and they don't manufacture ferroresonant
transformers greater than 1.0 kva.
	Since this is a military transformer, Freed could have made almost
anything under these circumstances.  Bear in mind that ferroresonant
transformers ALL require capacitors; if this is a ferroresonant device then
you are either going to find a removable plate which will reveal capacitors,
or you will find additional terminals for external capcitors (military
ferroresonant transformers made by Freed all have external capacitors,
according to their catalog).
	I have a funny feeling that this is NEITHER an isolation transformer
NOR a ferroresonant transformer.  I don't know what it is - it may even be
a high-voltage transformer which puts out 25 kilovolts.
	My advice is to forget it, since you don't know what you are
getting into.  If you are bound and determined to power it up, use extreme
CAUTION in case it is 25 kilovolts on the secondary!


> 	Output	    Xformer	Wall
> 	+ ---------- +|+ -------- +
> 	GND --        |        -- GND
> 	- ---------- -|- -------- -
> 
> Is this the right way to do it?

	That's a conventional method of connection, but the +/- business
bothers me.  Turns direction on AC transformers are usually designated by
a "dot" on the "start" of winding, or by "S" and "F" to indicate start and
finish terminals.

> Third, anyone got any idea on how much E&M (RFI, whatever) this
> thing will put out?

	Plain 'ole transformers generate no RFI, but do generate a weak
magnetic field.  I certainly wouldn't be concerned about anything at least
2 feet away.  Ferroresonant transformers do generate some RFI, but it
is rather low frequency, and limited to powerline harmonics (<< 1 KHz).

> Fourth, any ideas on heat output?

	A very rough guess is a 5% energy loss to heat on a conventional
transformers.  UNloaded ferroresonant transformers can get quite warm.

> Sixth (just for fun), how bad do you think the noise might be?

	Probably less noise than the fans on your PC, unless the transformer
is a real pig with loose laminations.

<>  Larry Lippman @ Recognition Research Corp., Clarence, New York
<>  UUCP:  {allegra|ames|boulder|decvax|rutgers|watmath}!sunybcs!kitty!larry
<>  VOICE: 716/688-1231       {hplabs|ihnp4|mtune|seismo|utzoo}!/
<>  FAX:   716/741-9635 {G1,G2,G3 modes}   "Have you hugged your cat today?" 

john@hpcvlo.HP.COM (John Eaton) (08/20/87)

<<<<
< There are some numbers on the top that read "5950-856-1800" 
Sounds like a FSN (federal stock number) to me.

< Second questions -- how to safely connect it?  
With what you know and your experience level there is no safe way to connect
it. If I were doing it I would have a long cord with a fuse to power it and
be prepared for either a fire or explosion.


John Eaton
!hplabs!hp-pcd!john
 

brice@pnet01.CTS.COM (Brice Fleckenstein) (08/25/87)

Sounds a little like you might have a 25 KiloVolt transformer, rather than an
isolation transformer. The part number sounds VERY suspiciously like a
standard Military part number, though the number of digits seems a bit off.
"59xx" IS in the Military Electronic/Electrical part number range.