What does the switching in a UPS?

[stefanv]

I have some ideas about how to turn a standby inverter into an online UPS
I got a circuit working on my breadboard to detect the power-out an
command to switch over within 2ms.
My question, what to switch? I’ve read some frightening stories abou
blowing up inverters when using a DPDT relay. Is this so? Where thei
relays to weak? What are the other options? It has to be fast to be
UPS, but it looks like this has his own problems. How does a “real” UPS d
it? Anyone know any links to schematics?

The answer is not "Why don't you just buy a real UPS" I have those,
don't need one, I just want to learn how to do this, I like the chalenge.

Thanks for brainstorming with me,… Merry Christmas

Stefan V

 

[Joerg]

Hello Stefan,

I have some ideas about how to turn a standby inverter into an online UPS.
I got a circuit working on my breadboard to detect the power-out and
command to switch over within 2ms.
My question, what to switch? I’ve read some frightening stories about
blowing up inverters when using a DPDT relay. Is this so? Where their
relays to weak? What are the other options? It has to be fast to be a
UPS, but it looks like this has his own problems. How does a “real” UPS do
it? Anyone know any links to schematics?

It depends on what your load is. An incandescent bulb may be harmless.
But imagine a motor: It will continue to spin. Since your inverter's
phase is most likely not in sync with the power grid you might, within
2msec, apply an AC voltage that can be up to 180 degrees out of phase
with what it saw just 2msec earlier. Kaboom. Maybe.

Then there are capacitors. Every larger motor has one, some fluorescents
do, and so on. So let's say the whole collection of capacitors hanging
on the circuit happens to be at -160V and 2msec later you apply +160V.
The output devices in your inverter might express their unhappiness with
a dazzling pyrotechnic show.

The answer is not "Why don't you just buy a real UPS" I have those, I
don't need one, I just want to learn how to do this, I like the chalenge.

Well, a good UPS tracks the phase of the line. When the line goes it can
take over at the correct phase angle. I have also seen some not so good
UPS that were more or less slapped together. Mostly the remains thereof.
So be careful.

Thanks for brainstorming with me,… Merry Christmas

Same to you.

Regards, Joerg

 

[John Crighton]

I have some ideas about how to turn a standby inverter into an online UPS.
I got a circuit working on my breadboard to detect the power-out and
command to switch over within 2ms.
My question, what to switch? I’ve read some frightening stories about
blowing up inverters when using a DPDT relay. Is this so? Where their
relays to weak? What are the other options? It has to be fast to be a
UPS, but it looks like this has his own problems. How does a “real” UPS do
it? Anyone know any links to schematics?

The answer is not "Why don't you just buy a real UPS" I have those, I
don't need one, I just want to learn how to do this, I like the chalenge.

Thanks for brainstorming with me,… Merry Christmas

Stefan V

Hello Stefan,
I worked on some old UPSs years ago. Maitec was the brand.

Dirty mains in ---> DC and batt --->AC ----> Clean AC out

While the dirty mains was present, the clean AC out was
synchronised to the dirty mains.
If the dirty mains dropped off then the batteries took over
powering the UPS which still produced clean AC out.
The UPS output frequency was xtal controlled.

When the dirty mains came back on the UPS would
synchronise its output to that of the dirty mains. I think
this took a few seconds for the xtal oscillator to be
pulled. I can't remember exactly.

If the UPS was over loaded a fast change over relay
would switch the UPSs output connector from clean
UPS output to that of the dirty mains until the overload
was removed then the relay switched back to clean
UPS output again. Since both the dirty mains and
UPS output were in phase with each other the tiny
break in the supply for a couple of milliseconds was
not noticed if powering light globes or computers.

Regards,
John Crighton
Sydney

 

[stefanv]

If not else, I am learning more on how UPS function

So my idea of using a "cheap" inverter and making a transfer switch that'
fast enough turning it into a UPS doesn't seem to fly... If I have to chec
the phase and put them in sync.. I can't change the phase of the main
comming in, so if I want to control the phase of the inverter, I woul
have to hook into it somehow to control its oscilator. Looks like a bi
step further then what I wnated to do in the first place with som
"universal inverter to in-line UPS adapter"

What about delaying the switch-over 10 ms. This would make that the powe
comes back after half a sine wave. I figure what ever value it comes bac
with at that time, it will not have the big "spike". How would this dea
with power supplies (computers is what we are switching here).

Stefan

 

[John Crighton]

If not else, I am learning more on how UPS function

So my idea of using a "cheap" inverter and making a transfer switch that's
fast enough turning it into a UPS doesn't seem to fly... If I have to check
the phase and put them in sync.. I can't change the phase of the mains
comming in, so if I want to control the phase of the inverter, I would
have to hook into it somehow to control its oscilator. Looks like a big
step further then what I wnated to do in the first place with some
"universal inverter to in-line UPS adapter"

What about delaying the switch-over 10 ms. This would make that the power
comes back after half a sine wave. I figure what ever value it comes back
with at that time, it will not have the big "spike". How would this deal
with power supplies (computers is what we are switching here).

Stefan

Hello Stefan,
why not keep things simple. You say you already have an inverter.
Let your inverter power your computer equipment continuously.
Don't overload the inverter, run it conservatively and forget about
change over relays.
Regards,
John Crighton
Sydney

 

[Ignoramus6309]

I have some ideas about how to turn a standby inverter into an online UPS.
I got a circuit working on my breadboard to detect the power-out and
command to switch over within 2ms.
My question, what to switch? I’ve read some frightening stories about
blowing up inverters when using a DPDT relay. Is this so? Where their
relays to weak? What are the other options? It has to be fast to be a
UPS, but it looks like this has his own problems. How does a “real” UPS do
it? Anyone know any links to schematics?

I just took apart a APC 2200 UPS.

I believe that the relays control the source. These are "on board
relays" that solder right into the board. Rated for 20 amps.

They use an array of mosfets to do switching. They have transformers
that serve as both down and up transformers, based on relays, I
think.

i

The answer is not "Why don't you just buy a real UPS" I have those, I
don't need one, I just want to learn how to do this, I like the chalenge.

Thanks for brainstorming with me,… Merry Christmas

Stefan V

--

 

[Mark]

I had a pump load wired to a DPDT BREAK BEFORE MAKE relay with one set
of contacts contacts powered by the power line and the other by a free
running modifed sine wave inverter. When I tested the system by
switching the relay (I don't remember now which way it was going line
to inverter or inverter to line) the inverter blew up!!!!!! I was
not happy :-(

Either one of two things happened....

1) The load was a motor that kept running on interia during the switch
over and was therefore out of phase with the inverter

or

2) the relay contacts arced enough to bridge the gap and momentarily
connected the inverter directly to the line.

I never figured out which problem it was and abandonded the plan using
a real UPS instead. I think I would have had to use two relays and
lots of time delays to make it work. (I didn't care if the pump went
off for a few seconds during the swittch)

Seamless switching is not easy. I think you do have to keep the
inverter running all the time and keep it in phase with the power line
so it is ready to take up the load at any time should the line go
off.....

and when the power returns, you have to slowly slew the inverter phase
to be back in phase with the line before you switch the load back over
to the line....

it is a very interesting problem....the solutions are MUCH easier if
you can afford to let the power go off at the load for a second or so,
which won't work too well for a computer....

let us know what you figure out...

Mark

 

[Rich Grise]

Hello Stefan,

It depends on what your load is. An incandescent bulb may be harmless.
But imagine a motor: It will continue to spin. Since your inverter's
phase is most likely not in sync with the power grid you might, within
2msec, apply an AC voltage that can be up to 180 degrees out of phase
with what it saw just 2msec earlier. Kaboom. Maybe.

Then there are capacitors. Every larger motor has one, some fluorescents
do, and so on. So let's say the whole collection of capacitors hanging
on the circuit happens to be at -160V and 2msec later you apply +160V.
The output devices in your inverter might express their unhappiness with
a dazzling pyrotechnic show.


Well, a good UPS tracks the phase of the line. When the line goes it can
take over at the correct phase angle. I have also seen some not so good
UPS that were more or less slapped together. Mostly the remains thereof.
So be careful.

I've seen a mongo inverter that had a sync input/output, so that they
could be sync'd together for higher power. I could see squaring up the
line, and using that for the sync signal, but you'd have to have something
that could keep up the phase during the outage, like a slow PLL or
something, then just gate the drive to the driver transistors.

Cheers!
Rich

 

[Rich Grise]

If not else, I am learning more on how UPS function

So my idea of using a "cheap" inverter and making a transfer switch that's
fast enough turning it into a UPS doesn't seem to fly... If I have to check
the phase and put them in sync.. I can't change the phase of the mains
comming in, so if I want to control the phase of the inverter, I would
have to hook into it somehow to control its oscilator. Looks like a big
step further then what I wnated to do in the first place with some
"universal inverter to in-line UPS adapter"

What about delaying the switch-over 10 ms. This would make that the power
comes back after half a sine wave. I figure what ever value it comes back
with at that time, it will not have the big "spike". How would this deal
with power supplies (computers is what we are switching here).

Stefan

Well, as an educational project, it sounds dandy, albeit I'd schedule
about 6 months to get the thing done.

If the goal is to power computers uninterruptably, I'd look into computer
power supplies that can run right off the batteries (and their float
charger, of course. )

Have Fun!
Rich

 

[Rich Grise]

Hello Stefan,
why not keep things simple. You say you already have an inverter.
Let your inverter power your computer equipment continuously.
Don't overload the inverter, run it conservatively and forget about
change over relays.
Regards,
John Crighton
Sydney

And find a deep-cycle battery that's happy with a continuous float
charge.

Cheers!
Rich

 

[Joerg]

Hello Rich,

I've seen a mongo inverter that had a sync input/output, so that they
could be sync'd together for higher power. I could see squaring up the
line, and using that for the sync signal, but you'd have to have something
that could keep up the phase during the outage, like a slow PLL or
something, then just gate the drive to the driver transistors.

It's no big deal. You have to keep running the 60Hz as accurately as
possible. Cumulative error is important. Then when the line comes back a
good UPS begins to adjust its phase towards what the line has while
monitoring the line for how solid the restored power is.

Sync is a tad more complicated that just squaring up the line voltage.

Regards, Joerg

 

[stefanv]

I appreciate the comments.

As mentioned I see this as a chalenge and an educational project more tha
a way to keep my computer going. What about this one,...
When the main power goes off, I sense this in about 1ms. I then wait fo
the next zero-crossing from the inverter in max 10ms (likely way less a
inverters have more like modified block waves that are zero 1/3 of th
time.)so in 10-11ms max I swith my relay.
When the power comes back, I switch the inverter off on a zero-crossin
and wait for the mains zero crossing to switch back to mains.
- Problem 1 would be, a relay is just to slow, I can't afford to ad
another 20ms of relay switching to my delay, and that would make me loos
the zero-crossing by about 200% Go to Solid state?
- Problem 2 could be, zero-crosings appear from neg-to-pos and fro
pos-neg. If I catch a neg-to-pos from the inverter and then swith to
pos-to-neg from the mains, meaning getting 2 negative sine halves (or
positive sine halves) to follow each other, what will be the result?

Unfortunately, this is something my old scope will never show me as i
doesn't have memory or storage. Looks like I will need to upgrade to fin
out what's really going on.

Any thoughts on this?

Thanks StefanV

 

[Joerg]

Hello Stefan,

- Problem 2 could be, zero-crosings appear from neg-to-pos and from
pos-neg. If I catch a neg-to-pos from the inverter and then swith to a
pos-to-neg from the mains, meaning getting 2 negative sine halves (or 2
positive sine halves) to follow each other, what will be the result?

Depending on what's connected, anything between 'nothing' and 'Kaboom'.
A motor or capacitive load might fry the inverter if it tries to pull it
into the wrong phase polarity.

Unfortunately, this is something my old scope will never show me as it
doesn't have memory or storage. Looks like I will need to upgrade to find
out what's really going on.

Oh, I scoped out that kind of stuff with my old 8MHz Hameg. 3" tube, no
trigger, just a 'synchronizer'. Back in those days I built dimmer type
circuits for inductive loads and you had to really know what was going
on or the mounting box would become one hellacious black hole.

Guess what, 30 years later that Hameg is still working fine. I just
can't part with it.

Regards, Joerg

 

[Ian Stirling]

And find a deep-cycle battery that's happy with a continuous float
charge.

Irrelevant.
If you design the charger appropriately, the normal voltage will not differ
from the other case.