Switching 240v power sources. Different phases.

[Ed.Kloonk]

Got a 800w sinewave 12vdc -> 240v inverter. And a big battery. I would like to be able to keep powering my computer desk when there is a storm and would like to use a DPDT  switch to change the source from mains to inverter when the weather looks dicey.

Computer desk has computer, audio amp, light intercom and cordless phone base all connected on a power strip. I want to not have to turn off computer when switching, if possible.

The two voltage sources are out of phase. Trying to keep the computer running, what problems occur when you quickly switch power like this with these devices. Or is it a no-go.

[Simon]

The problem your thinking of does not exist, your problem will be how long can your computer PSU keep going for without power. UPS's usually have a continual supply from the inverter and it's just a case of if there is power to charge the batteries or not.

[BradC]

The problem your thinking of does not exist, your problem will be how long can your computer PSU keep going for without power. UPS's usually have a continual supply from the inverter and it's just a case of if there is power to charge the batteries or not.

Oddly enough that's not *quite* correct. Full time on-Line UPS's do that (I have one here; an MGE Pulsar), most cheap-o UPS's actually have a relay in them to switch over, but they maintain rudimentary mains sync so there is little glitch (APC BackUPS for example; have one here).
Even the "line-interactive" APC SmartUPS (I have 2 of them here) series don't do the inverter full time. They watch the sine wave on the incoming mains and as soon as it starts to deviate from a certain tolerance they juice up the inverter and switch over the relay. The SmartUPS have extra windings on the transformer to allow them to boost and buck also (thus the moniker line-interactive), but it's still all relay switched.

[Simon]

Oh right ok, I suppose if it's done fast enough you can make a switch, explains why as soon as we have a glitch at work one persons UPS kicks in.

[BradC]

Oh right ok, I suppose if it's done fast enough you can make a switch, explains why as soon as we have a glitch at work one persons UPS kicks in.

Yeah, annoying isn't it. *clunk*, buzz, ..5seconds.. *click*. Most power supplies I've seen specify a holdup time in excess of 20ms (which is one mains cycle) at full load. So as long as you are not at full noise, then your hold up should be a couple of mains cycles. Enough for a pretty smartish relay to click in, but nowhere near enough for a toggle switch. I've been looking at this pretty closely as I wanted to built an auto-transfer switch to reliably cut over to a generator (and back again) when the mains fails/restores. Because it's on the input of the UPS time is not critical, and slower is probably better.

Now, the effect of switching to a source that is 180 degrees out of phase is another thing entirely. Certainly any input filters are likely to object pretty strongly.

On that note, but different my old man has some photos somewhere of a couple of megawatt alternator that was synced up 180 deg out of phase with the grid back in the 60's, and the alternator sized hole it left as it exited the side of the building. I'm constantly amazed at the relatively tiny source impedance of the mains grid and the damage it can cause when poked in anger.

[Simon]

On that note, but different my old man has some photos somewhere of a couple of megawatt alternator that was synced up 180 deg out of phase with the grid back in the 60's, and the alternator sized hole it left as it exited the side of the building. I'm constantly amazed at the relatively tiny source impedance of the mains grid and the damage it can cause when poked in anger.

A friends grandfather used to work in a power station, they'd have to manually sync generators as they brought them in looking at scope screens and feeling the bump on the floor if it was slightly out.

[Kjelt]

Now, the effect of switching to a source that is 180 degrees out of phase is another thing entirely. Certainly any input filters are likely to object pretty strongly.

Why? A PC SMPS rectifies the mains, so it would be no different than switch it on from power off to on at full wave? I could even argue if it is less hard on the input of the smps since the Vbus elcos are still charged.
But if you are using a relay I would test it a couple of times with a probe on the 12V output to see if it glitches. Esp. the harddrives will not like glitches on the 12V at all.

[Ed.Kloonk]

The problem your thinking of does not exist, your problem will be how long can your computer PSU keep going for without power.

Guessing several tens of milliseconds. A switch or relay controlled manually would be faster than that. Just not convinced that switching so quickly the inputs that are out of sync wouldn't damage computer or the inverter.

There is no UPS. I intend to isolate the incoming mains when the weather starts to look shakey. Lightning is giving me hell.

[Ed.Kloonk]

Now, the effect of switching to a source that is 180 degrees out of phase is another thing entirely. Certainly any input filters are likely to object pretty strongly.

Why? A PC SMPS rectifies the mains, so it would be no different than switch it on from power off to on at full wave? I could even argue if it is less hard on the input of the smps since the Vbus elcos are still charged.
But if you are using a relay I would test it a couple of times with a probe on the 12V output to see if it glitches. Esp. the harddrives will not like glitches on the 12V at all.

Sounds plausible. Stand by for loud bang...

[BradC]

A friends grandfather used to work in a power station, they'd have to manually sync generators as they brought them in looking at scope screens and feeling the bump on the floor if it was slightly out.

Here they did it with three lights connected across the phases between the generators. All lights out, all synced up. When I visited the power station ~30 years ago they had a big analog dial (the synchroscope), and I was lucky enough to be allowed in the room when they brought a generator on line. These guys did this every day, but they still held their breath when they flipped the switch. It's probably all automated these days.

[BradC]

Now, the effect of switching to a source that is 180 degrees out of phase is another thing entirely. Certainly any input filters are likely to object pretty strongly.

Why? A PC SMPS rectifies the mains, so it would be no different than switch it on from power off to on at full wave? I could even argue if it is less hard on the input of the smps since the Vbus elcos are still charged.

I was just wondering what you do to the common mode inductors when you (relatively) instantly reverse the current in them rather than the leisurely sine they are designed for.

[Ed.Kloonk]

I was just wondering what you do to the common mode inductors when you (relatively) instantly reverse the current in them rather than the leisurely sine they are designed for.

That's my question in a nutshell.

[Kjelt]

I was just wondering what you do to the common mode inductors when you (relatively) instantly reverse the current in them rather than the leisurely sine they are designed for.

Good question, please correct me cause I am not an expert at this but  I would think that there is a small (time) gap for the relay contact to switch from AC1 to AC2,
so lets assume that the switching of the relay between contacts will take 10ms?

The step would be as suggested (worst case) so AC1 at -325V and AC2 at 325V (Vtt for 230VAC)
t=0ms -325V  t=1ms 0V (contact open) t=10ms 325V from the input side.

Now with normal mains so no switching:
t=0ms -325V  t=5ms 0V  t=10ms 325V

The difference is that without switching the voltage will in the mean time gradually change and with the switching it will be two hard steps but I would think this is not different than switching it on from power off to power on at the voltage peak.
No idea what the CM inductors will do, although pretty small they would like to keep the voltage equal if i remember correctly, but since there is no mains input for almost 10ms they will be drained by the load or not?

Please fill in or correct me where wrong, I also would like to learn 

[SeanB]

UPS to power the PC and desk for at least 2 minutes, then you have time to start the inverter/generator/whatever and use a transfer switch to change over to it.

A regular switch is not going to cut the mustard with mains switching, you need a transfer switch that has 3 positions. Incoming mains, off and standby mains. you need the switch to spend a few cycles in the middle position to have no power, and the switch contacts must have at least 4mm separation in the off position, as required by the specs for a mains isolator. They must also switch both live and neutral leads both sides. At present I am waiting for stock to come from the manufacturer, as the wholesalers are all out of stock for some reason ( Eishkom and load shedding) and the switches are $80 each excluding VAT. Otherwise a plug and 2 labelled sockets next to each other works for light loads.

[Ed.Kloonk]

UPS to power the PC and desk for at least 2 minutes, then you have time to start the inverter/generator/whatever and use a transfer switch to change over to it.

A regular switch is not going to cut the mustard with mains switching, you need a transfer switch that has 3 positions. Incoming mains, off and standby mains. you need the switch to spend a few cycles in the middle position to have no power, and the switch contacts must have at least 4mm separation in the off position, as required by the specs for a mains isolator. They must also switch both live and neutral leads both sides. At present I am waiting for stock to come from the manufacturer, as the wholesalers are all out of stock for some reason ( Eishkom and load shedding) and the switches are $80 each excluding VAT. Otherwise a plug and 2 labelled sockets next to each other works for light loads.

Ok. So there must be a short delay. And 4mm of gap on switches. Though, I plan to pull the mains input from the socket once transfer has been done, I was wishing not to use a UPS. So it sounds as though what I am trying to achieve cant/shouldn't be done.

[denelec]

According to the ITIC curve, a computer should be able to tolerate 20ms without power.
http://en.wikipedia.org/wiki/International_Committee_for_Information_Technology_Standards

The curve is meant for 120V 60Hz equipments but it's probably be similar for 240V 50Hz.

[Ed.Kloonk]

According to the ITIC curve, a computer should be able to tolerate 20ms without power.
http://en.wikipedia.org/wiki/International_Committee_for_Information_Technology_Standards

The curve is meant for 120V 60Hz equipments but it's probably be similar for 240V 50Hz.

One cycle?

 

[Kjelt]

One cycle?   

As an exercise do the calculation how big an elco should be to power 12V at 500W for 20ms without dropping voltage

[Kjelt]

I just did a short calculation on the value of the capacitor that drops from 12V at t=0s to 11V at t=0.02s (20ms) with a resistor of 0.288 ohms (500W).
It is a staggering 1.25 F so 1250000 uF. Expensive power supply 

[Zero999]

I just did a short calculation on the value of the capacitor that drops from 12V at t=0s to 11V at t=0.02s (20ms) with a resistor of 0.288 ohms (500W).
It is a staggering 1.25 F so 1250000 uF. Expensive power supply 

It wouldn't have to be that big.

Your calculation is unrealistic. The filter capacitors on the primary side just need to be able to store enough energy, plus a bit more to fill in the gap which is 10J at 500W for 20ms.

[Kjelt]

Sure if the primary voltage bus capacitor is large enough but it was a reaction on

One cycle?   

to give an impression how much energy 20ms of 500W actually is.

[SeanB]

15000uF will do about a second on most PC power supplies, providing there is no PFC. That is 3 4700uF 450V cans. going to be a very big surge though, and a truly horrendous power factor in regular use, all short high current peaks means you will need 35A bridges just to keep them from blowing up at power on.

[Zero999]

Sure if the primary voltage bus capacitor is large enough but it was a reaction on

One cycle?   

to give an impression how much energy 20ms of 500W actually is.

Not much, a 220µF capacitor charged to 325V contains more than enough energy to provide 500W for 20ms. Of course, in real life it won't be possible to extract all of the energy from the reservoir capacitor because the SMPS will stop working at some point, so it will need to be larger but not much larger.