Register to reply

Effect of a capacitor on an AC circuit

by dE_logics
Tags: capacitor, circuit, effect
Share this thread:
dE_logics
#1
Apr9-14, 01:44 AM
P: 735
Hi!

I'm in a place where electricity goes out frequently; as a result I got a UPS which is powering my computer the total power input of which is less than 200W. The UPS is 300W or 600VA.

The transfer time (time to switch from AC mode to battery mode) of the UPS is around 4 ms; but in that time the capacitors of the PSU of the computer discharges, as a result it turns off. This doesn't happen when the computer is taking less power -- the PSU capacitors are able to hold for 15-20ms in that case.

The solution I found is to add a capacitor in parallel (rated 440V AC) to the output of UPS
(or input of the computer); that improves the transfer time of the UPS by half an AC cycle or 10ms (for 50 HZ input).

But I was wondering the consequence of doing that may be higher electricity bill; can it reduce the power factor in any way? Cause when the UPS goes in backup mode, for a few seconds regularly, it shows overload signal when the capacitor is connected. Or is it that the power measurement circuit of the UPS read the apparent power (that means the capacitor is reducing the power factor)? Also I heard the energy meter here reads the real power; but if that's not the case, I'm out of luck.

The output of the UPS in backup mode is 'pseudo sine wave'; an engineer said it's square wave.
Phys.Org News Partner Engineering news on Phys.org
UConn makes 3-D copies of antique instrument parts
Amazon worker piloted drone around Space Needle
Five next-generation technologies for positioning, navigation and timing
vk6kro
#2
Apr9-14, 08:57 AM
Sci Advisor
P: 4,016
Putting a capacitor at the output of the UPS just puts a higher load on the UPS. It may increase your power bill, but in any case you should remove it

When the AC from the UPS reaches your computer power supply, it is rectified and a large DC voltage is developed across a large electrolytic capacitor (or two capacitors in 110 volt systems).
It is these capacitors that hold the charge.
You could get someone with experience to test these capacitors and see if they need replacing.
These capacitors hold a lethal charge, even when the power is turned off, so stay away from them.

You could try running your monitor off a different circuit to reduce the current through the UPS. Maybe it could use a different UPS, but the important aim is to preserve the data in your computer to save whatever you are working on
dE_logics
#3
Apr9-14, 10:09 AM
P: 735
What do you mean 'rectified'? The output of the UPS is AC; it's an offline UPS which just passes the AC input (100% sine wave) to the PSU of the computer as long as there's input power; once the electricity is out it converts the input from it's internal 12V lead acid battery to 220V 50HZ square wave signal.

The UPS only powers the CPU; 200W is excluding the monitor.

Actually I got 2 UPS, one new and one old. Both have the same problem.

So this's unlikely a UPS problem.

But that 10F capacitor I applied really did the trick. Now the system never shuts down. Remember -- the capacitor is applied across an AC output in parallel and computer PSU takes in AC 220 V input.

I can provide a diagram if things are not clear.

vk6kro
#4
Apr9-14, 10:19 AM
Sci Advisor
P: 4,016
Effect of a capacitor on an AC circuit

Inside the power supply of the computer, there is a bridge rectifier or some power diodes to turn the AC input into DC.

This charges up a large capacitor (or two) which then supplies the rest of the circuit.

This is where you need to look to increase the time the computer will keep working without input power.
jim hardy
#5
Apr9-14, 11:19 AM
Sci Advisor
PF Gold
jim hardy's Avatar
P: 3,515
The transfer time (time to switch from AC mode to battery mode) of the UPS is around 4 ms; but in that time the capacitors of the PSU of the computer discharges,

i'd think it more likely the computer's power line monitor initiating shutdown well before the capacitors discharge.


the capacitors in the computer are sized to carry between power line peaks, at least 10ms for 50hz.

If your ups really does a 4msec transfer,
then what you have is a race between two power fail detectors - the one in your UPS that initiates transfer, and the one in your computer that initiates power down sequence.

Why 20 uf across the line delays the computer's power fail detect is a mystery to me.
But I trust it's a motor run capacitor not a motor start capacitor - else you will soon get smelly capacitor goo on the floor .

let us know what you find - it's interesting.

Does your ups have a switch that lets you run continuously from the battery while keeping it charged ? That'd be a cleaner fix.
dE_logics
#6
Apr9-14, 11:41 PM
P: 735
Quote Quote by vk6kro View Post
Inside the power supply of the computer, there is a bridge rectifier or some power diodes to turn the AC input into DC.

This charges up a large capacitor (or two) which then supplies the rest of the circuit.

This is where you need to look to increase the time the computer will keep working without input power.
It's a new PSU (from Gigabyte). It'll void warranty that way. Anyway I opened up an old smps to find the 2 large capacitors of 330F rated for 300V (peak in a 220V AC input?).
dE_logics
#7
Apr10-14, 12:23 AM
P: 735
Quote Quote by jim hardy View Post
i'd think it more likely the computer's power line monitor initiating shutdown well before the capacitors discharge.


the capacitors in the computer are sized to carry between power line peaks, at least 10ms for 50hz.

If your ups really does a 4msec transfer,
then what you have is a race between two power fail detectors - the one in your UPS that initiates transfer, and the one in your computer that initiates power down sequence.

Why 20 uf across the line delays the computer's power fail detect is a mystery to me.
But I trust it's a motor run capacitor not a motor start capacitor - else you will soon get smelly capacitor goo on the floor .

let us know what you find - it's interesting.

Does your ups have a switch that lets you run continuously from the battery while keeping it charged ? That'd be a cleaner fix.
According to ATX specifications, the PSU should hold for a minimum of 16ms. Since I verified this by 2 different UPS, probably the UPS is transferring in 4ms so the PSU has a design flaw; although they say it complies with ATX 2.3 specs.

I dont think the PSU does have a power down sequence when power is gone from the input, since there's no battery or any kind of significantly long storage or in that case any circuit to switch unlike the UPS (which switches to battery input). The only thing that's protecting it from short term power losses are those 2 capacitors which I wrote about above. Those are probably directly connected to the rectifier output which explains why they're rated for 300V (Vmax is 311 for 220V AC input).

It has a power off sequence after the rectifier which communicates with the motherboard via ATX standards.

As per vk6kro's analysis, the capacitors after the rectifier discharges powering off the PSU's electronics which makes complete sense.

Yes -- the capacitor is a motor run capacitor. My theory for applying it was that for a 50HZ output of the UPS it can provide half a cycle (or (1/50)/2 seconds or 10ms) of output which can easily cover up the 4ms delay. So when there's a power failure, the PSU doesn't feel a thing.
jim hardy
#8
Apr10-14, 04:00 AM
Sci Advisor
PF Gold
jim hardy's Avatar
P: 3,515
It has a power off sequence after the rectifier which communicates with the motherboard via ATX standards.
typically those things keep watch for a missing power line peak and upon finding one absent tell the motherboard to initiate its power down sequence , on the premise the CPU can get its house in order during the 16 milliseconds it has left while the filter capacitors (those 330's you found) are bleeding down.

Interesting fix you have found there, for while it sounds like it shouldn't work there must be a reason that it does.

If your UPS is a quasi sinewave with sharp edges your cap will draw an interesting current waveform. That is likely why the "overload" light illuminates. You might try a small inductor in series with the capacitor, a few turns through a hefty donut ferrite core.

Just be careful - no exposed terminals where tiny fingers can get at them.... kids are curious.

ps - it shouldn't show on your power bill. It's "wattless current".
sophiecentaur
#9
Apr10-14, 05:26 AM
Sci Advisor
PF Gold
sophiecentaur's Avatar
P: 11,919
It strikes me that the odd behaviour is probably due to the computer power supply, rather than the two UP Supplies (after all, the computer is the common factor in both cases)
The 10μF capacitor is only around j3kΩ at 50 Hz so it isn't going to stress a UPS. I think the computer supply must be a bit too sensitive to glitches on its mains supply and there may be a bit of phase change on the switch over from the mains to the UPS 50Hz. Your Capacitor is just providing a bit of smoothing- enough to let the UPS settle down without triggering the shutdown. I would bet you could solve your problem with a different computer PSU; that may not cost a lot and wouldn't involve any clever electronics re-designs.
dE_logics
#10
Apr10-14, 10:14 PM
P: 735
I'll surely try the inductor. Thanks for the clarification and advice!

Yes a better PSU is a solution. I upgrade from a cheap one to an expensive one, but I think need to buy top notch. Gigabyte brand of PSUs is not that reputable.

As an alternative I can replace the cheap PSU's internal rectifier capacitors with better ones.
vk6kro
#11
Apr11-14, 10:34 AM
Sci Advisor
P: 4,016
A 10 uF capacitor has a reactance of 318 ohms at 50 Hz. On a 230 volt system this would carry a current of about 1 Amp peak, depending on the waveform.

This could be causing a current of 10 amps or more to flow from the battery, in addition to the current caused by the computer's consumption.

This will reduce the time that the battery will supply power to the computer and may reduce the life of the battery.

It is not lossless either. The capacitor is not across the mains input and there will be losses in the battery and the inverter that generates the output sinewave.

All of this has to be paid for when the mains power comes back on and the battery charges up again.
sophiecentaur
#12
Apr11-14, 11:31 AM
Sci Advisor
PF Gold
sophiecentaur's Avatar
P: 11,919
Quote Quote by vk6kro View Post
A 10 uF capacitor has a reactance of 318 ohms at 50 Hz. On a 230 volt system this would carry a current of about 1 Amp peak, depending on the waveform.
Thanks for the correction. What kind on an eejit thinks 10μF is 10-6F?

On the practical side of this, I cannot imagine how any installation that really needs a UPS can be expected to work properly with a cheep computer PSU. If the system needs to be power supply tolerant then either spend some money strategically on hardware or spend some time sorting the software out so that it recovers from crashes. (Frequent automatic file saving etc) Even word processors do that.
jim hardy
#13
Apr11-14, 01:33 PM
Sci Advisor
PF Gold
jim hardy's Avatar
P: 3,515
I'm thinking as follows:

The cap across the line cannot supply more than 1/2 a missing cycle of AC, in fact it should backfeed current into the house system and follow line voltage on down, having no effect,.... but from OP's observations it doesn't do that..

which makes me ask "what's inside that UPS? Is it a regenerative inverter that can return instantaneous reactive power to the battery ? I have no idea. "

Extrapolating(speculating) from OP's observation that his unlikely contraption actually works makes me ask " What is the nature of the UPS's internal switch that accomplishes the transfer? Is it perhaps unidirectional, so blocks his capacitor from discharging back into the grid? "
That would explain a lot.
Presumably there's a "missing peak" detector ahead of the 330 uf energy storage capacitors. Since a sinewave spends 2/3 of its time above 50% of peak, such a detector could operate with a 50% setpoint and ~1/3 cycle decay time constant. That'd give it a 12 millisecond head start on energy storage capacitors' decay(17 msec by ATX standard) .
If his capacitor is blocked from backfeeding into his housepower by something inside the UPS, it explains the behavior.

Summary :
I suspect his motor run capacitor is holding charge trapped between his UPS's solid state transfer switch and his PSU's missing peak detector, maintaining voltage above some threshold long enough for the UPS transfer to win its race with the PSU's peak detector .

But it's only a guess.



old jim
dE_logics
#14
Apr13-14, 01:06 AM
P: 735
Quote Quote by vk6kro View Post
A 10 uF capacitor has a reactance of 318 ohms at 50 Hz. On a 230 volt system this would carry a current of about 1 Amp peak, depending on the waveform.

This could be causing a current of 10 amps or more to flow from the battery, in addition to the current caused by the computer's consumption.

This will reduce the time that the battery will supply power to the computer and may reduce the life of the battery.

It is not lossless either. The capacitor is not across the mains input and there will be losses in the battery and the inverter that generates the output sinewave.

All of this has to be paid for when the mains power comes back on and the battery charges up again.
10Amps at 220 V is 2200W. The UPS is rated at 300W.
vk6kro
#15
Apr13-14, 07:23 AM
Sci Advisor
P: 4,016
That is true, but these UPS devices use a battery of about 12 to 24 volts.

This would be where the 10 amps would be flowing, just to supply the smaller current taken by the capacitor.

12 volts times 10 amps is 120 watts.
Not so alarming, but a lot of power out of a small battery.

Your capacitor does work, though, so maybe it would be worth trying a smaller one?
I would try a 0.47 uF 600 volt capacitor.

In computer power supplies, there is a small circuit which supplies the switch on the front panel with power, so that you can turn the computer on and off.
If some interference reached this circuit, it may interpret this as a "turn off" signal.
dE_logics
#16
Apr13-14, 10:19 AM
P: 735
But the capacitor is applied across an AC circuit 220V. Battery is not connected in online mode.
vk6kro
#17
Apr13-14, 12:02 PM
Sci Advisor
P: 4,016
Quote Quote by dE_logics View Post
But the capacitor is applied across an AC circuit 220V. Battery is not connected in online mode.
When there is power, a portion of the UPS circuit acts as a battery charger and also mains power is fed out to the computer.

When the power fails, another part of the circuit acts as an inverter and produces a simulated mains voltage output to the computer. It does this by taking power from the battery.

Because the battery voltage is lower than the output voltage, the battery current must be higher than the output current for the battery to supply the same power as the output.

For example if the output voltage was 120 volts and the battery voltage was 12 volts, if the output power was 120 watts, the output current would be 1 amp (120 volts times 1 amp = 120 watts), but the battery current would be 10 amps (12 volts times 10 amps = 120 watts). This is neglecting losses in the inverter for clarity purposes.

The current flowing in your capacitor has to be supplied from the battery when the mains power is off and this will reduce the time you have to use your computer before the battery goes flat.
It will also probably reduce the lifetime of the battery if you have a lot of power failures.


Register to reply

Related Discussions
Capacitor voltage of series LC circuit (inductor and capacitor) Engineering, Comp Sci, & Technology Homework 5
Electric circuit - Why is the switch needed in this capacitor circuit? Classical Physics 1
Capacitor - Resistor || Capacitor - Capacitor AC circuit Introductory Physics Homework 0
Capacitor Circuit (Capacitor as source) Introductory Physics Homework 4
RC Circuit:placement of capacitor on circuit Introductory Physics Homework 3