Every time I start my server, my home UPS goes 'thud' !

That is the relay switching to battery, and after 2~3 seconds, goes back to mains. This is because the UPS detects voltage below 200VAC (voltage sag) for a short duration due to heavy inrush current of 1000W smps. I dont know how this is going to affect my 850VA UPS (upgrading to 1500VA later) regarding its lifespan. Even if it doesnt hurt it, I am annoyed by the sound it makes. I am thinking of doing something to limit the inrush current. This server has a 1000W smps.

I am also soon going to build another two servers, which will contain 24 numbers of 3.5 inch HDDs each. So a total of 48 HDDs. They both will have 850W of smps each. These are my backup servers which will be started 2~3 times per week, take in back up of my data lasting around 1 hour, and then shut down. I might even have to start these servers if there is any important data to be backed up everyday.

So now I have now become a bit more concerned about the blips of huge currents passing thru my UPS every now and then. I had thought of adding more NTC thermistors in series, but opening each smps and modifying the circuit is a bit of pain. Also eventually I will have more computers in my home office, so this circuit will be applicable to all.
With my limited knowledge I have come up with the following circuit. Please see if it is any worth ?



This circuit will sit in between the UPS and the server/computers.
Inductance can be 100u -> 10m -> 100m, anything will slowdown the current.
I am assuming the varistor will cut down the oscillations if it goes above 250V. It will act like an immediate damper !
And the X class capacitor will help with the noise/spike reduction or anything with a high slew rate.

So what do you say ? Does this circuit make sense or its a garbage design ?

Thank you for your time.

 

Wouldn't it be simpler/cheaper to use a suitably-rated NTC inrush-suppressing thermistor made for the job, in series with the mains input?

 

Explain what you think your circuit is doing, because it is not obvious to me.

Edited to add: It will certainly limit the current, but not only at startup and by far too much.

 

Why don't you build yourself a power supply sequencer - like audio touring companies use for their amplifiers? Then only one device creates an inrush current at any time.
Note: if you are starting up switched-mode supplies, then it is best to start the at mains zero-crossing, but if you are starting up traditional transformer power supplies, it is best to start at mains peak.

 

Why don't you build yourself a power supply sequencer - like audio touring companies use for their amplifiers? Then only one device creates an inrush current at any time.
Note: if you are starting up switched-mode supplies, then it is best to start the at mains zero-crossing, but if you are starting up traditional transformer power supplies, it is best to start at mains peak.

True, its called staggered powerup.
That technique will solve the problem of starting 24~48 HDDs all at once. But my problem is not just that, its also about switching on 1000W/850W/850W smps.

 

Edited to add: It will certainly limit the current, but not only at startup and by far too much.

Agreed.
Apart from inefficiency, do you think this will generate/induce noise at sub 250VAC levels and which cant be cleaned with that capacitor ?

 

Many switch mode power supplies use an NTC device to limit the inrush current at switch on. Typically these devices have a resistance of 4Ω - 10Ω at 25⁰C, which reduces to less than an ohm due to internal heating as a result of the current flow.

 

Agreed.
Apart from inefficiency, do you think this will generate/induce noise at sub 250VAC levels and which cant be cleaned with that capacitor ?

I think your 1000W power supply will not work at all with a 100mH inductor in series.

 

Hello,

With inductive loads, sometimes it is better to start at the peak of the sine wave rather than the zero crossing. That is because of the transient response of inductors. It seems unintuitive to do that, but if you start at the zero crossing it could simply mean that the maximum current is reached just a few milliseconds later even though it may start out at zero. This means starting at the peak leads to a lower current peak overall. It depends highly on the particular circuit though.

The traditional way to handle this is to incorporate a slow start mechanism. This is actually used in a lot of power supplies. I don't know how hard it would be to modify all of your power supplies though, but that's the best way because you have complete control over the startup sequence without resorting to unusual fixes like surge limiters, inductors, capacitors, and stuff like that.

 

With inductive loads, sometimes it is better to start at the peak of the sine wave rather than the zero crossing.That is because of the transient response of inductors

If the load is a transformer, it is because of the saturation. Flux is 90° out of phase with voltage, so the transformer should be started when the flux would naturally be zero (i.e. at voltage peak).
If it is started at zero-crossing, then the magnitude of the flux keeps increasing for a complete half-cycle whereas it should reach its peak after a quarter of a cycle and then start to return towards zero.
A transformer started at zero-crossing will almost certainly saturate just before the next zero-crossing, unless it is wound for low flux.

 

If the load is a transformer, it is because of the saturation. Flux is 90° out of phase with voltage, so the transformer should be started when the flux would naturally be zero (i.e. at voltage peak).
If it is started at zero-crossing, then the magnitude of the flux keeps increasing for a complete half-cycle whereas it should reach its peak after a quarter of a cycle and then start to return towards zero.
A transformer started at zero-crossing will almost certainly saturate just before the next zero-crossing, unless it is wound for low flux.

Hi,

But isn't that true with a general inductor?
We get rid of the exponential part (in some circuits) by exciting with a cosine wave (starts at the voltage peak at t=0) rather than a sine wave which starts at zero volts at t=0. This is true of a general inductance. The solution is a pure sine wave even at t=0. This is sort of intuitive because when we apply a step voltage to an inductor there is little or no current flow for a short time.

 

Hi,

But isn't that true with a general inductor?
We get rid of the exponential part (in some circuits) by exciting with a cosine wave (starts at the voltage peak at t=0) rather than a sine wave which starts at zero volts at t=0. This is true of a general inductance. The solution is a pure sine wave even at t=0. This is sort of intuitive because when we apply a step voltage to an inductor there is little or no current flow for a short time.

Indeed it is.
My experience comes from the disco pinspot of the 1990s. Basically, a #4515 lamp and a transformer in a case, Perfectly OK for lighting a mirror ball, but connect it to a light sequencer and turn up the speed and the transformer quickly dies.
They were made to compete on price and therefore to run the transformer as close to its limit as possible, and all the sequencers had zero-crossing sync'ed outputs.

 

Indeed it is.
My experience comes from the disco pinspot of the 1990s. Basically, a #4515 lamp and a transformer in a case, Perfectly OK for lighting a mirror ball, but connect it to a light sequencer and turn up the speed and the transformer quickly dies.
They were made to compete on price and therefore to run the transformer as close to its limit as possible, and all the sequencers had zero-crossing sync'ed outputs.

Hi,

Sounds like they took "All Microwave Ovens 101" at Walmart University
You know what they got on their SAT's right? Mustard, mayonnaise, and katsup.

 

Hello,

With inductive loads, sometimes it is better to start at the peak of the sine wave rather than the zero crossing. That is because of the transient response of inductors. It seems unintuitive to do that, but if you start at the zero crossing it could simply mean that the maximum current is reached just a few milliseconds later even though it may start out at zero. This means starting at the peak leads to a lower current peak overall. It depends highly on the particular circuit though.

The traditional way to handle this is to incorporate a slow start mechanism. This is actually used in a lot of power supplies. I don't know how hard it would be to modify all of your power supplies though, but that's the best way because you have complete control over the startup sequence without resorting to unusual fixes like surge limiters, inductors, capacitors, and stuff like that.

Hello sir, Everybody is giving the same advice as you are. I have found some tutorials on youtube on this same topic.
I have found some cheap circuit delay timers on amazon. Maybe I will try to make use of it to make my life easy.
Thanks for your advice sir.