I did a test: When my computer is off, the Osram LED filament at the corridor of my ceiling at my hostel, working without a problem, but when my computer is on, it vibrates in brightness.
Why?
https://www.youtube.com/watch?v=4BnWg9XQ9dE

I cannot see it in the video.

Is it behind a dimmer? Dimmers are very sensitive to other devices on the grid causing interference.

Your computer may be causing that interference. In earlier days people would quickly complain about that because it makes AM reception impossible, but today it can remain undetected until things started acting weird. This is also the reason why fluorescent fixtures are required to have suppression capacitors (at least in europe).

If it is not just your computer causing this but the issue also shows up when you connect a 300-500w incandescent lamp or a heater on the outlet where you normally plug in your computer, and your lamp starts flickering randomly, this means there is DANGER. With the small load of the lamp there is no problem, but once extra load is turned on, a wire nut or other mains voltage connection starts to heat up and spark, causing flickering. This is very dangerous.

The computer is the cause, as it don't shimmering when the computer is off. The voltage on that phase, is usually 200-210V.

If the voltage on that phase there is just 200..210V, computer is not the dause, faulty wiring is. The fault cause the PFC in thecomputer to be unstable, causing voltage amplitude to getmodulated by the instability and so the lamp starts to flicker.
So better ispect the wiring, such low voltage on a 220V, moreover 230 or 240V rated mains is everything but normal...

Modern PSUs are rated for 100-240V, 50-60hz.

They are, but are also rated only to work from a low impedance voltage source, so from a source where the voltage does not change with load change.
If you add a series impedance, the reduced voltage is by far not the only thing the circuit has to deal with, the fact the voltage droops when the PSU sucks more current is.

In a steady state power factor correction operation, the internal regulation loop needs to be slow, so it won't respond to the AC voltage (the sinewave), so the input current can follow the input voltage shape, so it won't cause distortion. But this creates a problem when the conditions, mainly the output load, it has to have extra mechanisms to respond quickly to prevent undervoltage or overvoltage on the DC filtering capacitor. Then the circuit settles into a new steady state operating point, where again the regulator responds rather slowly and so maintains low distortion. The over/undervoltage actions do lead to anything but a sinewave input current, but that is not that big deal, as it is only active during big transitions.

The problem is, when the supply source is weaker, has high impedance. When these under/overvoltage functions activate, the current surges/dips cause big change of the input voltage but mainly do not lead to sufficient recovery effect, so the controller goes into more severe action, so larger current surge and larger dip. And the whole thing starts to oscillate, at some fraction of the mains frequency, so some 1..30Hz.

It is technically possible to design a PSU able to work on weaker input power source (it mainly needs larger DC tank capacitor after the PFC and mainly extra controller functionality so it is able to find a stable operating point), but normally there is no reason for such extra complications for mains powered PSU.
So when the supply on a standard mains equipment becomes weak, these instabilities are quite normal, often triggering protection circuits or even not so rare is causing excursions leading to some component destruction (good example are cheepeese inverter welders operated on small gensets - these combinations tend to fry either welders, generators, damage the engine or any combination of them).

I forgot to say that my computer is connected to a Line Interactive UPS with an AVR which boost the voltage when it is low, so now my computer don't gets the low voltage of the phase it is connected. So why the LED lamps still shimmering?
My computer PSU is Seasonic 620W Active PFC 80+ Bronze.

The boosting UPS fixes just the steady voltage problem. It won't have any effect on the high impedance.
Even worse, the flickering could actually come fromthe UPS, when that is searching for a good boosting ratio, when the computer power supply (a constant power input, so higher voltage means lower current) may cause the UPS to overshoot, so starts "hunting".
Neither thecomputer power supply, nor that UPS are designed for the input power feed tobe so weak the load these things exhibit has impact onthe mains voltage.
They are designed to get the input voltage in a wide range, but it should not change in response to the load these things present.
If the impedance gets so high the mains voltage responds to the load, the regulation may become unstable, so starts oscillating, so the lamp to flicker...

249V output voltage looks strange.


Some years ago I did a repair on a large Furman voltage regulator/stabilizer, intended for on-stage application. The unit generally worked, but output voltage jumped chaotically and indeed it caused some strong interference and light flickering on the mains upstream. Regulator topology was a large toroidal autotransformer, very heavy, circa 3kVA, with about 15 taps. Taps were selected by a series of optocoupler controlled TRIACs, switched by a digital output of a reversible counter. Control loop idea was when output voltage goes out of spec, counter jumps one step up or down and selects a next tap on the transformer. Somewhat reminiscent of the very old-school idea of a reversible servo motor  rotating a VARIAC brush, but done in semiconductor implementation. I spent may be a half-day and the evening reverse-engineering (these jerks do not provide schematics) and looking for an actual defect. Long story short, an idiot who designed this professional and rather expensive unit cheapened to use a base-emitter junction of a general purpose silicon transistor as circa 7V main reference voltage zener diode!  Over the time this 'zener' degraded and become noisy and jumpy, causing transformer taps to be changed in a fast chaotic way, out of sync to line periods, and actually generating intense voltage sags and interference on the  power line!

But clearly this UPS is controlled by the microcontroller, which also measures the output voltage. And I doubt it would use separate voltage sensing for regulation and separate for displaying the voltages in the app. Most likely it uses a common V-meter (I mean divider, ADC and processing) for both displaying the output voltage, as well as controlling the transformer tap switches.
So if it displays 250V, it means either that is what the regulation aims for (within the tolerance), or the regulation loop itself got crazy (due to input feed impedance and load behavior, a combination the thing was apparently not designed for) and it is just hunting around to reach the correct output so much, the reading after averaging (i guess the regulation uses fast ADC readouts, but the display averages/filters them out to get somewhat stable reading on the "display") it shows the voltage being higher than actually intended in average.

And essentially the same principal functionality (minus the user readout display) is in the PFC of the computer PSU. So you have a cascade of two elements which both could get crazy under some not so usual conditions, so making the combination even way more likely to go crazy...

But it could also be fault in one of them. The computer PFC very likely uses analog regulation loop, so will use analog capacitors and resistors to program the dynamic response behavior. And the main power tank capacitor is also part of this loop. So once some of them dries out and degrades, the parameters may shift so it starts become unstable by itself, or at least more sensitive for higher mains impedance (and there the boosted voltage from the UPS will have higher impedance than the direct mains when boosting, due to the extra resistance of the transformer winding, even when the UPS itself is behaving correctly). And the computer PSU oscillations may then also upset back the UPS control, throwing everything into the deep mess of unstable regulation loops.

I doubt the control problem could be in the UPS itself, as there all the dynamics is most likely implemented in the processor, so not relying on any parameter of external timing components, except of the CPU clock generation. And because this clock generation is also used for the communication and the communication to the external world is working, the CPU clock will not be defective.
The only thing that remains is some faulty switching element on some regulation step. Then the voltage would disappear completely instead of e.g. getting reduced by a few volts. But on the other way I would expect the firmware will contain some detections of such "impossible" states (the voltage dropped to zero instead of just by few V) and trip some alarm and/or safe foldback mode.

Why would a cheap UPS AVR have any feedback at all ? I would expect it only measure the input voltage, and based off that select the transformer tap : V<A run of battery, A<V<B tap +20%, B<V<C straight through, C<V<D tap -20%

If the changing load vs. line z make it change taps repeatedly, it would produce some constant relay clicking sounds that Dor would notice. Most likely it was just staying on the +20% tap the whole time. That aligns pretty well with 249V output, for an input voltage around 207V

The feedback then pops out:
- Output voltage goes lower
- The mains voltage sags so the UPS sees it as input getting lower
- The UPS switches tap to higher output voltage.
- That cause increase of the input current, causing the input voltage to sag deeper
- The UPS makes another correction.
- So the voltage recovery takes twice as long. This may cause stronger current draw response from supplied systems than anticipated.

Plus for each step there is some kind of transient response of the computer power supply, mainly the PFC input stage, moreover next step comes before the response for the first settled, which may cause way stronger "swing" of the regulated output, so more likely triggering the "hard limiters", which form way more abrupt changes of the input current.

The UPS won't make another correction because it does not have any additional taps. If it goes out of the range that can be corrected by the first (and only) tap, it will switch to battery

Besides, even if it had, it would not have sufficient effect to reach to the next tap of the UPS :

Assume the only load on the line is the PC+UPS. The voltage measured on the line is 210V. The voltage supplied before the bad connection is 230V. The PC power (at input to UPS) is 60W. Assume UPS and PC PS efficiency is constant and not dependent on voltage

60W / 210V = 0.28A
(230V - 210V) / 0.28A = 70 Ohm (resistance of supply upstream, so the bad connection)
Now lets say the voltage dropped to 190V
60W / 190V = 0.31A
70 Ohm * 0.31A = 21.7V
230V - 21.7V = 208.3V
So even with the current draw of theoretical 190V, the line voltage still goes towards the 210V. We can go another iteration and see that it would still stabilize at 210V, as long as the fault resistance is the same and PC power is the same

Now as the PC is likely not the only load on the line (there are other loads that are on the same bad connection, but not on the UPS), the actual voltage fluctation on the line due to the PC alone will be even less. It does not take much voltage oscillation to make lamps visibly flicker

Then i guess that the LED lamp there is modulated at 100 Hz anyway (lighting only above specific momentary voltage of the sine wave), so smaall changes to the amplitude or distortion would change the lamp on duty cycle..

The observation, as I understand it from the description is, the lamp flickers when is turned on at the same time with the computer and the flickering stops once the computer is switched off. And the flicker is present whenthese things are on the UPS.

The lamp starts flicker once the mains stsrts to fluctuate, if I understand it the problem is not with the 100Hzitself, but with some lowe (subharmonic,...) frequency.
So that way the lamp can flicker only when the mains voltage is modulated by such lower frequency.
Because the flicker disappears once the computer is switched off. So the modulation is linked to the computer drawing power.
The computer does not draw more than low few amps, yet there is a significant impact on the mains voltage,
That means the impedance of the mains at that point is unusually high and it is most likely alsothe reason why the voltage is so low there in general.
The way the regulation of the PFC stages use to be designed means they tend to change current abruptly once the internal tank voltage reaches the limit. This is normally intended to keep the tank voltage within reasonable range even on an abrupt load change, but still allow the main feedback to be rather slow, so it won't respond to the 100Hz ripple and so cause current distortion.
This works pretty well when the mains is the usual low impedance source, so it won't jump up or down even when the limitter activates and the PFC suddenly stops drawing current.
But whenthe mains connection is weak, the sudden current change cause significant voltage jump and that then makes the PFC to respond excessively again, causing another jumps and so on, so you have pretty wildly oscillating system.
The voltage correction function of the UPS may start to respond to these as well, adding another layer to that mess.
Plus when in boost or buck mode, the UPS adds extra series resistance (the resistance of the winding section between input and output), making the mains impedance problemevenworse.

And the main question is,why is the impedance that high inthe first place. Normal wires would need to be higher 100's m long to get that. That does not seem like plausible explanation.
Anotherculpritcould be some bad connection, causing intermittent contact and arcing. Because that is something that responds to the load current (heating in the contact spot, with thermal expansion moving with the contact), so something that could be really contributing to the PFC upsets.
And there the main problem is not that much the lamp flickering alone, but the bad connection posing fire hazard... 

What i mean is : The LED lamp may (this is just a guess) be made as a series of LEDs with fairly high (say 270V++) forward voltage and a linear current regulator. The lamp will light only around the tops of the sine wave anyway. So it will normally flicker at 100 Hz with some duty cycle D

The low line voltage make the lamp already work at lower D than designed to, so near the very top, low dv/dt areas of the sine wave

Any additional rectifier load on the line without PFC (the other computer peripherals like monitor, maybe even the UPS transformer saturation current near the voltage peaks) make the top of the sine wave flat, so introduce an even lower dv/dt region in the sine wave

What for the voltage is low dv/dt, means for the time high dt/dv...

This mean, that if the minimum voltage for the LED lamp is near what the flat top of the sine wave (at the low voltage) is, then any additional small change in the voltage will make big change in the duty cycle D that the lamp can light with this voltage

So : The PC PFC may make an oscillation at say 5 Hz that has a minimal effect on the line voltage (+/- low few V rms between the top and bottom areas), which will not be sufficient to trigger the UPS changing modes, but may well be sufficient to change the duty cycle (of the lamp's natural 100 Hz flicker) to big extent

With some luck the line voltage may end up so precise near the lamp cutoff voltage that it will make the lamp full on/off flicker at the same 5 Hz...

And all this without requiring big 20%-order fluctations of the line voltage

Besides, i would expect an UPS to have sufficient time delay built into the AVR logic, to not be able to flicker at 5 Hz with its tap changing relays. (Though cannot expect much from an Advice UPS nowadays)



This means then, that the bad connection may not be so close to the PC, but quite a bit further upstream in the circuit. So the z is not very high, but the current loading on it is not just a PC but significant part of the house, which leads to the same "working point" of voltage drop

Which leads to question, how long would such bad connection last before it melts the insulation sufficiently to cause a short circuit somewhere ?

70 Ohm in a final circuit (local to the room with the PC) is a lot. At such resistance, at the PC load from example above (0.28W), it will dissipate 5W. I can imagine a hot connection with melted insulation (but by chance not touching anything) in a wall connection box last a few years dissipating 5W

But then it means, there will be very noticable effect on the line voltage from the activity of the PC - The PC booting or opening some program make the line voltage really dip, lights go out, UPS click taps etc. And this was not reported....

If the connection is more upstream, so part of the entire house is on it, the load may be averaged enough to make the voltage on that phase appear fairly constant low at around the 210V

This will mean a lower resistance bad connection, carrying much higher current so dissipating much higher power... But being on thick conductors mean they provide better heat sinking to the bad connection, so slow down the destruction of the insulation



The quantity of brownouts, low voltage conditions, etc described by Dor over the years does seem abnormal (on my experience the electrical grid here is fairly reliable and well maintained, i live in a different town but it maintained by the same company), so there may be several problems in and around the house specifically



Dor : What is the extent of circuits where the voltage is 210V ? Is it just that room and the nearby corridor, or big part (1/3) of the house ?

If you want to monitor the line voltage in different areas of the house, see if you can get some simple plug-in voltage meter with display for home use. I can't imagine the other people there being happy with you sticking an UPS or multimeter probes into sockets

(Besides, the common cheap multimeters in the <50..100 NIS range don't have adequate safety construction to measure line voltage with, even if they are rated for the voltage itself)