I am helping out a mate in a computer forum. Basically the story is :

"I bought new computer. I switch on the Fluorescent light and the screen goes blank for a moment"



Initially he mentioned an A/C unit along with the Fluorescent. I proposed the following possibilities :

Voltage drop in the line
Assume all is plugged in on the same circuit. There is significant brownout from the starting current draw of the A/C unit, and too long/thin circuit conductors to the room, or bad connection burning up somewhere upstream

Doubt : How bad the line would be to have voltage drop even from a Fluorescent (assume worst case 10A inrush into capacitor of HF ballast - I did not know yet the ballast type), yet still be capable to run the A/C unit at all ?

Surge
Assume there is a bad connection burning up somewhere upstream. Under the starting current draw of the A/C unit, the connection momentary opens, the back EMF from the air condition unit kicks everything else present on the line

Doubt : How bad the connection should be to not yet start arcing intermittently just in normal use. And how bad can it be affected by the Fluorescent

HF interference
From HF ballast in Fluroescent and/or Inverter type A/C unit

Doubt : How bad can the lantern be ? And why then the A/C would interfere only in startup ?

Unbalanced currents in home wiring
see here : http://www.lighting-gallery.net/gallery/displayimage.php?album=lastcom&cat=11192&pos=0&pid=111528

Open Earth / surge induced in Earth
Assume Earth wire is broken upstream. Inductive kick from Switch Start Fluorescent is coupled to Earth wire by capacitance or partial isolation breakdown somwhere in the lantern, and spreads to the rest of Earthed equipment, along with iterference emitted to the air



I asked to make some tests and provide more information :

1. Test again with all the conditions, and with other appliances - Vacuum cleaner (big motor) and kettle (big resistive load) : result : Only interference from light

2. Test on other circuit - He did not do exactly what i asked, no usefull data

3. If computer is laptop. test when it is isolated and running of battery : result : Still happens even when on battery (so i assume it is a laptop). Not clear to me whether with laptop LCD or external display though

4. Type of Fluorescent light : result : Switch Start
Asked him to test again with the laptop being posered from battery (isolated from mains). Problem still there - would be EMI



I think i can now eliminate all but the Unbalanced currents and Open Earth possibilities, and sent him off to run some more tests :

1. Take out lamps from lantern and test

2. Test metal lantern body with glow lamp tester when powered (Whether Earth is broken)

3. If it is external display, test the video cable (whether shield is intact end to end)



Any other thinks you think it might be ?

If it is just the monitor and it§s only responding to the fluorescent lights, two things came to my mind:
- If it is some high end monitor with ambient light color compensation, it is the monitor's light sensor (and the following algorythm), what may get confused, resulting the monitor needs some extra time till it adjusts itself for the new lighting condition. If the automatic adjustments is not necessary, switching it off would make the monitor ignoring all the light changes, so stop blacking out.

- The synchronization is lost, mainly due to the disturbance entering the signal path. Some monitors use rather slow response PLL's, so are rather immune towards such spurious glitches. But some are using fast acting synchronization and these tend to respond to any glitch by messing up the sync pattern (so causing it to go dark till the synchronization isn't reestablished; and that takes some seconds).

The problem with such disturbances uses to be either the connection between the PC and the monitor and their connection to the mains, in any case it is indicating quite severe problem, which may in it's more severe consequence end up by fried interface (seen quite some parallel ports blown that way).
So verify:
1) The PC and monitor are both connected to the same socket (good practice is to use one extension lead common for all the equipment belonging together - PC, monitor, printer,...; and do not connect anything else to that), so ensure all of theequipment connected by low voltage ports (VGA, DVI, USB, LPT,...) have single common PE with no voltage drops on it.
Plus run all the cables as much parallel to each other as possible (signal cable from the PC along the power one, till it meets the power cable from the monitor, then go along the monitor power cable; and from that meeting point both the PC and monitor power cables would go along each other till the socket)
Connecting any of the component to some different socket means a voltage may build up between the PE contacts. Either by a voltage drop on the common PEN (in older installations) or just because the loop formed by the PE connections and the signal cable may become rather large.

2) Check the signal cable, mainly it's shielding connection. It should reliably and with low impedance connect the schassis of both sides togehther and so prevent any voltage seen by the wiring inside from building up. Quite common problem is, the shielding is either not connected to both sides, or even not connected at all. Such cables should not be used (obviously except Ethernet UTP or equivalent interface rated for unshielded cable, but that interface is equipped by a 1kV insulation strength isolation transformers on each side of the cable, co can tolerate the spikes well)

And remember, the fluorescent lantern by itself is there just "the messenger", replacing (or fixing) that as the first place measure would mean the PC to monitor to whatsoever IT interface remains very vulnerable, so asking for way worse problems than just a monitor occasionally darkening for few seconds...

He does not describe things very accurately and so far i know that the computer is a laptop, but it is not clear to me whether the interference happen with the laptop LCD or an external connected monitor

Those things are allready being tested :

1 - I asked whether the computer does that when working off battery and disconnected from other Earthing sources like Ethernet cables, ie. computer is floating. He says that it does still happen. I guess we can eliminate 1 then

2 - I asked (on in "if" basis - "if the monitor is external") to test the shielding of the cable. He should answer this in his next post



If the computer is floating and the display is the laptop LCD or the cable is found to be intact, i would assume that it is either :

 - A fault with the computer or monitor itself. I seen cheap monitors that have NO internal EMI shielding - Everything just behind a plastic cover, so from EM point of view "floating in air", and with bad bonding of internal components - For example, the link between GND of the video board and PE being through internal signal cables only, as all screws are attached into plastic

 - Something like slight isolation breakdown in the lantern, disconnected PE upstream - energising the PE that an external monitor is connected to (with the laptop, even when floating, being an "antenna" to dissipate the pulse at switch ON), or other big objects on the same PE like a desktop PC next to the laptop being energized and emitting the EMI

But in the 2nd case, i wonder - If the isolation breakdown is after the ballast (that would mean the worst case for interference i guess), would not this sink the pulses to the point that the lamp does not start - but only keep flashing, or not even that but only starter keep firing ?

Update :

Laptop connected to external monitor by HDMI cable

Cable confirmed to be cheapish quality, likely to have disconnected shields. The guy does not have multimeter to verify. I asked to test continuity with a LED + battery or the like.....

He have left 1 EOL lamp in the lantern (looks like its a 2x36) for a long period. I think, the ballast might have burned the wiring in the lantern (including the wiring of the good lamp too) making an Earth fault. If PE is disconnected upstream (so, the PE wire does connect the energised lantern body and his monitor's PE connection), that may be why the light is the source of interference....

I got similar experiences too.
In my case, the keyboard of my PC usually ‘disconnects’ for 2-5 seconds upon turning my 1*F18T8 preheat fixture on or off, and back to normal working state.
Problem solved by adding a capacitor in the 2ft preheat fixture.

I don't think there would be any fault in the fixture, only it's design creates a few spikes to the surrounding. The problem there is, the cabling allows such spikes to enter the interface itself. By modifying the fixture you may solve the fixture causing the monitor going dark, but any other spike source may still destroy the interface. That's why I won't mess up with the fixture and focus mainly on the computer set.

With notebook the problem could be with the power brick:
To prevent the RF interference to radiate too much from the converter, the input to output has to be bypassed by a capacitor, so shorting out all RF signals that may "use" the cables as an antenna legs. This capacitor then forms a parasitic leakage from live terminals (effectively half of the phase voltage) to the output. That leakage is below the safety limit and the design of the capacitors is such, it fulfills the requirements of a class II insulation (= Y2 rated).
Normally the decent notebook adapters use 3-wire mains connection. The PE is there not because the safety needs it, but the capacitors mentioned above are connected with an extra few nF capacitors in series and the center point is connected to the PE. That makes all the secondary side decoupled from the mains, so the high frequency disturbances do not spread, so the whole set is then way less susceptible to e.g. cheepeese signal cables.
But many adapters just use 2-pin mains connection without the PE wire (with some cheepeese ones I've seen even the use of 3-wire cable, but the PE not connected anywhere inside of the adapter; not a safety problem, as the adapter is made as class II anyway, but definitely it is an EMC problem).

So first use decent signal cable. It is the onlything, that may prevent quite expensive damage from any spike source.
Second look at the notebook adapter and check it's PE connection. For that, you need an C-meter - the capacitance between the PE contact and the output ground should be in nF range or a dead short circuit (in models, which use direct PE connection instead - that's OK too).

Cable is allready looked at

Laptop adapter is completely eliminated - laptop does the same when disconnected from the adapter and working on the battery

But it pulls my attention why the light and not A/C unit, Vacuum cleaner, and other appliances that have EMI too.... I think there is something with the light or its wiring

The DVI cable is indeed the first one to be replaced for some decent one...

For the EMC does not depend that much what is the low frequency current of the appliance, but rather the high frequency components from the switching transients. And the way, how the wiring is made.
So:
If the fixture is connected to the same outlet as the computer set, the path to the computer is strong, so the pulse does not get that much distance attenuated.
But more important: The vacuum cleaner is an inductive load. This may made high voltae spikes on it's internal wiring (behind the switch), but towards the upstream wiring the dominant part is just an inductor, so rather slow current changes. The AC is again high current load, but again behind it's switch. And although the compressor motor has it's capacitor, it is still behind the inductance, so again no fast high current spikes towards the mains (the voltage dips are rather slow, so although they cause e.g. the incandescents to dim momentarily, it does not cause any disturbance to the signal cables or so).
On the other hand the fluorescent has most likely a PFC capacitor directly across the input. That means, once you switch it ON, it momentarily shorts out the mains (maybe for few us, so no big deal for fusing or so). And this causes many order of magnitude higher peak currents than the AC and vacuum cleaner summed up together.
And if the lantern is in a permanent "lighting" installation with the phase-switch-Neutral forming large loops (as it uses to be in old installations), you have nice antenna for that pulse to be radiated and so eventually captured by the DVI cable.

When switchig off the motor, the current is "suddenly" dropped from its value to 0, that may happen to be at the peak of it too..... That would make a high frequency burst too

Only thing i can see limiting di/dt is the arc in the switch (allowing the current to keep going), but hey, the voltage across the opening contact is in the kV range, so i'd expect to see interference of frequency about :

Vswitch / Vmains ~= (di/dt)switch / (di/dt)continuous ~= Finterference / 50Hz

 - How does that compare to the frequencies of Switch Start starting, or PFC capacitor hitting the mains ?

 - I asked about presence of capacitor, he have not opened it but the description fits a "sign strip" sort of light, those come without the capacitor. The starter is a switch and the lamp as spark gap that limit the voltage, so i'd expect the Switch Start action to behave like "behind the switch" as well



The switch loop is 1 option i am considering

The other option i am considering is an open upstream PE, and the inductive kick from the ballast being coupled - either capacitively (between ballast coil and core) or through some partial isolation breakdown (from a wire burned from the choke of the EOL lamp) to the lantern body. Then soething like :

 - Conducting it down the (open upstream) PE of the room and reaching the monitor' PE, and from there to the computer

 - Conducting it down the PE and from there to some other path - Ethernet or coax cables connected to equipment, so forming a "loop antenna"

 - Lantern body is an antenna (maybe just not connected to PE at the lantern itself, even if PE in the room is ok)

Im interested though, would any of those conditions sink enough of the energy to prevent the lamp itself from starting properly ?



"phase-switch-Neutral forming large loops (as it uses to be in old installations)"

From curiosity, how common that actually is ? I seen it only a feew times, and mostly done in hacks/additions/upgrades and not as original installation. but i cant say that is common, even in old installations

What sort of "Shortcut" were they doing in the installation you describe, and what age it is ?

I can imagine however such loop being formed by an isolation breakdown, PE, and external Earthing at some appliance (ie Ethernet cable)

Hi what country is your friend with the computer problem in? I ask because in UK installation lighting and power circuits are normally separate I believe that in Europe sockets and lights are on the same circuit. Also would it be worth him powering it thru a surge suppreser socket this will not help with voltages on the earth line but will deal with transients on phase and neutral

Israel - For the most part EU-like system, few small changes here and there. Detting of circuits in homes is most commonly done on a room basis, each room include its lights

Surge suppressor won't change anything at all. There are no HV spikes between P-N-PE involved, only a rapid changing voltage component (within range of normal voltages), or something going down the Earth wire, or induction through the air

If the cable is sloppy, there could be just small spike to cause signal error. The DVI itself uses symmetrical pairs, so the main signals do not need ground connection. So I won't be surprised the cable would seem to work even without any ground connection at all. But 1..2V common mode shift, e.g. as a voltage difference between the monitor vs PC grounds means it goes out of the input range of the receiver and so the signal gets corrupted.
So with the ground missing, even slight disturbance means problems with the signal.

What you think of the isolation breaksown / currents into PE possibilities i described above ?

Can such leakage/breakdown be limited wnough to not affect the starting of the lamp ?

It looks to me like unrealistically fine balance required to get so little losses to leakage that the lamp still starts at all (isolation breakdown or capcitance of big length of PE wiring would sink the energy too much), but i down want to dismiss the possibility right away either

The user says that the lamp flashes abit too much at startup, but starts nevertheless. I told him to remove the starter from the EOL lamp and when he did that, the starter plastic crumbled in his hand - So i guess the starter on the good lamp may be the same, then the flashing may well be just cheap starter and not related to leakage



"phase-switch-Neutral forming large loops (as it uses to be in old installations)"

From curiosity, how common that actually is ?

If the computer is connected to the same outlet as the monitor and the PE is really connected, the potential on both will be the same, so no difference from that (that is the reason, why it always should be in one socket in the first place, just to prevent disturbances coming from that direction).

Between the motors vs capacitor in the lantern is one huge difference: The inductive loads do create a high voltage overshoot, but just across the switch and/or the motor, but there is nothing else connected.
With the capacitor the high current spike is mainly in the mains upstream, so exactly where the computer is connected, or where the big loop from the sloppy wiring layout is present. But even with that, we are talking about the voltage going down, no excessive overvoltage or so, so no reasonable signal connection should ever be disturbed by that.
The fact the monitor gets disturbed by just a single light means the main problem is mainly (and I would even say only) between the monitor and the PC.

I think of cass like this : Note that in all of them the laptop is floating, so not directly connected to different PE than "its" monitor