Need some electronics help

[npbg6464]

I have a CFL backlight connected to the 12V rail of my HTPC and controlled by a physical switch:

The problem is, if I turn the backlight on while the PC is powered, sometimes the PC will hard reset. I suspect that powering up the backlight causes a voltage spike that will sometimes trigger short-circuit protection in the PSU.
I was told that adding a large capacitor BEFORE the switch may get rid of the voltage spike when the backlight is turned on. From what I understand it should be wired like this:

However, the spike still occurs sometimes. Does anyone here know if this is the right approach? The capacitor is not connected right next to the switch but a short distance away with jumper wires. Might that be the problem? Do I just need a bigger capacitor? Any help is appreciated, thanks!

 

[moerik]

Both the PSU and the Capacitor are parallel to the switch which could be at fault, what you should try is to put the PSU and the Capacitor in a series circuit.
Here is a cheap text-based diagram; PSU -> Capacitor -> Switch -> CFL Transformer -> PSU.
Please report back with your findings.

 

[FAST6191]

Are you sure it is a voltage spike/drop issue and not a current draw (inrush current or general overload) issue? Though a PSU might be several hundred Watts (though being a HTPC I doubt it is one of the beefier ones) they are often broken up into rails with substantially less power available to them, a 15W bulb could make a huge difference. If you have a spare rail on the PSU try that instead, I do not know what PSU you have so I can not say but there will usually be a couple and, limits allowing, it can be nice to have a secondary one used for this sort of thing.
If you are stuck with the setup you have it is usually capacitors, or maybe an inductor if you favour being a bit more classical, that are usually used to solve the same problem.

Alternatively what about an LED? Some of the white ones do very well these days.

 

[grossaffe]

Both the PSU and the Capacitor are parallel to the switch which could be at fault, what you should try is to put the PSU and the Capacitor in a series circuit.
Here is a cheap text-based diagram; PSU -> Capacitor -> Switch -> CFL Transformer -> PSU.
Please report back with your findings.

It's supposed to be in parallel. The voltage on a Capacitor cannot change instantaneously, making it useful for making for a cleaner voltage.

As for the overall concept of what's going on, I believe CFLs use AC electricity, so the CFL transformer should be converting that power from DC to AC, meaning it probably uses Inductors. Inductors can have some interesting properties when hooked up to switches as it suddenly introduces or removes a closed loop on the Inductor. Spark Plugs, for example, take advantage of the fact that current can't change instantaneously, so when you flip a switch that kills the loop the current is flowing through, it sparks across at a really high voltage because the current is forced to change almost instantaneously (V = L(di/dt), so as the current is forced to go from where it is to zero really quickly, it results in a short high voltage burst). You may also notice that when your AC kicks on in the summer that the lights dim briefly before returning to normal status. This is also a result of inductance as the inductor pulls extra current when it's first connected to the circuit to get up to it's final voltage before it settles back in to a steady state. So were I to guess, I'd say you are experiencing the result of self-inductance as the switch is thrown.

Now I'm conjecturing here, but I'm thinking that since the problem is when you close the switch, the capacitor may not be of much help as it is situated because it's voltage will be the same as the voltage on the power source, so it's still an instantaneous change in voltage that the inductor has to pull lots of current to meet. Perhaps if you place the capacitor on the other side of the switch, it'll start out at zero volts and steadily increase to the final voltage so the inductor can also steadily climb in voltage instead of having to try to jump directly to 12 V and drawing all that current.

 

[npbg6464]

Thanks for everyone's replies! I tried a number of things like putting the capacitor on the other side of the switch and adding an inductor before the capacitor - didn't help. What ultimately solved the problem was replacing the rocker switch I had to a push-button switch... go figure... I still have the capacitor installed like in my previous diagram just in case. No more crashes in over a week of use. Btw, my power supply is an OCZ 600W.