Troubleshooting a High-Frequency Noise in a Capacitor

[XiangYan]

Hi,
I am making a capacitor with Aluminum plate and Mylar film. Al plate surface has tiny groove arrays, each groove forms a cavity and the mylar film is sitting on top of the groove to allow its vibration under input AC voltage.
I am putting in AC of ~300Vpp with frequency of 60KHz. The AC signal input is pretty noise free (observed from OSC).
But I can hear some high frequency noise coming from the Capacitor, frequency ranges from 5K~10K. Is anybody know where this noise coming from ? Is this Al alloy dependent ? Because when I use 6061-T6 Al plate, the noise is gone, when I use 3003 (not sure the temper) and 6063-T5 plate, the noise is quite obvious.

If anybody has any related knowledge I would be greatly appreciated. Thanks a lot.

 

[Enthalpy]

When stacking in air the electrodes and an insulating film, you enclose an air film whose dielectric strength is far worse than the plastic. If the field suffices, that is far before the strength of the plastic, you obtain sparks in the air layer which are noisy.

To exploit the insulator's strength, capacitors are rolled within an oil bath, or the electrodes are evaporated on the plastic film under vacuum.

If these sparks are the cause of the noise, the alloy should not make a direct difference. Maybe some alloys are still fat or damp and the others are dry? That would influence the gas layer.

An other difference is that the native oxide layer is a good insulator for 3003 and 6063, less good for 6063. But would that have an effect?

 

[XiangYan]

Thank you Enthalpy.
Do you mean 3003 and 6063 have good insulator oxide layer, while 6061 less good ?
The air layer does not spark (we can hear and see if it sparks, which happens occasionally).

 

[Enthalpy]

Yes, 3003 and 6063 form an excellent oxide layer, very hermetic, that protects them efficiently against corrosion. The little bit of copper in 6061 degrades slightly the oxide layer, and accordingly, anodization isn't as good with it, nor is corrosion resistance.

Alloys in the 5000 (5083, 5457 resist marine corrosion) and possibly 1000 series (1050, 1080, 1099 increasing purity) could be even better than 6000, while copper in 2000 and zinc in 7000 are detrimental to the oxide layer.

-----

Could you observe the small HF component of the current or voltage at the capacitor that is outside the intended frequency? This would give indications about what happens and at what frequency.

Maybe you have some kind of Corona effect which isn't very bright. Also, discharge from a 60kHz AC voltage does not look and sound like at 50Hz or DC.

Are the dielectric and metal plates completely dried? Tiny vapour bubbles exploding would be hearable as well.

You have grooves in the aluminium plates: are these grooves open, or do they resist the expansion of warm air, whose bursting leak is noisy?

I'd try to assemble a capacitor within oil.

 

[Enthalpy]

One other difference: when you machine your grooves, 6061 makes sharper edges which concentrate the electric field and favour micro-discharges there. You could try to sandblast the grooved plate, or smoothen it (electro-) chemically, to get rounder edges.

Usually, capacitors tend to use very smooth metal plates.

If you press the sandwich together very strongly, maybe the softer 3003 and 6060 deform and match the irregularities of the Mylar film, while 6061 doesn't. Not quite convinced, because rolled Mylar isn't extremely hard neither.

 

[XiangYan]

the drive voltage input is quite clean, no HF components correspond to the observed noise. Corona discharge happens and I think should happen regardless of the Alloy, the E-field should be on the same order for all cases.
The plate is quite dry and the groove is open to air.
Actually, we use this particular grooved Al plate as one electrode, the Mylar film (electroded on one surface) as the other electrode, the film is very thin and compliant, which allows it to vibrate under the AC to create a dynamic Capacitor.