 |
Capacitor layers and margin organization |
This page is mostly intended as a repository for technical data about my capacitor and its development status. As this is the first rolled capacitor I am building and its robustness is not yet tested, I recommend not to take this as an example and not to build your own capacitor like mine. Hopefully this design will result in a robust capacitor and I will be able to remove all the above warnings...
The main target is to gather experience about rolled capacitors' construction. The built capacitor should stand about 20 kV DC, have capacitance in the range of 5 to 50 nF and be as small as possible.
Proven capacitor home-builders state that it is safer to build two 10 kV capacitors and connect them in series, because of the reduced corona effect. I am building two identical capacitors designed to stand 10 kV each: this reduced target voltage translates into a shorter margin to leave between the electrode plates.
To prevent dielectric foil impurities and imperfections to concentrate corona effect and eventually result into dielectric perforation, more than one dielectric layer should be used. The possibility for two or more imperfections to result aligned is very small. While for dielectric strength a single 0.8 mm HDPE layer should be enough, I also use a second thin layer of LPDE (0.2 mm). This should give me some additional security against dielectric impurities, without encreasing sensibly the capacitor's size.
|
Capacitor layers and margin organization |
An "extended foil" capacitor provides a reduced inductance compared to a traditional strip design, because all the plate edges extend out of the dielectric layers and provide contact. On the other hand, its construction is much more difficult because all the layers have to be correctly offset to obtain sufficient margins. My extended foil capacitors have got 4 cm margins between the electrode plates, on each side (see picture above). The LDPE layers provide 2 cm extra margin. The purpose of paper layers is to allow oil to be well absorbed and to avoid air bubbles between all the layers: anyway, they should not provide any surface discharge path to the copper plates. The paper layer width should then be exacly equal to the width of the surface common to the two enclosing layers.
Material |
Dimensions |
Characteristics |
HDPE 0.8 mm thick |
2 sheets 25 cm wide, 2 m long |
|
LDPE 0.2 mm thick |
2 sheets 27 cm wide, 2 m long |
|
Copper 0.1 mm thick |
2 sheets 27 cm wide, 2 m long |
|
Kraft paper 0.05 mm thick |
2 sheets 23 cm wide, 2 m long |
|
The capacitors are filled with NESTE 10.X transformer oil.
 I rolled the first capacitor: it was extremely difficult to get all the 12 layers in place and straigth. After re-rolling the fifth time I managed to have the layers positioned in an almost correct way. |
 In practice I had about 5 mm max. shift of the HDPE layers and about 1.5 cm max. shift of the LDPE layers, plus some minor paper layer shifts. Also the copper layers shifted a little bit, resulting in an effective electrode width of about 16 cm, instead of the planned 17 cm. I had also to correct the length of the copper plates near to the roll outer end, so that also the copper plate got shorter than 2 m (maybe 180 cm, actually). The roll measured about 10 cm outer diameter and a capacitance of 9700 pF. |
|
 I cut the copper plates edges, leaving for each layer "turn" only two strips. It was not so easy to cut the copper sheet once rolled but I managed to do it. Then I connected together all the copper strips on each side with screws and nuts. This probably compressed further together the inner roll turns so that the final roll capacitance measured after that 10300 pF. |
Then I arranged a PVC duct case for the first capacitor. From here on the procedure was almost "standard": from one inlet I pumped some level of vacuum from the capacitor, then from a second inlet I let flow in the transformer oil.
After that I closed the oil inlet and sucked again from the first inlet as much air as I could until the created oil foam was growing too much. I did that about 5-10 times. The target was to get out the air bubbles possibly trapped in between the capacitor sheets. Once I felt satisfied, I removed the vacuum pump, let air fill the free space inside the capacitor and tapped the inlets with two rubber taps (you never know...).
Current status (2.17.2000): I have tested the capacitor with DC up to 10 kV and everything is fine, but I haven't (yet) had a chance of testing it with a real TC. My time has simply gone in a much more interesting task: the Thor Project. Maybe sometime, during some spare time...