Request EE advice, please

[Anthony P]

I am attempting to construct a home-built capacitor for a dye laser flash lamp. I am using 5mil Al sheets 10"x 15" active area and 12"x18" 5mil mylar. I assembled and clamped 40 layers together. Capacitance read 0.74uF (great). Voltage will be 20kvdc.
The problem is that the resistance is not infinite. As a good capacitor should, resistance starts at "short" and quickly climbs... but only to about 3megs. When I test commercially made HV caps, the resistance reaches infinity (200meg meter limit). When connected to HV supply, it does not build up a charge or store one.
I expected it to be "leaky" but was hoping submerging in oil would prevent this. I have not yet tried the oil as I wanted to do as much testing as possible before making that mess.
Any advice would be greatly appreciated.

 

[Cyparagon]

Are you holding the test leads with your fingers during this test? Did you clean the mylar prior to assembly? Try peeling the layers apart slowly and see if/how the resistance changes.

 

[Anthony P]

Thank you for your reply, Cyparagon. I am no EE.
The test leads were clipped on. Metal and mylar were cleaned as I went. I did disassemble the whole thing layer by layer. Resistance climbs to infinity( over 200Meg) until I get to about 10 layers. I have just re-assembled with double mylar layers. At 40 layers I read a capacitance of 380nF, and resistance climbs very, very slowly(10min or more) to 8.4 meg... still not infinity. Have not placed in oil yet.
I have read that these types of caps tend to be "leaky". I wonder if leaking vs charging from ohms test is reaching an equilibrium point. Further I wonder If the oil bath will stop or reduce leakage.
I will be ready for HV test and oil this weekend.
As a plan "B", I ordered a Maxwell 20kv 1uF from e-bay. I wish they had posted their add a month ago, before I started.

 

[Cyparagon]

I don't see how oil will help here, although I am possibly missing something. If your dielectric is 3Meg, placing an infinite resistance in parallel with it changes nothing. You would need to put the oil in "series" with the existing dielectric, which would physically mean separating the metal layers further with the oil. I don't think that's how these are constructed.

As you probably know, 20kV on 8 megs is 50 watts. That's several orders of magnitude over what is acceptable for a capacitor. Sounds like you need a better dielectric.

 

[Anthony P]

I don't see how oil will help here, although I am possibly missing something. If your dielectric is 3Meg, placing an infinite resistance in parallel with it changes nothing. You would need to put the oil in "series" with the existing dielectric, which would physically mean separating the metal layers further with the oil. I don't think that's how these are constructed.

As you probably know, 20kV on 8 megs is 50 watts. That's several orders of magnitude over what is acceptable for a capacitor. Sounds like you need a better dielectric.

I agree. There are only 2 parts: The conductor and the insulator. It will not accept a hv charge at all. Acts like a short on the power supply. Even if it was some leakage situation I should still be able to get some voltage into it. I will start researching other dielectrics. I really thought mylar was a good choice.
Thank you for your help. I will let you know how it goes.

EDIT:
To clarify, a single layer of mylar reads infinite resistance from metal plate to plate on either side. When stacked more than about 10 layers, resistance decreases. My thoughts were that charge was bleeding off from the edges of the plates and contacts. I will rebuild with laminated polyester pouches for each layer.

 

[Anthony P]

I hope this does not count as double post. Maybe if I say bump first it's ok? The fact is, I have spent the last weeks modifying the capacitor.

As Cyp pointed out, the insulators were the problem. I came to find out that the polyester sheets I used originally were labelled "anti-static". This was the problem considering that a capacitor is essentially a bucket of static. The new version uses 3mil laminating pouches that are not "anti-static". They work perfectly! At 42 layers I have .56uF with no break down at 20KV. It charges, discharges, and holds charge very nicely.

I have discovered, however, a positively terrifying effect. While completely disconnected from power supply and only to HV probe(100meg), the cap slowly builds up a charge... over 200v in 5 min. I will have to add a HV switch to short the output for safe handling and storage.

I did finally find a Maxwell 1uF 20KV cap on e-bay from Thailand. It arrived within a week of order in spite of the fact that estimated shipping time was up to 2 months. I will try both the Maxwell and the home made cap on the dye laser. Also, I can add more layers to my HM cap to increase capacitance if necessary.

This is some real Dr Frankenstein stuff going on here!

 

[Cyparagon]

Most of the "self-charging" you're seeing is just dielectric absorption. The electrons embed themselves in the insulator to an extent. Even when the device is shorted momentarily to dissipate the charge, the electrons will eventually find their way out of the insulator and back on the plates.

 

[Anthony P]

Most of the "self-charging" you're seeing is just dielectric absorption. The electrons embed themselves in the insulator to an extent. Even when the device is shorted momentarily to dissipate the charge, the electrons will eventually find their way out of the insulator and back on the plates.

Thank you so much for your insight. It has proven very helpful.

One quick question: the cap will be connected to a 100meg voltage divider to run a volt meter. Would the resistors then eventually drain off the residual charge?

The quartz parts arrived today. With the electrical issues essentially resolved, it is time to shift gears into mechanical mode to construct the laser head.

EDIT:
With a 17KV charge, after 3 hours, it still read 40v. Overnight it did read 0v.

 

[Cyparagon]

17kV on 100MΩ is 3 watts. Are you sure you want to burn 3W of power running a meter? Yes, that will discharge the device rather quickly.

 

[Anthony P]

17kV on 100MΩ is 3 watts. Are you sure you want to burn 3W of power running a meter? Yes, that will discharge the device rather quickly.

To clarify, by "run", I meant that a divider would be connected across cap to monitor its charge level. I could increase Rt if you think that would be better. Also, the HV probe I used for bench test is actually 1000meg, not 100meg. I did intend to use 100meg in final build, but I can easily change that.
Again, thank you for your time and help.

Edit:
Just ordered parts for 1000meg.