I bought some cheap Chinese 630V capacitors and I wondered... How good are they?
So I built something to check leakage current.
The simple idea is to place a current meter in series with the capacitor and connect it to a high voltage power supply. But as you can imagine, that's probably not a great idea if the capacitor suddenly goes short circuit.
So the simple solution is to use a current limited supply and/or protect the meter with some diodes
But let's say you limit your power supply to 20mA and you're measuring leakage in fractions of a uA. This will protect the meter from high voltages, but not necessarily from high currents.
So that's a solution, ats a bit of series resistance to limit the current. Using 10, 100, and 1k ohms will limit the current to 70, 7, or 0.7mA depending on the burden voltage of your meter.
A quick check in my meter indicated the maximum readings would be closer to 50, 5, and 0.5 mA.
The circuit also contains a series resistor from the power supply, but I'm using a current limited supply, so I omitted it.
Here it is partially constructed
I had to replace 2 of the terminal posts, so this is actually during this step. The other side shows the keying of the holes.
The top plate is some white acrylic painted black and etched/cut. A similar MDF version was made to guide the pilot holes in the case
Fully built it looks like this
And put together
In operation, the capacitor is connected across the top pair of terminals, the power supply to the next ones, and the meter between the lower left black post and the H, M, or L red post, depending on the amount of leakage you expect.
I tested a small tantalum capacitor to destruction using a low voltage power supply, then used my 1kV variable supply to test some known good, and Chinesium capacitors.
A known good 0.1uF 1200V cap showed a leakage near zero at 1000V. My meter in the uA range shows 0.1uA or 0.2uA when the leads are open, and maybe it spent more time on 0.2uA at 1kV.
The Chinese caps measured pretty much the same up to about 700V -- I didn't go higher.
A 10uF 450V cap showed about 50uA leakage at 450V. I've yet to test the Chinese electrolytics.
This turns out to be a very simple and useful device.
So I built something to check leakage current.
The simple idea is to place a current meter in series with the capacitor and connect it to a high voltage power supply. But as you can imagine, that's probably not a great idea if the capacitor suddenly goes short circuit.
So the simple solution is to use a current limited supply and/or protect the meter with some diodes

But let's say you limit your power supply to 20mA and you're measuring leakage in fractions of a uA. This will protect the meter from high voltages, but not necessarily from high currents.

So that's a solution, ats a bit of series resistance to limit the current. Using 10, 100, and 1k ohms will limit the current to 70, 7, or 0.7mA depending on the burden voltage of your meter.
A quick check in my meter indicated the maximum readings would be closer to 50, 5, and 0.5 mA.
The circuit also contains a series resistor from the power supply, but I'm using a current limited supply, so I omitted it.
Here it is partially constructed

I had to replace 2 of the terminal posts, so this is actually during this step. The other side shows the keying of the holes.

The top plate is some white acrylic painted black and etched/cut. A similar MDF version was made to guide the pilot holes in the case
Fully built it looks like this

And put together


In operation, the capacitor is connected across the top pair of terminals, the power supply to the next ones, and the meter between the lower left black post and the H, M, or L red post, depending on the amount of leakage you expect.
I tested a small tantalum capacitor to destruction using a low voltage power supply, then used my 1kV variable supply to test some known good, and Chinesium capacitors.
A known good 0.1uF 1200V cap showed a leakage near zero at 1000V. My meter in the uA range shows 0.1uA or 0.2uA when the leads are open, and maybe it spent more time on 0.2uA at 1kV.
The Chinese caps measured pretty much the same up to about 700V -- I didn't go higher.
A 10uF 450V cap showed about 50uA leakage at 450V. I've yet to test the Chinese electrolytics.
This turns out to be a very simple and useful device.