Leakage is predictable and can easily be distinguished from DA.
For example, after 100 time constants, the charging current should be 20µA/e^100=1.49e-48A.
That's 100s for the 1µ, and 330s for the 3.3µ. Anything above that is caused by DA
For example, after 100 time constants, the charging current should be 20µA/e^100=1.49e-48A.
That's 100s for the 1µ, and 330s for the 3.3µ. Anything above that is caused by DA
You are right of course, I messed it up completely.
To clarify:
If after 30 or 50 time constants you are still able to measure a current, wait another 30 tau.
If the current remains ~the same, it is the actual leakage current. If it is reduced, it is caused by DA, which does not follow the simple exponential law and you need to wait until it doesn't change significantly any more.
The final, stable value is the leakage, but even with moderately good dielectrics like PET, it can take a very long time, because the leakage current is extremely small and DA dominates for a very long time.
PP, PS, PE, PTFE caps should settle more quickly, but that is not always the case: a poor construction or contamination can cause an abnormally high DA
To clarify:
If after 30 or 50 time constants you are still able to measure a current, wait another 30 tau.
If the current remains ~the same, it is the actual leakage current. If it is reduced, it is caused by DA, which does not follow the simple exponential law and you need to wait until it doesn't change significantly any more.
The final, stable value is the leakage, but even with moderately good dielectrics like PET, it can take a very long time, because the leakage current is extremely small and DA dominates for a very long time.
PP, PS, PE, PTFE caps should settle more quickly, but that is not always the case: a poor construction or contamination can cause an abnormally high DA
Thanks for the confidence. 🙂
That said,Keithley 3401A is TEN Megohm input impedance so results are TEN TIMES BETTER than you think 😀
About what I expected, that´s why I suggested the test 🙂
Glad you found perfectly usable caps at home, no need to pay gold dust for them.
When I was looking for a timing cap for the compressor I was building ( I wanted at least 3 seconds of decay from 4.6v to 4.48v, 120mv (20db of gain reduction)) I tested them by charging them to 5v and letting them sit for a minute then measured there voltage again. I was surprised that the 5u film caps would drop very little voltage, the tau was far greater than I needed.
@JMFahey: you are absolutely right, I was looking at the AC spec, that's 1M.
10M it is at DC.
Jan
10M it is at DC.
Jan
You can calculate the RC time-constant of an insulator direct from its intrinsic electrical properties, resistivity (rho) and permittivity (eta)
R = d/A * rho
C = A/d * eta0 * eta
Thus RC = rho * eta0 * eta, where eta0 = 8.85e-12 (permittivity of free space).
For instance polypropylene (PP) has rho listed in the range 1e14--1e16 ohm-m, dielectric constant as 2.1, so worst case RC = 8.85e-12 * 2.1 * 1e14 ~= 2000s
The resistivity values for insulators are a bit variable, as it depends on humidity, temperature and electric field strength and manufacturing process, but there is a clear difference between mylar and PP/PTFE/PS - all of the latter are significantly lower leakage than mylar, and would be chosen for very low-leakage applications (time constants in hours rather than minutes).