> If your design constrains allow for a relatively high PS output
> internal resistance (5 - 100OHM) it will be wise to insert in SERIES
> with every capacitor a limiting resistor/fuse. A thyristor crossbar
In a series circuit, one fuse or one resistor is enough. One at the
each component would do absolutely no good at all.
The real problem with this suggestion, however, is that peak current
in a capacitor input supply is very very high during ever cycle while
the supply is under load. Any fuse that would be low enough current
rating to protect the capacitors would quickly undergo metallurgical
changes (often called fatigue) and fail. Any series resistance would
increase ESR and have a deleterious effect on supply regulation and
performance. Of course any fuse would also have to be a high-voltage
type fuse, and cost several dollars!
> can then be triggered by the failure of the cap, thus the other caps
> will be protected.
Average MTBF of a typical electrolytic operated at full voltage is 7-
10 years.
It makes little sense to use a cost>$500 circuit with considerably
less MTBF to protect $5 components.
That's why no one bothers with such overly complex and unreliable
schemes.
The real problems are easy to correct, and inexpensive to correct.
The primary reason most series electrolytic strings fail after
abnormally short time periods is use of carbon-based equalizing
resistors that are not low enough in resistance value. For some odd
reason, many people just blindly use 100k equalizers, and worse yet
some use carbon resistors. Carbon should NEVER be used in a critical
equalizing system.
You would have to take the worse-case leakage current differential
and calculate the safe equalizing resistor value based on that worse-
case condition and allowable voltage tolerances. Many times required
resistance is much lower than 100k ohms.
Compounding the problem, "technicians" often never bother checking or
replacing the resistors that often cause the initial failures. They
change the capacitors, while ignoring the root-cause of the
failure!73, Tom W8JI
W8JI@contesting.com
|