A POOR MAN'S ESR METER      01-20-2K

     One type of failure in an electrolytic capacitor is increased
Effective Series Resistance or ESR. It happens as a result of the
capacitor's electrolyte drying out with heat (both internal and external)
over time. High ESR makes the capacitor less effective, especially in
critical circuits such as switching power supplies. Standard capacitance
checkers may fail to show a problem. They may show normal readings for
capacitance values and leakages. True ESR testers are usually very
expensive.
     A simple method of checking an electrolytic cap for higher than normal
ESR involves the use of a square wave generator and an oscilloscope. Most
scopes have a 1KHz square wave calibrator built into them for setting up
the scope probe, and that works as a signal source if a generator is not
available. The idea is to run a square wave through a series limiting
resistor to the capacitor under test, then monitor the waveform across the
cap with the scope. This tester can be used in-circuit as long as all power
and grounds are disconnected from the device under test (or caps may
appear "shorted") and all caps are discharged.

Test setup (example): a 1KHz 0.5VPP square wave generator (or the 1KHz
calibrator output of a scope) is fed through a 1K resistor to a good 1uF
capacitor. The scope shows the expected triangle wave across the capacitor,
indicating it is charging and discharging linearly. If large value caps are
tested, the waveform is reduced to near zero as expected, requiring either
the signal source or scope gain to be increased so a waveform is visible. I
use a 100 ohm limiting resistor to check caps of 100uF or higher.

Simulated capacitor failures: If one leg of the cap is opened and a bit of
series resistance (10 ohms) is introduced (simulating higher than normal
ESR), the peak positive and negative excursions of the triangle wave have a
"gap" or vertical component to them. This indicates that as the cap starts
charging/discharging, some resistance is "seen" by the voltage source. As
the added series resistance is increased (100 ohms), the waveform changes
to a modified triangle with a greatly increased amount of vertical
displacement. If the capacitor series resistance is very high (1K ohm),
the waveform remains a square wave but is just reduced in amplitude.

     In actual practice, the waveforms are somewhat different than the
above simulated ones. Bad caps will show varying amounts of slope and/or
rolloff in the top and bottom excursions and many will only reduce the
amplitude of the square wave without altering it's waveshape. Caps that bad
will usually show reduced capacitance on a standard cap checker. If a
capacitor is leaky (actually quite rare), the effect on the waveform is to
cause the triangle waveform to "bend" into a sawtooth because the ramp up
and down is no longer linear. Combinations of high ESR and leakage will
further distort the waveform. Anything that deviates from a straight-line
triangle wave should be investigated further. 
     If this method is used to check capacitors in-circuit, keep in mind
that other circuit components can affect the readings. Therefore, it's best
to keep the source AC square wave signal under 0.5VPP. This will prevent
standard diode conduction from distorting the readings. Note that some
diodes in switchers can conduct at only 0.2 volts. If the waveform is
"clipped" at the top or bottom, suspect diode conduction, reverse the leads
and check the cap again. The clipped signal should be on the opposite half
cycle if that is what is happening. Inductors can introduce ripples,
ringing, or spikes to the waveform. If in doubt, disconnect one end of the
cap from the circuit and check it again. Note that in switching supplies,
often two caps will be paralleled with an inductor in between them, such as
in the 5 volt line in a typical VCR supply. You may get a false good ESR
reading unless one cap is disconnected. That kind of "error" will fool any
ESR meter.
     A few problems arise when checking caps this way. The scope calibrator
output impedence is a factor as it was not intended to drive such a load.
If it is excessively loaded, it's output will distort unpredictably and
change the resulting test waveform as well. That's why an external
generator with a low output impedance is preferable. As the value of the
cap under test increases, the energy required to charge and discharge it
increases also. For caps over 100uF, a lower value of series limiter
resistance to the calibrator will be necessary, but again, the calibrator
output may be overloaded. As long as you know what you're looking at (by
comparing to known good caps), you can use the simple setup described.

Ray Carlsen
CARLSEN ELECTRONICS... a leader in trailing-edge technology.