TESTING ESR OF ELECTROLYTIC
CAPACITORS
I recently found an easy and cheap way to test ESR (Equivalent Series Resistance) of
electrolytic capacitors, in circuit, that might save some people a lot of time. It requires
only an oscillosope and a simple signal generator.
I had an oscilloscope that I was trying to repair (Intensity control had little effect. Horiz
sweep was only halfway across screen at higher freqs. One power supply rail was too
low and others were too high.) and I had already checked every electrolytic capacitor in
several/many different ways (all in-circuit), and even compared each of the readings to
those from an identical unit: Powered off: Looked at signature from component tester
(single-curve tracer) across each cap, and from each end of cap to ground, did resistance
check with DMM, did capacitance check with DMM, checked resistance from each end
of cap to ground. Powered on: Put scope across each cap, and scope from each end of
cap to ground, used DMM and measured DC and AC voltages across each cap and from
each end of cap to ground. I did find some bad caps (and some other bad components)
and replaced them. But the problems were still there!
I had been wanting to order an ESR meter, but hadn't done it yet, and needed to get this
scope repaired immediately. I went to Sam Goldwasser's excellent repair-FAQ site, at
http://www.repairfaq.org/ and found a GREAT method for testing ESR of capacitors in-
circuit that requires only a signal generator and an oscilloscope (and some cables), and
had found and fixed the problem within about ten minutes!! Here's what I did (This
technique is basically directly from Sam's repair faq site):
I used a signal generator and an oscilloscope to set up what I now call an "ESR Scope":
At the output of the generator, I connected a BNC "tee" adapter. I ran one 50 Ohm BNC
cable from the tee to a good (Tek 2465A) scope (with a 50-ohm BNC terminator on the
scope input). On the other side of the tee, I connected another BNC cable that had
alligator clips on its other end (It might have been a 75 ohm cable; shouldn't matter too
much?), which I clipped onto the banana plugs of a set of cheap DMM-type probes.
(Terminator note: I used a Tektronix 50 Ohm "pass-through" terminator, on the scope
end of the BNC cable. But, you should also be able to use, instead, another BNC "tee"
on the scope input, with an "endcap" terminator on one side and the cable coming in on
the other side of the tee. A standard 10BaseT Ethernet 50 Ohm coax terminator (and 50
Ohm Ethernet BNC coax cables) should work fine. And they're available at Radio
Shack, and probably Staples, et al.)
I set up the signal generator to produce square waves at about 100 kHz, with about 100
mv peak-to-peak amplitude as seen on the attached scope, and no DC offset (A simple
555 timer circuit would do the job, too!). Then, I turned the scope's v/div to 5 mv/div,
with time/div at 1 microsec, with AC coupling of the input.
Shorting the probes together gave me a display on the scope that was about one division
high. It was basically a square wave, with large narrow peaks at each leading edge. But
I only looked at the horizontal part's p-p amplitude.
That's the whole setup! No resistors. No nothing. Just cables (and a terminator). I did
also try it with a decade resistor box in series with the probes, just to see what it would
look like. I could clearly see each one-ohm increase, on the scope display, with the
probes shorted together as well as with the probes across a good electrolytic capacitor.
When I applied the probes across a GOOD capacitor in-circuit, there was little, if any,
change in the scope display, compared to when the probes were shorted (since,
depending on the frequency, a capacitor should look more-or-less like a short circuit, to
AC). But, when I tried it across a BAD capacitor, usually the display would be almost-
totally off the screen. And, there were some caps that looked marginal, making the
display go from about one div p-p up to about 3 to 5 divs (which probably corresponded
with somewhere between 5 ohms and 20 ohms of ESR, if I recall correctly.)
Anyway, within just a few minutes I had found one more bad electrolytic filter cap in
the power supply, two smaller bad electrolytics in the P.S., a bad one on the horizontal
sweep switch's board, four bad ones near the middle of the main board, and a couple
more that I can't remember right now.
I made a note of each one. When I was all done checking, the first thing I did was
replace the filter cap in the power supply, and then power it on and check the power
supply rails' voltages. BINGO!!!!! YESSS!!! They were all normal again! Not only that,
but the horizontal sweep problem and the Intensity control problem were both GONE!!
Yippee!
That filter cap had checked out as perfectly OK, using every one of the other methods
that I described above (all were "in-circuit", though), and compared OK to the other
identical scope's same cap, in all of those cases. But with this "ESR Scope" method, it
was totally obvious, immediately. And the same cap on the other scope tested good,
with this method (So, the earlier comparisons WERE bad cap vs. good cap, but showed
nothing!). [I also noted that after the bad cap was removed, it tested bad in the same
way that it had while it was in-circuit, with a basically identical scope display. And all
of the other ones that I replaced also tested bad, when OUT of the circuit, even with the
other methods.]
This " ESR Scope " method isn't a perfect panacea, of course: There were some cases
where, without an identical unit to compare to, the displays would have been difficult
for me to interpet, and possibly misleading. (However, it *always* worked with every
*electrolytic* that I tried it on, IIRC, from 10 uF 10v to at least 1000 uF 100v, with no
need for an identical unit to compare to.) But, then again, I haven't played around with it
enough, yet, either. I assume that adjusting the frequency for different capacitances
might be helpful, especially if non-electrolytics were to be tested. I also seem to
remember that a DC offset in the signal is usually used, when testing ESR. I'll try that,
later. And maybe increasing the amplitude of the square wave would be useful,
sometimes, too. But, usually, I think I'll want it to be low-amplitude, probably less than
+/- 0.4v, so the signal doesn't turn on any semiconductor junctions.
Well, that's it! I hope that this can save some of you some time, sometime...
Tom Gootee
[email protected]
http://www.fullnet.com/u/tomg (Good used Electronic Test Equipment)
--------------------------------------------------------------------------------------------
ESR testing without an ESR meter
While, the techniques described below can in principle be applied to any capacitor, they
will be most useful for electrolytic types. Of course, make sure to observe the polarity
and voltage rating of the capacitor during testing! In addition, take care with the
maximum voltage applied to other components if you attempt to test caps in-circuit. It
should be small enough that semiconductor junctions do not get forward biased (a few
tenths of a volt max) and the impedance should be such that low value resistors don't
smoke!
The ultimate in cheap if you have an oscilloscope would be: 99 Cent ESR Test Adapter.
----------------------------------------------------------------------------------------------------------
99 Cent ESR Test Adapter
[ Main Page | ESR Testing ]
As a hobbyist, I rarely have any need to check a capacitor's ESR, certainly not often
enough to shell out hard earned money for a fancy "store bought" meter. After a little
research and thinking, it became obvious that by using a suitable sine wave generator
and an oscilloscope, it was possible to test for ESR with a minimal outlay of cash.
Before finishing the sine wave version, the thought crossed my mind that using a crude
square wave generator would be simpler and might just be good enough. Much to my
surprise, the square wave design worked just fine. In fact, under some conditions it
would work better than one designed around sine waves.
ESR meters tend to use test signals of around 100 kHz and 100 mVAC. The low voltage
prevents problems with applying reverse voltage to electrolytic capacitors and is low
enough to keep semiconductors in associated circuits from effecting the results. At 100
kHz, ideal capacitors of 10uF or greater are essentially short circuits to the AC portion
of the test signal. For such capacitors, the ESR can be found by either comparing the
test results with those from resistors of similar value, or by calculating using the
formula for parallel resistors.
The simple test adapter shown in the first figure does the job. For capacitors above
10uF, the signal should remain square wave, just check the peak to peak voltages under
three conditions:
With the test leads shorted, so you can compensate for any resistance in the
leads!
The open circuit voltage: Vo
The voltage across the capacitor under test: V
A chart is provided at the bottom of this page to convert the voltage ratio to ESR.
Capacitors down to about 1 uF will give meaningful results, but you will need to
compare the signals to known good parts to interpret the signals
There are a couple of limitations to the simple version. If you were to accidentally test a
capacitor with a few hundred volts of charge on it, the IC and some of the other parts
would go up in smoke. The oscilloscope could also be damaged. The second limitation
also applies to some commercial testers. Basically, a shorted capacitor and a low ESR
capacitor test the same! Both of these limitations can be gotten around with a few
modifications to the design.
More information than just ESR
If you test some capacitors in the .1 uF to 10 uF range at 100 kHz, you will note that as
the value of the capacitor drops, the test signal distorts from a square wave. At 3.3 uF,
the top and bottom of the wave form will have a distinct tilt, with a 1 uF capacitor, this
effect is even more pronounced. The tilting effect is just capacitor charging up as it
integrates the test signal. If you just want to check ESR, this is a nuisance, on the other
hand if you want to know
if the capacitor or
associated circuitry is
shorted..... a short does
not integrate charge. By
making the frequency
variable, this effect can be
used over a wide range of
capacitors.
A second source of information is that with a DC coupled probe, the display will show a
shift between a short and a low ESR capacitor. A good capacitor under test will develop
a DC+ bias at about 1/2 Vo which will shift the waveform. A plain resistor or shorted
component will not store charge, and so when the test signal drops to zero, the signal on
the oscilloscope will also show zero volts.
Other Additions
It's also possible to add protection to the tester to prevent inadvertent exposure to
charged HV caps from damaging the tester and oscilloscope. Since the test voltage is so
low, a couple pairs of back to back Si power diodes (D), combined with using power
resistors for R5, R6, R7, and R8 should do the trick. It's probably a good idea to socket
the IC anyway.
The third change to the second circuit is that the comparison resistor (R6) has a
switchable parallel resistor (R7). The 2 ohm setting {switch closed) is better for
checking low ESR caps, while the 10 ohm setting (switch open) is better for caps with
higher ESR.
Using an AC millivoltmeter instead of an Oscilloscope
I tried the using a couple of DMM's to see if they could be used instead of an
oscilloscope. In both cases, the results were quite usable with the following limitations.
Both meters worked fine at 1 kHZ and 10 kHz, but at 100 kHz the results were dismal.
Typical results (10 kHz, 100 mV P-P, with R6 = 5 ohms): Open circuit V= 38.4 mVAC,
large cap with 2 ohm ESR V=10.8 mVAC, large cap with 5 ohm ESR V=18.8 mVAC.
These results are pretty much what one would expect from theory. The main limitation
was operating frequency, at 10 kHz, the minimum capacitor that could be reasonably
tested was 100 uF. If you have an old meter with a 1.5 volt range, it should be possible
to redesign the circuit to make use of it. It would be a matter of using a 200ma min.
power supply in the 12 - 15 volt range with R4 near the minimum value (75 ohm min@
15 Volts to keep within the 555's current limit).
Additional comments:
The square wave generators in the two schematics (everything to the left of R4)
are interchangeable. Since a 555 is smaller, cheaper and more robust than a
4049, it's probably the best way to go.
In the circuit using a 555 timer, R1 adjusts the duty cycle. It is optional, but the
waveform will be pretty unsymmetrical without it.
The 4049 version can be changed to frequency variable placing a 1 Meg pot in
series with a timing resistor (R2) (Schematic). It is possible to adjust the
frequency down to about 500 Hz. At 500Hz, a 2200 uF capacitor will show the
tilting effect on the square wave. . By adding a .1 uF capacitor and switch in
parallel to the C2 timing capacitor, the frequency range drops down to 30 Hz to
3200 Hz, in case you have bigger caps to look at. The frequency of the 555
version can be lowered to 20-3500 Hz by adding a 47nF capacitor in parallel
with the C2 timing capacitor.
Construction Ideas: If you have an extra scope probe laying around, build the
adapter into the cable. Salvage connectors from old coax Ethernet equipment
The supply voltage of +9 VDC is just a matter of batteries being handy.
Anything from 5 to 15 volts should work. Changing the supply voltage will shift
the frequency and my require changing R4 to keep the current draw & test
voltage reasonable.
R7 (2.5 ohm) can be four 10 ohm resistors in parallel.
See my earlier page "ESR Testing" for additional information and references.
Info on Hex Inverter Oscillator
[ Main Page | ESR Testing ]
The following table shows the ratio of peak to peak voltages for various values of ESR
and R6-R7. Since R4 provides a current signal and V=IR, V is proportional to R. Using
the ordinary formula for parallel resistors it is possible to construct a table. Note that R6
refers to the actual resistance of R6 and R7 in parallel.
ESR V/Vo V/Vo V/Vo
(Ohm) (R6=2 (R6=5 Ohm) (R6=10
Ohm) Ohm)
0.1 5% 2% 1%
0.2 9% 4% 2%
0.3 13% 6% 3%
0.4 17% 7% 4%
0.5 20% 9% 5%
0.6 23% 11% 6%
0.7 26% 12% 7%
0.8 29% 14% 7%
0.9 31% 15% 8%
1 33% 17% 9%
2 50% 29% 17%
3 60% 38% 23%
4 67% 44% 29%
5 71% 50% 33%
6 75% 55% 38%
7 78% 58% 41%
8 80% 62% 44%
9 82% 64% 47%
10 83% 67% 50%
20 91% 80% 67%
30 94% 86% 75%
40 95% 89% 80%
50 96% 91% 83%
60 97% 92% 86%
70 97% 93% 88%
80 98% 94% 89%
90 98% 95% 90%
100 98% 95% 91%
--------------------------------------------------------------------------------------------------------
An inexpensive way (for the cost of a resistor) to measure the ESR of a capacitor is to
apply a squarewave signal through a resistor in series with the capacitor under test.
Monitor the waveform on the capacitor using an oscilloscope. When using a sensible
squarewave frequency (a few kHz - not one where the inductance of the circuit becomes
an issue) there will be a triangle waveform with a step at the squarewave transition
times. The amplitude of the step will proportional to the ESR of the capacitor. Calibrate
things by adding a known small value ESR simulating resistor in series with the
capacitor. This doesn't have to cost anything if you have a squarewave generator, or can
build one cheaply.
Motivated by the discussions on the virtues of ESR testing, I ordered a genuine ESR
meter. While waiting for it's arrival, a large pile of dogs were accumulating in my shop.
To crank out these repairs quickly in the meantime, I constructed an 'ESR meter' by
cabling a (50 ohm) function generator output to the scope input and, via a T-connector,
on to a set of test leads.
With the test leads shorted, mere millivolts displayed on the scope. Across a good
capacitor, mere millivolts. Across a sick capacitor, mucho volts. The defective caps
stuck out like a sore thumb.
Wow, this is too easy. Instant in-circuit (power off) fool-proof testing of electrolytics. I
wish I had thought of this 50 years ago.
I used 100 kHz and 5 V p-p. With scope set at 0.2 V/div you can also check diodes
surrounded by low ohm transformer or inductor windings.
(Editor's note: to avoid the possibility of damage to semiconductors due to excessive
voltage, use a lower amplitude signal - say 0.5 V p-p - for in-circuit testing. This will
also prevent the most semiconductor junctions from conducting and confusing your
readings.
I have been reading the various messages about ESR checkers and while I don't doubt
their value in electronic servicing, I think that the use of these devices adds an extra and
IMHO unneeded step. My method of diagnosing possible electrolytic fault is to use just
a scope. Remembering that electrolytics pass AC or signals through them, a scope
should show *the same* waveshape on both sides of the cap. If the cap is a bypass cap
to ground, then the waveshape should just be a flat line on both sides; if it is a coupling
cap, the waveshape should be the same on both sides.
There are some exceptions, one being a cap that is used for waveshaping in a vertical
circuit but such applications are few. Most electrolytics are either coupling or bypass.
Using 'my' scope method has several advantages. The main one is that it tests caps
dynamically in the circuit they are used in and using the actual signals applied to them
in real life. The method is fast because you just have to go from one to another (if you
are using the scatter-gun approach) using just the scope prod. But, best of all, it
seamlessly integrates a total dynamic approach to servicing using the set's own signals
or lack thereof. If you are tracing a video circuit, you can find an open cap, an open
transistor, or a defective IC using the same piece of equipment.
I have been running a service business for over 40 years. Most of my business today is
doing tough-dog service for other service companies.
But, I must admit that sometimes I fix sets just by changing the caps that are
swollen. ;-}
I still do just enough work that I'll one day break down and buy an ESR meter (I always
give in and indulge myself with the toys of my "trade"). For now, though, the quickie
method I use is the oscilloscope. It goes something like this:
1. Scope positive lead. Any significant AC? If not, go to next cap.
2. Is the AC more than about 5% of the DC? If not, note this location and go to
next cap.
3. Scope negative lead. AC here roughly the same as on positive lead? If so, go to
next cap. (If this lead is *obviously* grounded, skip this step.)
4. Set off; note value; jumper in roughly same value at safe voltage rating. (Note:
make sure both caps are discharged! --- sam)
Set on; scope positive lead. Significant difference? If not, note this location and
go to next cap.
5. Replace cap. Test set. If not okay, go to next cap.
If that doesn't catch it, a quick review of the "noted locations" often does. This fixes
98% of cap problems. Not exhaustive or perfect, nor is it intended to be. Close cover
before striking. Probably causes cancer in laboratory rats. Your mileage may vary.
It is always best when measuring a component parameter to lean the measurement
method towards some sort of emulation of the application to which the parameter is
important. This is particularly true of power components, because the parameter-value
may vary with operating conditions. It is essential with magnetics, less so with
electrolytics, but a good habit anyway.
Hold the cap charged up and find some way of applying repetitive square *current*
pulses to it, an Amp or more each time, depending on the expected ESR.
If the cap has no ESR then a scope across its terminals will show that each current pulse
results in a nice smooth ramping triangle. If the cap has an ESR then each triangle will
be preceded by a small vertical step. If the current is known, measurement of that step
gives you the ESR-value. You can cross-check the accuracy of the method by seeing the
effect of increasing the "ESR" as low-value R's are put in series with the cap, 0.01 to 0.1
ohms.
Be careful about the placement of the scope leads, you don't want to measure the IR-
drop in the wiring.
If the size of each step+triangle is small compared to the steady voltage on the cap then
the known "constant-I" discharge pulse can be approximated with no more than a
resistor and switching Fet.
If you want to the decoupling capability, you maybe want to know the ESR only at the
series resonant frequency. It's quite simple:
Use a sine generator, connect some coax cable to its output, at the end of the cable put
47 ohms in series and connect the resistor to one end of the cap, connect a kind of
detector in the same way (47R - cable - detector) to the same lead. Other end of the cap
(and coax shields) to a small ground plane. Detector can be voltmeter, scope or
spectrum analyzer, depends on your equipment and resonant frequency. Spectrum
analyzer with tracking generator eliminates need for separate generator, makes
measurement easy and enables you to measure even very small capacitor values.
Tune to minimum signal at the detector. With a scope you can check also phase shift
(thanks for this hint, Winfield!), cap should be only resistive (no phase shift). ESR can
be easily calculated now.
After seeing all the glowing recommendations for ESR meters on the
sci.electronics.repair newsgroup, I decided to look into this. Being a cheap sort, I first
tried setting up my own ESR meter using stuff lying around the shop: Function
generator set to 2 volts p-p, 100 kHz sine wave, hooked up to a BNC tee, one side of the
tee goes to some clip leads, other side to the scope, set for 0.1 volts/cm, 10 us/cm
sweep.
With the clip leads dangling free, the scope trace is almost invisible, as it's zooming up
and down 20 cm 10 times across the screen. With the clip leads shorted, I get about 0.3
cm of a sine wave. With a 1 ohm resistor across the clip leads, I get about 1 cm of sine
wave.
I put a GOOD 2 uF capacitor across the clip leads, we see about 0.5 cm of sine wave.
Tests with various good electrolytics all give less than 1 cm of sine wave.
Now we can just hop along a circuit board bridging the electrolytics as we go along. A
good electrolytic is going to show 1 cm or less of deflection. Many old ones with 1970's
date codes will show 2 or 3 cm. Probing around a suspect old pc board revealed that
80% of the caps gave more than 2 cm of deflection!
Now this isn't always a bad thing. You have to use a little judgement. If the electrolytic
is in a high-impedance circuit, such as coupling two voltage amp stages, a few ohms
isn't going to hurt too much. But if it's a bypass capacitor on a Vcc line, it could be
significant. Just realize that a circuit may appear to work just fine even with caps with
many ohms of ESR. I usually replace these caps anyway, as they're only going to go
downhill from here.
I can't tell you how much time this little set-up has already saved me!. Before I had to
unsolder one lead of the capacitor, hook it up to the cap bridge, twiddle the dials until I
got a semblance of balance, or if it was a bad capacitor, I'd waste even more time trying
to find the missing null. Now I can just probe the caps in-circuit, and mark the bad ones
with a big red magic-marker for later replacement. It's quick and great for morale.
This method works well with caps in the range of 1 to 500uf, with medium or high ESR.
But it doesn't have enough oomph to drive BIG caps. For this you'd need a generator
with a lower output impedance.
Next experiment-- we'll hook up a tranny from an old dead SMPS to lower the
generator's output impedance so we can test those big PS capacitors. Stay tuned....
BTW, this isn't meant to take away from built ESR meter sales! It may even increase
them as once you see how wonderful this technique is, you may want to buy a dedicated
ESR meter
--------------------------------------------------------------------------------------------
