Frequency of ESR measurements

[[email protected]]

I have a question about ESR measurements.

I have been doing some reading about ESR and the literature indicates
it can be a strong function of frequency. In the design a SMPS I
assume the ESR of importance is at switching frequencies (at least
for the line side capacitor).

I have also noticed that often the measurement made in trouble
shooting are low frequency measurements (step function - looking at
the abrupt change in voltage).

Is the above accurate and if so when a cap goes bad do the low
frequency measurements catch the majority of the high frequency ESR
failures. Are there significant failure modes where a low frequency
ESR measurement would miss the higher frequency ESR failure?


Thanks Much,

 

[Leonard Caillouet]

I have a question about ESR measurements.

I have been doing some reading about ESR and the literature indicates
it can be a strong function of frequency. In the design a SMPS I
assume the ESR of importance is at switching frequencies (at least
for the line side capacitor).

I have also noticed that often the measurement made in trouble
shooting are low frequency measurements (step function - looking at
the abrupt change in voltage).

Is the above accurate and if so when a cap goes bad do the low
frequency measurements catch the majority of the high frequency ESR
failures. Are there significant failure modes where a low frequency
ESR measurement would miss the higher frequency ESR failure?


Thanks Much,

Most of the ESR meters that I have seen test in the 50-200 kHz range. This
pretty much covers the switching frequencies of most power supplies, at
least in consumer equipment that I am familiar with.

Leonard

 

[Arfa Daily]

Most of the ESR meters that I have seen test in the 50-200 kHz range.
This pretty much covers the switching frequencies of most power supplies,
at least in consumer equipment that I am familiar with.

Leonard

Exactly ditto.My Bob Parker meter works at around 100kHz, and I have never
had a problem with it picking up any (several per week) faulty caps.

Arfa

 

[Jim Yanik]

Exactly ditto.My Bob Parker meter works at around 100kHz, and I have
never had a problem with it picking up any (several per week) faulty
caps.

Arfa

one thing to consider is the waveform of the usual switcher;the cap has to
deal with a fast rise waveform rich in harmonics.2nd and 3rd harmonics will
still have a lot of energy.

your usual 60hz XFMR supplies dealt with a sinewave.

 

[[email protected]]

Most of the ESR meters that I have seen test in the 50-200 kHz range.  This
pretty much covers the switching frequencies of most power supplies, at
least in consumer equipment that I am familiar with.

Leonard

Thanks everyone for your replies, you have answered my questions.

 

[Franc Zabkar]

My Bob Parker meter works at around 100kHz ...

Arfa

The following diagram and flowchart indicate that, at least for the
Mark 2 design, the capacitor under test is pulsed with 8us constant
current pulses repeated every 500us:

http://us1.webpublications.com.au/static/images/articles/i1008/100895_3mg.jpg
http://us1.webpublications.com.au/static/images/articles/i1008/100895_2mg.jpg

- Franc Zabkar

 

[Arfa Daily]

The following diagram and flowchart indicate that, at least for the
Mark 2 design, the capacitor under test is pulsed with 8us constant
current pulses repeated every 500us:

http://us1.webpublications.com.au/static/images/articles/i1008/100895_3mg.jpg
http://us1.webpublications.com.au/static/images/articles/i1008/100895_2mg.jpg

- Franc Zabkar

That does appear to be so. Without looking it up, I just had it fixed in my
head that it was around 100kHz. With the fact that it is a highly asymmetric
waveform, duty factor-wise, I guess that it's hard to quote it as an actual
frequency. It's sort of 100+ kHz 'width' pulses, repeated at a frequency of
2kHz ...

Arfa

 

[Jamie]

That does appear to be so. Without looking it up, I just had it fixed in my
head that it was around 100kHz. With the fact that it is a highly asymmetric
waveform, duty factor-wise, I guess that it's hard to quote it as an actual
frequency. It's sort of 100+ kHz 'width' pulses, repeated at a frequency of
2kHz ...

Arfa

So, its a 2 Khz signal with 2000/100000 = 2 % duty cycle =

1/2000*0.02*0.5 = 5 uS(+) pulse


http://webpages.charter.net/jamie_5"

 

[Franc Zabkar]

That does appear to be so. Without looking it up, I just had it fixed in my
head that it was around 100kHz. With the fact that it is a highly asymmetric
waveform, duty factor-wise, I guess that it's hard to quote it as an actual
frequency. It's sort of 100+ kHz 'width' pulses, repeated at a frequency of
2kHz ...

Arfa

Remember that they're also current pulses, not voltage pulses, so the
voltage appearing across the cap is a trapezoid. I don't know how this
compares with a meter that generates a sinusoidal test voltage,
assuming that's how others do it.

- Franc Zabkar

 

[John Bachman]

Remember that they're also current pulses, not voltage pulses, so the
voltage appearing across the cap is a trapezoid. I don't know how this
compares with a meter that generates a sinusoidal test voltage,
assuming that's how others do it.

The Capacitor Wizard uses a 100Khz sinusoidal signal. The Bob Parker
design is as described - that is what is used in our Blue ESR meter.
Bob always represented that his method is the equivalent of a 100Khz
signal.

Other designs are described in our comparison at
www.anatekcorp.com/esr_compare.htm

John
AnaTek Corporation

 

[Arfa Daily]

The Capacitor Wizard uses a 100Khz sinusoidal signal. The Bob Parker
design is as described - that is what is used in our Blue ESR meter.
Bob always represented that his method is the equivalent of a 100Khz
signal.

Ah. So that's where I got it from. Direct from Bob .... Arfa

 

[Jeroni Paul]

That does appear to be so. Without looking it up, I just had it fixed in my

head that it was around 100kHz. With the fact that it is a highly asymmetric
waveform, duty factor-wise, I guess that it's hard to quote it as an actual
frequency. It's sort of 100+ kHz 'width' pulses, repeated at a frequency of
2kHz ...

I have this meter:
http://es.geocities.com/podernixie/ESR/index-en.htm
and it outputs a 60 mVpp sine wave at 80 kHz. It works great.

 

[Ross Herbert]

On Wed, 31 Dec 2008 09:30:11 -0800 (PST), [email protected] wrote:

:>
:> :>
:>
:>
:> >I have a question about ESR measurements.
:>
:> > I have been doing some reading about ESR and the literature indicates
:> > it can be a strong function of frequency.  In the design a SMPS  I
:> > assume the ESR of importance is at  switching frequencies (at least
:> > for the line side capacitor).
:>
:> > I have also noticed that often the measurement made in trouble
:> > shooting are low frequency measurements (step function - looking at
:> > the abrupt change in voltage).
:>
:> > Is the above accurate and if so when a cap goes bad do the low
:> > frequency measurements catch the majority of the high frequency ESR
:> > failures.  Are there significant failure modes where a low frequency
:> > ESR measurement would miss the higher frequency ESR failure?
:>
:> > Thanks Much,
:>
:> Most of the ESR meters that I have seen test in the 50-200 kHz range.  This
:> pretty much covers the switching frequencies of most power supplies, at
:> least in consumer equipment that I am familiar with.
:>
:> Leonard
:
:Thanks everyone for your replies, you have answered my questions.


Now to throw in a curve ball...

In designing his Tan-Delta meter, Cyril Bateman used 100Hz sine. His theory was
that using 100KHz could cause the measuring lead inductance to exceed the self
inductance of the capacitor under test and thus complicate the accuracy of the
result.

 

[Arfa Daily]

On Wed, 31 Dec 2008 09:30:11 -0800 (PST), [email protected] wrote:

:>
:>
:>
:>
:>
:> >I have a question about ESR measurements.
:>
:> > I have been doing some reading about ESR and the literature indicates
:> > it can be a strong function of frequency. In the design a SMPS I
:> > assume the ESR of importance is at switching frequencies (at least
:> > for the line side capacitor).
:>
:> > I have also noticed that often the measurement made in trouble
:> > shooting are low frequency measurements (step function - looking at
:> > the abrupt change in voltage).
:>
:> > Is the above accurate and if so when a cap goes bad do the low
:> > frequency measurements catch the majority of the high frequency ESR
:> > failures. Are there significant failure modes where a low frequency
:> > ESR measurement would miss the higher frequency ESR failure?
:>
:> > Thanks Much,
:>
:> Most of the ESR meters that I have seen test in the 50-200 kHz range.
This
:> pretty much covers the switching frequencies of most power supplies, at
:> least in consumer equipment that I am familiar with.
:>
:> Leonard
:
:Thanks everyone for your replies, you have answered my questions.


Now to throw in a curve ball...

In designing his Tan-Delta meter, Cyril Bateman used 100Hz sine. His
theory was
that using 100KHz could cause the measuring lead inductance to exceed the
self
inductance of the capacitor under test and thus complicate the accuracy of
the
result.

The thing is with a bog-standard ESR meter, it's all about relative
measurements, 'feel', experience, and intuition rather than accuracy.
Determining whether or not an electrolytic is faulty by way of its ESR, is a
bit of a black art, and is with any ordinary 'in-circuit' ESR meter. The
reading just gives you some help and 'feel good' backup. This is why I have
contended on many occasions that such an instrument is not one of absolutes,
and cannot be just picked up and used to give 'go / no go' definitive
answers about the state of a cap, by anyone who chooses to buy one. It's
just a helper that gives you one more pointer, allbeit a mostly pretty good
one, without having to remove the cap (generally) from the circuitry that
it's in.

The thing is with the inductance argument, I would have thought that given
that the self inductance of a 'standard' electrolytic is quite high due to
the way they are constructed, the inductance of the test leads would have
been pretty insignificant in comparison ??

Certainly, in practice, if it is a 'problem', it has never caused me any
trouble with using my Bob Parker to help identify bad 'uns ...

Arfa

 

[Leonard Caillouet]

The thing is with a bog-standard ESR meter, it's all about relative
measurements, 'feel', experience, and intuition rather than accuracy.
Determining whether or not an electrolytic is faulty by way of its ESR, is
a bit of a black art, and is with any ordinary 'in-circuit' ESR meter. The
reading just gives you some help and 'feel good' backup. This is why I
have contended on many occasions that such an instrument is not one of
absolutes, and cannot be just picked up and used to give 'go / no go'
definitive answers about the state of a cap, by anyone who chooses to buy
one. It's just a helper that gives you one more pointer, allbeit a mostly
pretty good one, without having to remove the cap (generally) from the
circuitry that it's in.

The thing is with the inductance argument, I would have thought that given
that the self inductance of a 'standard' electrolytic is quite high due to
the way they are constructed, the inductance of the test leads would have
been pretty insignificant in comparison ??

Certainly, in practice, if it is a 'problem', it has never caused me any
trouble with using my Bob Parker to help identify bad 'uns ...

Arfa

Ditto Arfa's comments. I have three ESR meters and they all read
differently, but they giv me enough info to sort out which caps are bad and
which not. And when there is doubt, change them. Caps are cheaper than the
time to worrry about sorting out the accurate measure, which is somewhat
meaningless anyway, as circuits vary so much. What might be bad for one
application may be ok for another. Regardless, if the ESR seems high I just
change them.

Leonard

 

[[email protected]]

Ditto Arfa's comments. I have three ESR meters and they all read
differently, but they giv me enough info to sort out which caps are bad and
which not. And when there is doubt, change them. Caps are cheaper than the
time to worrry about sorting out the accurate measure, which is somewhat
meaningless anyway, as circuits vary so much. What might be bad for one
application may be ok for another. Regardless, if the ESR seems high I just
change them.

Leonard

Just out of curiosity, has anyone tried using an infrared
thermometer to try and find bad or failing caps. I did a quick search
on the on the internet and found units you can get for as little as
$20. The spot size is very small and I believe the top of the cap is
an ideal target. Could use a black sharpie to make a small black spot
for even better measurements. A SMPS cap which had a High ESR should
read high in temp, might be more sensitive than using your finger to
test for temperature. An open or failed cap should read low. Of
course the unit being tested couldn't be completely dead. If it
worked one nice feature would be being able to identify bad caps
before even removing the PC board.

The power dissipated in the cap should be proportional to the
increase in ESR ,the temperature increase should be on the fist order
equal to the power dissipated in the cap. So if the original
temperature rise due to the new caps ERS was 1 degree C and the ESR
increased by a factor of 10 the old caps temperature should be about
10 deg higher, easy to measure.

Might be able to find bad IC's as well. Perhaps the increase in
sensitivity using this device would help find other failing or failed
components as well. Might also be useful in identifing failing or
failed components in very high voltage circuits. Those components you
would not want to touch with power on.

Don't know if the idea has any merit but thought I would mention it.

Gordon ,

 

[[email protected]]

 Ditto Arfa's comments.  I have three ESR meters and they all read
differently, but they giv me enough info to sort out which caps are bad and
which not.  And when there is doubt, change them.  Caps are cheaper than the
time to worrry about sorting out the accurate measure, which is somewhat
meaningless anyway, as circuits vary so much.  What might be bad for one
application may be ok for another.  Regardless, if the ESR seems high Ijust
change them.

Leonard

Back to ESR measurements,

In the past I have had any test equipment available to me I
wanted. At this point I am trying to collect test equipment on a
very limited budget. So my interest in ESR testers.

Thanks again everyone for all the info. I have summarized
below what I think I am hearing here and other places. I would
appreciate any comments to the validity of the statements below.

Inexpensive “bog - standard”, In Circuit ESR measurements of
Caps in trouble shooting “consumer electronics” came about due to the
Common and Sever Failure Mode of Dried Out Electrolytic Capacitors.
The ESR changed by a factor of 10X to 30X so was easy to pick up with
simple inexpensive in circuit ESR testers. These meters are a great
tool to have in a persons toolbox to expedite greatly trouble shooting
of failed or troubled circuits due to this capacitor failure mode.

If used for this purpose, due of the severity of the failure mode
(10X to 30X increase in ESR), you don't necessarily need to be too
concerned with accuracy. Many different meters designs out there
which may give different readings between themselves, each do a great
job helping to finding a large majority of capacitors that are causing
circuit problems due to this failure mode.

In trouble shooting consumer electronics for this failure mode,
what is more important is that you get a feel for what your particular
meter readings mean for electrolytic capacitors with different
parameters such as Capacitance, Temperature Rating and the importance
of low ESR in the particular type of circuit you are trouble shooting
(for example some caps may measure higher ESR's because they were not
low ESR Caps to begin with).

I can imagine that good calibration between the SAME design of
meters might come into play if people want to compare results to
assess each others expertise in trouble shooting this failure mode for
different electrolytics and circuits . Even here the requirements
may be fairly low since the failures are so sever (10X to 30X)

This is an exaggeration but an INEXPENSIVE IN CIRCUIT ESR tester
could be compared to a 12 volt sensor in trouble shooting a cars
electrical system. It does not need to be a very accurate tool and
may not be the right tool to use in all situations but it is A VERY
USEFULL TOOL. Many different simple designs of the voltage sensor work
well.

For other lower failure rate capacitor problems other In or Out of
circuit testers such as capacitance meters, capacitance leakage
meters, testers that can test capacitor parameters nearer their
operating conditions such as voltage of operation, imposed voltage (or
current) waveform and magnitude and frequency, are useful tools.

For these lower failure rate capacitor failures, another low
overhead and perhaps time saving approach is to pin point the circuit
that is having problems and replace a suspected capacitor with a known
good capacitor to determine if the capacitor is part of the problem.

Thanks again for all of your responses, information and sharing your
experience,

Jeroni Paul, I really like the LM324 ESR meter

 

[Ross Herbert]

:
::> On Wed, 31 Dec 2008 09:30:11 -0800 (PST), [email protected] wrote:
:>
:> :>
:> :>
:> :> :>
:> :>
:> :>
:> :> >I have a question about ESR measurements.
:> :>
:> :> > I have been doing some reading about ESR and the literature indicates
:> :> > it can be a strong function of frequency. In the design a SMPS I
:> :> > assume the ESR of importance is at switching frequencies (at least
:> :> > for the line side capacitor).
:> :>
:> :> > I have also noticed that often the measurement made in trouble
:> :> > shooting are low frequency measurements (step function - looking at
:> :> > the abrupt change in voltage).
:> :>
:> :> > Is the above accurate and if so when a cap goes bad do the low
:> :> > frequency measurements catch the majority of the high frequency ESR
:> :> > failures. Are there significant failure modes where a low frequency
:> :> > ESR measurement would miss the higher frequency ESR failure?
:> :>
:> :> > Thanks Much,
:> :>
:> :> Most of the ESR meters that I have seen test in the 50-200 kHz range.
:> This
:> :> pretty much covers the switching frequencies of most power supplies, at
:> :> least in consumer equipment that I am familiar with.
:> :>
:> :> Leonard
:> :
:> :Thanks everyone for your replies, you have answered my questions.
:>
:>
:> Now to throw in a curve ball...
:>
:> In designing his Tan-Delta meter, Cyril Bateman used 100Hz sine. His
:> theory was
:> that using 100KHz could cause the measuring lead inductance to exceed the
:> self
:> inductance of the capacitor under test and thus complicate the accuracy of
:> the
:> result.
:
:The thing is with a bog-standard ESR meter, it's all about relative
:measurements, 'feel', experience, and intuition rather than accuracy.
etermining whether or not an electrolytic is faulty by way of its ESR, is a
:bit of a black art, and is with any ordinary 'in-circuit' ESR meter. The
:reading just gives you some help and 'feel good' backup. This is why I have
:contended on many occasions that such an instrument is not one of absolutes,
:and cannot be just picked up and used to give 'go / no go' definitive
:answers about the state of a cap, by anyone who chooses to buy one. It's
:just a helper that gives you one more pointer, allbeit a mostly pretty good
ne, without having to remove the cap (generally) from the circuitry that
:it's in.
:
:The thing is with the inductance argument, I would have thought that given
:that the self inductance of a 'standard' electrolytic is quite high due to
:the way they are constructed, the inductance of the test leads would have
:been pretty insignificant in comparison ??
:
:Certainly, in practice, if it is a 'problem', it has never caused me any
:trouble with using my Bob Parker to help identify bad 'uns ...
:
:Arfa
:


I agree entirely. With regard to everyday servicing you only really require a
"relative" indication. In some cases some intuitive interpretation on the part
of the user is required to determine whether a cap is "bad" and requires
replacing. Obviously, if the indication is way too high it doesn't take any
intuition to decide to replace.

 

[Arfa Daily]

:
::> On Wed, 31 Dec 2008 09:30:11 -0800 (PST), [email protected] wrote:
:>
:> :>
:> :>
:>
:> :>
:> :>
:> :>
:> :> >I have a question about ESR measurements.
:> :>
:> :> > I have been doing some reading about ESR and the literature
indicates
:> :> > it can be a strong function of frequency. In the design a SMPS I
:> :> > assume the ESR of importance is at switching frequencies (at least
:> :> > for the line side capacitor).
:> :>
:> :> > I have also noticed that often the measurement made in trouble
:> :> > shooting are low frequency measurements (step function - looking
at
:> :> > the abrupt change in voltage).
:> :>
:> :> > Is the above accurate and if so when a cap goes bad do the low
:> :> > frequency measurements catch the majority of the high frequency
ESR
:> :> > failures. Are there significant failure modes where a low
frequency
:> :> > ESR measurement would miss the higher frequency ESR failure?
:> :>
:> :> > Thanks Much,
:> :>
:> :> Most of the ESR meters that I have seen test in the 50-200 kHz
range.
:> This
:> :> pretty much covers the switching frequencies of most power supplies,
at
:> :> least in consumer equipment that I am familiar with.
:> :>
:> :> Leonard
:> :
:> :Thanks everyone for your replies, you have answered my questions.
:>
:>
:> Now to throw in a curve ball...
:>
:> In designing his Tan-Delta meter, Cyril Bateman used 100Hz sine. His
:> theory was
:> that using 100KHz could cause the measuring lead inductance to exceed
the
:> self
:> inductance of the capacitor under test and thus complicate the accuracy
of
:> the
:> result.
:
:The thing is with a bog-standard ESR meter, it's all about relative
:measurements, 'feel', experience, and intuition rather than accuracy.
etermining whether or not an electrolytic is faulty by way of its ESR,
is a
:bit of a black art, and is with any ordinary 'in-circuit' ESR meter. The
:reading just gives you some help and 'feel good' backup. This is why I
have
:contended on many occasions that such an instrument is not one of
absolutes,
:and cannot be just picked up and used to give 'go / no go' definitive
:answers about the state of a cap, by anyone who chooses to buy one. It's
:just a helper that gives you one more pointer, allbeit a mostly pretty
good
ne, without having to remove the cap (generally) from the circuitry that
:it's in.
:
:The thing is with the inductance argument, I would have thought that
given
:that the self inductance of a 'standard' electrolytic is quite high due
to
:the way they are constructed, the inductance of the test leads would have
:been pretty insignificant in comparison ??
:
:Certainly, in practice, if it is a 'problem', it has never caused me any
:trouble with using my Bob Parker to help identify bad 'uns ...
:
:Arfa
:


I agree entirely. With regard to everyday servicing you only really
require a
"relative" indication. In some cases some intuitive interpretation on the
part
of the user is required to determine whether a cap is "bad" and requires
replacing. Obviously, if the indication is way too high it doesn't take
any
intuition to decide to replace.

A very good example of this occured on my bench last Friday. I had a Sony
DVD / HC that was very 'iffy' at the beginning of a disc. It would freeze
and pixellate pretty much at random, but if you left it and it managed to
get past the intro without totally falling over, it would play ok for the
rest of the disc. On auto setup, it would proceed ok on single layer and CD
discs, but fail every time on a dual layer. I had a laser for the model on
the shelf, so tried it, but it was just as bad. I then turned my attention
to the little servo board under the deck, where there were several surface
mount electrolytics. For no other reason than that similar caps in a similar
location give all sorts of odd problems on many of the Sony CD players from
a few years ago, I got out the ESR meter, and ran it over these caps. Now I
don't know about your experience of ESR and SM electros, but I've always
found that even with brand new ones, the ESR is significantly higher than
you would expect for the same value and voltage of a through-hole type.

There were two 22uF 10v ones on the board, and the first one measured 5
ohms. Now if that was a through-hole one, you'd immediately say it was bad,
but for a SM, that is quite possibly a satisfactory figure. When I measured
the other one, it went 3 ohms, so I found another similar one on another
board, and measured that. It also went about 3 ohms. So, based on the fact
that two of them read one value, and a third read a slightly different
amount, I applied experience, gut feeling and the measurements, and went
ahead and replaced it. Result ? Total cure, and the ESR meter played only a
supporting role in achieving that.

Arfa

 

[Fred McKenzie]

The Capacitor Wizard uses a 100Khz sinusoidal signal. The Bob Parker
design is as described - that is what is used in our Blue ESR meter.
Bob always represented that his method is the equivalent of a 100Khz
signal.

John-

I built one of your Blue ESR meter kits and am impressed with it.

Somewhere I read that there is an industry specification or common
practice, that requires ESR measurement to be made at 100 KHz. If it is
in fact a "standard" method, you should petition that standards
committee to agree that your method is equivalent.

As others commented, approximate methods for measuring ESR may be
adequate for detecting common failures. However, I do not think the Bob
Parker method is approximate. As long as there is no significant
inductance involved, it seems to be quite accurate. Try putting
resistors in series with low-ESR capacitors. Taking the low ESR into
account, the meter should measure close to the same value as when
connected across the resistors alone. I expect it will agree with other
methods of measuring the resistors. Certainly better than ten percent!

Fred