Cap testing question

[Phil Allison]

Jeff Liebermann wrote:
Phil Allison

** You went to lot of trouble to see the obvious.

Cap-B was one that had failed with high ESR. There was plenty of data
and anecdotal evidence on how a normal capacitor would act. I wanted
to see how a defective capacitor acted.

** Well, that depends how *defective* it actually is !!!

An electro that has lost nearly all of its fluid and reads off-scale on Bob Parker's meter is not likely to come good with a bit of heat.

I also wanted to see how the
ESR meter functioned with small value and size (0.22uF 50v) caps,
which was Cap-D. The change in ESR with temp was far less radical
than the others.

** Even a 0.22uF film cap reads 7 ohms on the Bob parker meter - IOW the impedance of the thing at 100kHz. Bob's meter does not read actual ESR values, rather it reads impedance vales at 100kHz.

This *fact* is pointed out in the instructions and the lowest value electro you can reliably test is about 1uF - as shown on the front panel table.

A man has gotta know the limitations of his ESR meter.
------------------------------------------------------

All I did was try a few 450V electros, heat them with a hot air
gun until they were darn uncomfortable to hold and note that in
every case the ESR reading had plumeted by 5 to 10 times.

I wasn't interested in high voltage electrolytics.

** They have higher ESR values than low voltage caps, so there is plenty of room for ESR readings to come down without approaching the low reading limit of Bob's meter.

I used a value of 47uF, so its reactance at 100kHz was negligible.

Pays to think ahead, you know.


..... Phil

 

[Dimitrij Klingbeil]

On 06.05.2016 21:14, ohger1s@gmail.com wrote:

On Friday, May 6, 2016 at 3:02:30 PM UTC-4, et...@whidbey.com wrote:
I have a computer that won't turn on unless it is warm enough. The
problem is in the power supply. If I use a blow drier to blow some
warm air through the exhaust holes in the power supply for a
minute or two the power supply starts working. I was told by
someone here or on the basics group that there is probably a bad
electrolytic capacitor causing the problem. I also have a CNC
machine that has a servo amp that acts the same. The machine must
be on for a while and when the cabinet that holds the servo amps
gets warm enough inside the amp start working and will work fine as
long as the machine is kept powered up. If the machine is turned
off for an hour or so and then powered back up the amp still works.
But if the machine is off overnight the amp acts the same as above.
I replaced the amp so there is no problem running the machine but I
would like to repair this amp so I can have a spare on the shelf.
Anyway, it has only 3 physically large electrolytic caps. I
unsoldered one lead from each cap and measured the capacitance with
my Extech multimeter. The two 10 uf caps measured 11. something uf
and the 33 uf cap measured 37. something. Could these caps be bad
anyway? There are also three blue colored dipped caps that have
printed on them: .068K 250. I don't what the dot in the .068K means
but I don't think it's a decimal point. Maybe it means lead free.
There are also a couple 47 uf wound polymer caps. Could either of
these types of caps need to be warm before they work properly? I
have no schematic for this amp and have not been able to get one.
Furthermore, the company that makes this servo amp has ground the
markings off of any device with more than three leads. Thanks,
Eric

Yes, caps that read good value can still be bad. Most electrolytics
that are bad will show high ESR readings (bad) before they show low
capacitance. ESR meters are cheap enough that anyone who screws
with electronics should have one. Despite the advice of others, I
always remove the capacitor to test ESR with an "in-circuit" ESR
tester.

Also, a good percentage of wonky electrolytics will show signs of
venting at the top of the cap, if not an actual electrolyte leakage,
then a discernible bulge. If you see any bulged caps, change them.

Most small disc and film caps will not become temperature sensitive.
For a device to be warm up sensitive like your device it's usually a
weak electrolytic , so concentrate there.

If there's only a few electrolytics, change them all with good
quality (I like Panasonic) caps and they'll probably run.

Good advice. But there is one more small-ish pitfall for the unwary. If
a power supply refuses to start when cold, it's usually an electrolytic
cap. But in many cases it's not one of the big caps but the small one
that is located in the start-up / auxiliary section.

Mains power supplies usually have an auxiliary section that is used to
power the internal needs of the primary side switching circuitry. This
section is powered by a small auxiliary winding on the transformer in
steady-state. But because it needs to start up first, and because at
first the main transformer is not yet operating, it is precharged with a
high value resistor (or sometimes a string of 2 or 3 resistors in
series) directly from the rectified mains. There is a small (47 uF or
similar, 25 to 35 V typically) electrolytic. At first it gets precharged
through the resistor up to the point where the controller chip can start
up. Once the controller starts, it would rapidly drain the small cap, so
the main switching circuitry has to come on line fast in order to top up
the the auxiliary supply cap and keep it from discharging.

Now this small cap only provides little reserve. It's usually sized very
spartanically and it lasts for just one try and only if the start-up
timing of the switching circuitry is not delayed for any reason. If it
drains before the switcher can fully start, the switcher will stop and
wait for another retry at a later time.

Often this small cap is considered "non-critical" and carelessly placed
by the power supply designer into whatever corner had whatever little
free space. This may happen to literally be a "corner" of a heatsink!

When this cap dries out either from overheating of just being selected
from the cheapest manufacturer, it will rise in ESR and thus no longer
provide the peak current that the controller needs in order to start.
This either delays the start to the point where the cap is drained or
causes it to outright abort prematurely because the auxiliary supply
rail tanks right at the first few switching cycles.

When warmer it has lower resistance, thus it can provide higher peak
currents and have better chances of starting the supply successfully.
Also if the supply is working continuously for many years and never
switched off, the condition of this capacitor is of no consequence
during operation. Even if somebody just cut it out of a working power
supply, there are usually some small ceramics on the rail and the
auxiliary transformer winding will keep them topped up, so that the
supply would quite likely keep on working. But if the supply ever gets
switched off, even for a very short time, it will never start up again.

If you see this type of behavior, find and replace the cap in the
auxiliary section. You can normally identify it as being physically
small, a low voltage type (from 25 to 50 V) and being connected to a
sizeable precharge resistor that gets supplied from the high voltage
DC rail.

If you don't have an exact replacement, it's normally OK to use the
next higher capacitance as long as the voltage rating fits. If you
find that yours sits in the corner of a heatsink, use a low ESR high
temperature rated replacement (no matter what the original has been)
and some reasonable reserve capacitance-wise won't hurt either.

Dimitrij

 

[Ralph Mowery]

In article <0001HW.1CDE42940005391D11ECB93CF@news.eternal-
september.org>, not@home.cow says...

That doesn't mean it could not have high ESR, but it was definitely not the
cause of the problem. There are people who like to change ALL the caps in old
hifi amps, I think they're wrong. While I will agree with changing the
corresponding cap in the other channel when you find a bad one, or even weak
one (significant ESR compared to Xc) it does no good to just change them all
arbitrarily.

Sure it does. If the problem is with one of these caps, changing them
arbitrarily FIXES the problem.

?Penny wise, pound foolish.? Yes, you?re right, finding the ONE cap
that is the cause is technically proficient. But just to say ?I found the
culprit!? does not help the customer (or yourself, if it?s you). The unit
will be back on your bench very soon if you don?t replace all electros now,
if you find one or more below spec.

Unless the capacitors are very expensive, it is best to just replace
them all even if only one or two is showing up to be the problem.

When working a factory repair man came in to repair a 300 HP electric
motor varitable speed drive. This was a 3 phase AC input. He determined
2 out of 3 large diodes were bad. I asked him to replace the 3 rd one
and he said those were about $ 100 each. I said so what, the equipment
beind down was costing several thousand dollars an hour and another $
20,000 to start it up and shut it down. At that ratio of dollars to
production I thought it would be a good idea to replace it even if it
was good.
I am sure it would have started, and may have ran from then on, but why
take a chance it would crap out a week or even a year later.

 

[Jeff Liebermann]

On Sat, 7 May 2016 18:18:16 -0700 (PDT), Phil Allison
<pallison49@gmail.com> wrote:

Jeff Liebermann wrote:
Phil Allison


** You went to lot of trouble to see the obvious.

Cap-B was one that had failed with high ESR. There was plenty of data
and anecdotal evidence on how a normal capacitor would act. I wanted
to see how a defective capacitor acted.


** Well, that depends how *defective* it actually is !!!

Well, cap-B was labeled 2200uF 10v and was bulging at the top.
Room temp ESR was measured at about 2.3 ohms:
<http://802.11junk.com/jeffl/Electrolytic-cap-test/Cap-B.jpg>
and should have been about 0.1 ohms according to the latest chart for
my ESR meter:
<http://802.11junk.com/jeffl/crud/ESR.txt>
It's the 2nd from the left in:
<http://802.11junk.com/jeffl/Electrolytic-cap-test/caps.jpg>
and shows no signs of leakage except for the bulging top.

An electro that has lost nearly all of its fluid and reads
off-scale on Bob Parker's meter is not likely to come good
with a bit of heat.

True. However at 100C, the ESR dropped to 0.25ohms which is not
perfect but would probably be adequate for most purposes.
<http://802.11junk.com/jeffl/Electrolytic-cap-test/Cap-B.jpg>

I also wanted to see how the
ESR meter functioned with small value and size (0.22uF 50v) caps,
which was Cap-D. The change in ESR with temp was far less radical
than the others.

** Even a 0.22uF film cap reads 7 ohms on the Bob parker meter -
IOW the impedance of the thing at 100kHz. Bob's meter does not
read actual ESR values, rather it reads impedance vales at 100kHz.

Right. The impedance is the vector sum of the resistive (ESR)
component and the capacitive reactance of the capacitor. At 100KHz:
Xc = 1 / (2*Pi*MHz*uF) = 1 / 2*Pi * 0.1 * 0.22 = 7.23 ohms
If the ESR were the typical less than 1 ohm, the bulk of what the
meter is reading is from the capacitive reactance.

The meter was reading exactly 7.00 ohms, I can get a rough idea of the
resistive ESR from:
R = sqrt(Xc^2 - Z^2) = sqrt(7.2^2 - 7.0^2)
= sqrt(52.3-49) = 1.8 ohms

However, let's pretend that this capacitor was defective and it
exhibited a higher ESR, such as 6 ohms. What would the ESR meter
read? Same formula:
Z = sqrt(R^2 + Xc^2) = sqrt(6.0^2) + 7.2^2)
= sqrt(36+52.3) = 9.4 ohms
You would be able to see the difference between a good 7.0 ohm reading
and a bad 9.4 ohm reading, but only if you did the calculations in
advance. I once played with an HP/Agilent LRC meter that did all this
automatically. it would be nice if the next generation of ESR meters
could also do that.

Raising the frequency to 1MHz would reduce the capacitive reactance to
0.723 ohms, which would produce something closer to the real ESR.
However, the ESR changes with frequency, so accuracy will suffer.

This *fact* is pointed out in the instructions and the lowest
value electro you can reliably test is about 1uF - as shown on
the front panel table.

A man has gotta know the limitations of his ESR meter.

I just hate reading the instructions. However, the 1uF lower limit is
printed on the front of almost every ESR meter. My Capacitor Wizard
has it boldly displayed on the front panel.
<http://www.howardelectronics.com/images/thumbnails/1138/1138/detailed/1/CAP-WIZARD.jpg?t=1440164290>

All I did was try a few 450V electros, heat them with a hot air
gun until they were darn uncomfortable to hold and note that in
every case the ESR reading had plumeted by 5 to 10 times.

I wasn't interested in high voltage electrolytics.

** They have higher ESR values than low voltage caps, so there is
plenty of room for ESR readings to come down without approaching
the low reading limit of Bob's meter.

I used a value of 47uF, so its reactance at 100kHz was negligible.

Pays to think ahead, you know.

Yeah, but I think you missed my point. I don't do tube amplifiers and
therefore do not run into too many high voltage capacitors. The one's
I see most of are low voltage and high capacitance electrolytics as
found on PC motherboards, ATX power supplies, LCD monitors, wall
warts, and various consumer electronic devices.

Incidentally, speaking of tubes (valves), you might be amused by this:
<https://www.kickstarter.com/projects/1815752970/hybrid-tube-amp-for-the-raspberry-pi>

>.... Phil

--
Jeff Liebermann jeffl@cruzio.com
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558

 

[Phil Allison]

Jeff Liebermann wrote:

I also wanted to see how the
ESR meter functioned with small value and size (0.22uF 50v) caps,
which was Cap-D. The change in ESR with temp was far less radical
than the others.

** Even a 0.22uF film cap reads 7 ohms on the Bob parker meter -
IOW the impedance of the thing at 100kHz. Bob's meter does not
read actual ESR values, rather it reads impedance vales at 100kHz.

Right.

** Then you should not have posted what you just did. The actual change in ESR of a 0.22uF 50V electro was not readable on a Bob Parker meter.

What is really happening is the electrolyte becomes more conductive at higher temps, many times more in the range from 20C to 100C. It is the resistance of this fluid that dominates the ESR value.

A man has gotta know the limitations of his ESR meter.

I just hate reading the instructions.

** Right - you take zero notice of anything others tell you.

I wasn't interested in high voltage electrolytics.

** They have higher ESR values than low voltage caps, so there is
plenty of room for ESR readings to come down without approaching
the low reading limit of Bob's meter.

I used a value of 47uF, so its reactance at 100kHz was negligible.

Pays to think ahead, you know.

Yeah, but I think you missed my point.

** No I didn't.

You missed mine, as usual.


..... Phil

 

On Saturday, May 7, 2016 at 8:18:08 AM UTC-4, jeanyves wrote:

On 2016-05-06 19:05:35 +0000, etpm@whidbey.com said:

I have a computer that won't turn on unless it is warm enough. The
problem is in the power supply. If I use a blow drier to blow some
warm air through the exhaust holes in the power supply for a minute or
two the power supply starts working. I was told by someone here or on
the basics group that there is probably a bad electrolytic capacitor
causing the problem. I also have a CNC machine that has a servo amp
that acts the same. The machine must be on for a while and when the
cabinet that holds the servo amps gets warm enough inside the amp
start working and will work fine as long as the machine is kept
powered up. If the machine is turned off for an hour or so and then
powered back up the amp still works. But if the machine is off
overnight the amp acts the same as above. I replaced the amp so there
is no problem running the machine but I would like to repair this amp
so I can have a spare on the shelf. Anyway, it has only 3 physically
large electrolytic caps. I unsoldered one lead from each cap and
measured the capacitance with my Extech multimeter. The two 10 uf caps
measured 11. something uf and the 33 uf cap measured 37. something.
Could these caps be bad anyway? There are also three blue colored
dipped caps that have printed on them: .068K 250. I don't what the dot
in the .068K means but I don't think it's a decimal point. Maybe it
means lead free. There are also a couple 47 uf wound polymer caps.
Could either of these types of caps need to be warm before they work
properly? I have no schematic for this amp and have not been able to
get one. Furthermore, the company that makes this servo amp has ground
the markings off of any device with more than three leads.
Thanks,
Eric

lots of esr meters to choose from : to build or to buy
http://kripton2035.free.fr/esr-repository.html

--

Jean-Yves.

(Nice link, thanks.)
George H.