Elektor Electronics

"Eric the Kiwi"



**  A waste of perfectly good trees.





**  Trees 1,  Elektor  0.





**  Who bloody cares ??

Every second DMM on sale measures capacitance well enough.

For PSU electros an ESR meter is all you need and Bob Parker's one is a
winner.

If an electro passes the ESR test -  then it has its original (marked)
capacitance value.

If an electro fails the ESR test -  then it goes straight in the bin.


Capice ?




.........  Phil

On Sun, 11 Sep 2005 13:11:45 +1000, "Phil Allison"



Why is that Phil? Is it because the dont publish designs based on
pre-historic components?

Perhaps the ex roadies dont like mags that publish mostly digital
designs

<snip>

Because you can get same thing in a pdf

Now why did I think as I wrote the question, "I would most unlikely get a
answer, just pathic replys from children that have nothing better to do"

This is aus.e;ectronics, and it does reflect the nature of the
industry in australia :)

I used to be an elektor subscriber and they continue to send me mags.
If this one turns up in the post I will email you.

to keyboard and composed:



With respect, I can't see that this question makes any sense. An ESR
meter measures ESR, not capacitance. Ideally the capacitor's C should
not affect the result because it should look like a S/C to the test
signal.

I don't know anything about the Elektor meter, but Bob Parker's meter
applies charging pulses of three different amplitudes to the capacitor
under test, and then compares the voltage to an internal reference
capacitor. The assumption is that the pulse is of sufficently short
duration so as not to appreciably charge the test capacitor. AFAICS,
the larger the cap, the more accurate the measurement. As for very
small caps, I would think that the meter's internal time constants
would be selected so as to accommodate the smallest expected values of
capacitance. At the very worst, the meter would overestimate the ESR.
In this case the user could check the reading against a known good cap
of the same value.

-- Franc Zabkar

Please remove one 'i' from my address when replying by email.

Looking at a review of ESR meters I note that they do vary in respect of the
range of caps they can accurately check for ESR - especially at the lower end of
the range I gather

David

Franc Zabkar wrote:

On Wed, 14 Sep 2005 23:57:28 GMT, quietguy
keyboard and composed:



Well, that's what I expected in the worst case.

Anyway, I notice that the ESR table on Bob Parker's Mark 1 meter goes
down to 1uF, so I tried the following test using five Hitano EXR
series 105 degC 1uF 100V low ESR electrolytics (WES code 1EXR100).

 individual ESR measurements (1uF):   3.1, 3.3, 3.6, 3.8, 3.8 ohms
 all five in series (0.2uF):          18 ohms
 all five in parallel (5uF):          0.67 ohms

The results suggest that the meter is reasonably linear in the range
0.2uF - 5uF.

The next readings are for used, 250V, film, mains rated X2 caps:

 C (uF)   ESR (ohms)
-----------------------
  1.0     1.4  1.4
  0.82    1.5  1.5
  0.47    3.3  2.9  3.5
  0.22    7.7
  0.10    19   19

I would have thought that the ESR for this type of cap was negligible
(< 0.5 ohm ???). If so, then the readings must reflect the
capacitance's contribution to the measurement process. If I understand
Bob's design correctly, the capacitance should contribute a
"resistance" value of dt/C (= dV/Ipulse), where dt is the duration of
the charging pulse. For example, a 1us pulse into an ideal 1uF cap
should make it test like a 1 ohm resistance, whereas an 0.1uF cap
would measure 10 ohms.



-- Franc Zabkar

Please remove one 'i' from my address when replying by email.

"Franc Zabkar"
 quietguy



**  Bob's brilliant ESR meter is very clearly intended to test  ** ELECTROS
**  - while in circuit, in faulty gear !!

The whole reason for testing the electros is because at the end of their
useful life,  the " juice " inside dries out sending the ESR value high.

The principle of operation relies on the fact that at a frequency of about
100kHz, an electro's impedance it at its minimum value  -   so nearly equal
to the intrinsic internal resistance or  ESR  !!

Very low value electros, like 1 uF, need a slightly higher test frequency
for best accuracy.

However,  plastic film capacitors need to be tested from 1 to 10 MHz to find
their true ESR values  -  only a variable frequency ESR meter could do that
as you need to find the series resonance .

Film caps do not vary in ESR throughout their life  -  so no need for such
testing exists.





.........    Phil

On Thu, 15 Sep 2005 18:41:49 +1000, "Phil Allison"



The test signal is not sinusoidal. While the Xc of a 1uF cap at 100kHz
is a negligible 1.6 ohms, that is not what the meter sees. The test
signal, at least for the Mark 2 version, is an 8us current pulse with
a frequency of 2kHz. The voltage seen at the meter's terminals would
be given by ...

 V = (I * R) + (I * dt/C)

The meter assumes it is seeing a pure resistance (R'), so it applies
Ohm's Law and comes up with ...

 R' = V/I
    = R + dt/C

This reduces to R' = R + 8 in the above case.

So if one were to apply the design principle described by Bob in his
SC article, one would expect the meter to read 8 ohms for a perfect
1uF capacitance. This is not what I observe in practice, but it may at
least partly explain the higher than expected readings for the film
caps.

-- Franc Zabkar

Please remove one 'i' from my address when replying by email.

"Franc Zabkar"
 "Phil Allison"



 **  No need to be,  since the impedance minimum of an electro has a very
broad range.






 ** 1.6 ohms is very far from negligible for a film cap.

  It is,  however,  what the ESR meter reads with a 1uF film cap.





 **  It is an coupled AC signal, remember.




.............  Phil

On Thu, 15 Sep 2005 18:41:49 +1000, "Phil Allison"







I did not suggest that such testing was required. I merely wanted to
see how the meter would evaluate "perfect" caps. To paraphrase Dirty
Harry, a man's gotta know his test equipment's limitations.

Regardless, manufacturers quote a loss tangent figure for these types
of caps.

See this datasheet:
 http://www.vishay.com/docs/28119/mkp3382.pdf

Tangent of loss angle at  1 kHz    10 kHz    100 kHz
----------------------------------------------------
C <= 470 nF               0.001    0.002     0.010

The loss tangent is given by:

 tan d = 2pi * f * C * R

So R (in ohms) in the above three cases for an 0.47uF film cap is as
follows:

 R at  1 kHz    10 kHz    100 kHz
---------------------------------
       0.34     0.068     0.034
 

-- Franc Zabkar

Please remove one 'i' from my address when replying by email.

"Franc Zabkar"
 "Phil Allison"



 **  That is a tedious irrelevance.

The fact explains why Bob's ERS meter is set up to test electros ONLY.

 It is  NOT,  repeat  NOT , a precision ESR  tester.

-------------------------------------------------------------






** Another tedious irrelevance.

Bob's ESR meter is clearly stated to be ONLY for electros.

Its only job is to pick out the ones that have gone bad in service.

-------------------------------------------------------------







** More tedious bloody irrelevances.

 BTW  The result figures are jumbled.


 What a colossal , pedantic PITA you are Zabkar.



.......   Phil

On Sat, 17 Sep 2005 12:37:44 +1000, "Phil Allison"





Using a "perfect" cap demonstrates how the meter responds to the C
component in a real cap. The designer's assumption is that the C
component does not contribute significantly to the measurement
process. This is demonstrably true for 1uF caps, but not for smaller
ones.



The OP wanted to know the range of an ESR meter. I answered the
question, both by actual testing, and by analysis of the design
methodology.




Nope. Read the datasheet and apply the formula. If you still don't
like the result, then complain to the manufacturer. Maybe he will give
a damn. I don't.



The OP's question was in regard to the *limits* of measurement of the
ESR meter. In the absence of explicit specs from the manufacturer or
designer, the answer to such a question requires an understanding of
the measurement process ... or access to perfect caps. Therefore, the
question of itself *demands* a pedantic answer.

-- Franc Zabkar

Please remove one 'i' from my address when replying by email.

"Franc Zabkar"
"Phil Allison"



 **   WRONG  !!!!!!!!!!!!!!!!!!!!!

Here is the post:

" Anyone get Elektor Electronics magazine
http://213.222.12.205/Default.aspx?tabid27%&year20%05&month=9&art52%895&PN=O
Hopeless NZ bookshops have not put it on the shelf yet,
Does anyone know what capacitors range the ESR meter does? "


The new, Elektor meter is a COMBINED capacitance and ESR meter.

The OP was asking what range of capacitors it could test for capacitance.

Bob Parker's ESR meter has  NOTHING  to do with the OP's question !!!!!!!!!!






**  But for the WRONG  meter and you got it all WRONG as well.





** But  NOT  Bob Parker's one.





  **   You are an autistic lunatic -  piss off.



 .........  Phil

"Franc Zabkar"
 "Phil Allison"




** Please do  -  look at the impedance curves on page 7.

A 470nF cap has a resistance of  **40 milliohms** at 2 MHz.

This represents the intrinsic, series resistance of the cap.

Bob Parker's ESR meter indicates the same parameter for an electro.

The big difference with electros is their impedance minimum is very broad  -
ie the curve has a low Q.



.........   Phil

On Sat, 17 Sep 2005 12:22:05 +1000, "Phil Allison"



Whether or not the test signal needs to be sinusoidal is irrelevant.
The fact is that it isn't. It is in fact an 8us 2kHz *pulse*. The
principle of operation of the meter involves driving such a pulse into
a pure C in series with a pure R. A discussion of a real capacitor's
performance at 100kHz is just pointless obfuscation.

All you need to know is that I = C * dV/dt. From this it follows that
a constant current produces a voltage ramp.

That's elementary circuit theory. Quite simple really.



Irrelevant. You were referring to electros. Anyway, nobody cares about
the behaviour of the test cap at 100kHz.



The meter does not read Xc. It reads *ESR*, or at least it thinks it
does. In any case, the fact that the reading is "far from negligible"
for a "pure" capacitance demonstrates that the meter has limitations,
albeit acceptable ones, which was essentially the point of the OP's
question.



Capacitors C5 and C6 have no effect since they are in series with a
current source.

Capacitor C7 also has no impact because the input impedance of the
meter (neglecting R23) is appropriately high.

You can quite easily calculate the expected degradation of the pulse
at the junction of C7 and R12. Assume you are testing a pure resistor,
and that it produces a voltage, Vt, at the meter's test leads.

The voltage developed across C7 by the test pulse would be:

 dV = Iin . dt / C7

where Iin is the current flowing through C7 and R12.

Now the impedance looking into the junction of R13 and R14 is 69K.

This gives:

 Iin = Vt / 69K

So the degradation, dV, of the pulse, Vt, is given by:

 dV/Vt = (Vt /69K) . (dt / C7) / Vt
       = 8us /(69K . 33nF)
       = 0.35%

Hardly worth worrying about.

-- Franc Zabkar

Please remove one 'i' from my address when replying by email.

"Franc Zabkar"
 "Phil Allison"



 ** Strange then how YOU raised the issue.

Go back and read the first para re the PRINCIPLE of operation.






**  Which is AC coupled to the cap under test  AND is a signal with energy
concentrated in the range above 100kHz.





**  You are DEAD  WRONG   !!

The PRINCIPLE of operation is that measuring an electo's  IMPEDANCE  at a
high frequency finds its ESR - because they are essentially the same number.






**  Completely irrelevant to has Bob's meter works.

The voltage across the cap under test is assumed to be entirely due to ESR -
and so it is for electos above a few uF.





 **  Piss off    -   dickhead.





 **  YOUR  comment above refers to a film cap - dickhead.





**  WRONG   -  that is how the ESR is being measured.

 At that frequency, electro cap impedance and ESR are the same or similar.





**  WRONG   -  it reads Xc.

  The PRINCIPLE of operation is that Xc = ESR over a range of high
frequencies.





  **  Gobbledegook.





 **  The OP never asked any such thing -  dickhead.






**  The remove any DC component -  you stubborn bloody ass.

 The cap under tests cannot "charge up"

  You are so UTTERLY wrong,  Zabkar  - there is no way to get anything RIGHT
into your fat head.




............  Phil

Ive tested faulty electro caps which still give a 'pass' according to the
table. Ive also got some caps quite visibly leaking & buldging that test
'ok 'with my ESR & a CAP meter.
Just something to bear in mind for amatures (like me).