changed.

An ESR meter gets a r o u n dt h i s p r o b l e mb y u s i n ga t e s t fiequency (or pulse rate) that's high enough for the capacitive reactance to be almost zero [Link] just the ESR as the measurement.

Meosuring ESR
ESR obviously cannot be measured directly, using a c()nventional ohmmeter. so a means has to be found 'get to to'the ESR that's hidden inside the capacitor. A number of ready-made meter designs and kits are now availablefi)r measurincthe ESR of an electrolyticcapacitoi incircuit, using [Link] achievesuccess and simplicity in varying [Link] of a suitable meter enables cold checks to be made without the risk of any further damage occurring. Strictly speaking it is the irnpedancethat's being measured but. over a parlicular rangeof test pararneters. it can be shown that this presentsmore or lessthe same value. It is all too easy to over-complicate things when it comes to ESR. There are those who argue that ESR meters don't rneasurethe ESR precisely. But, in the real world, how precise does the reading need to be? It really doesn't matter. The serviceengineerjust needsto know, as quickly as possible,which capacitoris causingthe [Link] ESR meter doesjust that! Once technicians have become accustomed to using an E,SRmeter, they wonder how they ever managed without one. Although the ESR varies somewhat with frequency, we can in practiceregartjil as being a constantin-phasecomponent, and calibrateour meter using fixed resistors. This is useful. as the meter r'vill also serve well as a lowohm meter. So. if an ESR meter doesn't measurecapacitance,what readings can we expect from good and bad capacitors using such a meter? [Link] you know rvhether a capacitoris OK or not? The curve shown in Fig. 2 gives a practical idea of the sort of ESR readings that should be obtained rvith good capacitors of diff'erent values. There is no hard-and-fastrule however - it's not an exact science. All it needsis a bit of getting used to. This doesn't take long: just measure the ESR of a few new caoacitors. Try | . 10. .17. 100.470 and 1,[Link] rvill find that values of 47pF and above measurequite low, 0'5Q or less. rvith the buzzer coming on (the buzzerturn-on point can be varied, see later). The

The equivolent series resistqnce (ESR)of qn electroly'ic cqpqcitor is o reliqble indicotion of its condition. Alon Willcox's first ESR meter design, published in these pqges some five yeors qgo, wqs very populor. This Mork 2 version incorporoles seyerol improvemenls, in porticulqr q simpler oscillqtor design ond single-bottery operotion
simplified equivalent circuit of an electrolytic capacitor, seeFig. l. is usually shown
rvhen providing a brief explanation of what ESR is all [Link] addition to the ideal capacitor Xc, a second component is present. [t has a significanteffect orr the capacitor's performance, and is referred t o a s l h e e q u i v a l e ns t c r i e sr e s i s t ance (ESR ). Electrolyticcapacitors are used mainly tbr decouplingand, to a lesserextent,for signal coupling. This being so it's important that, for optimunr pelfbrmance, the impedance (AC rc-sistance) of the capacitoris as low as possible. An electrolytic capacitor's ESR is mainly determined by the condition of the electrolyte (paste) that separatesits foils. The electrolyte increases the component'scapacitance but. when it [Link] increasesthe impedancepresent. The lolver the ESR. the better the capacitorl E,lectrolytic capacitors used in switch-mode(chopper)power supplies and those mountedclose to heatsinkstend to run hot. Heat is inclined to dry out the electrolyte and, in tirne, a capacitor may develop a high ESR. This will itself introduce a power loss and more heat!The effect of failure of this type in a power supply can be [Link] example,if the capaci t o r i s i n l h e H T m o n i t o r i n cs c c l i o n ol a TV set's power supply the HT voltage might rise, damaging the line output transistor and maybe the field output IC. This type of problem is more significant rvhen a set is started up from cold. as the condition of a [aul1yelectrolyticcapacitoris worse rvhencold. Thus cold checks are by far the best approach to testing! Problemswith electrolytic capacitors,particularly those used in switch-mode power supplies, are causednot so much by a changeof value as by an increase capacitance in the component'sESR. Thus rernovalof a suspect capacilorlo check it with a conventionalcapacitance nteter is largely a waste of time. Furthermorea faulty capacitor may he overlookedbecause its capacitance value has hardly

Fig. 1: Simplified equivalent circuit of an electrolytic capacitor. Xc represents an ideal capacitor. lts reactance moves towards zero ohms as the frequency increases. Xc = 1/(2n fc). The value of the equivalent series resistance ESRis determined mainly by the condition of the electrolyte.

76

December 2 0 0 4T E L E V I S I O N

important point is just how low capacitors with values of 47pF and above [Link] 47OpF and over, the reading is close to zero ohms. l[. in practice.a 1.0004Fcapacitor producesa readingas high as 0'5Q, it's no good!

ESB((D

Anqloque or diqirql

displof?

Digital methods of measurement and display are generally regarded as providing more accurate results. F o r m a n y a p p l i c a t i o n lsh i s i s t r u e . but it's not necessarily so rvith ESR [Link] are two ways of interfacing the capacitor being The eapacitor testedand the nreter. can be connected in parallel with the test signal source, as shown in Fig. 3(a), or in [Link] Fig. 3(b). lf a digital ESR readoutis used, the capacitor must shunt the test the [Link] an ideal capacitor, result rvill be zero ohnrs and zero display: with a bad capacitor there will be a high readlng,with little shuntingaway of the test-signal. We are interested in ESR values of 3f2 or less, but a digital meter gives readings far higher than this. It's not a major problem, but can give rise to superlluous information as far as ESR measurementsare concerned.A more important factor is that the parallel rnethod results in a significant increase in the readings obtained because ol' excessive sensitivity to the inductanceof the test leads. This can be overcome by fitting two leads to each probe. It provides cancellationto a large extent,but this solufion is a bit clumsy in use and to [Link] series method of interfacins does not sutTerfrom this drawbalk. so the use of conventional test leads becomes acceptable.I don't know why this effect occurs - I just found out the hard way. Although we have become used to digital readouts norvadays,fbr ESR measurementthere is little d o u b t t h a t a m o v i n g - c o i lm e t e ri s the best type of [Link] gives a rapid, easy-to-interpretindication of the conditir-rn of'the capacitor. After some experience of using it, one gets to know where the pointer should approximately be rvith good capacitors of difTerent values. Indeed a meter scale becomes almost unnecessary. I know of people who have used this type of meter quite satisfactorily without ever having taken the trouble to fit a scale dedicated to ESR measurement. To repeat: il's nol an eract science but. rvith sortreexperience.

220 1tF +

Fig. 2: Plot of ESR vs. capacitance. The curve is typical of 63V working capacitors. With higher voltage ratings the ESRis somewhat higher. ESR

Constantcurrent generalot

Constantvortage generator

Rsample. Value chosen tor oplmum meter scale

(a)

(b)

Fig. 3: The shunt type of interface used with a digital display is shown at (a). Zero ESR= zero display. The output is directly proportional to the ESB. This method is sensitive to lead inductance. The series type of inbrtace used with an analogue display is shown at (b). Vout is non-linear, which is exactly what we require to avoid range-switching. The scale is expanded in the lowohms region. Vout is proportional to Rsample/(Rsample + ESR). So, if the value of Rsample is 3K),this would be the mid-scale reading. Just about ideal!

;i. 9;

t: '4

..*

.q 'F 4:'
, * F

a
q

c.*

-s

-*
.&

'* n
,a{
|l

f
rll

tr.

8.,

+s'

**

*?

,tl

It

*
I

U :..*
t

*'
i1

4l

3':r*

5'',*

t-;ff

**
*1 3,*

*trn

f'

.,'r,.;[r3]a

* nr*i

*'* * **

Loyovt oJ AIan Willcox's Mork 2 ESR mefer on stripboord. The Mo,rk 2 incorporates severol improvements, including o simpler oscillator design and single-bdl+ery operotion.

TELEVISIOD Necembe2 r 004

77

Fig. 4: The basic ESRmeter circuit, new version. ICI and lC2 require positive and negative supplies at pins I and 4 respectively. Further circuitry is required, see nert month, to generate the split-rail supply and provide a buzzer comparator. one soon gets to know which capacitor is causing the trouble by knowing roughly where the ESR meter's pointer should be with a good one. One argument that's sometimes put forward against the u s e o f a m o v i n g - c o i lm o v e m e n ti s that the movement will be damaged should the meter fall off the bench. I have a solution to this problem. lf you are inclined to be clumsy. attach a piece of string between the meter and the bench. long enough so that the meter comes to a halt just before it hits the deck but not so short that you don't get a bit of a jolt to serve as a reminder. sinewave is to incorporate in the oscillator's positive-feedback path a network originated by Max Wien in [Link] ensuresthat the feedWaveform: Analysis of a squareback is an in-phase component at wave has been used. and I've tried only one fiequency, which is fixed this myself. The results are a bit by the RC values used. At very unpredictablehowever. I had a lot high frequenciesC2 presents a lowof trouble trying to preservethe impedance path. rvhile at lorver frervaveform intact at the mV level to quencies Cl becomes an effective give a useful, predictable indicaopen-circuit. At some point in tion. It's not worth the effort. A betweenthere will be maximum sinewave is so easy to generateand output from the network Cl, Rl, amplily that I cannot see any justiC2.R.2. which is referred to as a fication for incorporating squareWien bridge. The RC values used lvave analysis into the design of an here result in oscillationat about ESR nreter. l00kHz. An imprcved ESR. mener There are two simple methods The meter presentedhere is similar of stabilisingthe output level rvith TesFsignolponometers an op-amp Wien-bridge oscillator. to the one described in the The method traditionally used is to March/April 1999 issues of Amplitude: To test for ESR or for i n c l u d ea b u l b i n t h e n e - e a t i v e [Link] then the use of an actual capacitancethere are no cont-eedbackpath (between pins l and ESR meter to check quickly for straints on how low the test signal 2 here). This was the approach faulty capacitors has become well can be as f,aras the capacitor is used in my 1999 [Link] established T.h e o r i g i n a lc i r c u i t [Link] into account has been some misunderstanding works well and has stood the test power consumption, and the probabout the bulb. because of the difof time, but feedback from the lems associatedwith low-level sigferent specifications used in US trade has prompted me to make nals of the order of microvolts, literature. I went into the matter some improvements. These include noise considerationsetc., a level of in some detail in the 1999 articles. single-battery operation and about 5mV peak-to-peakseemsto The corrcct specification is improved temperature stability. be a good compromise. 28Y,24nlA. The new circuit remains stable The principle behind the use of down to 6'2Y, and consumes about Frequency: The time period bulb is that its resista tungsten l3mA. Becauseof the improved should be short enough to zero out ance increases with the current temperature stability of the new the capacitive reactance.100kHz is that flows through it. If a bulb's oscillator circuit, there is no need a popular choice. At about this freis measured,the small resistance for an externally-available set-zero quency the ESR and impedance current from the ohmmeter will control. converge with the type of capaciWhen increaseits resistance. tors in which we are interested. The oscillqror cirruit employed correctly in this applicaApart from that consideration, or pulsegenerator tion the bulb does not come anyAn HF oscillator l00kHz is about the top frequency with a stable outputover a reason- rvhere near incandescence.I decidat which a predictable gain of up to ed to use the lamp then becauseof ablesupplyvoltagerangeis the ten times can be obtained with ICla in its elegant simplicity and extremeheartof everyIISR meter. readily-available, low-cost operaly low distortion figure (O'0O25o/o). the new circuit,seeFig. 4, is the tional-amplifier ICs. With the new design I wanted to in thisdesign. sinewave oscillator Remember that if the rate-ofrvayof generating a reduce the operating voltage of the change of the voltage is fast enough, The sinrplest will provide negliany capacitance (Xc) to it. g i b l eo p p o s i t i o n

/o

Decembe2 r 0 0 4T E L E V I S I O N

whole meter so that a single PP3 battery could be used as the power source while still achieving very low power [Link] problem with the use of a bulb is that in these conditions the bulb's temperature is not sufficiently far away from the ambient temperature to ensure good stability. So this time I decided to use the brute-force method of diode stabili s a t i o n( D 1 . D 2 ) . The idea here is that when the output from the oscillator rises above the conduction point of the diodes the negative feedback increases, the output settling at an amplitude which dependson the characteristics of the diodes. In this case the net result is a sinewave source signal acrossR-5 with an amplitude of about 6rnV peak-to-peak. Diode stabilisation introduces greater distortion. but this is not important here. To maintain oscillation. the value of R3 must be over twice that of R4. A preset resistor, adjusted to just sustainoscillation (lowest distortion), is usually used in the R3 position. I decided to use a fixed value that's a few ohms on the high side. to ensurereliable oscillation regardless of distortion.

lnierfoce with the copocitor being lested

The interfacewith the capacitor being tested is the crucial part of the [Link] took me a long time to get this right. See Fig. 5. The waveform across R5 (the source resistor) is not an ideal, constant-voltage source, becauseof the need to include the sample resistor (R6). whose value must be comparable to the ESR values in which we are [Link] ESR of the capacitor being tested forms part of a potential divider with R6. Thus if, for example, a good 1,0004F capacitor with an ESR of about 0'lQ is connectedfor test.R6 is effectively in parallel with R5. This means that the supply-signal source is less because,when a capacitor is being tested, the constant-current source to R5 is shared with the ESR and R6 in parallel. The voltage waveform developed acrossR6 as a result of the current through the capacitor being tested is amplified and then detectedby the rest of the circuit. If the ESR of the capacitor being tested is equal to the value of R6. half of the supply waveform will be passedon. The supply waveform is not independent of the load however. If the ESR is less

than the value of R6. as in most casesit is. the wavefbrm voltage acrossR6 increases. As the ESR rises above the value of R6. the latter becomes less eff'[Link] result of all this is a non-linear scale,expandedat the lower range and somervhat logged out. This is ideal for the p r e s e n ta p p l i c a t i o n . because in s o m e c a s e si l i s i m p o r t a n tt o b e able to distinguish betrveena very low value, close to zero, and one of about 0'5Q.

lnlernol view I have achievedlow current conof the new sunrptionand circuit sinplicity by ESR meter. using the feedback current and The test leod compromisingsomervhat on the connections ore ol ideal, constant-voltagesource. ihe boltom, with protection diodes Next month between lhem. In the concluding instalrnent next m o n t h I ' l l c o m p l e t et h e c i r c u i t d e s c r i p t i o nd . e a l w i t h s o m ep r a c t i cal points and protectiorrrnethods, provide a detailed comporrentslist I and a stripboard lavout.

6mVp-punder 0.6Vp-p Constant open-circuit (lCla pin2) current conditions I

t
Constant-voltage source (fromosc. feedback loop)

I
/ R+ roosl

I I

rsn

Xc=o

H5

1o

R6 2"7{2

Fig. 5: The method of test capacitor intertace used in this meter. The source voltage is not a true constant voltage because of the need for R6, whose value must be comparable to the ESB of the capacitors in which we are interested. Current economy and circuit simplicity are achieved by using feedback current from the oscillator circuit as the source fed to the capacitor under test. The output is inversely'logged' out in relation to ESR.

T E L E V I S I OD Ne c e m b e r 2004

79