ESR-meter problems

[ChildOfVision]

Hi! First of all, sorry for my horrrrrible English!
Can anybody explain me why would some small ceramic capacitor in parallel with CUT causes problems with some ESR-meters? Allegedly, some, say 100kHz ESR-meters can't cope with such a situation. On the other hand, some others that use "series of short pulses" or something like that, can measure ESR of that capacitor-configuration. I tried to contact via PM Hero999 who wrote something about, but he didn't answer (yet). So if somebody knows something about this problem please let me know (some link will be sufficient)!

Thanks in advance!

[amspire]

I am not sure they cause problems.

It is just that when capacitors are in parallel, then their ESR's are in parallel. So if you have a 4.7 ceramic capacitor in parallel with a 10uF electrolytic, then you would probably read less then 1 ohm ESR, even though the electrolytic may be bad with a 100 ohms ESR.

The ESR reading is correct - it is correct for two capacitors in parallel which probably is not what was wanted.

In this case, you would have to remove the electrolytic to test.

There is always an issue when testing components in-circuit while they are connected to other components. It often does not work well.

Richard.

[ChildOfVision]

Yes, you are completely right! Of course, there is no such a meter that is capable to measure ESR of only one particular capacitor if ANY other cap is connected in parallel. But there is something else, I really can't remember where I read this: now I'm in doubt: maybe I just misunderstood something - if I think better all ceramic capacitors have very low ESR - true? If so, there is no "problem", you have simply no choice but isolate CUT and measure ESR! However some authors specifically highlight ability of their ESR-meters to cope with "that problem" - see example (note 9th "point") so I'm a little confused!

http://members.upc.hu/lethanh.hung/LCFESRmero/en/

[amspire]

All ESR meters have to keep the test voltage low.  The 300mV in point 9 is really too much as some Schottky diodes can turn on below. that. 150mV peak to peak test voltage is as high as the test voltage should go. 100mV would be better..

Also  you can test in parallel with small ceramics.  Anything below 1uF will have a minimum total impedance of several ohms or more. So you can easily test a 100uF capacitor in parallel with a 0.1uF ceramic - the ceramic will not be noticeable.  Once the ceramic capacitors get over 1uF, they can definitely confuse the readings.

[ChildOfVision]

But, ESR-meters supply AC test signal to CUT and then measure voltage drop across it, neglect AC-part (reactance) - remains DC resistance - true?
If so, and if I think well, if we have two capacitors, say, 100uF and 1nF, meter should see the sum of two parallel resistors, say, 100uF=>10mOhm and 1nf=>1mOhm - sum=approx 0,9mOhm, correct me if I wrong. Maybe meters that use "short pulses method" do some kind of trick, so that test signal have no time to "reach" smaller cap or something like that, I'm not an expert?

[BravoV]

All ESR meters have to keep the test voltage low.  The 300mV in point 9 is really too much as some Schottky diodes can turn on below. that. 150mV peak to peak test voltage is as high as the test voltage should go. 100mV would be better..

Richard, since you mentioned the test voltage, this really rings a bell, just curious and like to know from your experience, so say if an esr meter can be adjusted down at it's pulsing test signal down to 100 mV peak to peak, the chance of biasing modern semiconductor assuming they're silicon based is almost none, correct ?

Now, just want to make sure if this voltage level is low enough for an in-circuit cap testing which might have a really old semiconductors like germanium type instead of silicon that is connected parallel with it ?

Its just I'm planning to make one of those simple diy analog esr meter for my friend which is an antique radio collector, he just starting it.

[amspire]

100mV is very safe.  You would be unlucky to have a problem with 200mv peak-peak, but at 100mV, I can't think of anything that could cause a problem.

Germanium devices have a more gradual turn on then silicon so they do start conducting mush sooner then silicon, but at 100mV or less, we are still talking uA of current.

I also cannot think of any reason you would find an electrolytic cap in parallel with a germanium junction - diode or transistor - in 60's radio gear. I don't think you will ever see it happening. A zener diode often has an electrolytic capacitor in parallel, but I have never heard of a germanium zener diode.

You will be fine.

Richard

[BravoV]

100mV is very safe.  You would be unlucky to have a problem with 200mv peak-peak, but at 100mV, I can't think of anything that could cause a problem.

Germanium devices have a more gradual turn on then silicon so they do start conducting mush sooner then silicon, but at 100mV or less, we are still talking uA of current.

I also cannot think of any reason you would find an electrolytic cap in parallel with a germanium junction - diode or transistor - in 60's radio gear. I don't think you will ever see it happening. A zener diode often has an electrolytic capacitor in parallel, but I have never heard of a germanium zener diode.

You will be fine.

Richard

Actually I never seen or experienced at any old those old radio circuit, or used germanium semiconductor before, honestly I have no idea at all.

Because once the meter built and tested with scope that it has the right test voltage, it will be sent hundreds km away, and once its there, it ain't easy to help him if its is too high. Yeah, he just learning in electronic & planning to test those old electrolytic caps at the power supply section in those radios. 

Thanks for the confirmation.

[jimmc]

But, ESR-meters supply AC test signal to CUT and then measure voltage drop across it, neglect AC-part (reactance) - remains DC resistance - true?
If so, and if I think well, if we have two capacitors, say, 100uF and 1nF, meter should see the sum of two parallel resistors, say, 100uF=>10mOhm and 1nf=>1mOhm - sum=approx 0,9mOhm, correct me if I wrong. Maybe meters that use "short pulses method" do some kind of trick, so that test signal have no time to "reach" smaller cap or something like that, I'm not an expert?

The total impedance of a capacitor is made up of its reactance (=1/(2.Pi.F.C) in series with its ESR so at 100kHz the 1nF capacitor will have an impedance comprising  of 1.6k reactive in series with 1mOhm resistive.
For the 100uF the impedance  will be 16mOhm reactive in series with 10mOhm which is about 19mOhm resistive.
The ESR meter will measure the 19mOhm (Z= SQRT(X² + R²).
This is the reason why ESR meters use a relatively high frequency - to make the capacitive reactance small. You can't use an ESR meter on a capacitor of less than about 1uF (reactance = 1.6Ohm.
More complex meters use a phase sensitive detector to distinguish between ESR and capacitive reactance.

It relatively easy to design an ESR meter with a maximum signal of <10mV.
Several designs use general purpose Op-Amps as high gain amplifiers, these have a gain- bandwidth product of around
3MHz (eg TL081) which limits their open-loop gain at 100kHz to 30!
A simple discrete amplifier can easily out perform them enabling lower drive voltages to be used.

Jim

[alm]

A simple discrete amplifier or a less ancient op-amp, not sure if I see a reason to go discrete here.

[ChildOfVision]

The ESR meter will measure the 19mOhm (Z= SQRT(X2 + R2).

Sorry, but I don't understand!
Z is impedance (in Ohms);
R is (pure ohmic) resistance - i.e. ESR what we want to measure, right?
X is reactance (in Ohms) - i.e. complex AC resistance that depends on frequency, true?
I can't sketch this, but R and X are two vectors on complex plane under 90° angle, true?
So, if we have, say, 100uF capacitor, its reactance will be 0,0159 Ohm (or 16mOhm as you said).
Impedance will be 18,78 mOhm (if we assume that ESR=10mOhm), again as you said.
BUT IMO ESR meter WILL NOT (or better to say: SHOULD NOT) measure 19 mOhm! Isn't the main feature of every ESR-meter to separate X and R and to show on its display only value of R, i.e. ESR? It should show 10mOhm.
Sorry if I said something stupid, I repeat: I'm not some expert!

[jimmc]

The ESR meter will measure the 19mOhm (Z= SQRT(X2 + R2).

Sorry, but I don't understand!
Z is impedance (in Ohms);
R is (pure ohmic) resistance - i.e. ESR what we want to measure, right?
X is reactance (in Ohms) - i.e. complex AC resistance that depends on frequency, true?
I can't sketch this, but R and X are two vectors on complex plane under 90° angle, true?
So, if we have, say, 100uF capacitor, its reactance will be 0,0159 Ohm (or 16mOhm as you said).
Impedance will be 18,78 mOhm (if we assume that ESR=10mOhm), again as you said.
BUT IMO ESR meter WILL NOT (or better to say: SHOULD NOT) measure 19 mOhm! Isn't the main feature of every ESR-meter to separate X and R and to show on its display only value of R, i.e. ESR? It should show 10mOhm.
Sorry if I said something stupid, I repeat: I'm not some expert!

Your figures are correct, but not a practical example.
An normal electrolytic capacitor of 100uF will have an ESR of 100mOhm or more making the effect of the reactive component insignificant.  (A random example http://www.farnell.com/datasheets/2161.pdf)
Only a few specialised types have an ESR anywhere near 10mOhm.
Non electrolytic capacitors do not normally fail with high ESR, they don't dry out.

You will find most ESR meters actually measure  impedance.
As I said the use a relatively high frequency to minimise the reactive component.
100kHz is a compromise, a higher frequency would reduce the capacitive reactance but (parasitic) inductive effects would start to become a problem.

ESR meters are use to perform a quick test on capacitors, they are not a precision instrument; failing capacitors normally have an ESR a lot higher than expected.
Few of the simple designs around will distinguish between a short circuit and 10 or 20 mOhm.

Jim

[kripton2035]

elektor esr meter get rid of the reactance and display only the esr

details here:
http://kripton2035.free.fr/commercial%20esr/esr-elektorsept2.html

[ChildOfVision]

You will find most ESR meters actually measure  impedance.

Yes, that make sense! I understand now what you mean!

@Kripton2035: Dear Mr.Kripton: I think we have a common friend - he helps me a lot about building of ESR-meter, well, I'll just say he is from Canada. I have only partial PDF of that article: that meter have two PICs isn't it? Can you please give me a link to a full article? Also, can you tell me (I didn't find this info on my pdf) can this meter do "in-circuit" measurements, and what is the range and resolution of ESR?
I believe you know for member of many forums - Hero999 - I strongly believe that he posts something on some forum about "problem with small ceramic cap in parallel". If you know where is that post please let me know!

Regards!

[kripton2035]

see my post just before your post

[BravoV]

Friend of mine built that elektor version while ago, I might be wrong though, I think it can not be used for in-circuit cap testing since the test voltage is quite high, can't recall how high, but definitely its not capable.

[ChildOfVision]

@Kripton2035: I already bookmarked it !
   And what about mysterious "Hero999"?

@BravoV: As I could understand from my corrupted PDF - I know that meter is somewhat obsolete. Also, as much as I can recall, there are some extremely long time to measure big capacitors (over 5 min or so). Kripton, if this is not true, please correct us both!

I found recently some schematic of "simple" ESR DMM-add-on meter without any MPU, but it has too low resolution for me (10mOhm). It can also measure internal resistance of a battery. If somebody knows for such a type of meter, but with better resolution (1mOhm) - please!

EDIT: I just found on my partial documentation this sentence: "The pins marked Signal+, Signal–, Sense+ and Sense– are for connecting the additional ESR test leads with their separate sense lines in order to measure capacitors while they are still connected in a circuit..."

[jimmc]

Did you mean this thread? https://www.eevblog.com/forum/index.php?topic=2960.0

If you really want 1mOhm resolution then you are out of the realm of simple ESR meters and into vector impedance measurements with 4wire connections and phase sensitive detectors!

Jim

[ChildOfVision]

Yes, reply #9! On THIS forum. But he wrote about that issue on some other forum which I can not remember !

And who said that it must be simple? I even prefer measurements with Kelvin probes!

[jimmc]

A simple discrete amplifier or a less ancient op-amp, not sure if I see a reason to go discrete here.

Assuming that a diode rectifier is used somewhere it will need 1volt to drive it (in round numbers)
Using a maximum measuring voltage of 10mV and requiring linearity down to 1% FSD gives 100uV input (all voltages pk).
So gain required at 100kHz is 1/100u = 10000 giving a  closed loop GBW of 1GHz

Unless you're talking quite an exotic beast that's going to take two cascaded OP-Amps and require
care with layout and decoupling.

The simplest Op-Amp single supply amplifier I can think of has two resistors to set the gain, two for bias, 3 capacitors for coupling plus 1 for a HF roll off.

A two transistor feedback amplifier requires 6R and 4C to do the same job.
(Plus supply decoupling in both cases).

So why use Op-Amps?

Jim

[kripton2035]

Friend of mine built that elektor version while ago, I might be wrong though, I think it can not be used for in-circuit cap testing since the test voltage is quite high, can't recall how high, but definitely its not capable.

test voltage of elektor esr meter is 100KHz ±5v so too high for in circuit 

[kripton2035]

@BravoV: As I could understand from my corrupted PDF - I know that meter is somewhat obsolete. Also, as much as I can recall, there are some extremely long time to measure big capacitors (over 5 min or so). Kripton, if this is not true, please correct us both!

I dont like elektor c-esr meter for another reason : they use a 16f84 pic "only" to generate a 100khz signal
they could really use another method...
elektor esr (only) meter is quite precise, even if it doesnt measure in circuit
no in circuit means no use for a technician definitely...