ESR Meter Adapter Design and Construction

[The Electrician]

"Be a true ESR meter as oppose to an impedance meter."

That will depend on how you define a "true ESR meter".

I would define it to mean "a meter which displays the real part of the capacitor's impedance, as measured at a single frequency with sinusoidal excitation".  This is the desired functionality, and it is what professional grade impedance analyzers do.

Newer commercial units generally don't use that approach: they are mostly D/FFT driven or follows 3/4 parameter 1057 algorithms.

Could you please provide a link to such an instrument?  I'm unaware of them, and I'd like to have a look at the performance specs.

[dannyf]

"a meter which displays the real part of the capacitor's impedance, as measured at a single frequency with sinusoidal excitation"

I got the first part. Not sure why one would care how it is measured.

As to meters, Agilent makes a few network / impedance analyzers and I am sure others do (Hioki?)

Analog has had a few such chips that integrate signal generation, adc + dft on a single chip, with limited capabilities at both ends of the measurement range. They just announced a new chip that further integrates a cortex-m3 + associated afe together. No detailed datasheet yet but it would be interesting to see what can be done there.

[The Electrician]

"a meter which displays the real part of the capacitor's impedance, as measured at a single frequency with sinusoidal excitation"

I got the first part. Not sure why one would care how it is measured.

Because if it is measured with a square wave or pulse as most of the low cost "esr" meters do, that's not measuring impedance as it is understood in network analysis.  It's measuring some summation of "impedances" at the fundamental and harmonics of the applied excitation.

As to meters, Agilent makes a few network / impedance analyzers and I am sure others do (Hioki?)

What I see from Agilent using FFT techniques are combination VNA/impedance analyzers.  I'm aware that recent VNAs use FFT Techniques; my R&S portable VNA does.  Those instruments don't have the dynamic range a swept sine type analyzer has.  Hioki doesn't use FFT techniques in its analyzers.  The best performing impedance analyzers for frequencies in the low MHz range use swept sines and don't try to also be VNAs.

I've seen some special purpose impedance analyzers aimed at the audio business.  They use sound cards for excitation and FFT techniques to produce the results, but here wide dynamic range isn't needed.

[dannyf]

if it is measured with a square wave or pulse as most of the low cost "esr" meters do, that's not measuring impedance as it is understood in network analysis.  It's measuring some summation of "impedances" at the fundamental and harmonics of the applied excitation.

That's not necessarily true.

Synchronized detectors (typically runs off a square wave), for example, are true measurement of the real element of the impedance - with the caveats with discussed that are entirely due to implementation, not fundamental to the approach itself.

A pulse excitation can also be used to measure both real and imaginary parts of the impedance - think of it as a Dirac excitation on the system, and by analyzing how the system reacts to the excitation, you can indeed measure the impedance. Mathematically, a Dirac excitation is all you need to know to describe any linear system.

[BravoV]

@The Electrician, as requested, the measurement on the red cap at the ESR meter adapter, also results using the LCR meter MS5308.

First, the DMM was zeroed using one of the cap pin shorted across the two beefy measurement probes.




The red cap measurement result, and I didn't use 4 wires since the results were very close when using these two gold plated beefy banana connectors which conduct very good and also really near the circuit. 4 wires only needed when I use long wired alligator clips.




The red cap in series with a 100 mili Ohm resistor, to verify the resistance measurements consistency.





Results on the red cap using the LCR meter MS5308 at various frequencies.

[The Electrician]

Thank you, BravoV

That's a very good result.  6 milliohms for the Jay_Diddy_P meter, versus 9 milliohms for the MS5308.  The sweeps I show for the 8 remaining caps I have, suggest that 9 milliohms is likely a correct result.

Jay_Diddy_P's meter clearly distinguished the true ESR from the 239 milliohm impedance  He achieved his goal.

For comparison, the Atlas ESR70 reads 20 milliohms for this capacitor, and the MESR100 reads 89 milliohms

[Jay_Diddy_B]

Hi,

I have also done some testing with a film capacitor. I do not have the 'official' 6.8uF film capacitor. I used a 4.7uF 250V Metalized polypropylene film capacitor.

Here is a picture of the capacitor used for test:



This is has an extremely low ESR. I first measured the capacitor using my HP4274A LCR meter with a HP 16047A test fixture. Here are the measurements:



Note: That the device impedance is very reactive, >89.6 degrees at 100 kHz

Here a couple of sample pictures of the measurements:







When I connected the capacitor to the ESR adapter using the 4W configuration, I zeroed the meter using the relative feature on  the DMM.
I measured 5.355mV which corresponds to 53 m Ohms.



This is 0.26% of the ESR meter adapters full scale, and significantly lower than the impedance of the capacitor 340.7 m Ohms.



Regards,
Jay_Diddy_B

[The Electrician]

When looking at BravoV's results, I forgot that the output from your meter is not a one-to-one correspondence of millivolts to millohms.

His result was 61 millohms, not 6 millohms.

Still pretty good, though.  A lot better than just displaying the "impedance".

[BravoV]

Yeah, my mistake not mentioning its 61 instead of 6 miliohms, sorry.

With the full scale of ESR measurement capability max at 20 ohm, the resolution and accuracy at 100 miliohms as demonstrated in my measurement means it has 0.5% accuracy. By looking at the circuit cost and relatively simple to build, I think this performance is very good indeed.

[BravoV]

The main thing I see, if you can arrange the switches to come on for the central part of the square wave in each half cycle, the circuit will not see the effects of the caused by the inductance. I am trying to think of a simple way to generate the required signals.

Here is the model:



The switches S3 and S4 are turned on in the center of each half cycle. This is timing required for the switches:



The result:



The circuit is no longer sensitive to the inductor. The capacitive part is left, but it averages to zero.

Re-reading the fine thread, bumped on this again.

Academically, I am very-very curious and really like to build this someday just for fun. 

[AJ4OM]

Does this design to in-circuit readings?

[BravoV]

Does this design to in-circuit readings?

Yes.

[3roomlab]

hmmm so to trigger in between edges or blanking. use a 1Mhz clocking and trigger off from a decade counter? or maybe even 2Mhz with 2 decade counters?

also i got kinda lost looking at the HP readout chart vs the ESR adapter 5.3mv ... what is going on there?

[Jay_Diddy_B]

also i got kinda lost looking at the HP readout chart vs the ESR adapter 5.3mv ... what is going on there?

The impedance of a pure capacitor at frequency F can be expressed as:

Xc = 1/(2 x Pi x F x C)

So an ideal 4.7uF capacitor at 100 kHz has an impedance of:

Xc = 1/ (2 x Pi x 100,000 x 4.7E-6) = 339 mOhm

This is a vector quantity and can be written as:

Z=339 mOhm, Angle = -90 degrees

or

Z= -339m  jOhm


The capacitor has a resistive component, ESR, which is at 90 degrees to reactive component.

*** See later post. There are some maths errors. JDB ***

The total impedance


Z = 4m - 339mj Ohms

The magnitude is

|Z| = Sqrt ( R² + Xc²)

|Z|= Sqrt (4m + (339m)²) = 334.8 m Ohms

The phase angle = ArcTan -339m / 4m = -89.32 degrees

If the ESR meter was to read impedance, it would read |Z| which is 334.8 m Ohms.

If the ESR meter could read the ESR accurately it would read 2m Ohms.

The ESR adapter reads 53.55 m Ohms

Note: That is a particularly difficult ESR measurement. A 4.7uF electrolytic capacitor will typically have an ESR around 1 Ohm, which is much easier to measure.


Jay_Diddy_B

[3roomlab]

thanks for the very nice breakdown Jay

i understand the first part, ideal Z of the 4.7uF cap. straight forward equation.

i got lost again at the second equation, (R²+Xc²), the value of R comes from ... ? (but if R² = 4m, wouldnt R = 0.063? ... yep im really lost    hahaha)

i kind of understand where the equation is getting at, but i dont understand where the value 4milliohm (4m) is taken from? (as usual, its something right there that i couldnt comprehend lol)

i keep thinking there is something about the casing in bravoV pic ... OF COURSE ! its 3D printed ! wonderful arnt they? 3dprinting ... no longer would you need to hunt for a matching box ...

[Jay_Diddy_B]

thanks for the very nice breakdown Jay

i understand the first part, ideal Z of the 4.7uF cap. straight forward equation.

i got lost again at the second equation, (R²+Xc²), the value of R comes from ... ? (but if R² = 4m, wouldnt R = 0.063? ... yep im really lost    hahaha)

i kind of understand where the equation is getting at, but i dont understand where the value 4milliohm (4m) is taken from? (as usual, its something right there that i couldnt comprehend lol)

Hi,

The equation (R²+Xc²) comes from vector maths. The impedance of the capacitance is at 90 degrees to the resistance so you use Pythagoras' theorem to obtain the impedance.

The measured value of the ESR, the resistive part is 2m Ohms, (from the HP 4274A LCR meter).

I made a some mistake in the original calculations.

The impedance should be:

|Z| = sqrt ((2m x 2m) + (339m)(339m)) = 339m

This is because 2E-3 x 2E-3 = 4E-6, very small

The angle should have been:

ArcTan(-339m/2m) = 89.66^o  This agrees with the HP 4274A meter now 


This just demonstrates how hard it is to measure the ESR of the film capacitor.

Sorry for the mistakes in the math.

Jay_Diddy_B

[dannyf]

His math is messed up.

ESR=4m is presummed / measured earlier on that capacitor - it is very close to the 1-2mohm list earlier.

(serial) Impedance of the (ideal) capacitor is 339ohm, per calculation on the (presummed / measured?) capacitance. It is 90 degrees away from the ESR.

So the "combined" impedance of the ESR + ideal capacitor = squrt (4m^2 + 339m^2) = 339.0236mohm -> Xc so overwhelms the ESR.

[The Electrician]

Jay_Diddy_B, here's something you may want to try on your 4274.

I've noticed that even the best professional LCR meters begin to have trouble making an accurate measurement of the smaller part of an impedance when the ratio of the larger part to the smaller part begins to exceed 1000.  An example of this is when measuring the ESR of a very good quality polypropylene capacitor.

It would be good to have a way to test a meter, but with capacitors there's no way to be sure just how low the ESR is at low frequencies (which is where that ratio gets large (this is the Q in the case of a capacitor).

An alternative is to measure the inductance of a wirewound resistor.  A high value wirewound resistor measured at a low frequency provides a test.  Here's a picture of a suitable resistor I used.  It is a 10k ohm resistor, wound with very small diameter resistance wire.  This wire is so small that skin effect will not be noticeable up to several MHz:



Here's the result of a sweep of this resistor on the impedance analyzer.  The real part of the impedance is shown in yellow, dead flat at 10000 ohms.  The imaginary part (X) is shown in green and shows the expected behavior as we go down in frequency until we reach about 10 kHz, the frequency where Rs is 1000 times larger than X.  Below 10 kHz, the reactance curve is no longer the straight line we would expect--the meter is having trouble making the measurement.



This resistor has no ferromagnetic core and we can be sure that the inductance does not rise by several orders of magnitude at low frequencies, so if we plot the inductance vs. frequency over a range from 100 Hz to 1 MHz, the result should be a flat line.You can see that the "measured" inductance rises drastically at low frequencies.  Notice that the inductance begins its drastic rise about the time that the ratio Rs/X reaches 1000.  At larger ratios, the meter can't make an accurate measurement.

[dannyf]

when the ratio of the larger part to the smaller part begins to exceed 1000.

When the phase angle approaches 90 degrees, the calculation becomes tricky, particularly for 24-bit floating point math, and algorithm that uses tan()/atan(): the derivative of tan() at 89.9999 for example is close to 600K. A tiny bit of rounding will kill the precision.

If the integration is done in hardware, a tiny bit of timing off from turning on / off those switches will throw off the calculation.

This is where DFT/FFT or 1057 type algorithm is powerful - they are practically immune to that.

[The Electrician]

when the ratio of the larger part to the smaller part begins to exceed 1000.

When the phase angle approaches 90 degrees, the calculation becomes tricky, particularly for 24-bit floating point math, and algorithm that uses tan()/atan(): the derivative of tan() at 89.9999 for example is close to 600K. A tiny bit of rounding will kill the precision.

If the integration is done in hardware, a tiny bit of timing off from turning on / off those switches will throw off the calculation.

This is where DFT/FFT or 1057 type algorithm is powerful - they are practically immune to that.

Current Professional meters don't just use 24 bit floating point arithmetic; they use a decent processor with IEEE floating point--56 bit double precision, 80 bit extended precision.

The derivative of the tangent at 89.9999 degrees is 3.3*10^11; the tangent itself of 89.9999 degrees is 572958 (close to 600k).  No LCR meter tries to measure a phase angle of 89.9999 degrees.

Professional LCR meters don't use gated analog switches for phase detectors.  The schematics I've looked at show double balanced mixers with very accurate zero degree and 90 degree references.

[The Electrician]

The Wayne-Kerr 6440B is described by the manufacturer as able to measure very low D capacitors.

I did a sweep of the inductance of the same wirewound resistor as in the previous thread.  One can see that the 6440B doesn't begin to have problems until the frequency is below 1 kHz, which is about 10 times lower than the Hioki.  A good measurement is obtained even when the larger part of the impedance is 10,000 times the smaller part.  Note that the vertical scale is not logarithmic, but the frequency axis is:



Not many LCR meters can do this.  However, the 6440B is a low frequency instrument only; no RF measurements with it.

[Bryan]

Hi TorqueRanger and the group,

I have attached pdfs that should help DIY construction of the ESR meter adapter.

There is copper layer which has been mirrored for toner transfer. I have included a scale so that you can make sure the board is the correct size.
There is pdf showing the silk screen layer so you know where the components go.
I have also attached the latest version of the schematic.

Jay_Diddy_B

Hi Jay_Diddy_B:

I have a couple build questions. Is this the latest schematic?

What is the purpose of the potentiometer?. Is it to zero the DMM if the meter does not have a Relative feature.?

Can the LT6241CS8 be substituted with the  LT1498CS8. The LT6241CS8 is not available from Newark.

Are you able to provide the source for the PCB board, not sure what program you used. but would like to modify the PCB board. 

Cheers and thanks.
 

[xwarp]

Has anyone built one of the two shown in the quote below?

I'm interested in bread boarding one for fun, but can't make sense of some of the things pictured.

Oldway stated in a later post that ESR4 is updated to correct some things in ESR3, but unless I am reading something wrong, the dates on the schematics show that ESR3 is a later version than ESR4.

Also, ESR4 doesn't have the 10/100 khz switch as in ESR3.

Also, where do the leads connect in for testing a cap and where is the output from the unit to the meter?

Apologies if these are dumb questions.

My ESD tester works well.
Thank you to Jay_Diddy_B !

I made some changes: (see diagram ESR3)
- Test frequencies of 100 kHz and 10 Khz
- Only 2W...4W is not practical for such an ESD tester.
- Offset and the resistance of test cables are compensated  by setting to zero with a 20K potentiometer. It's no longer needed to use the "relative function" of the multimeter. (cheap 5 bucks multimeter can be used )
-choice of easy to buy and low cost components.
- fully protected to 100V (seems to be enough)
The PCB is single sided and the components are SMD.
Only one jumper was needed.
Consumption is about 7 mA.

I'am preparing a more sophisticated version without problem of delay and two ranges, one from 0 to 20R and a more accurate one from 0 to 2R. (see diagram)

[Shock]

Anyone have a spare or populated PCB made up?

[oldway]

As nobody has reacted anymore, I thought this was a dead topic and i did not make updates anymore.

Schematic ESR4 is the good one but modifications has still been made.

As schematic and layout must be consistent, I have not placed on the schematic the components who are not mounted on the board as potentiometers, terminals, switches, ...

That's the reason why you have some difficulties to understand this schematic.