East Tester ET4410 ESR Measure

[mawyatt]

Any quality somewhat larger value Polystyrene cap should suffice for experimenting and duplicating the high "Q" issue with ESR.

Here's an surplus example from TRW:

https://www.surplussales.com/Capacitors/Poly-Unelco.html

Best,

[indman]

mawyatt,it's not a secret to me. I wanted to find out the marking of the capacitor that measured The Electrician

[Martin72]

In the next days I´ll measure a electrolytic cap on all the three, if there are remarkable differences between them.

Today a electrolytic cap on my ET4410, same cap then on monday with the lcrs on work.
It´s a low esr type from panasonic, 100µF/35V.

Results with 120hz, 1Vrms, series:

C: 102.96µF
Xc : -12.882 Ohm
D: 0.0208
Q: 47.388
Phase: -88.79°
ESR: 0.2699 Ohm (stable.. )

So far, so good, everything seems plausible, you can take the values and calculate, everything is coherant.
But I´ve choose this type of cap, because a spec sheet is avaible.
And the sheet "says" at 120Hz the D-factor will be 0.12 for the 35V type....
Of course there are tolerances, but appx 6-times less..
I trust the value in the spec more than the measure.
Taking the D-factor from the spec, taking 100µF, calculations are Xc -13.26 Ohm, Q 8.33, ESR appx 1.59 Ohm
The measured Xc seems plausible, so why are D and Q so different..
I think it got to do with the here already mentioned signal to noise ratio.
Actually I read the very good keysight handbook, makes things clearer for me - Thanks for the hint !

[TimFox]

Normally, the specified D value on a data sheet is a maximum, not a "bogie" value.  Was there a percent tolerance indicated on the data sheet?
I would look it up myself, but you didn't state a Panasonic part number.

[Martin72]

As far as I can see not for the tan-phi.
But maybe you see more:
https://cdn-reichelt.de/documents/datenblatt/B300/EEUFR_ENG_TDS.pdf

A few minutes ago I doubled the output voltage to 2Vrms, no effect.
Same when using the mode with 1.5V bias.

[TimFox]

On that data sheet, there is a specified tolerance for the capacitance value (+/- 20% when measured at 120 Hz), but no tolerance on the tan delta (D) value.
They do not control the actual value of D, which is not practicable, but they test for maximum D.
This is normal for a capacitor data sheet.

[TimFox]

If you are trying to test your meter, you might do the following:
The largest polypropylene capacitor in my hobby inventory is 20 uF.
Larger ones are available.  100 uF/400 V from Wima costs about $22.00 in singles, and is specified at only 10 kHz (due to its use in switching devices).
The 20 uF capacitor, measured at 120 Hz on my trusty DE-5000 reads "OL" on Q.
Assuming Q > 1000, the ESR < 0.066 ohms.
To increase the ESR to 1.59 ohms (as per your example) requires an external series resistor of about 1.5 ohms.
Connect such a network to your meter and check the readings at 120 Hz.

[Martin72]

Hi Tim,

Good advice, got several caps here at home with several materials build up, let´s see.

If you are trying to test your meter

Exactly this is the point.
When you don´t know the manufacturer, there´s no sheet avaible, you must trust your meter.
What the main parameters concerns I have no doubt about it.
In case of the DE5000 you shouldn´t have no doubt about it too, I´ve tested it so much in comparison to official calibrated ones, when you "only" want to know the LCR values, its more than enough with the DE5000.
But sometimes I need to know at least one of the "secondary" parameters and it would be good when you can trust your meter.

[The Electrician]

Here is an example of how extremely low loss capacitors can overwhelm even the best LCR meters.  On eBay are offers for very high performance capacitors from Ukraine left over from the USSR military.  I bought some of them and this is a .5 uF polystyrene capacitor.

Name the marking of this capacitor or show a photo of it, if possible. I'll look for the same one in my stocks..

They look this: https://www.ebay.com/itm/193196126203?hash=item2cfb62effb:g:FLgAAOSwZL5dwu1n
but mine are .5 uF

I also have some like this: https://www.ebay.com/itm/203495673318?hash=item2f61498de6:g:MFoAAOSwXtFiNb1g

Notice that both of these sellers are in Ukraine.

[The Electrician]

Hi Tim,

Good advice, got several caps here at home with several materials build up, let´s see.

If you are trying to test your meter

Exactly this is the point.
When you don´t know the manufacturer, there´s no sheet avaible, you must trust your meter.
What the main parameters concerns I have no doubt about it.
In case of the DE5000 you shouldn´t have no doubt about it too, I´ve tested it so much in comparison to official calibrated ones, when you "only" want to know the LCR values, its more than enough with the DE5000.
But sometimes I need to know at least one of the "secondary" parameters and it would be good when you can trust your meter.

For some reason the data sheet you found for the Panasonic cap has deviated from the industry standard.  Data sheets for small electrolytics nearly always specify that the given value of D is "max".

Have a look at any of these data sheets: https://www.rubycon.co.jp/en/products/al-radial-lead/series/

You should be able to trust your meter for electrolytics because their D won't ever get close to .001.  The one thing that leads to the DE5000's inability to get a good measurement at frequencies above 100/120 Hz is that the electrolytic is large value.  For example, when I try to measure a 33000 uF cap, I get OL on every frequency except 100/120 Hz.

[Martin72]

To increase the ESR to 1.59 ohms (as per your example) requires an external series resistor of about 1.5 ohms.
Connect such a network to your meter and check the readings at 120 Hz.

Tested it with the same electrolytic cap and with 2.7Ohm.
Q decreases, phase is under 80° and ESR value is exactly plus 2.7Ohm(2.96Ohm).

[The Electrician]

Page 41 of that manual shows the adjustment.  If that adjustment is taken into account, the behavior of the DE5000 is accounted for.

I've read this guide before and it introduces even more questions than answers!

Az is not an impedance value--it's a correction factor.

I should have said that it "attempts" to account for behavior.  I didn't read all that info carefully because I saw a couple of expressions that one would find in a list of correction factors, but upon reading carefully I see that they have not done a good job.  The manual for the ET4410 linked in reply #xx has a much better set of correction factors.

The DE5000 is quite good for ~$100, but it has some serious limitations.

For example I have an 1800 uF 16v cap from this series: www.nichicon.co.jp/english/products/pdf/e-hz.pdf

This is a very low ESR capacitor typical of the sort found on computer motherboards.

Have a look at the spec for impedance at 100 kHz; it's 6.5 milliohms and my Hioki measures it as 6.54 milliohms.  The Hioki measures the capacitance as -493.8 uF since 100 kHz is above the self resonance frequency.  For measurements of such a low ESR, it's important to insert the capacitor leads fully into the fixture.  If I insert the leads so that only the very ends of the leads are in the fixture, the ESR reads 7.78 milliohms.

If I measure it with the DE5000 at 100 kHz, the capacitance value is OL, the ESR value is ------, a bunch of dashes.  However, if I set the DE5000 to measure Rs, I get a value of .014, not .0065 as it should be, but at least it gets something. 

I have a some more handheld LCR meters to try.

The Tonghui TH2822C measures the capacitance at 100 kHz as -372 uF, and the ESR as 7.0 milliohms.  I had to do the short compensation to get that value.  The value of 7.0 milliohms is .5 milliohms higher than the expected 6.5 milliohms because the plastic case of the meter prevents the capacitor leads from being inserted flush with the contact blades.  This is excellent performance.

The Applent AT826 measures the capacitance at 100 kHz as -391 uF, and the ESR as 7.1 milliohms.  I did not have to do the short compensation with this meter.  This meter has the same problem as the Tonghui--the capacitor leads can't be inserted all the way into the contact blades, giving a reading a few tenths of a milliohm higher.

These two meters are quite a bit more expensive than the DE5000, but they perform much better.  The DE5000 performance is not bad for $100, but isn't there some old saying about getting what you pay for?

[TimFox]

When measuring such large capacitance values at such high frequencies, fixturing is critical to get an accurate result.
What exact fixture did you use on the DE-5000?  (By the way, I would not trust the ESR value under these extreme conditions on my DE-5000.)
I find it perplexing how many people are concerned about accurate measurements of such ESR values;  maximum values are normally more important.

[Martin72]

If I measure it with the DE5000 at 100 kHz, the capacitance value is OL, the ESR value is ------, a bunch of dashes.  However, if I set the DE5000 to measure Rs, I get a value of .014, not .0065 as it should be, but at least it gets something.

Doesn´t wonder me, the chipset the DE5000 use did not specify 100Khz as a testfrequency for such a "large" value.
In the 2000µF range, max. 1khz.

[TimFox]

You cheated by reading the datasheet.
It's like affecting a physical situation by doing a measurement.

[The Electrician]

When measuring such large capacitance values at such high frequencies, fixturing is critical to get an accurate result.
What exact fixture did you use on the DE-5000?  (By the way, I would not trust the ESR value under these extreme conditions on my DE-5000.)
I find it perplexing how many people are concerned about accurate measurements of such ESR values;  maximum values are normally more important.

The best that can be done with these handheld meters is to use the built-in four blade fixture.  The point of my post was to show that there are other handhelds that can do much better than the DE5000 under certain conditions.  The measurements the Tonghui and Applent got were within a few percent of the Hioki's result, unlike the DE5000's result.  But, of course they cost several times what the DE5000 cost.

100 kHz isn't such a high frequency for this measurement.

[Martin72]

You cheated by reading the datasheet.
It's like affecting a physical situation by doing a measurement.

Me ?

Please explain why.

[The Electrician]

If I measure it with the DE5000 at 100 kHz, the capacitance value is OL, the ESR value is ------, a bunch of dashes.  However, if I set the DE5000 to measure Rs, I get a value of .014, not .0065 as it should be, but at least it gets something.

Doesn´t wonder me, the chipset the DE5000 use did not specify 100Khz as a testfrequency for such a "large" value.
In the 2000µF range, max. 1khz.

I wasn't expecting the DE5000 to get a good value for the capacitance.  I only mentioned what the DE5000 got for the capacitance of the 1800 uF cap (which was OL) to compare it to what other meters get.  The other meters got a negative value which shows that 100 kHz is above the self resonant frequency; but they didn't just give up and say OL.

  What I want an LCR meter to do is be able to measure the ESR of a large capacitance at 100 kHz.  This shouldn't be a difficult measurement due to a low value of D (high Q).  Switching the DE5000 from capacitance measuring mode to the Rs measuring mode at least gives a value which is substantially in error.

Another thread from some time ago did some tests that showed that the DE5000 has difficulty making measurements when the voltage across the DUT is about 4 millivolts or less, which can be the case when measuring capacitors of this size capacitance.  That appears to be the case here.  But the Tonghui and Applent don't have that limitation.

[Martin72]

but they didn't just give up and say OL.

Hmm..What will be "better", don´t "giving up" and display anything or show clearly it´s not the range I can measure something like this...
The smart way would be to "obey" frequencies above(E.g., at 2000µF range testfrequencies up to 1khz max are possible), but this would require a manual mode which the DE5000 and many others don´t got.

[The Electrician]

but they didn't just give up and say OL.

Hmm..What will be "better", don´t "giving up" and display anything or show clearly it´s not the range I can measure something like this...

Did you not see the emoticon following that sentence ( )?

The smart way would be to "obey" frequencies above(E.g., at 2000µF range testfrequencies up to 1khz max are possible), but this would require a manual mode which the DE5000 and many others don´t got.

[TimFox]

If I measure it with the DE5000 at 100 kHz, the capacitance value is OL, the ESR value is ------, a bunch of dashes.  However, if I set the DE5000 to measure Rs, I get a value of .014, not .0065 as it should be, but at least it gets something.

Doesn´t wonder me, the chipset the DE5000 use did not specify 100Khz as a testfrequency for such a "large" value.
In the 2000µF range, max. 1khz.

I wasn't expecting the DE5000 to get a good value for the capacitance.  I only mentioned what the DE5000 got for the capacitance of the 1800 uF cap (which was OL) to compare it to what other meters get.  The other meters got a negative value which shows that 100 kHz is above the self resonant frequency; but they didn't just give up and say OL.

  What I want an LCR meter to do is be able to measure the ESR of a large capacitance at 100 kHz.  This shouldn't be a difficult measurement due to a low value of D (high Q).  Switching the DE5000 from capacitance measuring mode to the Rs measuring mode at least gives a value which is substantially in error.

Another thread from some time ago did some tests that showed that the DE5000 has difficulty making measurements when the voltage across the DUT is about 4 millivolts or less, which can be the case when measuring capacitors of this size capacitance.  That appears to be the case here.  But the Tonghui and Applent don't have that limitation.

Measuring the ESR of a large capacitance with a high Q (low D) is, in fact, a very difficult measurement.  [Emphasis added in bold face above to original comment.]

[The Electrician]

If I measure it with the DE5000 at 100 kHz, the capacitance value is OL, the ESR value is ------, a bunch of dashes.  However, if I set the DE5000 to measure Rs, I get a value of .014, not .0065 as it should be, but at least it gets something.

Doesn´t wonder me, the chipset the DE5000 use did not specify 100Khz as a testfrequency for such a "large" value.
In the 2000µF range, max. 1khz.

I wasn't expecting the DE5000 to get a good value for the capacitance.  I only mentioned what the DE5000 got for the capacitance of the 1800 uF cap (which was OL) to compare it to what other meters get.  The other meters got a negative value which shows that 100 kHz is above the self resonant frequency; but they didn't just give up and say OL.

  What I want an LCR meter to do is be able to measure the ESR of a large capacitance at 100 kHz.  This shouldn't be a difficult measurement due to a low value of D (high Q).  Switching the DE5000 from capacitance measuring mode to the Rs measuring mode at least gives a value which is substantially in error.

Another thread from some time ago did some tests that showed that the DE5000 has difficulty making measurements when the voltage across the DUT is about 4 millivolts or less, which can be the case when measuring capacitors of this size capacitance.  That appears to be the case here.  But the Tonghui and Applent don't have that limitation.

Measuring the ESR of a large capacitance with a high Q (low D) is, in fact, a very difficult measurement.  [Emphasis added in bold face above to original comment.]

I worded what I was saying in an ambiguous way.  Let me rephrase it; replace that sentence with this one:

"Since in this case (of a large electrolytic) we won't have a low value of D (high Q), this shouldn't be a difficult measurement."

[mawyatt]

One should have a good handle on how these LCR meters work and the ranges that are within "acceptable" limits of such. Some of the higher end LCR meters display (selective) DUT current and applied voltage, these parameters are beneficial to access the viability of the DUT measurement.

One should also consider the impedance magnitude involved, a few milliohms will be difficult to accurately measure and requires a large measuring DUT current which a handheld battery powered device may have difficulty supplying.

If you consider an ideal 1000uF produces just 0.0016 ohms at 100KHz, with a DUT current of 0.1 amp this only produces 160uV across the DUT. However, a more practical measurement at 100Hz produces 1.6 ohms and 160mV across the DUT at 0.1 amp.

The DE-5000 we have produces just ~5ma maximum available current which limits the ability to sense low impedance magnitudes, the better LCR meters produce much more available current for measurements.

Best, 

[The Electrician]

One should have a good handle on how these LCR meters work and the ranges that are within "acceptable" limits of such. Some of the higher end LCR meters display (selective) DUT current and applied voltage, these parameters are beneficial to access the viability of the DUT measurement.

One should also consider the impedance magnitude involved, a few milliohms will be difficult to accurately measure and requires a large measuring DUT current which a handheld battery powered device may have difficulty supplying.

If you consider an ideal 1000uF produces just 0.0016 ohms at 100KHz, with a DUT current of 0.1 amp this only produces 160uV across the DUT. However, a more practical measurement at 100Hz produces 1.6 ohms and 160mV across the DUT at 0.1 amp.

The DE-5000 we have produces just ~5ma maximum available current which limits the ability to sense low impedance magnitudes, the better LCR meters produce much more available current for measurements.

Best,

The first thing I said in reply #20, my first post in this thread was: "Your problem is undoubtedly a matter of signal to noise ratio."

I've been harping on this for the rest of my posts, and it's good to have some older, experienced EEs confirm.

It hasn't been mentioned yet, but one of the limitations of the fact that the DE5000 has integrated the entire functionality of an LCR meter is that compromises were necessary.  The IC process they had to use was not the best suited for low noise performance.  A meter not fully integrated can use a low noise opamp, or even a fully discrete design, for the front end gain stage.

The marketing literature for the Hioki top end meters, and the Keysight E4980A meter, describe how their engineering teams expended considerable effort to reduce the front end noise.  Here's some of the Keysight description of how the noise reduction affects low impedance measurements:

[mawyatt]

It hasn't been mentioned yet, but one of the limitations of the fact that the DE5000 has integrated the entire functionality of an LCR meter is that compromises were necessary.  The IC process they had to use was not the best suited for low noise performance.  A meter not fully integrated can use a low noise opamp, or even a fully discrete design, for the front end gain stage.

The marketing literature for the Hioki top end meters, and the Keysight E4980A meter, describe how their engineering teams expended considerable effort to reduce the front end noise.  Here's some of the Keysight description of how the noise reduction affects low impedance measurements:

No doubt the chip set utilized within the DE-5000 is CMOS since it must support significant digital as well as the analog front end. Enormous strides in CMOS noise levels have been achieved over the decades including some clever signal processing to "help" with the noise. Apparently the chip set utilized is two separate chips, suspect one is analog and other is digital and they may not be produced in the same CMOS process. From experience I can tell you this is no trivial feat to implement an entire LCR meter within a couple chips, even if only modest performance....so hats off to the chip set designers!

If you check the latest CMOS Op-Amps from TI and AD you'll find some pretty good noise performers, even tho they are CMOS. This is a result of superb design, clever techniques and a good CMOS noise friendly process that's likely proprietary and not really available to the outside. Of course there are excellent bipolar Op-Amps available also, so a LCR design team would have the luxury of selecting the best Op-Amps for noise performance, the best SD ADC, the best DDS and so on, whereas the custom IC LCR designers were quite restricted and thus performance is compromised.

One area that might be beneficial to the custom chip set LCR meter types, where one could significantly "Boost" the output current for lower level impedances, which as you correctly indicate would improve the SNR. A high speed/current discrete buffer amp or something like a OPA-633 could extend the range of the chip set significantly, but you would still want this in a mains powered LCR meter, not a handheld battery powered since the power consumption would be high (why our Hioki has a very loud fan, the TH2830 fan is not as bad tho, but it only goes to 100KHz).

Anyway, that's a nice document from KS, thanks for showing!!

Best,