East Tester ET4410 ESR Measure

[Martin72]

What is a "glimmer" capacitor?

The german word for "Mica", sorry for not translating that.

https://www.reichelt.de/de/en/mica-capacitor-10-nf-100-v--1--cy-22-5-10n-p42453.html?GROUPID=3158&START=0&OFFSET=16&SID=94c8b836af2ba4924e602b770031c41804c22b97609adc734fdca&LANGUAGE=EN&&r=1

These ones ans then 10nF.

[mawyatt]

A few days ago we ordered some high "Q" film capacitors since we had none on hand, these were some Polystyrene, Molded Mica, Dipped Mica and few of the Polyproplyene and a film mixture. We had some Polyester or Mylar types already, and various ceramic types including NPO/COG.

To evaluate these high "Q" types we had just received we use the Hioki IM3536, Tonghui TH2830 and DER-EE DE-5000 LCR meters. Attached in the page from our notebook to browse thru, and all measurements were done at 100KHz within each meter's range.

The 1nF Polystrene has a very high Q of ~10,000 (D of 0.0001) and even caused the IM3536 to flicker into a Display Out of Range condition which occurs at D of 0.0001. Although the IM3536 remained reasonably stable with all the tests, the TH2830 was't quite as good with a bit more randomness and not reading as high a "Q" as the IM3536 (averaging was employed to help stabilizes the IM3536 & TH2830 meters, the DE-5000 doesn't have this option setting).

You can see the good correlation between all meters in capacitance readings everywhere, but no so well with the ESR and subsequently D and Q with high "Q" types. At the bottom are the Mylar and Polypropylene types and the meters agree reasonably well with these lower Q capacitors as has been previously mentioned.

Best,

[indman]

Therefore I said not again...
We can discuss about in one of the DE5000 threads, here it´s off-topic.

Now it's allowed to mention DE-5000 again in this thread, several times in a row? Was it not long ago that I was forbidden to do this?

[Martin72]

I think you know how it´s meant. 

@mawyatt:
Interesting that even the hioki won´t be "flawless".
The Tonghui is the same LCR like the ST2830 we got.
I must read the ST manual again, also the one of the keysight - But i´m pretty sure that there are no remarks in it about the limitation of the ESR (Rs) measure.
Same on the ET4410 and of course DE5000.
So the conclusion could be that we can´t expect proper ESR/Q readings for every cap-type on the "cheapos" up to the costs of a TH/ST LCR...Hmm.



 

[indman]

So the conclusion could be that we can´t expect proper ESR/Q readings for every cap-type on the "cheapos" up to the costs of a TH/ST LCR...Hmm.

In this case, non-TH/ST LCR devices cannot be considered full-fledged METERS, but should be transferred to the category of AMATEUR DISPLAY METERS with random readings.

[mawyatt]

So the conclusion could be that we can´t expect proper ESR/Q readings for every cap-type on the "cheapos" up to the costs of a TH/ST LCR...Hmm.

In this case, non-TH/ST LCR devices cannot be considered full-fledged METERS, but should be transferred to the category of AMATEUR DISPLAY METERS with random readings.

That's totally ridiculous!!

Like saying that a HP34401, KS34465, DMM6500 are "AMATEUR DISPLAY METERS" because they can't read a waveform with very high peak to RMS values as well as the Keysight 3458A!!

These are instruments that must be used and understood properly rather than blindly applied, just like the high resolution DMMs mentioned 

Best,

[indman]

That's totally ridiculous!!
Like saying that a HP34401, KS34465, DMM6500 are "AMATEUR DISPLAY METERS" because they can't read a waveform with very high peak to RMS values as well as the Keysight 3458A!!
 

Not convincingly.
It is ridiculous, this is when the specification for the device specifies the range and the possibility of a full-fledged measurement of secondary parameters with a certain stability and reliability, but in fact this is not provided.
If this is stipulated in the documentation, as for HP34401, KS34465, DMM6500, then this is another matter 

[The Electrician]

A few days ago we ordered some high "Q" film capacitors since we had none on hand, these were some Polystyrene, Molded Mica, Dipped Mica and few of the Polyproplyene and a film mixture. We had some Polyester or Mylar types already, and various ceramic types including NPO/COG.

To evaluate these high "Q" types we had just received we use the Hioki IM3536, Tonghui TH2830 and DER-EE DE-5000 LCR meters. Attached in the page from our notebook to browse thru, and all measurements were done at 100KHz within each meter's range.

The 1nF Polystrene has a very high Q of ~10,000 (D of 0.0001) and even caused the IM3536 to flicker into a Display Out of Range condition which occurs at D of 0.0001. Although the IM3536 remained reasonably stable with all the tests, the TH2830 was't quite as good with a bit more randomness and not reading as high a "Q" as the IM3536 (averaging was employed to help stabilizes the IM3536 & TH2830 meters, the DE-5000 doesn't have this option setting).

You can see the good correlation between all meters in capacitance readings everywhere, but no so well with the ESR and subsequently D and Q with high "Q" types. At the bottom are the Mylar and Polypropylene types and the meters agree reasonably well with these lower Q capacitors as has been previously mentioned.

Best,

One of the things I like about the Hioki meters is their ability to display 4 parameters at once.  Try setting yours to match how I did it in reply #20.  Then choose the "set" mode, followed by "freq".  This allows you to quickly switch frequencies and watch the effect on all those parameters.  The polystyrene cap I used has its minimum D around 1 kHz; the Q is way over 10000, so the indication is "Disp out".  The capacitance value has 6 stable digits, but ESR has only 1 stable digit--all the less significant digits are bobbling like crazy if the speed is set to "slow".  This "bobbling" of the less significant digits is a clear indication that the reading is down in the noise because that is what bobbling digits are--noise.

The most stable and accurate readings can be obtained if the speed is set to "slow2", and the level is set to "5V".  Also setting averaging to 256 helps a lot for accuracy.  Without averaging, I get 1 stable digit for the ESR of the polystyrene capacitor, if I average for 256 times, I now have 3 stable digits, but their accuracy is still subject to the instrument specification.

So, Martin72, there's nothing wrong with your meters, you just have to know what it means that for some parameters like ESR, when the less significant digits are bobbling.  Those digits are simply measuring noise; the meter has been asked to make a measurement near its capability limit.  It's too bad the meter isn't smart enough to always know when to not show noise digits.  Your East Tester tries--only showing 2 digits for the ESR is its attempt to not mislead the user into thinking the result is better than it is.

[Martin72]

Hi,

I understand this, you, mawyatt, TimFox and other help to make clear the priniciples behind, now I KNOW it´s not a failure or somekind of mystic behaviour.
The conclusions are, that you could use the mentioned "cheaper" meters here well for all purposes, but you should have in mind that they couldn´t measure all possible cases, in this case caps with high Q, their ESR.
Even more expensive one can´t do that proper, it needs "more"..
To say oh if they can´t do it in every case, they shouldn´t do this in general is not OK.


 

[The Electrician]

That's totally ridiculous!!
Like saying that a HP34401, KS34465, DMM6500 are "AMATEUR DISPLAY METERS" because they can't read a waveform with very high peak to RMS values as well as the Keysight 3458A!!
 

Not convincingly.
It is ridiculous, this is when the specification for the device specifies the range and the possibility of a full-fledged measurement of secondary parameters with a certain stability and reliability, but in fact this is not provided.
If this is stipulated in the documentation, as for HP34401, KS34465, DMM6500, then this is another matter 

The manual that came with my DE5000 does not do a good job of specifying accuracy under all conditions.  It does not provide an accuracy adjustment to deal with the effect of low loss on capacitor measurements.  However, the manual for the IET labs version of the DE5000 does show that.  Their manual is at: https://ietlabs.com/pdf/Manuals/DE_5000_im.pdf

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

The low cost meters are not as well specified as the more expensive ones for sure, and that is a downvote against them.  I wish all users of low cost LCR meters would read this thread and not fool themselves into thinking that just because their meter shows a lot of digits that all those digits are meaningful. 

[mawyatt]

One of the things I like about the Hioki meters is their ability to display 4 parameters at once.  Try setting yours to match how I did it in reply #20.  Then choose the "set" mode, followed by "freq".  This allows you to quickly switch frequencies and watch the effect on all those parameters.  The polystyrene cap I used has its minimum D around 1 kHz; the Q is way over 10000, so the indication is "Disp out".  The capacitance value has 6 stable digits, but ESR has only 1 stable digit--all the less significant digits are bobbling like crazy if the speed is set to "slow".  This "bobbling" of the less significant digits is a clear indication that the reading is down in the noise because that is what bobbling digits are--noise.

The most stable and accurate readings can be obtained if the speed is set to "slow2", and the level is set to "5V".  Also setting averaging to 256 helps a lot for accuracy.  Without averaging, I get 1 stable digit for the ESR of the polystyrene capacitor, if I average for 256 times, I now have 3 stable digits.

This is generally pretty much how we have the IM3536 setup with Cs, Rs, D and Q for capacitors, but with 16 averages (trying to keep the IM3536 and TH2830 similar in settings). Haven't tried 256 averages yet tho.

Really like the IM3536 but also miss the ability to have a display that shows a frequency sweep on the instrument. Not a fan of having to hang a laptop off everything to "see" decent plots, and since we are dominantly Mac based, much of the support software is not Mac compatible!!

These polystyrene caps are generally really high "Q" types, wonder why they are not more popular now (know about the temperature limit), recall they were more popular back a few decades ago.

Best,

[The Electrician]

I think you know how it´s meant. 

@mawyatt:
Interesting that even the hioki won´t be "flawless".
The Tonghui is the same LCR like the ST2830 we got.
I must read the ST manual again, also the one of the keysight - But i´m pretty sure that there are no remarks in it about the limitation of the ESR (Rs) measure.
Same on the ET4410 and of course DE5000.
So the conclusion could be that we can´t expect proper ESR/Q readings for every cap-type on the "cheapos" up to the costs of a TH/ST LCR...Hmm.

On page 18 of the manual: https://www.easttester-cn.com/uploads/ET44-Benchtop-LCR-METER-for-Component-Measurement-User-Manual.pdf

are the adjustments for ESR (R) measurement when the Q of the capacitor is taken into account.

[The Electrician]

One of the things I like about the Hioki meters is their ability to display 4 parameters at once.  Try setting yours to match how I did it in reply #20.  Then choose the "set" mode, followed by "freq".  This allows you to quickly switch frequencies and watch the effect on all those parameters.  The polystyrene cap I used has its minimum D around 1 kHz; the Q is way over 10000, so the indication is "Disp out".  The capacitance value has 6 stable digits, but ESR has only 1 stable digit--all the less significant digits are bobbling like crazy if the speed is set to "slow".  This "bobbling" of the less significant digits is a clear indication that the reading is down in the noise because that is what bobbling digits are--noise.

The most stable and accurate readings can be obtained if the speed is set to "slow2", and the level is set to "5V".  Also setting averaging to 256 helps a lot for accuracy.  Without averaging, I get 1 stable digit for the ESR of the polystyrene capacitor, if I average for 256 times, I now have 3 stable digits.

This is generally pretty much how we have the IM3536 setup with Cs, Rs, D and Q for capacitors, but with 16 averages (trying to keep the IM3536 and TH2830 similar in settings). Haven't tried 256 averages yet tho.

Really like the IM3536 but also miss the ability to have a display that shows a frequency sweep on the instrument. Not a fan of having to hang a laptop off everything to "see" decent plots, and since we are dominantly Mac based, much of the support software is not Mac compatible!!

These polystyrene caps are generally really high "Q" types, wonder why they are not more popular now (know about the temperature limit), recall they were more popular back a few decades ago.

Best,

The IM3570 only costs $5000 more. 

Have a look at this thread of mine from a while ago when the IM3570 was new to me.  I should have titled it "Fun with a Hioki IM3570".

https://www.eevblog.com/forum/projects/capacitor-measurements-on-an-impedance-analyzer/msg178362/#msg178362

[The Electrician]

These polystyrene caps are generally really high "Q" types, wonder why they are not more popular now (know about the temperature limit), recall they were more popular back a few decades ago.

Best,

I think the manufacturers of the polystyrene film decided it wasn't worth it to keep manufacturing.  I remember when the only manufacturer of polycarbonate film also decided to get out of that business.  I still have an assortment of stacked film polycarbonate caps made by Siemens that the rep gave me years ago.  Those were good caps for their time with respect to ripple current capability.

[Martin72]

Well...

The IM3570 only costs $5000 more.

High quality equipment has its justification, to say the least.

[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.  One half microfarad is rather large for a polystyrene capacitor.  Here is what the Hioki IM3570 gets for a measurement at 10 Hz:



At first glance it looks like the measurements of a very low loss cap.  But, wait a minute!  The ESR is negative!!  The D value is negative!?  The IM3570 just can't do it!  And, even though the D is negative, look at its value: .000017.  WOW!!

This is a situation like TimFox described where the only hope of getting a value is with calorimetry.

[mawyatt]

Your frequency is only 10Hz, try a higher frequency.

Best,

[Martin72]

Oh, indeed.
Nanofarad and some Hertz...won´t fit.

[The Electrician]

Your frequency is only 10Hz, try a higher frequency.

Best,

As I said in the sentence just before the image, the measurement is at 10 Hz.  I showed that because that's the frequency where the IM3570 got negative values for ESR and D.

Here are the measurements at 1 kHz and 10 kHz.  This is the highest Q capacitor I have:

[TimFox]

Regardless of whether data sheets have been written carefully, it's important to note the complex (mathematically) nature of the impedance of a capacitor-looking component.
Since there is a small (hopefully) but finite coupling (or leakage) between the in-phase (real) and quadrature (imaginary) components in a real machine, the capacitance and ESR of a high-Q component are going to affect each other, specifically the displayed (small) resistance will suffer a small effect from the (larger) measured reactance, which in turn is inversely proportional to the displayed capacitance.
A practicable specification for a physically realizable meter (not a legal fiction) is the maximum resolvable Q factor (or minimum D), or the maximum Q for a given accuracy of the displayed Q.
Too many people look at a data sheet and assume that the component can meet any combination of the line items simultaneously, without reading or understanding the relationships (e.g., maximum voltage, maximum current, and maximum power that cannot be safely applied simultaneously.)

The negative D value reminds me of my missed Nobel prize when I measured a negative noise factor on a very-low-noise RF amplifier due to a slight miscalibration of the apparatus.

[mawyatt]

Getting negative values for ESR and D is not difficult, we just did so with a 10nF Polystyrene at 10, 100 and 1KHz.

If you look at the impedance angle, the phase drops slightly below -90.0 degrees which is obviously impossible for a real capacitor with a simple series C and ESR model. So this is just dancing around the limits of measurement for a particular instrument and DUT, where a higher measurement frequency increases the DUT current and lowers the instrument transimpedance gain (lower feedback resistor that senses the DUT current) moving back into a region where the instrument can provide a meaningful measurement, and why we mentioned such.

Best,

[mawyatt]

The negative D value reminds me of my missed Nobel prize when I measured a negative noise factor on a very-low-noise RF amplifier due to a slight miscalibration of the apparatus.

Reminds me of a couple physicist searching for the magnetic monopole long ago. They sent up a detector in a balloon and when developing the photographic plates from the detector found a trace and claimed it was the elusive magnetic monopole. They also missed the Nobel when later it was discovered the plate was installed 90 degrees out of phase...oops   

On a closer note, we had a similar case when IBM provided a new HICUM SiGe bipolar model, under certain conditions this model would negate some of the noise from the driving source. They didn't win a Noble either, more like a pie in the face when we notified them about the magic transistor!!

Best,

[indman]

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!

"Secondary Parameters Accuracy:
Az = impedance (Z) accuracy
Definition: Q = 1/D & Rp = ESR * (1+1/D2)
D value accuracy: Dz = ± Az *(1+D)
ESR accuracy: Rz= ± Zm * Az (Ω)
ie., Zm = impedance calculated by 1÷(2πƒC) or 2πƒL
Phase angle Θ accuracy: Θz= ±(180/ππππ)*Az (deg)"

What should I understand from this? What is Az - an unknown impedance value, which, under the condition D<0.1, is characterized by the reactance of the capacitor? In what units is it calculated? And how does this explain the "fantastic" results secondary parameters when measuring high-quality polypropylene capacitors? Why does the manual not specify the accuracy and tolerance in % as it is done for the primary parameters L-C-R?
It turns out that I myself have to conduct experiments and determine the degree of reliability of the testimony?
Or does the device manufacturer have to do it?
And if I don't have a "TH/ST" class device to compare readings with?

[Martin72]

And if I don't have a "TH/ST" class device to compare readings with?

Even when you got one, it won´t bring you any further as they also can´t measure/displaying it correctly.
Example, mentioned before, the 10nF mica cap I´ve measured it with the ET4410, the U1733C from keysight and the ST2830 from sourcetronic.
All meters measured the capacity correctly inbetween their accuracies and inbetween the tolerance of the cap.
All meters measured the Xc correctly.
But none measured the expected ESR correctly.
Expected means in this case the calculated one, based on the values of the Tan phi from the spec.
It should be in the range of appx 16ohms at 1khz.
The ET4410 displays appx 5 ohms, the ST2830 appx 3 Ohms and the keysight "freaks out", displaying something and blinking all the time.
The most probably reasons why you´ll find it here in this thread.
So for me there are several conclusions "avaible".
1.) Can´t expect to have it all even when you spend 1000 bucks or more for a LCR.
2.) The "secondary" measure functions of the "cheapos" (including the 1000 bucks or more) like D or ESR are (probably, will test it soon) trustable for caps with "bad" losses like electrolytic ones.
3.) When you wil get in a situation (as a hobbyist) to know exactly the losses/ESR of a high Q cap? Isn´t it more practical to know the ESR of a electrolytic cap which the meters could display trustable?
4.) Developers choosing a cap for their circuit design according to the specs of the manufacturer, not by measuring it before.
The only exception is to know the exact capacity (for filters) - And capacity every not too cheap meter can measure it correctly.
So why we private ones will go further...Buy a cap from a known manufacturer according to the specs it got and you need and that´s it.
5.) Knowing the exact ESR is in the most cases interesting by electrolytic caps (switching power supplies) when you must repair something. And the ESR of these ones even a cheaper meter should measure it.

[indman]

2.) The "secondary" measure functions of the "cheapos" (including the 1000 bucks or more) like D or ESR are (probably, will test it soon) trustable for caps with "bad" losses like electrolytic ones.

I agree with this 100%! This information should have been placed in the documentation for the device, and not to invent fantastic formulas with unknown variables that confuse the user even more.
If we follow your logic, then the measurement of secondary parameters for high-quality non-electrolytic capacitors for devices of this class is NOT RELIABLE and is only for evaluation and familiarization purposes,what does the status of an amateur meter mean!
For me, this conclusion is obvious and so far no one has convinced me otherwise.

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..