EEVblog Electronics Community Forum
Electronics => Projects, Designs, and Technical Stuff => Topic started by: mawyatt on March 24, 2022, 08:32:44 pm
-
Many folks have various DIY and lower cost LCR meters that can measure capacitor ESR, but without a reference ESR device, the question of a good ESR measurement remains.
Inductor ESR can have both an AC and DC measurements, which gives some level of confidence, however capacitor DC measurements for ESR are not possible. Simple technique which can give a better feeling regarding measuring capacitor ESR is to use an additional series resistor and make a pair of AC ESR measurements, one with the series R and one without. The difference in ESR readings should be very close to the added series resistor, if not then the measurements are in question.
Adding a series resistor of say 1 to 10 ohms should suffice for most electrolytic type capacitors. For the best results the resistor should have a small form factor (SMD) with low residual inductance and capacitance. The individual added resistance value can be measured with a DMM, or even the LCR meter if the meter has a good low resistance range, either AC or DC.
Anyway, this might help when trying to validate ESR measurements as a sanity check!!
Best,
-
low cost ESR meters can measure reliably nothing below 0.1ohm ESR and most modern caps are way below that
5transistor ESR will do it better and cheaper
inductor ESR is just DC copper resistance
-
Inductor ESR can have both an AC and DC measurements, which gives some level of confidence, however capacitor DC measurements for ESR are not possible. Simple technique which can give a better feeling regarding measuring capacitor ESR is to use an additional series resistor and make a pair of AC ESR measurements, one with the series R and one without. The difference in ESR readings should be very close to the added series resistor, if not then the measurements are in question.
...
this might help when trying to validate ESR measurements as a sanity check
Nice trick, thanks!
About the DC measurements of a capacitor's ESR, I've read once about another trick, where one can see the ESR with an oscilloscope. The idea is to apply an extra current step on the capacitor, while watching with an oscilloscope the voltage across the capacitor.
- In theory, the voltage on the capacitor should start to increase smoothly.
- In practice, the oscilloscope will first show a step-jump in voltage, then the voltage will start increasing smoothly from there, because the resistive part (the ESR) will respond instantly to the step.
The size of the voltage jump-step observed on the oscilloscope is proportional with the ESR. :D
This method might not be very accurate because of using only the visuals on the oscilloscope instead of the many digits of a dedicated instrument, but in practice this method can be good enough as a sanity check, and good enough to sort out bad capacitors. I've used it a few times.
W2AEW made a video about observing the ESR with a DC step, with very clean explanations and a demo:
#135: Measure Capacitor ESR with an Oscilloscope and Function Generator (https://youtu.be/115erzCCxgE)
-
Yes the scope transient technique is a good option if one doesn't have a dedicated LCR meter at hand!! Thanks for reminding of that technique, and posting the video :-+
We've used the resistor swap (we have 0.1, 1 and 10 ohms for this) to verify readings from the T7 Multi-Tester and DE5000 LCR meter before or when a quality lab grade instrument wasn't available. The T7 does surprisingly well, and has 0.01 ohm ESR resolution as does the DE5000. The DE5000 is well known for providing reliable readings, and has demonstrated such.
BTW recall the T7 uses the pulse technique for ESR, also recall that the T7 can give incorrect readings without warning when the battery voltage drops.
We are in the middle of evaluating some components & instruments for very demanding applications, and decided to test some resistors, capacitors, and inductors to compare instruments for starters. We have some aged Caddock low TC 0.01% resistors as reference, and some self-generated reference film capacitors, also have some 0.1% Venkel SMD devices. During the testing we grabbed some good quality electroyltic capacitors we use for SMPS work just to "see" how well things lined up. The little T7 did an admirable job, as of course the DE5000 did compared to a pair of lab LCR instruments (Tonghui TH2830 and Hioki IM3536).
Best,
-
low cost ESR meters can measure reliably nothing below 0.1ohm ESR and most modern caps are way below that
5transistor ESR will do it better and cheaper
inductor ESR is just DC copper resistance
We have a low cost T7 and DE5000, the T7 does fairly well with low values of ESR for capacitors (and inductors), as does the DE5000.
If the DUT inductor is a higher value and uses a core, then the DCR may give an optimistic reading, as it does not include core loss, which will show in higher frequency measurements. Most lab LCR meters use a sine wave excitation where the measurement frequency can be varied, and measure the voltage and current in the DUT. The phase between the voltage and current indicates the real and imaginary components, and inductor core loss increases the real component just as DC wire resistance does, thus increasing the apparent ESR. Generally skin effect isn't of concern since the measurement frequencies are low compared to where skin effect becomes significant.
Recall the T7 uses a pulse technique which works well, however the DE5000 uses the same technique as the expensive Lab LCR meters. The reason it's so affordable is because it uses a special chip-set that does all the signal generation, amplification, complex I and V detection and requires just a few external components. This is why it compares very favorably with quality lab meters such as the Tonghui TH2830 or Hioki IM3536. At first we were reluctant to blindly rely on the DE5000 measurements, but after many thousands of measurements, no reluctance now!!
YMMV.
Best
-
If the DUT inductor is a higher value and uses a core, then the DCR may give an optimistic reading, as it does not include core loss, which will show in higher frequency measurements. Most lab LCR meters use a sine wave excitation where the measurement frequency can be varied, and measure the voltage and current in the DUT. The phase between the voltage and current indicates the real and imaginary components, and inductor core loss increases the real component just as DC wire resistance does, thus increasing the apparent ESR. Generally skin effect isn't of concern since the measurement frequencies are low compared to where skin effect becomes significant.
skin effect yes. core loss is parallel resistance not ESR
LCR meters like DE5000 is not designed for ESR measurement, good enough for testing leaking capacitors. tested this function.. meh
-
Why do you say the LCR meter DE-5000 is not designed for ESR measurements? Within its published specifications, it is perfectly capable of measuring the equivalent series resistance of capacitors and inductors at a discrete set of frequencies, as well as the equivalent parallel resistance. Inherently, it measures the two components R and X for the series combination R + jX by measuring the in-phase and quadrature components of the ratio of the voltage across the part to the current through the part, and calculates the other displayed parameters from those two components.
-
skin effect yes. core loss is parallel resistance not ESR
LCR meters like DE5000 is not designed for ESR measurement, good enough for testing leaking capacitors. tested this function.. meh
Wrong on both accounts!!!
Maybe you don't understand the relationship between parallel and series Equivalent circuits. ESR means Equivalent Series Resistance, parallel loses can be translated between Equivalent Series and Equivalent Parallel circuits!!
Don't believe this, then take a simple length of wire and wind it around a nail. Measure the inductance and ESR without the nail inserted in the coil at 100KHz, now insert the nail and measure again. Guess what, the inductance increased and so did the ESR, but no series element was added, the core is in parallel!! Why the increases, because the nail core relative permeability is much greater than air (1), and the nail core loss (air has no loss) shows up in the Equivalent Series Circuit as an increased ESR!!
You should also understand that ESR is usually frequency dependent, meaning it isn't a constant with frequency and usually increases with frequency. So if one is designing a SMPS then it's a good idea to measure the ESR near the switchmode frequency, if the design is for a Mains filter then measure ESR at twice the mains frequency due to the full wave frequency doubling effect. These are reasons some capacitors are specified at 100Hz and 120Hz, and others at 100KHz.
Also, the DE5000 is quite capable of making decent capacitive and/or inductive ESR measurements within its' designated range. Certainly works well for us and many others, but does require some instrument & test knowledge to be utilized properly:-+
Edit: Just for fun measured a couple inductors with the DE5000 and a TH2830 and Hioki IM3536 at 100KHz. Here's the results.
DE5000 TH2830 IM3563
9.301uH @ ESR 0.11 ohms (DCR 0.04) 9.3086uH @ ESR 0.1663 ohms 9.27509uH @ ESR 0.167 ohms
446.6uH @ ESR 4.03 ohms (DCR 0.55) 449.319uH @ 3.79 ohms 449.186uH @ ESR 3.96 ohms
Best,
-
Robs old video on this.
https://www.youtube.com/watch?v=D111PZqtAis (https://www.youtube.com/watch?v=D111PZqtAis)
-
Hi,
I use a few different techniques for testing the ESR meters that I have designed.
(https://www.eevblog.com/forum/projects/esr-meter-adapter-design-and-construction/?action=dlattach;attach=70047;image)
(https://www.eevblog.com/forum/projects/esr-meter-adapter-design-and-construction/?action=dlattach;attach=70049;image)
This a series of 4 wire resistors. The project is here:
https://www.eevblog.com/forum/projects/esr-meter-adapter-design-and-construction/msg343855/#msg343855 (https://www.eevblog.com/forum/projects/esr-meter-adapter-design-and-construction/msg343855/#msg343855)
That is almost pure resistance.
This is more difficult test, a film capacitor in series with a resistor:
(https://www.eevblog.com/forum/projects/esr-meter-adapter-design-and-construction/?action=dlattach;attach=89397;image)
This image was shared by forum member BravoV.
I also compare the readings on the ESR with an HP4274A LCR meter.
In some case an LCR meter is not suitable for testing in-circuit capacitors or they are not as well protected against charged capacitors or applied voltage.
Regards,
Jay_Diddy_B
-
@ Shock,
That's a great video, thanks for posting, exactly what I was discussing. Those grabber type Kelvin clips look interesting, we made some out of the larger type Kelvin clips, but those grabber types would be better for smaller DUT.
There seems to be so much mis-information being spewed out on here and other places, it's nice to see a true quality hands on video like this :-+
Best,
-
yeah Rp 10kohm(leaky capacitor) will do much to ESR 0,1ohm but to impedance Xc could
DE5000 measures impedance (most resolution) and angle and do math to extract other parameters
inductor winding capacitance EPC cancels phase a bit , therefore more ESR + skin effect at 100kHz
-
The hardware in the DE-5000 is based on the old General Radio "Digibridge" design without standard capacitors or inductors.
A good description of the DigibridgeTM theory of operation is in section 4, especially pp 91-94 of the PDF, in the current IET manual https://www.ietlabs.com/pdf/Manuals/1659_im.pdf (https://www.ietlabs.com/pdf/Manuals/1659_im.pdf)
The data sheet for the chipset used in the DE-5000 (and several competitors) can be found at http://www.cyrustek.com.tw/wp-content/uploads/ES51920.pdf (http://www.cyrustek.com.tw/wp-content/uploads/ES51920.pdf)
An AC voltage is applied to the DUT, with the "Low" end connected to a current measuring circuit that uses calibrated standard resistors to measure the current.
A separate measurement measures the voltage across the DUT from the source.
In terms of that standard, using phase-sensitive detection, the fundamental measurement is of the parameters R and X in the complex impedance R + jX.
Knowing the AC frequency and the standard resistor, these two values can be used to compute the following display values:
The equivalent series model, where X is converted to either capacitance or inductance.
The equivalent parallel model, where R + jX is used to calculate the complex (inverse) admittance G + jB, from which they compute the equivalent parallel resistance and either inductance or capacitance.
The magnitude and phase angle.
The capacitance or inductance that corresponds to either equivalent, and the Q or D value that corresponds to the ratio of components.
My only problem with this choice of display values is that it does not show explicitly if the C or L value is positive (e.g., when the inductor is above self resonance, the reactance is capacitive with the opposite sign).
In order to check that, the magnitude and phase display will show the polarity.
-
I use the same technique as post #9. Precision low value resistors are easy to find - from current shunts (down at the 0.1 milliohm to 20 milliohm range), and then smt and leaded or power package devices with typically 1% tolerance, but down to 0.1% or 0.05%. Also a zero ohm bar to confirm kelvin probes are zeroed. Then the additive method with a series capacitor for assurance (a film with low temp and humidity and freq and tan delta variation like MKP). KISS.
-
I designed my own precision ESR meter.
To keep it accurate it has a calibrate mode where it measures a known resistor for the reference point.
-
Mentioning the GR Digibridge, one should also mention one of the giants of that era, Henry P. Hall, who came up with the design.
I use the film cap and resistor method for my bridges, which seems to be entirely adequate, though not for large values. One interesting thing about leakage is it only takes a small amount of DC leakage to completely mess up a tube stage by offsetting the grid, but that same amount of leakage can go unnoticed with an esr test, or even a good bridge measurement. Sometimes the circuits are more sensitive than the test equipment!
I've never understood why anybody would take the trouble to build an esr tester of moderate complexity, when the same amount of effort will produce a good LCR bridge that gives far more reliable answers.
-
Took a look at DE5000 just out of curiosity, and it's the only instrument I've seen with Comic Sans in its datasheet. ;D
https://www.ietlabs.com/pdf/Datasheets/DE_5000.pdf (https://www.ietlabs.com/pdf/Datasheets/DE_5000.pdf)
Decent specs, though. A nice to have instrument. The resolution for the lowest resistance range and ESR, 1m\$\Omega\$ could have been better, IMO. Saying this because only this week I've measured 8m\$\Omega\$ for a 10 years old Ni-MH rechargeable AA. A brand new battery might have even lower ESR, and so the need for a resolution lower than 1m\$\Omega\$.
I've measured those 8m\$\Omega\$ with nothing but a $400 oscilloscope (Rigol DS1054Z set to 500uV/div and the acquisition mode set to average), a 1k\$\Omega\$ resistor and a 1.5uF. The AC signal used to measure was coming from a DDS. The setup was like this:
(https://www.eevblog.com/forum/projects/esr-evaluation-and-verification/?action=dlattach;attach=1448128;image)
I was a little unhappy with the m\$\Omega\$ ESR resolution in my setup.
With a dedicated synchronous detector it would be possible to go much lower than m\$\Omega\$. For example, Analog Devices' AN-306 (see the attachement) shows a microohmmeter made with a 1mA AC generator, a 100 000 x amplifier and an AD630 as synchronous detector + filter for a DC voltage output. The resulting sensitivity is ~15m\$\Omega\$/V.
-
Mentioning the GR Digibridge, one should also mention one of the giants of that era, Henry P. Hall, who came up with the design.
Brilliant design approach by Henry Hall, did not know he conceived the bridge concept!!
I've never understood why anybody would take the trouble to build an esr tester of moderate complexity, when the same amount of effort will produce a good LCR bridge that gives far more reliable answers.
Or just get the DE-5000 for $110~120!! Classic example of how specialized chip(s) can significantly reduce the cost of a complex electronic item without suffering performance loss :-+
Edit: At RoGeorge. Recall AD have a dedicated chip for high resolution impedance measurements, but don't remember how low a Z it could resolve? Some DMMs (DMM6500 has 1 ohm scale) are capable of reasonable sub milliohms measurements. Haven't designed an LCR meter but seems to be successful you want lots of stimulus current, synchronous demodulation, and low offset, low noise high gain stages.
Best,
-
AD630 was the sync demod IC. It can be used in a battery ESR meter to achieve 1-2% accuracy down to 100 microohm FS with 1 microohm resolution.
-
Edit: Just for fun measured a couple inductors with the DE5000 and a TH2830 and Hioki IM3536 at 100KHz. Here's the results.
DE5000 TH2830 IM3563
9.301uH @ ESR 0.11 ohms (DCR 0.04) 9.3086uH @ ESR 0.1663 ohms 9.27509uH @ ESR 0.167 ohms
446.6uH @ ESR 4.03 ohms (DCR 0.55) 449.319uH @ 3.79 ohms 449.186uH @ ESR 3.96 ohms
Best,
Does this mean you are now the proud owner of an IM3536?
-
Does this mean you are now the proud owner of an IM3536?
Yes! However not very happy with the fan noise, it's much louder than anything we have in the lab! Took a video and sent it to Hioki USA and they compared it and said this is same as the one they have :o
Best,
-
That scope method is quite useful and can be enhanced just by having enough gain. I do it with a high gain diff plug-in on my old boat anchor Tek, but easy enough to build something. Can't remember the details, but I feed the RC with a square wave and look at the triangle output. The signature of the triangle tells you a lot about the cap and there are some interesting differences between say, tantalum and aluminum electrolytics.
-
Agree about the Scope Method noted by RoGeorge. This is very effective at showing the effects for SMPS work since the "Test Waveform" is similar to what the cap will "see" in actual SMPS application. With a carefully crafted controlled setup and proper instruments one can resolve the effective ESL effects as well. We found in the past that the larger physical size caps have the larger ESL, even if they have a lower ESR.
Haven't evaluated any of the Polymer types yet, so will be interesting to see how they behave compared to the various Al and Ta types.
Best,
-
Using a cheap VNA to measure ESR. Starts about 4 minutes in.
https://www.youtube.com/watch?v=scZ3kZ4Q2sQ (https://www.youtube.com/watch?v=scZ3kZ4Q2sQ)
-
Using a cheap VNA to measure ESR. Starts about 4 minutes in.
https://www.youtube.com/watch?v=scZ3kZ4Q2sQ (https://www.youtube.com/watch?v=scZ3kZ4Q2sQ)
At 3:20, you say "is that capacitor really 320 ohms? Of course it isn't; it's a ceramic. It's going to be less than an ohm."
You have failed to take into account that the measurement frequency you're using is 1 kHz. That ceramic disc cap is most likely not using a RF capable delectric. At 1 kHz such a high ESR is quite reasonable.
Here are a series of images showing measured ESR of a similar ceramic disc cap at frequencies of 1 kHz, 10 kHz, and 1 MHz.
As can be seen, the ESR decreases rapidly as the frequency increases, so at moderately high RF frequencies the result you got with the NanoVNA is not unexpected. But, at 1 kHz, an ESR of hundreds of ohms is not wrong.
(https://www.eevblog.com/forum/projects/esr-evaluation-and-verification/?action=dlattach;attach=1448782)
(https://www.eevblog.com/forum/projects/esr-evaluation-and-verification/?action=dlattach;attach=1448788)
(https://www.eevblog.com/forum/projects/esr-evaluation-and-verification/?action=dlattach;attach=1448806)
-
I was going to mention a network analyzer, as well as series/paralleling the components for a quick and dirty test and low and behold Joe has made a video on it.
-
Using a cheap VNA to measure ESR. Starts about 4 minutes in.
...
At 3:20, you say "is that capacitor really 320 ohms? Of course it isn't; it's a ceramic. It's going to be less than an ohm."
You have failed to take into account that the measurement frequency you're using is 1 kHz. That ceramic disc cap is most likely not using a RF capable delectric. At 1 kHz such a high ESR is quite reasonable.
...
Looking at a 10nF part, the data sheet shows a max dissipation factor of 2.5%, 1kHz, 1V. ESR = 0.025/2*pi*1000*10n = 400 ohms (check my math).
https://www.vishay.com/docs/23110/561r562r565r.pdf (https://www.vishay.com/docs/23110/561r562r565r.pdf)
If I measure this part, at 1kHz, 1V. The dissipation factor is 0.008, 9.3nF, ESR 136 ohm.
I believe you are correct.
-
Shown measuring the CM 10nF disc on the BK. Fixture used to measure the device with the LiteVNA. Notice how cutting the leads roughly doubled the resonance and reduced the ESR by 150 mOhms.
-
I've got an old HP vector analyzer and yeah, the effect of long leads at high frequencies is astounding.
-
Using the BK to look at the ATC 330pF, we can see even at 10Khz, it measures over 10 ohms. At 1KHz, the readings will fluctuate +/- a few hundred ohms.
Using the Lite, we can see it resonates around 250MHz and it measures an ESR of roughly 32mohms. Looks about 2mohm higher than the last time I looked with the V2Plus4 using different cables.
-
... That ceramic disc cap is most likely not using a RF capable delectric. ...
Could you please expand on the above comment?
-
... That ceramic disc cap is most likely not using a RF capable delectric. ...
Could you please expand on the above comment?
I meant that generally disc ceramic caps are old technology, often with high K dielectrics. The cap I used to produce the images I posted earlier has a marking "Y5P". Some others from my parts store are marked "Z5U" and "Z5P". Modern caps intended for RF use are typically surface mount parts, and use a better dielectric, as I'm sure you are aware. American Technical Ceramics says that a dielectric they use is porcelain, which is not a titanate based dielectric like the high K dielectrics usually are. They have other, "proprietary", dielectrics, which might be referred to as "RF capable".
See: https://rfs.kyocera-avx.com/userFiles/uploads/pdfs/esrlosses_appnote.pdf
-
I have some through hole ceramic capacitors with porcelain dielectric that I bought from the Boeing surplus store in Seattle decades ago:
(https://www.eevblog.com/forum/projects/esr-evaluation-and-verification/?action=dlattach;attach=1449703)
The legend "BAC.C1" denotes "Boeing Aircraft Corporation", and you can see the date code, 62 07, the 7th week of 1962.
Here is what my Hioki IM3570 says about it:
(https://www.eevblog.com/forum/projects/esr-evaluation-and-verification/?action=dlattach;attach=1449715)
At 1 kHz, the value of D, .000198, is right at the limit of what the instrument can do. The ESR is exceptionally low, and the Q, of course, is exceptionally high.
This is what you can get with a very low loss dielectric.
-
C0G and NP0 dielectrics generally give good RF results (of course depends on package) and are somewhat stable.
Best,
-
Also, disc ceramic capacitors in NP0/C0G have been available for decades, but not in very large values (non-disc SMT MLCC and TH MLCC capacitors are now available above 10 nF).
The Vishay Ceramite (originally Sprague) 1 kV NP0 discs are available up to 270 pF, and the non-zero tempco N750 1 kV disc capacitors up to 1000 pF.
Other low-loss ceramics with even higher tempcos are available past that.
Disc capacitors are limited by their construction: metallization on both sides of a single disc.
The ATC porcelain capacitors have absurdly high Q values, up to 10,000.
-
Perhaps another old ceramic capacitor.
Paper showing the resonant tube.
-
Nice article by Knowles, shows the subtleties & nuances of making these measurements as frequency increases, which conveys there's nothing trivial about this!!!
The article references Modelithics Inc. at the end, great knowledgable resource for these kind of efforts (know the founders Drs Weller and Dunleavy very well, in fact will be giving a presentation April 27th at the IEEE WAMI Conference they helped create long ago).
Best,
-
Attached datasheet is for an MKT1822 series metallized polyester film. Shown is a 14uF part. They spec the tangent of loss angle at 10X10^-3 @ 1kHz. ESR should be a maximum of 0.114 ohms (check my math).
The BK measures 0.057 ohms. Also shown are 0.1 and 0.05 ohm resistors along with a copper shorting bar.