AC Rs measuring is what old bridges do, this is indeed a good method and a accurate bridge can measure accurate down to a few mOhm. Good idea, i will add this kind of measurement too ( ease using impedance bridges or RF-IV meter)
ESR must change over freuency, so this is a good thing. Only a pure resistor will change almost nothing. But ESR is part chemical and part electronic so not a constant. It becomes leer if frequency rises
The datasheet gives you D or ESR or both. This for one or more frequencies. And indeed 100KHz is a much used value. So for repairing a smps with the small high frequency caps this is a good frequency.
But many people do not know what they are measuring. And also use them on large filter caps. The 47 uF I used had specs for 120Hz and 100 KHz. A RF-IV based C meter has no problems with 100KHz because they measure the loss angle and that can be easy converted to ESR or D. But meauring esr with the simple schematics is a total different way. Using a scope you can see the ESR part if you can zoom in enough, but the schematics I have seen use 100KHz for the sole reason that in that case they measure the total impedance. As you know impedance is R and jX. JX is the reactance of the capacitance and R is the ESR.
If the reactance is very small, and it is for most electrolitics at 100KHz, the R part is sooner dominating. And if a cap is bad esr is all you see. But i could not measure the low ESR of a 1uF or 10uF standard cap. However if I just use the total Z I get a value that looks pretty low. But in real it is about 10x higher or more as the ESR realy is. But again, real bad caps show up. ( but in real, many good caps will show bad to because of the ESL, however, you will never know, if you replace ten caps that measure bad and only two of them were really bad so 8 good, then the repair was a succes and you will never know you replaced 8 caps that were good. And you tell everyone how good that esr meter was. )
Just draw the schematic and do the math and you know what I mean. Or use spice. Take a 5Vt squarewave, put a 100 ohm resistors in serie then, a try combinations of 10-100 nH coils, 10mOhm to 10 Ohm series resistors and 1 to 1000 uF caps. Measure current by using the inline current probe.
For instance take 47 uF, 1 Ohm, 10KHz and 10 nH 50 mA, then look at the voltage over the LCR combination. The voltdrop from top to where the curve starts is the ESR. Lets say this is 50 mV. Now you divide 50 mV by 50 mA and the answer is 1 Ohm.
But then take 100KHz , 1000 uF and 10 mOhm Now try to measure the ESR and look if you still manage to zoom in far enough to get 10 mOhm. And then enlarge 10 nH to 100 nH. This is still a very low value is you are not using test leads. Then you know what I mean.
These pictures will be in my article too, also scope measurements made on a 350MHz DSO. Also bridge measurements madecwith. 0.01% GR bridge, a 0.1% GR bridge, a peak atlas ESRr60, my esr very simple homebrew meter that use a total different concept and several VNA and RF-IV measurements that are able to measure phase, D, ESR upto 1500 MHz.
I will show SRF curves, ESR curves, D curves , capacitance curves and compare them with simulations and the rest.
If you use 100KHz only for smps caps, so the onces that actually work at those frequency ( many ( older) smps work at much lower frequency ) then its OK, but most people use them for all caps because they do not know what they measure, or what selfresonance is or that ESR is frequency dependend ect. My ESR meter meaures ESR down to 100 nF and with reduced accuracy even lower. And the ESR of a 100 nF can be is still close to a few mOhm. The 1uF standard cap I could not measure, but the tabels on things like the Parker meter tell different.
10 uF at 100KHz is 0.16 Ohm. If ESR is 0.16 Ohm this means that the DF is 1. A bad cap. The power dissipated in the cap is as large as the power through the reactnce part, so another way is to say is dissipation is 100 %, a bit confusing because not 100% of total power is dissipated. An other way is to use the loss angle. A good cap has a 90 degree phase difference between current and voltage. A D of 1 is arc tan 1 or a 45 degrees loss angle.
A good 10 uF would have a for instance a 5 degree loss angle, or tan 5 is a D of 0.087
We know Xc is 0.16 ohm so 0.16 x 0.087 is an ESR of 14 mOhm.
The current though you 100 Ohm and the ~ 0.16 Ohm impedance is about 50 mA.
The voltdop over the ESR will be 0.014 X 0.05 = 700 uV. You can not measure this whit home made stuff or a simple build in uP ADC. Not only because the resolution but also because it can not sepate R from Xc so the meter will tell the ESR is 0.16 Ohm and that sound pretty low. If the cap is bad and ha an ESR of 2 Ohm there is no problem , ESR will show as 2.16 Ohm. But if the dap is above self resonance skineffect and ESL peak make the total impedance much higher and so the ESR is a useless value. Lucky the processors are slow so with some luck nd bad enough designed it will not see the ESL and still give a value you think is correct.
Then the test often is used , put 1 or 10 ohm in serie with your cap and it will show this, so it must be correct. Now you know why that test will not tell you if initial ESR is correct. My ESR meter is not able to measure a resistor direct because it samples only that the voltdrop by switching between to sample caps and the opapm only sees that drop. Whit only a resistor this is not possible, no phase angle.
It does not cost 100 dollar, just an analog meter or your multimeter, and 4 sub dollar junkbox ICs, not even matched parts. Sample frequency i most times use is 50KHz but I used a potmeter to adjusted it. In practice it is allways on 50KHz. The advantage of analog is you see the meter climbing up and this tell you also something about the cap, it must go smooth.
In circuit measurements are not possible and most times useless. On powerrails a lot of caps are parallel.
Pff, i never learn to type short stories.
Home
Help
Search
Login
Register
