Capacitors, inductors and resistors measurements

oliver

Member
Joined
May 24, 2017
Messages
59
Location
Italy
Hi John,
I deal with small SMD power inductors.
I have made a set of measurements for values from 15uH up to 4700uH ( few values are missing),maybe it can be useful.
Attached also the DCR for each inductance value.
 

Attachments

John Mulcahy

REW Author
Joined
Apr 3, 2017
Messages
3,039
Very useful Oliver, thanks. The measured DC R values all seem to be about 1.2 ohm higher than the specified values.
 

oliver

Member
Joined
May 24, 2017
Messages
59
Location
Italy
Hi John,
just for more info, my system has been calibrated before the measurements.
Attached the result for a 4.7 Ω resistor.
Regarding the previous measures, below 18uH the "component model" button doesn't become not active.
If you want me to perform other measures just let me know.
 

Attachments

John Mulcahy

REW Author
Joined
Apr 3, 2017
Messages
3,039
just for more info, my system has been calibrated before the measurements.
Attached the result for a 4.7 Ω resistor.
I don't doubt themeasurements, just thought it interesting that the components all seemed to have an additional resistive component beyond the specified value.
Regarding the previous measures, below 18uH the "component model" button doesn't become not active.
Yes, the phase for the smaller values wasn't reaching the threshold at the check frequency, I have adjusted that.
 

Antonio Di Motta

Active Member
Joined
Feb 8, 2018
Messages
164
Location
Italy
Finally I managed to test with the latest version of REW, I measured a 1mH inductance with a series resistance of 0.55 ohms. I noticed that by positioning the cursor on the curve the effective value of the inductance varies from 1.02mH to 1.03mH while the series resistance is at 10Hz and 0.557 ohm in line with what is declared by the component. On the model of the component the inductive value is approximate to the value of 1mH but the series resistance is much higher 1.45 ohm instead of 0.557ohm reported when I position the cursor.
Is there something wrong or wrong with interpreting the component model function?
 

Attachments

Jimmy Venema

Member
Joined
Apr 25, 2019
Messages
17
I like the new component model option. But in version 50 the capacitors model seemed ok, but the serie resistance of inductor was way too high. With version 51 the inductance seems oke, but now the capacitors seems way off, no resistance en the fit is not ok.
Cap.png
Inductance.png
 

Jimmy Venema

Member
Joined
Apr 25, 2019
Messages
17
Here a zip file of the mdat file with one inductor and a few capacitors. A 7z compressed file was not allowed to upload. Perhaps an idea to allow that, since the file size was 11 MB vs 19.5 MB. It is not that 7z is uncommon, in most places it replaced zip years ago.
 

Attachments

Jimmy Venema

Member
Joined
Apr 25, 2019
Messages
17
I've just tested with the 52 version and the capacitors are still off as in the 51 version, while they were ok in the 50 version.
 

John Mulcahy

REW Author
Joined
Apr 3, 2017
Messages
3,039
I've just tested with the 52 version and the capacitors are still off as in the 51 version, while they were ok in the 50 version.
The underlying problem is the accuracy of results after calibration. You may have noticed that the real part of your capacitor measurements is zero over much of the range, whilst it should never be less than the series resistance component. The curve also deviates from the expected response at the lowest frequencies. I can accommodate that by only fitting capacitors over the range from 100 Hz upwards and using the last part of the response to establish bounds for the series resistance element, that gives the results below for your 390 nF measurement. You can see the response deviation below about 30 Hz.

31767


I was being a bit unrealistic in my expectations of measurement accuracy, the setup I am using (with an ADI-2 Fs Pro) is a lot more accurate than I expected it to be but that is likely an outlier. By way of example, even a 10 nF capacitor is accurately captured to the point the impedance gets to about 3Mohm, I had to increase the precision of the calibration file entries to fully reproduce the equipment accuracy (that change will be in the next build, along with the capacitor match changes).

10nf.jpg
 

Jimmy Venema

Member
Joined
Apr 25, 2019
Messages
17
A 390nF capacitor is normally used a bit higher in frequency then 30 Hz ;-). In fact in my case it was only used in parallel with an another higher valued capacitor to get at a needed value that wasn't available in the normal E series. At the boundaries there will always be deviations. But I guess a lot of people are going to use this feature to verify their components for either a passive or active filter. So they are more interested at what the value is at their target crossover point than what the behaviour at the extremes is. Perhaps give an option to give a center frequency en over how many octaves the best fit should work. I guess the value 3 octaves away from the target working point is less important.

But anyway this is a very nice feature and the accuracy quite impressive. I used it before the model option and I moved the cursor around the crossover area and then read the value at the left bottom corner.
 

Jimmy Venema

Member
Joined
Apr 25, 2019
Messages
17
Sure, but just to be clear for anyone reading, it is not the capacitor that is deviating from expected behaviour, it is the impedance measurement.
Sure the deviations for a capacitor even small ones are more to the higher end of the frequency range, because of the inevitabel inductance. So indeed the deviation at low frequencies is in the measurement of the impedance. The measurement rig has also coupling capacitors etc, so that would influence measurements at the boundary of the rig.
My point is that a designer is interested in the component value in the frequency range were it should do its filter function. Deviations from that value outside that area are not that relevant.
 

Antonio Di Motta

Active Member
Joined
Feb 8, 2018
Messages
164
Location
Italy
John, I made many measurements to verify the accuracy of the values detected by the Component model function and I must say that it behaves quite well, even if sometimes it makes mistakes between the values that are detected by positioning the cursor on the graph and those calculated with said function . For example, I measured a 10uF non-electrolytic capacitor with 5% accuracy and the value calculated by the function was 12.3uF while the cursor detects 10.2uF. I also noticed that almost always the values of the series resistance of the paper capacitors or polyester is equal to 0.000 ohms, perhaps it is particularly generous in the evaluation of the series resistance.
Again for capacities, I noticed that under the values of 0.2uF the component model does not activate while on the cursor it continues to detect much smaller values, for example I measured a capacity of 12.1nF 0.625%.
As for the inductances it doesn't seem to have these limits, I measured an inductance of 1uH without particular problems.
I also carried out a test with a speaker and the model considers only the inductance of the moving coil excluding the values of the circuit oscillating on the cone's resonance frequency. This choice seems to me absolutely correct. I attach the tests I have carried out for any checks.
Thanks for the attention
 

Attachments

John Mulcahy

REW Author
Joined
Apr 3, 2017
Messages
3,039
The calibration file resolution increase in the next build will address the sawtooth impedance traces you are getting at high impedances, which should give better measurement results. In the meantime here is how the new capacitor model does with those caps, bearing in mind the lowest values should be better after remeasuring with the next version:

31911


31912


31913


31914


31915


31916
 
Top Bottom