Step by step on measuring inductor and self resonance


 

Have watched several videos on this and they all seem to leave out most of the setup. Looking to measure inductance but most interested in self resonance frequency. Resonance will be about 2 MHz I am guessing. Have it across Port1/S11 as seen in the videos but don't have display of anything like in the videos. I have looked at the Absolute Beginners Guide but doesn't cover this.


 

Hi! I sent a post showing how to measure resonance and self resonance and I calculated all parameters of a resonant circuit using almost any VNA. The self resonance is the parallel resonance when you connect a series LC to VNA port. The resonance conditions becomes when the imaginary part of S11 becomes zero at a determined frequency… in the attached picture there are both the serial resonance frequency on marker 1 and parallel on marker 2. Parallel resonance result of parallel connection between L and Cp with is the parasitic capacitance , this is the self resonance frequency. The method is absolutely sensitive to calibration I recommend very good standards.
I don´t know about your application but self resonance are very unstable depending on dielectric properties of the form ( many times varies with temperature and moisture) and changes every time. The best is to have this resonance frequency very far from working frequency using a very well known capacitor.
If you need to know the self resonance to have the value of Cp parasitic capacitance its a very good method but there are other methods that gives it value at the work frequency. Cp frequency dependent. I showed a method to do it on other post , this is derived from a procedure with Q-meter.

If you need more details just let me know more about your interest.


Ing. Patricio A. Greco
Laboratorio de Calibración ISO 17025 AREA: RF/MW
Gral. Martín Rodríguez 2159
San Miguel (1663)
Buenos Aires
T: +5411-4455-2557
F: +5411-4032-0072
www.servicios-electronicos.com.ar

On 10 Dec 2024, at 15:09, jskinner58 via groups.io <jskinner58@...> wrote:

Have watched several videos on this and they all seem to leave out most of the setup. Looking to measure inductance but most interested in self resonance frequency. Resonance will be about 2 MHz I am guessing. Have it across Port1/S11 as seen in the videos but don't have display of anything like in the videos. I have looked at the Absolute Beginners Guide but doesn't cover this.





 

Thanks but I am looking for the specific setting steps to be able to get this sort of information with a NanoVNA. And what did you use for the display you posted? Looks like some sort of PC display or at least not NanoVNA.

Jim


 

I suspect your problem is not your setup (set 2 traces, one to LOGMAG and the other to Smith Chart) but rather your ability to interpret the Smith Chart. Suggest you search YouTube for "Smith chart tutorials" and ground yourself on those. Most of these videos consider a single point ONE FREQUENCY as it appears on the chart. Once you realize all the lines traced on the chart are just a collection of individual points tracing positions on the chart for each frequency in the range of frequencies you are sweeping, it will cease to be a mystery


 

On Tue, Dec 10, 2024 at 10:09 AM, <jskinner58@...> wrote:


Looking to measure inductance but most interested in self resonance frequency.
Resonance will be about 2 MHz I am guessing. Have it across Port1/S11 as seen
in the videos but don't have display of anything like in the videos.
In order to make a measurement of an inductor's parameters like inductance, reactance and resistance vs frequency and self-resonant frequency the easiest method is the S11 shunt method using just CH0 (Port 1) on the NanoVNA. You need to do the following:
- Construct a suitable test jig for the type of device you are measuring (SMD, leaded part, large part etc.)
- Connect a short cable from CH0 on the NanoVNA to the test jig
- Calibrate right at the test jig where the device under test (DUT) is attached to the test jig
- Measure the parameters of interest. For an inductor the self resonant frequency is where the reactance X=0

Measuring inductors and capacitors has been discussed many times in this group and if you search the archives you will find all kinds of discussions about ways to build test jigs, calibration loads for the test jig and how to interpret the results. To get you started here is one I wrote some time ago.

https://groups.io/g/nanovna-users/message/20848

Roger


 

No, although I am not that familiar with Smith charts my question remains the settings and not how to read a Smith chart - I'm not really interested in that chart at this point (but likely will be later on). You gave me two of them: set 2 traces, one to LOGMAG and the other to Smith Chart. And of course, set the start and stop frequency. My issue is also not my fixture that is very short and connected to the right port. I would like to see something like the R and X vs. frequency graph that is next to the last picture in this link that Roger Need posted.
https://groups.io/g/nanovna-users/message/20848
I think I might be able to get the other pieces from this but what other settings are needed?


 

You could also run it through the COIL software. It will spit out SRF. It also allows for several different types of former being used, and also the wall thickness of the former.


 

The Smith Chat is best suited for revealing both resonant frequencies and
values of an inductor. All questions are answered in the single Smith
Chart display of the measured data.

RESONANCE: Where the trace crosses the centrao horizontal line. This and
only this line represents pure resistance on the Smith Chart. And remember
the indication of resonance is pure resistance in the total absence of any
reactive component.

VALUE: Set the cursor to the intended operating frequency for the
inductor. Read the value either in + j X or the actual value in XHenries.

Dave - WØLEV

On Tue, Dec 10, 2024 at 10:35 PM jskinner58 via groups.io <jskinner58=
sbcglobal.net@groups.io> wrote:

Thanks but I am looking for the specific setting steps to be able to get
this sort of information with a NanoVNA. And what did you use for the
display you posted? Looks like some sort of PC display or at least not
NanoVNA.

Jim





--

*Dave - WØLEV*


--
Dave - WØLEV


 

On Tue, Dec 10, 2024 at 04:16 PM, JimLS wrote:


I would like to see something like the R and X vs. frequency graph that is
next to the last picture in this link that Roger Need posted.
You need to set your traces to read R and X. You have to go into the Display Menu and select a Trace and set it to S11. Then go back and set the Format to Resistance or Reactance. Then set the Scale to have the proper range on the display. Here is an example of an inductor with the traces set to R and X


 

This would indicate parallel resonance of the inductor as at resonance, the
measurement is indicating a very high resistance. The Smith Chart with the
cursor would give you the same information.

Dave - WØLEV

On Thu, Dec 12, 2024 at 11:48 PM Roger Need via groups.io <sailtamarack=
yahoo.ca@groups.io> wrote:

On Tue, Dec 10, 2024 at 04:16 PM, JimLS wrote:


I would like to see something like the R and X vs. frequency graph that
is
next to the last picture in this link that Roger Need posted.
You need to set your traces to read R and X. You have to go into the
Display Menu and select a Trace and set it to S11. Then go back and set
the Format to Resistance or Reactance. Then set the Scale to have the
proper range on the display. Here is an example of an inductor with the
traces set to R and X






--

*Dave - WØLEV*


--
Dave - WØLEV


 

Still, thanks Roger! I've never tried those settings! Didn't understand them..


 

I think the best way to measure inductor self-resonant frequency is to place it between the VNA ports and find the transmission loss peak. Using just a single port parallels its capacitance with the inductor. This can greatly affect measured SRF. For example, the measured capacitance of 1.9 pF for port 1 of a NanoVNA-H4 exceeds the 1.2 pF calculated self-capacitance of a 10 uH air-wound coil. SRF will measure 28.7 MHz instead of the true 46.1 MHz. No VNA calibration is needed for this measurement.

Brian


 

VNA must to be calibrated to perform any measurement. You measure at reference plane no matter the internal impedance of VNA , this is the part of magic of this instrument.

Ing. Patricio A. Greco
Gral. Martín Rodríguez 2159
San Miguel (1663)
Buenos Aires
T: +5411-4455-2557 ( tel:+5411-4455-2557 )
F: +5411-4032-0072 ( tel:+5411-4032-0072 )
www.servicios-electronicos.com ( http://www.servicios-electronicos.com/ )


On 13 Dec 2024, at 00:22, Brian Beezley via groups.io
<k6sti@...> wrote:


 I think the best way to measure inductor self-resonant frequency is to
place it between the VNA ports and find the transmission loss peak. Using
just a single port parallels its capacitance with the inductor. This can
greatly affect measured SRF. For example, the measured capacitance of 1.9
pF for port 1 of a NanoVNA-H4 exceeds the 1.2 pF calculated
self-capacitance of a 10 uH air-wound coil. SRF will measure 28.7 MHz
instead of the true 46.1 MHz. No VNA calibration is needed for this
measurement.

Brian







 

That's correct, Patricio, but the OP expects a resonance at 2 MHz which means a Lambda of 150 m, thus the reference plane is not that critical.
Especially as the component is directly connected to the VNA ports without any cables if I understood that right.

- Bonzo -


 

Bonzo , this is a different concept. I say reference plane because is the way to define the measurement point. Its clear that the wavelength is bigger than normal working dimensions in this case , but the problem are the accuracy of measurement and the parasitic components, you can for example add a fixed 100pF capacitor at the VNA port and then perform the OSL calibration after that, over the fixed capacitor you connect another capacitor let 100pF the VNA will measure close to 100pF ignoring the fixed value. This is the VNA calibration. If you use 2 ports to perform a good measurement you need to perform a FULL 2 PORTS calibration to completely define the reference planes or the connection points. Many nanovna has a only 2 parameters , in this case I recommend to use only one port (port 1 ) very good calibrated.
Other example is a RLC meter , you have an open and short definition to calibrate the instrument and lambda is kilometric… you need to null the parasitic components prior to measure.
Again, all depends on your accuracy requirements.

Regards, Patricio.


Ing. Patricio A. Greco
Laboratorio de Calibración ISO 17025 AREA: RF/MW
Gral. Martín Rodríguez 2159
San Miguel (1663)
Buenos Aires
T: +5411-4455-2557
F: +5411-4032-0072
www.servicios-electronicos.com.ar

On 13 Dec 2024, at 06:45, Bonzo via groups.io <mora@...> wrote:

That's correct, Patricio, but the OP expects a resonance at 2 MHz which means a Lambda of 150 m, thus the reference plane is not that critical.
Especially as the component is directly connected to the VNA ports without any cables if I understood that right.

- Bonzo -





 

Patricio, thanks for correcting me about calibration nullifying port capacitance. However, I don't think it can nullify stray capacitance to ground due to a fixture or large inductor size. When using a single VNA port, this capacitance appears in parallel with the inductor and alters its SRF. But if the inductor is placed between ports, the capacitance appears from the ports to ground. This does not affect SRF if measured as the frequency of maximum transmission loss. To verify this, I modeled the circuit shown below with stray shunt capacitance and stray lead inductance. C2, C3, L2, and L3 had no effect on the transmission null frequency.

I didn't make the measurements myself, but SRF values near 1 MHz for 2.2 mH inductors differed for single-port and two-port VNA measurements.

Though not needed to measure SRF, this writeup describes a method to avoid the effects of stray capacitance when measuring inductor resistance and reactance:

https://k6jca.blogspot.com/2020/07/the-y21-method-of-measuring-common-mode.html

I've implemented the Y21 method here:

http://ham-radio.com/k6sti/stray.zip

Brian


 

On Fri, Dec 13, 2024 at 06:44 AM, Brian Beezley wrote:

However, I don't think it can nullify stray capacitance to ground due to a
fixture or large inductor size.
Maybe just lifting the inductor ground lead while doing the open calibration would work.

Brian


 

Brian: All depends on your accuracy requirement , how you build your calibration standards … for special case of serial resonance you can use 2 ports but you can only measure this frequency. The self resonance frequency is not very useful on RF applications, this is a problem. Is good to know it to have it far from working frequencies.For this reason my approach are more focused to measure other parameters more useful in my point of view. We measure a parameter for a reason, my reason is to design an inductor ro my application or validate a calculus. In general this is one more of parameters under analysis to measure. I posted an example in other subject I will repeat it :

Finally I got some time to perform measurements. I used an E5061A VNA (a nanovna can be used as direct replacement, this is nothing special). I started calibrating my port 1 on SMA male. I build a simple coil over a plastic form and I used a 160pF mica plate capacitor in series. I got a series resonance frequency of fs= 2.718 MHz and a series resonance due pasitic capacitance (Cp) of fp=15.757 Mhz. I used segmented sweep to have more frequency resolution at fs.

If I short the coil and I measure the capacitance this is @fs C= 162.14 pF performing some calculations* Cp = 4.97 pF , then we can compute L = 20.518 microH at fs the series resistance was Rs= 2.314 ohm and this is Q= 151. There are more information on attached pictures.
If we not consider the Cp(=0) … the results are : L= 21.147 microH Q= 156.

*The series capacitance is close to the ratio of fp and fs squared minus one times Cp : C= ((fp/fs)^2 -1) Cp

In my opinion this is a complete set of measurements for example to design an inductor of a resonant LC for a trap or part or any major design. Is interesting the idea to shunt the inductor to measure the capacitor C and then compute the rest of values. Measure only the self resonance would be interesting for a quartz resonator but not for an inductor where you need to know his self inductance value and the Q factor … again : with error on results would be acceptable in your project ? it defines how to measure. A Q factor with an error of 10% is very good for the most applications, this is a reference value , in actual designs it don´t defines the bandwidth of an amplifier for example , modern filters are used now.


Ing. Patricio A. Greco
Laboratorio de Calibración ISO 17025 AREA: RF/MW
Gral. Martín Rodríguez 2159
San Miguel (1663)
Buenos Aires
T: +5411-4455-2557
F: +5411-4032-0072
www.servicios-electronicos.com.ar



On 13 Dec 2024, at 11:44, Brian Beezley via groups.io <k6sti@...> wrote:

Patricio, thanks for correcting me about calibration nullifying port capacitance. However, I don't think it can nullify stray capacitance to ground due to a fixture or large inductor size. When using a single VNA port, this capacitance appears in parallel with the inductor and alters its SRF. But if the inductor is placed between ports, the capacitance appears from the ports to ground. This does not affect SRF if measured as the frequency of maximum transmission loss. To verify this, I modeled the circuit shown below with stray shunt capacitance and stray lead inductance. C2, C3, L2, and L3 had no effect on the transmission null frequency.

I didn't make the measurements myself, but SRF values near 1 MHz for 2.2 mH inductors differed for single-port and two-port VNA measurements.

Though not needed to measure SRF, this writeup describes a method to avoid the effects of stray capacitance when measuring inductor resistance and reactance:

https://k6jca.blogspot.com/2020/07/the-y21-method-of-measuring-common-mode.html

I've implemented the Y21 method here:

http://ham-radio.com/k6sti/stray.zip

Brian





<ckt.png><resp.png>


 

On Fri, Dec 13, 2024 at 08:44 AM, Patricio A. Greco wrote:


A Q factor with an error of 10% is very good for the most applications
Patricio, this calculator may interest you for inductor design. Its Q accuracy is within a few percent on average for test coils measured with an HP 4342A Q meter. Worst case Q error is < 10%. It has an optimizer for maximizing Q. It can also optimize a coil for minimum loss in a trap. Much to my surprise, this does not occur at maximum coil Q.

http://ham-radio.com/k6sti/coil.htm

Optimizing wideband antenna models with coils or traps:

http://ham-radio.com/k6sti/wbc.htm

Brian


 

Biran : I downloaded the program and I entered the values of my measured inductor Leq= 21.2 michoH , this is very good result. I suppose the Resistance would be computed via skin effect. For any reason looks like includes the parasite capacitance, this is ok if you want to compute the parallel capacitor for a trap for example or if you with to use the inductor on a circuit design ( for this reason I assigned it to (Leq)uivalent).
L varies on frequency if you see it goes up this is because you have Cp in parallel then que equivalent Leq= L1/(1- Omega^2 L1 Cp) , L1 is the pure self inductance. Q is ok ! Freq/SRF measured is 0,17 and this is also good.

The Q mismatch you mention comes from this error in my opinion , I would need to verify via calculus… but if you read my post you will see that Q goes down if I consider Cp the capacitor used on trap also reduces Q.

This is a very good program!!

In this example (I posted a capture) I set 2.7 MHz as reference Leq= 21.2 microH , the capacitor to made a trap is C= 163.9 pF we can approximate Cp = 2.81 pF (this is using the program F/SRF ratio) , L= 20,84 microH then If we ignore Cp Q= 169.8 with Cp Q= 166,75. I recommend to perform this calculus and check because in the most of cases the coil Q defines the resonator Q . In my measurement the capacitor Q is ~ 20 times the inductor … you can see when I measured C ESR= 0,13 Ohm… this is not perfect but Q functions like parallel resistors … Qtrap= QL QC/(QL+QC) then Qtrap will go down by 5%

Other source of error is because Cp is in general a very bad capacitor , in general results of form dielectric constant and wire enamel… with the most of energy in a lossy capacitor Q goes down… it also affects the self resonance frequency measurement because the dielectric constants varies with frequency for this reason there are another method to compute Cp at the work frequency I shown it in other post.

Of course you have optimum values for higher Q it depends on wire length and geometry.




Ing. Patricio A. Greco
Laboratorio de Calibración ISO 17025 AREA: RF/MW
Gral. Martín Rodríguez 2159
San Miguel (1663)
Buenos Aires
T: +5411-4455-2557
F: +5411-4032-0072
www.servicios-electronicos.com.ar

On 13 Dec 2024, at 14:06, Brian Beezley via groups.io <k6sti@...> wrote:

On Fri, Dec 13, 2024 at 08:44 AM, Patricio A. Greco wrote:


A Q factor with an error of 10% is very good for the most applications
Patricio, this calculator may interest you for inductor design. Its Q accuracy is within a few percent on average for test coils measured with an HP 4342A Q meter. Worst case Q error is < 10%. It has an optimizer for maximizing Q. It can also optimize a coil for minimum loss in a trap. Much to my surprise, this does not occur at maximum coil Q.

http://ham-radio.com/k6sti/coil.htm

Optimizing wideband antenna models with coils or traps:

http://ham-radio.com/k6sti/wbc.htm

Brian