New tool - toroidal ferrite core inductor calculator

Discussion in 'Antennas, Feedlines, Towers & Rotors' started by SP3L, Jun 14, 2021.

1. SP3LHam MemberQRZ Page

No problem. The same sentence is now also repeated in the initial section of the page.

Click the blue link below which will take you to my page:
Contents - Toroidal Ferrite Core Inductor Calculator (google.com)
and go to the page.
and click the word "here".

Visit the page from time to time to see if you have the latest version. The current version is 1.2.

2. SP3LHam MemberQRZ Page

You are right. This sentence results from my bad habit of making shortcuts in explanations. So, I am going to be more systematic now.

Let's start with an inductor having a single core in which you have adjusted the number of turns so that it has the highest possible impedance over the frequency range you are interested in. Let's say its is

|Z| >2000 and R>500 ohms for F=2...30 MHz

Material 43, size 2.4" OD, number of turns = 11.

Now let's add to the plot an inductor with 3 cores, the same material and the same number of turns. This is the blue plot.

For the three-core inductor, the impedance and resistance increased but the self-resonance frequency (SRF) decreased significantly. As you can see in the picture above, it is no longer suitable for the 2...30 MHz range because its impedance and resistance are too small at higher frequencies.
We must reduce the number of turns to move the SRF towards higher frequencies. We have to go down to 5 turns to bring the multicore choke to the desired frequency range - see below.

We have now a multicore choke for the 2...30 MHz range with LOWER common-mode impedance and resistance than the single core choke!
The only benefit is higher power rating.

I hope it is clear now what I meant.

3. EA1DDOHam MemberQRZ Page

Jacek, Dan, I can see the tool is able to calculate X too (among others, including Rs, Ls, Cp) for a single frequency.
Would be good to plot X as well.
Is that option on your list for new releases?

Thank you

73, Maximo

4. W9IQHam MemberQRZ Page

The graphic showing three cores vs a single core underscores the often ineffective use of multiple cores. By putting two single cores in series, you double the real resistance and typically more than halve the common mode power dissipation per core. And the goal should always be more resistance, not more common mode power handling (cure vs treatment of symptoms). You do need to consider the differential mode power losses in some applications but these tend to be outliers.

- Glenn W9IQ

5. SP3LHam MemberQRZ Page

Thank you Glenn for concluding my previous post. I absolutely agree with you.

6. SP3LHam MemberQRZ Page

Maximo,

well, it could be done although not without a little problem. X can be positive or negative. When you plot |Z| and R they are always positive numbers so you can use linear as well as logarithmic Y scale. Logarithmic scale can not be used for negative values.

I have seen charts showing |X| (the module of X) rather than X. They used logarithmic scale. You need to be smart enough to know which part of the plot you must negate - treat as negative numbers. You can get used to it after a while.

Moreover, we would have to add a small number to |X| to avoid zero which is not permitted either when you use logarithmic scale. So, we could plot |X|+0.01 ohm or something like that when logarithmic Y scale is selected and simply X when linear scale is selected.

But, I am reluctant to add extra plots that contribute very little to the choke design/selection process. If you have access to the QEX Magazine please read my article mentioned in the very first post in this thread. QEX latest issue is already available to the ARRL members. I explained in the article why the X component of the common-mode impedance is of little significance. In short - the X component of the choke can be cancelled by the coax feeder reactance of opposite sign if only the feeder has the proper length between the choke and the grounding point. What cannot be cancelled is the resistance of the choke. If you design the common-mode choke focus on its resistance. It tells you what you can expect in the worst case - when the coax length is most unfortunate. Maximize R, do not care about the X component.

EA1DDO likes this.
7. EA1DDOHam MemberQRZ Page

Not sure if eveyone will agree that... but I'll take it into account.

Thank you Jacek.

73, Maximo

8. SP3LHam MemberQRZ Page

Yes, this is not very obvious.

When calculating the common-mode current (which also directly impacts the power dissipated in the choke) you must take into account not only the impedance of the choke itself but also the impedance of the coax feeder. The common-mode impedance of the choke and the common-mode impedance of the coax are connected in series.

The common-mode impedance of the feeder depends on its length - measured from the choke to the point where the coax can be regarded as grounded (for example: where it touches ground for the first time and in this way is grounded via capacitance).

When the feeder length is equal to the even number of a quarter wave (L=2/4, 4/4, 6/4.... lambda), its impedance is purely resistive and very small (a few tens of ohms). So, it practically does not contribute to the total common-mode impedance of the choke-feeder network. In such a network, the magnitude of total impedance is practically equal to |Z| because we neglect the small impedance of the feeder.

If the feeder length is slightly smaller or slightly greater than the even number of a quarter wave (L=2/4, 4/4, 6/4.... lambda), the resistive component of the feeder impedance is still very low (tens of ohms) but the reactive component is no longer zero. This reactive component of the feeder impedance can be of opposite sign to the reactive component of the choke impedance and in such a case they may cancel one another. In such a case, the total impedance of the choke-feeder network drops almost to R (small resistive component of the feeder impedance neglected). And this is the worst case.

When the feeder length is equal to the odd number of a quarter wave (L=1/4, 3/4, 5/4, ... lambda), the common-mode impedance of the feeder is very high (about 3 kohm) so it contributes quite a lot to the total impedance of the choke-feeder network. In such a case, the magnitude of total impedance is more or less equal to |Z| + 3 kohm.

To put it most simply:
• when the feeder length changes, the total common-mode impedance magnitude of the choke-feeder network changes from about R to about |Z| + 3 kohm
• |Z| is the median value
• R is the minimum value
Thatâ€™s not 100% accurate but good enough for practical considerations.

9. SV1IYFHam MemberQRZ Page

I sent to Jacek by e-mail a new set of measurements. The graphs previously uploaded here (at page 4) have a very poor accuracy.
There were two reasons for this:
First, the OSL meter calibration doesn't seem to be accurate enough. In today's measurements I did the corrections manually, subtracting the real and imaginary parts of the measured cores from the naked measurement harness (empty loop).
Second, RigExpert shows quite a variance among isolated measurements (low repeatability). Instead, "NanoVNA saver" piece of software has the ability to average multiple measurements while at the same time discarding outliers which filters out a lot of the measurement variance.
My tests prove, within my instruments' accuracy, that combining cores of different materials is a linear process, that is, not better or worse than stacking two cores of the same material.
A totally linear and hence anticipated difference is that on a frequency range where the 43 core has already become capacitive while 61 is still inductive, they mutually counteract each other and X is reduced until both cores become capacitive again. On a 2x43 core stack this X cross-over appears somewhere between 50-70* MHz while for a 43+61 combo this moves higher to somewhere 70-100* MHz.
The benefit of such a combo starts becoming apparent above 50* MHz.
In numbers, 2x43 |Z| for 50, 70, 100 MHz: 117, 127, 135 Ohms and 105, 139, 228 Ohms for 43+61.
Naturally, the real part (R), of interest to CM filtering, is not affected.
(*) Mind, that these numbers are for a single turn (N=1) and as turns increase so does self capacitance, pushing frequencies downwards.
73 Nikos

Last edited: Jun 22, 2021
10. SP3LHam MemberQRZ Page

Version 1.3

It is available now. Dan improved mainly the workbook behavior when the calculator is downloaded and opened for the first time. The changes are as follows:
• you cannot go beyond the opening page if macros are not enabled,
• you cannot go beyond the opening page if any version of Excel older than 2007 is being used,
• you cannot save the workbook in *.xls format (to prevent creating non-working variants of the calculator)
• the initial content for the 2xRLC sheet has been changed to match the Help file pg 8 example, where the same target shows a better correlation with 2xRLC vs 1xRLC.