#80: High Power Balun with #31 Ferrite Material

[@jonathancristole6074]

Thanks for the video. I’m operating at 100w so based on this I think I’m good even with a random wire antenna, but definitely with a correctly measured dipole. Have a 240-31 which I’m going to wrap with some RG316 and will just test myself it to make sure. Again, thanks for the video.

[@w0qe]

You are welcome! Larry

[@Stovetopcookie]

Would love to see more videos on core selection based on determining flux requirements and power level.

[@galileo_rs]

Very interesting, would also like to see a simulation with a 43 mix core as that mix is also commonly recommended for baluns.

[@w0qe]

See my reply above.

[@tomthompson6911]

Very interesting and practical, Larry. I would like to see the multiple core video done.

[@Stovetopcookie]

This is very helpful and presented easy to follow. I am so glad to have found your channel. Thank you.

[@jimbrown1559]

Very interesting analysis of power handling. Several observations. First, NA6O has done some careful measurements of temperature rise with chokes like this for his upgrading of the N6RO contesting station, and his results are not at all out of line with yours. I've done extensive work on the design of common mode chokes for the HF bands -- in the last year, I've built and measured the impedance of nearly 500 chokes on my stash of more than 160 of the same core you used here. The chokes have wound with RG400, a #12 pair of Teflon insulated wire, and with THHN. All ferrite parts designed for suppression have a data sheet tolerance of 20%, so before winding the chokes, I first characterized all 160 of the cores themselves with a test winding of 10 turns of RG58, placing cursors at the HF bands, and tabulating complex Z in a spreadsheet at those cursor points. So my second observation is that the variations in complex Z from one core to another have significant effect on the resulting impedance curves. Third, for the reasons your work clearly shows, the choke must have rather large value of Rs at the operating frequency, which is achieved by adjusting the turns ratio and the winding style. Fourth, a good choke USES the resistance around the parallel resonance between the inductance and the interwinding capacitance -- the latter will depend on conductor size, turns spacing, number for turns, and dielectric. For practical chokes the resonance is broad, and in #31, because it is an MnZn ferrite mix, has both the resonance of the winding AND a dimensional resonance that occurs within the core due to standing waves within the core, the two resonances can combine to further broaden the impedance curve. The equivalent circuit of a choke with dimensional resonance is two parallel resonant circuits in series, and the dimensional resonance tends to be rather high Q. Depending on the number of turns and the core size, the two resonances act much like the "stagger-tuned" IF transformers used when my hair was brown instead of grey. Fifth, because common mode chokes are inserted in antennas, and because current and voltage varies along an antenna, common mode dissipation in chokes must be modeled as part of the antenna. I do this by inserting them as loads in a wire having the geometry and insulation of the transmission line (the outer jacket in the case of coax). Sixth, mu' and mu' ', which account for Rs and Xs in a single turn choke, both vary over frequency, usually by a order of magnitude. Seventh, Rs is far more important than Xs or Zmag, because Xs of the choke can be fully or partially cancelled by the Xs of the rest of the common mode circuit AT THE POINT where the choke is placed, but legal limit power handling can be achieved if Rs is made sufficiently high, and if chokes (not cores) are used in series. Eighth, as your work here clearly shows, these chokes are not suitable for use in lines with significant mismatch or significant imbalance. Either condition significantly reduce power handling. For legal limit chokes, my latest choke cookbook recommends Rs values of 10-15K ohms, and provides a matrix of designs and Rs values for the three line types. N6BV authored a great piece in QST several years ago called "Don't Blow Up Your Balun" that pointed out the dissipation issues associated with mismatch. Ninth -- if two identical chokes are placed in series, common mode current and total dissipation will be divided by two, and the dissipation in each choke divided by four. If the chokes are not identical, those relationships depend on their individual Rs and Xs values. I chose cores at the limits of those I'd measured for Rs and Xs of the ten-turn test winding, and wound chokes on those cores with the various line types. Initially, I measured a lot of chokes with other line types before settling on those types as the most practical. The tables are based on the worst case Rs values for chokes wound on those four "limits" cores. The new cookbook is at k9yc.com/2018Cookbook.pdf An analysis of how chokes work is in k9yc.com/RFI-Ham.pdf , which includes a reference to E. C. Snelling's classic text from which I learned about dimensional resonance. Finally, someone asked about #43 for HF chokes. #43 is a NiZn, so it has no dimensional resonance, and the Q of the resonance of chokes wound on it are much higher than #31, and thus much narrower. This higher Q exacerbates the effects of component tolerance, so the "limits" approach I used for the cookbook for #31 cores would not work for #43 -- the individual ham would have measure every choke and tweak turns to place the resonance. The #52 material proposed by G3TXQ has the same shortcoming. How do I know ? I bought 40 cores this winter, splitting the order between four Fair-Rite dealers, characterized the cores, and wound chokes using limits cores. 73, Jim K9YC

[@w0qe]

Hi Jim, Thanks for the comments and I have previously read your 2018 cookbook. Keep up the good work! I have also read many of the references you often mention and initially tried to analyze the conclusions with programs such as LTSPice. However I was never as successful as I wanted to be until SimSmith version 16 became available. I mention this all the time, but it's a good day when you can build something that measures closely to what you simulate. 73, Larry, W0QE

[@hepcat67]

Hi Jim: In your Choke Cookbook, you mentioned that a pair of #12 teflon wires has an impedance of about 96 Ohms. I plugged the figures into a couple of online calculators, and they give a much lower result. Can you explain the discrepancy please? Also, the link in your paper for the TL data doesn't work. Thanks.

[@jimbrown1559]

@@hepcat67 It''s measured data -- open and short circuit impedance sweeps, post-processed using AC6LA's ZPlots. Remember that TL parameters are affected by spacing, and the dielectric material.

[@hepcat67]

@@jimbrown1559 Yes, I entered the dielectric constant for teflon and 109 mils as the spacing, and then tried 80 and 88 mils as the conductor diameter to vary the insulation thickness. I got either 56 or 68 ohms. To get 96 ohms, I had to increase the wire separation to 140 or 150 mils, which would mean that there would be air between the wires. This paper mentions the risk of measurement errors by the mating of the balanced line to the unbalanced VNA: www.mwrf.com/test-and-measurement/determine-twisted-line-characteristic-impedance I'm wondering if that's the reason for the discrepancy? This is the calculator I used: www.easycalculation.com/engineering/electrical/parallel-wire-impedance-calculator.php Thanks.

[@jimbrown1559]

@@hepcat67 The same test setup produces TL parameters for CAT5 pairs that agree with published specs. You're welcome to build some line and measure it.

[@no5x937]

Larry, can you explain what VNA, PC platform, and PC software you used to create the two .csv files for the 11 and 14 turn chokes? Where did you come up with RCM = 290 ohm and XCM = j600? Can you provide a link to Kevin Smith W9CF's paper on common mode impedance? Thank, JT

[@w0qe]

I used an AIM 4170 analyzer in this video. However SimSmith/SimNec can read several other file formats most notably .s1p which can be created by the VNWA and the NanoVNAs. What VNA do you have? Larry, W0QE

[@no5x937]

@@w0qe I don't own any VNA's today, but Christmas is right around the corner. I'm looking at the nanoVNA V2 1.5 GHz (or higher) versions or the LibreVNA. What I'm wrestling with is how to calculate then simulate the internal termperature of an RF toroidal core (1uH inductor using T-68 powdered iron). I understand how to calculate the flux density of a core based on the voltage across the core, #turns, frequency, and Ae. But how do you calculate the Curie-Weiss temp effects on the core's permeability starting to drop? At 5:30 in this video, you state the heating effects on the coil are related to the voltage across it, not the current through it. What formula is this based on?

[@martinsole7597]

Great stuff Larry. and very interesting. Please do the same with a 43 core and show some comparisons.

[@galileo_rs]

+1

[@w0qe]

Martin and Dragan, The puzzle here is that you have two choices: first is to keep R very low in the common mode impedance so that I^2 * R does not burn up the core or second is to make R so high that the current is so low that again I^2 * R does not burn up the core. #61 material works pretty well for the first case and #31 for the second case. Unfortunately #43 material is right in the middle and I don't think the results for it will be good. I am becoming more of a #31 material fan all the time. :) Larry

[@IZ4HDM_Chris]

Larry, you're definitely exposing arguments that are obscure for most of the hams. Please don't stop in doing this, we all appreciate your contribution. IZ4HDM Cristiano

[@w0qe]

Cristiano, The videos I make are a mix of topics that I feel are important to the understanding of RF, topics that I think would be interesting to the user community, topics of interest to me, and finally topics that are suggested to me by viewers. I hope that you will continue to find content that I produce to be useful. Feel free to comment anytime. Larry, W0QE

[@wb5mgr]

Larry, is this also going to be an appropriate way to model a ferrite bead or snap on core that is on the transmission line? Or do you have a better way?

[@w0qe]

There really isn't any difference in modelling a single turn device or a multiple turn device. However remember that SimSmith is a circuit analysis program so that things like determining the common mode source and load impedance will need to be determined by measurements and estimation. Larry, W0QE

[@tommiller1710]

Very interesting Larry. I would like to see more especially with 43 mix and taking it to the extrees like you did here and see what happens with higher load impedances like you started to do here with the 400 Ohm simulation. I will be looking for the files to play with it also. Keep up the GOOD work. WB8ICT

[@w0qe]

Tom, Below the video is a link to the files in a zip format. I replied to Martin below about #43 material. Larry, W0QE

[@BritishBeachcomber]

Also, 8 turns seems to be optimum for HF.

[@w0qe]

Peter, If only it was this easy! 8 turns may seem like a reasonable level but I can show that a balun placed in an open wire feed of a dipole used for multiple bands at a 1500W power level will be challenging to build without overheating. This is why the common mode impedance of the balun needs to be measured and an analysis done to see if the core heating is acceptable. SimSmith can do these calculations easily. We live in a wonderful time to be able to have inexpensive hardware to measure RF parameters and software to do analysis of problems. Larry, W0QE

[@w0qe]

Below the video is a link to the files if you wish to download them.

[@PixelSchnitzel]

Dang it, Larry, you've been reading my mind! I'm just starting to tinker with mix 31 and have some beads & toroids coming in the mail soon. I'm a little glazed over after the first pass at the video, trying to digest the "why's". But like most, I'll watch a few more times and the lights will click on. I think I'm "getting it" with the transformer model, though. So I'm confident the rest will click soon. Thanks again! The wonders of this stuff are endless.