Transmitter Band Switching Issue

[AB2EZ]

Yesterday evening, I was having one of many very pleasant conversations with Carl, WA2UJX.

Carl asked me what, at first, seemed to be a simple enough question:

“Does it matter if the band switch in a multi-band transmitter shorts together all of the taps of the unused portion of the tank coil, or is it just as good to short together just the two ends of the unused portion of the coil.”

As is often the case, what seemed, at first, to be a simple enough question turned out to be more complicated than I thought.

After thinking about this for a few hours, I thought I would share the answer that I came up with.

Bottom line: Yes, it is much better to short together all of the taps of the unused portion of the tank coil.

There are also some surprising (to me) issues related to how multi-band tank circuits really work

Here is why:

Attachment 1 shows the simple case of a tank coil with 20 turns, 12 of which are unused.

A circuit model of this simple case is shown in Attachment 2.

The two portions of the tank coil are, in fact, an air coupled transformer, with a coupling coefficient that is determined by the details of how the coil is constructed.

The unused portion of the tank coil is basically shorted out with a resistance (R) that depends upon (among other things): the diameter of the wires (skin effect), the diameter of the coil, the number of turns, and the details of how long the wires between the coil and the switch are.

Note that in attachment 2, the leakage inductance is less than the inductance of a 8 turn coil, even though 8 turns of the coil remain un-shorted. The sum of the leakage inductance and the magnetizing inductance is equal to the inductance of an 8 turn coil. Therefore, with the unused portion of the coil shorted out, the inductance of the remaining (not shorted) portion of the coil is significantly less than it would be if one were to remove the unused section of the coil (rather than just shorting it out).

Attachment 3 shows why it is much better to short together all of the taps of the unused portion of the coil.

If you have (for simplicity) a coil with 12 unused turns, with a tap every 3 turns… then the power dissipated in the shorted turns (heating of the turns, and reduced output from the transmitter) is 4 times larger if only the ends of the unused portion of the coil are shorted together vs. shorting each of the 3-turn sections.

[ab3al]

only one comment to make.

Its really nice having a professor among us to give really detailed answers.  there is almost no question left to ask on the covered topic after DR. S gives his lecture.  good job pal...

now if'n he culd just help me wif my spellin

[WD8BIL]

I ain't seein' nufin rong wif yor spelin, Mike !

[The Slab Bacon]

I have just about always built stuff with the bandswitch shorting out the unused part of the coil. I know that it does make a difference in the shunt capacitance needed to resonate the circuit. I have seen this when testing with a GDO.

But shorting the coil has been the accepted way of doing it for many years, there must be a reason for it. Every major transmitter MFR has done it that way since time started.

Sometimes it is just better to accept the fact that something works than beating your brain up trying to figure out why it works. (if it aint broke, dont fix it)

                                                                    The Slab Bacon

[steve_qix]

Interesting !

I know from the class E work, if there are a FEW turns remaining at the end of a coil, it is better not to short it out (get much more coil heating with shorted turns than with not).  This is strictly by observation - I have done no calculations whatsoever involving this.  I have about 5 turns hanging out there on my 75 meter big class E rig (the total coil is about 25 turns total), and if I short out these 5 turns, the coil gets noticably hotter.

When I build multi-band transmitters, I tend to use seperate coils if there are a lot of turns involved to avoid the coil heating from shorted turns.

Another trick is to spread the coil out for the higher bands - reducing the coupling and therefore the heat produced in the shorted turns.

I'm wondering if the impedance of the circuit (single-digit values for the class E rigs as opposed to thousands of ohms for tube circuits) makes a difference, and the number of turns "out there" (not used) as a function of the total turns of the coil.  That would be a very useful analysis.

Nice evaluation!

Regards,

Steve

[AB2EZ]

I appreciate the compliments... but I think I owe everyone an apology.

I didn't do the arithmetic right. ["Do I get part credit for having the correct equation and just adding up the terms incorrectly"] The answer I came up with (see attachment 3 of my original post) didn't make any sense to me after my second cup of coffee this morning.

The analysis I did would suggest that it doesn't make any difference whether you short the unused portion of the coil end-to-end, or whether you short all of the taps of the unused portion of the coil together.

I think that there are, indeed, more subtle issues that impact on whether it makes a difference if you short all of the taps together... but I have not identified what they are.

The corrected attachment 3 is attached here.

P.S. I also agree with Steve that leaving the unused portion of the coil disconnected will eliminate the losses in the unused section... provided it doesn't have too much capacitance between its own turns, and provided it doesn't have too much capacitance to to nearby conductors.

Best regards
Stu

[K3ZS]

I have learned that when building L-tuners, that if you leave coil turns unshorted, it some situations the coil acts like an unterminated autotransformer.  This sometimes results in arcing and Tesla coil effects.

[WA1GFZ]

My 4X3 rig covers all bands from 80 to 10 incluning WARC bands so it wasn't possible to use a shorting switch. I have never had a problem doing this as long as the switch can handle the voltages along the coil taps. I've blown a number of smaller switches. So a shorting switch will allow you to use a smaller part in my view. I'm glad the smart guy provids deeper thought on all these interesting subjects.

[nu2b]

Hi Folks...Great thread!

I had scratched my head a while back concerning this subject, with the goal of estimating tapped coil parameters for use in spice modeling. As mentioned by the group, mutual inductance and coupling coeffs are not exactly intuitive concepts.

So, below is a link to a spreadsheet which can be used to estimate Q, mutual-L and coupling coeff for tapped airwound coils. Just input Dia, N-turns,TPI,selected tap and frequency.

http://www.qsl.net/nu2b/xls/TapCoil20080518.xls


The results were used to model an 80-40 meter bandswitch. A schematic is also posted to show some results. A look at the schematic notes will help folks gain insight into the dissipation losses as a function of coupling coeff K.

If you play with the spreadsheet, it's interesting to note that (for a given inductance value) that short, fat coils have a greater coupling coeff than longer, skinny coils. This might lead to greater losses for short, fat coils with shorted turns.

Don't miss the note on incorporating the coil as part of a PiNet and possibly noting dramatic power loss increase as the output "loading cap" value is increased. Remember, on 80Meters, an 800pf loading cap will shunt the same amount of current as the 50 ohm load. This would double the coil loss.

On the subject of multiple-shorted taps...Maybe someone wants to come up with a NEC finite-element model of an inductor (using hundreds of descriptor lines)...probably not! My initial guess is that since a shorted "turn 2" would tend to shield the adjacent "turn 1" that there is not much difference in loss.

Regards, BobbyT NU2B...Is this stuff fun or what?