Common mode chokes

From Contesting

Jump to: navigation, search

It is conventional wisdom that low-gain receiving antennas, such as flags, pennants and K9AY loops require careful attention to preventing common mode current flow on their feedlines. The explanation is that the feedline shield serves as an antenna, and can couple local noise as well as signals from unwanted directions back through the antenna's matching transformer to the receiver.



I recently installed a Clifton Laboratories active antenna for use with my SDR-IQ and CW Skimmer. It works well, but since the whip is only 8 feet long, noise pickup could be a problem. Here is a baseline picture of the noise level in the SDR-IQ with no antenna connected.
160M noise floor2.JPG

Note that the average noise level is approximately -122 dBM (on the scale to the left).

No choke

Next, I connected the active antenna to the receiver through a length of quad-shield RG-6, with no common mode choking. The result is shown below.
No choke.jpg

Quite a difference, huh? I've been told that the double humps ~70 KHz apart are probably a switching power supply somewhere, but that aside, the average noise level between the humps is about 30 dB higher than with no antenna. This was recorded in broad daylight on 160M, and I watched carefully for any indication of electrical storm activity within ground-wave distance. Wind was almost dead calm.

Conventional bead choke

Now let's add a common mode choke. For my first effort in this respect I am going to add a 100-bead choke of #73 beads right at the antenna.

100-bead conventional choke.jpg

I've got to be careful not to overanalyze this, because today is cloudy and threatening rain, so the ambient noise level may be a bit higher than before. It looks like the 70 KHz humps are down about 5 dB, and the noise between the humps is down 2-3 dB. Not great return on the choke. My next trick will be to wait till clear sunny weather and retest.

One added point - the spike just to the right of the first hump is a reference signal that I produced by connecting my MFJ-259B to my shunt fed tower. Skimmer says the signal-to-noise ratio of that signal is 51 dB - very loud. Spectravue only shows it to be about 25 dB out of the noise - quite a discrepancy. Alex warns that the on-the-fly SNRs reported by Skimmer are not to be trusted, and that only spots have SNR really well computed. Oh well...

Pete Smith, N4ZR 15:11, 28 June 2009 (UTC)

Center-tapped bead choke

Wow! I just tried the same 100-bead choke, but with a center tap off the coax shield connected to ground about 20 feet from the antenna. This is per the ON4UN low-band DXing book, figure 7-88. Same reference signal, same everything, and look!

Center-tapped choke.jpg

The base noise level is down to about -115dB, only 7-8 dB above the receiver with no antenna connected, but 20 dB better than with no choke. The humps are also much reduced, by about the same amount. As a result, the desired signal jumps out of the noise.

Next step, I'll make some screen shots at various frequencies, and then the piece de resistance - I'm going to substitute a Clifton Laboratories super-duper center-tapped choke, and compare.

Pete Smith, N4ZR 21:10, 3 July 2009 (UTC)

Center-tapped bead choke details

Several readers have asked for a photo of the common mode choke.
Center-tapped bead choke.jpg
Click on the picture for a larger version. The same is true for all the spectrograms above, as well.

The continue to be a lot of questions about the common-mode choke arrangement, and many of them exceed my limited technical competence. Rather than get it wrong, I've decided to quote ON4UN's description from the 4th edition of Low-Band DXing, figure 7-88 and related text. I trust this falls within the fair use copyright exception.

Center-tapped choke schematic and explanation.jpg


Now you can take care of the transformer, as explained in Sections 2.8. Make sure the antenna ground (at the bottom of the transformer's high-Z winding) is grounded some distance away (at least 5 meters) from the closest ground rod grounding the feed line. This will introduce several thousand ohms of reactance in the common-mode signal path, as well as provide another path to earth for common-mode noise. Keep the connections to these grounds as far away from one another as possible. Note that the primary (low-Z) winding of the transformer is not grounded but merely connected to the shield of the coax, which itself is grounded some distance away (minimum 5 meters). Fig 7-88 shows how the feed line should be connected to the transformer to achieve the greatest possible attenuation of signals picked up by the outside of the feed line. The stack of beads (B1) will form a high impedance (Z1 = typical 1500 ohms on 1.8 MHz for 100 stacked Wireman beads) for common-mode currents. In the equivalent schematic Z1 and RG2 (the ground resistance of the coaxial cable grounding rod) form a voltage divider. Assume this ground resistance RG2) is 100 ohms(that's a fairly good ground). This means that we have a voltage divider of 100:1100 or 1:11.1, which results in a signal attenuation of 20 log (1/11.1) = -20.8 dB. The spurious signals left across RG2 will now be fed into the feed line via another voltage divider made up by the impedance of the second bead stack (B2) in series with the impedance of the low Z winding of the transformer (Z3) and in series with the impedance of the coaxial cable (Zk = 75 ohms, assuming the cable is terminated in its own characteristic impedance at the receiver end). This time we have a voltage divider made up by Z2 = 1000 ohms(bead stack) + Z3 = 250 ohms (transformer low Z winding) and Zcoax = 75 ohms. The attenuation of the voltage divider is 20 log (75/(1000+250+75)) = -25 dB. The total attenuation of this setup is about 46 dB for common-mode signals captured by the outside of the coaxial feed line. If that does not cure the problem, you can ground the coax at regular intervals or look for a different mechanism of common-mode signal ingress into your receiving setup. Note that the stacks of ferrite cores across the coaxial feed line can be replaced by coiled-up lengths of coax. In order to achieve a choking impedance of approximately 1500 ohms on 160 meters, the inductance must be 125 uH. This requires a coil of coaxial cable measuring 30 cm in diameter and having 20 close-wound turns. Such a coil requires not less than 20 meters of coax, and two such coils are required in the system! If you use this approach, make sure the two coils are at right angles, to minimize coupling between them. Another alternative is to wind miniature type coaxial cable on a large high-permeability core. Five turns of miniature coax through a FT150A-F core (mu = 3000 and AL = 5000) achieves an impedance of 1500 ohms(7 turns = 3 kohms on 160 meters). You should now have a "quiet" feed system to the Beverage. Connecting everything to the system except for the Beverage wire itself should yield no signals.

A note on the test environment

I probably should have mentioned this earlier. Partly by chance and partly by choice, the antenna location used for these tests is about as bad as anyone could imagine. The short vertical is about 25 feet from a power-line with ~7500 volts on a single phase. The feedline for the antenna runs on the ground right under that power line for about 75 feet, then turns 90 degrees to run under another 7500-volt circuit back to my hpouse some 150 feet away. The whole feedline is lying on the ground.

Pete Smith, N4ZR 10:54, 8 July 2009 (UTC)

Other frequency spectrograms

Here are a few spectrograms that may be of interest.

3.550 mhz.jpg
7.050 MHz.jpg
14.050 MHz.jpg

Well, at least we are finally seeing some signals! But the real point is that with the spectrograms you can see that the band noise is approaching the level of the internal receiver noise (without antenna), so probably above 14 MHz the SDR-IQ isn't sensitive enough to show an audible increase in receiver noise when the antenna is connected.

Clifton Labs choke

Finally got some F to BNC adapters and installed the Clifton Labs choke. To my surprise it was much better than the coax and bead design only at VLF (if that's the correct term for ~200 KHz). At 160 it was not nearly as good. Jack Smith at Clifton surmises that maybe the design was biased too much toward the low end. He's going to run some tests of his own once I mail the prototype back to him. In the meantime, I have reinstalled the coax bead choke.

One of the things you notice, being able to view 100 KHz at a time, is things like the 70-KHz-spaced "humps." For a while, they were almost gone, but today they are back - much less severe than without the choke, but still there. I wonder what the source is...

Pete Smith, N4ZR 10:51, 17 July 2009 (UTC)
Personal tools