Optimize EWE- F/B (details of wide-band JB-Terminator)

Most DXers use a simple resistor as the terminator on a EWE antenna to get it directional. However the F/B (Front / Back) ratio is frequency dependent, and the resistor needs to have a different value for the lower and higher MW band. Some use a remote controlled resistor for that reason. But that still improves F/B only for a section of the MW band. For wide-spectrum recordings (like Perseus) you need good F/B over the entire band simultaneously.

My EWE size 30 x 4m is dictated by trees. Grazing cattle underneath forbid a smaller length.
I used a remote controlled resistor but only got poor F/B for this size. So there was need for improvement.

Extensive 4nec2
-analysis and optimization runs revealed that an EWE terminated with a simple resistor achieves maximal F/B only for a specific size and at a specific frequency. For all other sizes and frequencies you only get maximal F/B when using a complex termination (R/L or R/C in series).
But a complex termination makes F/B even more frequency dependent, and it is more difficult to remote control.

Why not try to build a single wideband complex termination that achieves best F/B for every channel across the MW band at the same time?

That's what I eventually managed to do and the result is called: JB-Terminator

It requires some effort and technical skills. You cannot simulate it, you cannot copy an existing JB-Terminator of another antenna.
You need to take measurements on your actual real EWE.

You need:
1. a variable R and a variable L (or variable C) connected in series (see photo below)  
2. an accurate R / L / C meter  
3. a curve-best-fit calculating program like Octave or Matlab  
This variable R and L was used for doing the measurements  
The DIP switches shortcut the SMD inductors 5, 10, 10, 10, 47, 47uH  
That covers a range of 5-130 uH  

·Pick 8 ..10 channels across the MW band that have a tx sitting on the EWE rear. Make sure they are a bit strong, best time is short before the greyzone when levels are good, but don't yet suffer fading from skywave propagation.  
·For each channel find the best F/B by varying the R and L (or C) of the termination  
·Write down the found R/L/C values per frequency, convert L or C to its respective reactive resistance for that frequency (column "XL" in table below)  
·Repeat it the following days to rule out errors or propagation oddities.  
·Enter the found values into a curve-best-fit calculator and let it find a network that fits best all your measured values.  
This task was done by Christoph Mayer ( Thanks ! ) using Octave. I ended up with the result seen below.  
When finally using the JB-Terminator I tried all channels to see if I could improve F/B a bit more. That wasn't possible for most channels, only above 1500 kHz it could be improved. Compared to the variable R that I used before, the F/B improved substantially even for the 1500+ kHz range.

The JB-Terminator is in use since 2011 on the EWE beaming to my northwest (Iceland - CAN - USA). Over years the F/B has dropped down slightly, possibly due to EWE property (soil?) changes. But recently (Jan 2016) I compared it against the North-Holland remote Rx that is also near the coast, and still noticed huge differences, I have Iran tx much weaker here. The same we found when a ALA was set up here for a test 2 weeks ago with its bidirectional property.

So all the effort really paid off.
The F/B of the EWE is far superior than the F/B of my 300m beverage. That might have to do with the wet soil under the beverage here.

Measured on 15. April 2011

kHz   Ohm    L/C    XL
540   1718   4,3nF   -70
630   1500   34uH   135
756   1364   85uH   403
801   1294   74uH   377
972    903   27uH   164
1134    873   94uH   670
1431    536   58uH   521
1458    562   59uH   540
1521    493   58uH   554

Some of these frequencies were entered in Octave to find a RLC network that best represents the measured R and XL of each channel (solid line)

Since F/B is more sensitive to resistor changes than L variations, we forced the blue line to deviate only slightly, and gave the red line more freedom.

Resulting circuit:

Result of best-fit aproximation (created by Christoph Mayer)

P1 = 2,2 k
P2 = 77 pF
P3 = 192 uH (I use 2x 100uH fixed inductors in series)

Unfortunately I don't have the Octave script, because Christoph did that task, so I never needed to run Octave myself.

  ©  3S design GmbH   Zeiterfassungs-Systeme   Zugangskontrolle-Systeme   BDE-Systeme   Ferienwohnung südliche Nordsee