Monday, July 6, 2015

Antenna Auto-tuner Design, Part 5: Directional Coupler Design

This, my fifth post for my Automatic Antenna Tuner, deals with the design of the Directional Coupler.  (Its construction and testing will follow in a later blog post.  And the previous post in this series can be found here.)

I plan to use a "Tandem Match" style coupler  (for background, see here) which uses two transformers -- one senses the voltage across the transmission-line, the other senses the current traveling in the transmission-line:

The two transformers will be wound on Ferrite cores (because I can use fewer turns for a high inductance, compared to Iron Powder cores).

Let's get started...

First -- the specifications I'd like this coupler to meet:
  • Frequency Range:  3.5 - 30 MHz (and possibly 54 MHz)
  • Forward Power In, Continuous operation (e.g. carrier)  at a maximum SWR of 3:1:  200 watts. (To ensure that the cores do not overheat).
  • Forward Power In, Peak, at an Infinite SWR:  800 watts.  (To ensure that the cores do not saturate).
  • Directivity:  40 dB, min.

These specs might change as I design and test the coupler, but they're a good starting point.

Next, I would like to select core size and type (e.g. mix 43, mix 61, or...?).

But how do I select the right cores?  Research was required.  And here are the pertinent facts that I found:
  • With a normal sine-wave signal above 100 KHz, both Iron Powder and Ferrite type cores will first be affected by overheating caused by core losses, rather than saturation (Amidon). 
  • Per Amidon, a core's operating temperature (ambient temperature plus core temperature rise) should be limited to 100°C for intermittent operation and 75°C for continuous operation (Amidon).
  • Core saturation is a secondary cause of coil failure.  Saturation will decrease the permeability of the core causing it to have impaired performance or to become inoperative.  The safe operating total flux density level for most Ferrite Materials is typically 2000 gauss and 5000 gauss for Powdered-Iron. (Amidon).
  • The extrapolated AC flux density limits (in the table below) can be used for BOTH Iron Powder and Ferrite type cores as a guideline to avoid excessive heating.  These figures may vary slightly according to the type of the material being used. (Amidon)
(click on image to enlarge)
  • The permeability is at its maximum just below the Curie temperature.  Above the Curie temperature, the core permeability sharply disappears, and the material is no longer magnetic while it is above the Curie Temperature. (Magnetics)
  • In all ferrites, saturation flux density (Bsat) declines as the temperature goes up. (Magnetics)
  • Per Fair-rite:  "Strong magnetic fields or excessive mechanical stresses may result in irreversible changes in permeability or losses."  (This warning is explicitly stated for materials 61 and 67).
And per VK2WB:
  • The permeability of all ferrite materials is frequency dependent (it decreases with frequency), so the manufacturer's quoted AL values (determined at approximately 10 KHz), should not be use used for calculations involving RF frequencies. (VKHam)
  • Many ferrite cores can permanently change their permeability after being subjected to relatively high power (flux) levels. (VKHam)

Calculating Inductance and Flux Density:

This design uses two transformers.  One senses current through the transmission line, the other sense voltage across the transmission line.

Because the Directional Coupler precedes the Tuning Network in the RF path from transmitter to antenna, I'm assuming it will see the following voltages and currents:
  • During normal operation, I'll specify that the SWR seen by the transmitter will be no worse than 3:1, thus the 3:1 SWR requirement for core heating at the average power of 200 watts.
  • When the tuner first senses SWR (e.g. with the Tuning Network untuned), the SWR could be very large.  Assume Power is at its maximum (800 watts) when this occurs, for calculating a worst-case Flux Densities (to verify that the cores do not oversaturate). 
  • Note that Vmax = |Vf| + |Vr| and Imax = |If| + |Ir|.  Therefore, at infinite SWR, because |Vf| = |Vr| and |If| = |Ir|, the maximum voltage and current will be 2*Vf and 2*If, where Vf and If are the voltage and current for a 1:1 SWR (i.e. into 50 ohms).
  These power and SWR levels result in the following voltage and current requirements :

The "shaded" boxes contain the levels I will use.  To summarize, they are:
  • For core heating (at 200 watts):  Vmax = 150 Vrms, Imax = 3.0 Arms
  • For core saturation (at 800 watts):  Vmax = 400 Vrms (= 566 Vpk), Imax = 8.0 Arms (= 11.3 Apk)
Note:  For saturation I will use Vpk and Ipk values (=1.414 * Vrms and 1.414 * Irms) for flux-density calculations.

Voltage-sense Transformer Requirements:

The N-turn primary of the voltage-sense transformer connects across the transmission line, so I would like its impedance at the lowest frequency of 3.5 MHz to have an inductive reactance of at least 500 ohms (10 times greater than 50 ohms, the transmission line impedance -- note that this reactance, when placed in parallel across a 50 ohm load, would result in a Return Loss of 26 dB).  So the inductance should be at least 23 uH.

As I just mentioned,  I want to ensure that:
  • The max RMS voltage across the voltage-sense transformer (at 200 watts, SWR = 3:1, tuner in bypass) doesn't overheat the transformer. (Flux Density to be below the recommended maximum).
  • The max peak voltage across the voltage-sense transformer (at 800 watts, SWR = Infinite, tuner in bypass) doesn't come close to saturating the transformer. (peak Flux Density to be below Saturation Density).
Here are recommended Maximum Flux Density Limits to ensure that cores do not overheat (these are the limits that the Mini Ring Core Calculator uses and which are based on (and interpolated from) Amidon's recommendations.)

And to verify that the cores do not saturate, I will assume that the Saturation Flux Density is 2000 Gauss for Ferrite material, per the Amidon note, above.  Therefore, I would like the maximum Flux Density in this design, under peak conditions, to be well under 2000 Gauss.

Current-sense Transformer Requirements:

I want to ensure that:
  • The max RMS current through the current-sense transformer (at 200 watts, SWR = 3:1, tuner in bypass) doesn't overheat the transformer. (Flux Density to be below the recommended maximum). 
  • The max peak current through the current-sense transformer (at 800 watts, SWR = infinite, tuner in bypass) doesn't come close to saturating the transformer. (peak Flux Density to be below Saturation Density).
Important Note, 19 Feb 2018:  My original text stated that the secondary-impedance reflected into the current-sense transformer's primary was 25/N^2 ohms.  This is wrong (or rather, it is only correct when Zload is a short!), as I have shown in this blog post:  

Instead, the reflected impedance is a much more complicated function of N, Zload, R3, and R4.  

However, assuming that |Vmax| and |Vmin| on the transmission line are calculable according to the following equations:

|Vmax| = (1 + |Γ|)*|Vforward|

|Vmin| = (1 - |Γ|)*|Vforward|

Then, the current-sense transformer core's Flux Density can be shown to be independent of Zload (if Zload is resistive -- I haven't verified this for complex loads), and we can simplify the calculation of Flux Density to the case when Zload = 50 ohms, resistive, in which case the impedance reflected into the current-sense transformer's primary, from its secondary, is simply 50/N^2 ohms.

Here's the much simplified equivalent circuit using the assumptions, above:

(click on image to enlarge)

L and Rloss are simply the inductance and loss (core) of the 1-turn primary.  Note that if the 50 ohm termination resistor is not placed across the secondary, there would be no "50 / N2" ohm reflected resistance across the primary.  Without this shunt, all of the transmission line current will instead pass through the primary winding's Rloss and L -- and (from personal experience), if the current is high, the core can easily become too hot to touch. 

But with the 50 ohm resistor attached, and assuming the "reflected" resistance is significantly smaller than the reactance of the 1-turn primary, almost all of the transmission-line current will shunt through the 50 / N2 resistance and the core shouldn't overheat.

And thus, the voltage across the transformer's primary is essentially equal to Irms * 50 / N2.  I'll use this voltage to calculate the current-sense transformer's Flux Density.

And for these calculations I will use Excel, rather than the "Mini Ring Core Calculator" program.


For Type-43 material, the  Mini Ring Core Calculator (V1.2) uses smaller values of AL than those published by Fair-rite or Amidon.  

Therefore, the Calculator's Inductance and Flux Density results will be lower (by a factor of about 0.85, per my experience) from those calculated using the manufacturer's AL.  

Also -- although the Calculator's results for mix-61 material generally track my results obtained via published equations, I did notice that the Calculator's Flux results specifically for the FT114A-61 core were lower by a factor of about 0.83.

The Design Equations:

(Note: For saturation Bmax calculations I use Vpk as the voltage in the equation, above, where Vpk = 1.414 * Vrms, Vrms having been calculated elsewhere)

The Calculated Results:

Here's the spreadsheet data for 3.5 MHz:

(click on image to enlarge)

Values in RED signify a failure to meet my criteria (e.g. Flux Density is above 80 Gauss on 3.5 MHz).

And here's the data at 28 MHz:

(click on image to enlarge)

Much better results at 28 MHz, even though the Flux Density Limit has dropped from 80 to 30 Gauss.

Some additional notes on the spreadsheets:
  • The numbers in parenthesis, such as "L(2)" or "B(1)", simply refer to the "number" of the equation used (e.g. equation L2 or equation B1 -- see the equations, above).  
  • The "Vrms" value for the Current-Sense transformer is the voltage drop across this transformer's single-turn primary (i.e. Vrms = Irms * (Xl || Rf), where Rf is the secondary's load (Rtermination) reflected to the transformer's primary side).
  • Please note that for "Peak Power" calculations, the SWR is assumed to be infinite.
  • For "average power" operation, core heating is our concern.  Thus, for Pavg, Flux Density, B(1), is compared against "Bmax" (e.g. 80 gauss at 3.5 MHz).  If the calculated Flux Density exceeds Bmax, the result appears RED.
  • For "peak power" calculations (i.e. worst-case) heating isn't really the concern (because this condition should be a short-lived event).  Instead, I check that the core does not saturate.  Therefore, for Ppk, Flux Density is compared against Bsat (e.g. 2000 gauss).  If the calculated Flux Density exceeds Bsat, the result appears RED.
 Let's focus on 3.5 MHz, where more cores fail.  Here's a reduced "summary" table comparing six of the most likely candidate cores.  These results assume an "operational" SWR (max SWR of 3:1 during normal operation):

(click on image to enlarge)

If the maximum operational SWR is assumed to be 2:1 (rather than 3:1), then the results are:

(click on image to enlarge)

(A slight correction:  in the two tables above, the core listed as "26435002" should instead be "2643625002".)


Maximum Flux Density (for saturation determination) is under 2000 gauss in all cases.

If the Tuner is assumed to tune to at least a 2:1 SWR (a reasonable assumption), then five of the six cores in the last table above would be acceptable for the Voltage-sense transformer, using a winding ratio of 25:1 (i.e. Coupling Factor of -28 dB).  Five cores decreases to three cores if the tuner cannot tune to an SWR under 3:1.

Inductances for all six voltage-sense transformer candidates are well over 23 uH, so their reactance will be well above my minimum requirement of "10 times 50 ohms".  But the real test will be after I build the coupler and can check its return-loss.

The Current-sense transformer Flux Densities are all well under the recommended limits.  Note that they are low enough that an FT50-43 core (and perhaps a core even smaller) could be used.

Wire Gauge:

Next, I'd like to look at wire gauge for the transformer windings.

Current-sense Transformer:

1-Turn primary:  14, 12 or 10 AWG for its primary.  Note: for a 3 Amp RMS current, the power loss on, say, 2 inches of 14AWG wire at 30 MHz (worst loss due to skin effect) would be 1/8 watt.

25-Turn secondary:  Assume:  3 Arms primary current, 36" wire length, 26 AWG, 30 MHz.  Secondary current is therefore 0.12A (i.e. 3A/25).  Using an on-line AC Resistance Calculator with wire length of 914.4 mm and diameter to 0.4049 mm (using this AWG to mm Converter), the secondary's resistance would be 1.044 ohms.  Power loss is (I^2)*R and therefore the Secondary's Loss = 0.015 watt.

So the Current-sense primary winding would be 14 AWG (and probably larger).
And the secondary winding would be 26 AWG.

Voltage-sense Transformer:

25-Turn primary:
  • Average current is simply 150Vrms / XL (the reactance of the transformer primary).  Let's take the FT87A-43 core as an example:
  • At 3.5 MHz, Irms = 150 / (2 * pi * 3.5MHz * 605 uH) = 11 mA.
  • Assume wire length = 36", wire gauge = 26 AWG:  Total R at 3.5MHz = 0.38 ohms.
  • Loss at 3.5 MHz: 0.011^2 * 0.38 = Negligible.
  • Loss at 30 MHz:  Irms = 1.3 mA.  Total R, 26AWG @ 36" = 1 ohm.  Loss = Negligible.
1-Turn Secondary:
  • Load across secondary is essentially 100 ohms (two 50 ohms in series).
  • Vsecondary = 150 Vrms / N = 150 / 25 = 6 volts.
  • Isecondary = 6 / 100 = 60 mA.
  • Assume wire length = 2", 26 AWG.
  • Loss at 3.5 MHz: Wire R = 0.02 ohms.  Wire Loss = 0.06^2 * 0.02 = Negligible.
  • Loss at 30 MHz: Wire R = 0.06 ohms.  Wire Loss = 0.06^2 * 0.06 = Negligible.
So the windings of the Voltage-sense transformer can both be 26 AWG.

Terminating Resistors:

Finally, a quick calculation to determine the power rating of the two 50 ohm terminating resistors.

For 25 turns, the Directional Coupler's "Coupling Factor" will be -28 dB.

Therefore, for continuous operation at a Power level of 200 watts (53 dBm), 25 dBm of power (0.32 watts) will need to be dissipated by the 50 ohm resistor on the "Vforward" side.

I'll spec the two terminating resistors to be identical, with this value:  50 ohms, 0.5 watts minimum, each.

Next Steps:

The next steps will include experimenting with cores and winding ratios, to see if I can get acceptable performance (e.g. I'd like Directivity to be at least 40 dB from 3.5 - 30 MHz).  But this will have to wait until I get purchase some cores!

And an important further note regarding implementation:

Directional Couplers using the Tandem-Match topology (two toroidal transformers) typically use identical large cores for their two transformers.  Per my spreadsheet, it is the voltage-sense transformer that dictates this large core size.  The current-sense transformer's core can be much smaller.  Going to a smaller core would save both money and circuit space.

In fact, if I instead use a capacitive voltage-divider (i.e. Bruene coupler) rather than the voltage-sense transformer to sense transmission line voltage, I believe even more space and money could be saved. for thought!


The previous post in this series can be found here.

And the next post in this series, part 6 (Notes on Match Detection), can be found here.


  • Book, Ferromagnetic-Core Design and Application Handbook, 2nd Edition, M.F. Demaw (Available from Amidon)

Links to my blog posts in this Auto-tuner series:

Part 1:  Preliminary Specification

Part 2:  Network Capacitor Selection

Part 3:  Network Inductor Selection

Part 4:  Relays and L-Network Schematic (Preliminary)

Part 5:  Directional Coupler Design

Part 6:  Notes on Match Detection

Part 7:  The Build, Phase 1

Part 8:  The Build, Phase 2 (Integration of Match Detection)

Part 9:  The Build, Phase 3 (Incorporating a Microcontroller)

Part 10:  The Final Schematics

Links to my Directional Coupler blog posts:

Notes on the Bruene Coupler, Part 2

Notes on the Bruene Coupler, Part 1

Notes on HF Directional Couplers

Building an HF Directional Coupler

Notes on the Bird Wattmeter

Notes on the Monimatch

Notes on the Twin-lead "Twin-Lamp" SWR Indicator

Calculating Flux Density in Tandem-Match Transformers

Other Tuner posts:

A quick tutorial on Smith Chart basics:

The L-network:

A correction to the usual L-network design constraints:

A look at highpass T-Networks:

More on the W8ZR EZ-Tuner:  (Note that this tuner is also discussed in the highpass T-Network post).

The Elecraft KAT-500:
 The Nye-Viking MB-V-A and the Rohde Coupler  :

Standard Caveat:

As always, I might have made a mistake in my equations, assumptions, drawings, or interpretations.  If you see anything you believe to be in error or if anything is confusing, please feel free to contact me or comment below.

And so I should add -- this design and any associated information is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.

No comments: