This is a refinement to an earlier 40 dB coupler that I built in 2013.  That first one was taken directly from the May/June 2013 issue of QEX Magazine.  Inside, there was an article from Loftur Jonasson (TF3LJ / VE2LJX) about building a simple digital RF power meter.  Jonasson's article referenced and built on earlier work by W7ZOI and W7PUA (published in the June 2001 issue of QST).

It was a great article.  Without much understanding of the math or design process (but that's how we learn!), I built a unit that, when later tested, proved to be quite usable.  It's not great at UHF, but is good at VHF and great throughout the HF bands.

What It Is

A tap (or non-directional coupler) is used to take a very small portion of the signal on a transmission line so it can be measured by sensitive instruments.  If you send a high-power signal directly into a measuring device, you can quickly destroy the device.  This tap is designed to pass the signal with minimal loss from the input to the output (the mainline) while diverting a very small portion to a sampling port.

The sample is 40 dB less than the main signal, or 1/10,000th the power of the original.  That allows, for example, the output of a 100 watt transmitter (+50 dBm) to be measured at a level of 10 milliwatts (+10 dBm).  10 mW is safe for most test equipment.  It also allows small signals to be measured.  For example, a 1 mW signal (+0 dBm) appears as a 0.1 microwatt signal (-40 dBm) at the sample port, which is still plenty strong for most test equipment.

My first build of this project followed Jonasson's design and was built into a Hammond 1490A box.  It had some SWR problems above ~250 MHz, but was very usable on the HF bands and at VHF with some compensation.

New Approach

Trying to grasp how RF travels through the device's internals, I took a suggestion to consider it as a microstrip transmission line; that is, an RF conductor encased in a ground.  The thought process to arrive at this design went through several iterations and arrived at these ideas:

  • By reducing the size so the two mainline connectors butt up against each other, we reduce the exposure of the signal to potential impedance shifts from the instrument and reduce resistive losses.
  • By using surface mount resistors instead of large through-hole resistors, we reduce the inductance of the resistor chain and create a more dimensionally uniform signal path.
  • By treating the internals as a microstrip transmission line, we get more control over the impedances.
  • By using a combination of parallel and series resistances, we can achieve the exact non-standard total resistance that's needed and also keep the signal path symmetrical.

I found the desired 2,475Ω attenuator resistance by putting two series strips of 1.65K resistors in parallel with each other.  The 1/2W rated resistors give a total capacity of 3W on the attenuator which should be adequate to measure >150W on the mainline.  Taking the dimensions of the resistor pack into consideration and using an online impedance calculator, we find that we need a air dielectric gap of 0.042" between the resistors and the housing to maintain 50Ω impedance along the resistor pack.

I wanted to machine the housing from pure copper, but the right size wasn't in stock at my supplier, so we settle for brass for now.  After machining the brass housing and installing the parts, here is what we have:

Completed Power Tap
Completed Power Tap
(click for a large view)

Internal Parts
Internal Parts
(click for a large view)

Taking Measurements
Taking Measurements
(click for a large view)
Compared To The Original Version
(click for a large view)
Plan Diagram
Internal Layout Diagram
(click for a large view)
Circuit Diagram
Circuit Diagram
(click for a large view)

So how does it perform?  Well, insertion loss is very good, the flatness of the coupling factor is very good, but we still have some return loss problem at UHF.

Network Analyzer Trace
Network Analyzer Trace -- Looks Great!
(click for a large view)
Insertion Loss
Insertion Loss -- Looks Great!
(click for a large view)
Coupling Factor
Coupling Factor -- Looks Great!
(click for a large view)
Return Loss
Return Loss / VSWR -- Not So Great!
(click for a large view)

Next Steps

The reason it looks so good on the network analyzer is because that instrument only goes up to 180 MHz.  The spectrum analyzer's 1.5 GHz bandwidth reveals the problem at high frequency.  The next step is to install a capacitance stub along the resistor pack to provide some high frequency compensation.

A future version will have the following features:

  • Pure copper machined housing for better conductivity
  • Higher quality N connectors with Teflon insulators for heat resistance during assembly
  • Fastener-free construction -- soldered case halves and connector flanges
  • Hot plate reflow soldering of the shunt resistors
  • Teflon internal dielectric for:
    • Better dimensional control along the resistor pack
    • Heat resistance during assembly
    • Higher power handling
  • 1 watt resistors instead of 1/2 watt for better power handling
  • Whatever changes are needed to ensure better UHF performance



+1 #1 KB5EO 2014-10-13 01:05
A little addendum: The VSWR measurements are flawed due to a problem in my return loss bridge. Chances are, it is much better at UHF than it appears.

Add comment

Security code