This is the 3rd in a series of continually improving designs for a high quality sampling head for making RF signal measurements.  The 1st and 2nd versions (and what they do) are described in this article.  This version has the best frequency range and power handling capability.

After machining the previous one from brass bar stock, I decided to make this one from copper.  Copper has much better conductivity than brass, but it really makes little difference in this application.

The biggest difference is the difficulty that copper presents during machining operations.  It has a tendency to grab at cutting tools.  In fact, I snapped two drill bits, even though I was proceeding slowly and using ample lubricant.  Threading worked well, however.  Like the brass version, I used thread forming taps rather than cutting taps.  They work by displacing and forming the metal instead of cutting & removing it.  That went well here.

The other major difference is in the internal dielectric.  The resistor chain that transfers and attenuates power from the main line to the sampling port is surrounded by the housing of the device and together, they make a stripline transmission line, similar to coax.  Therefore, it's important to carefully size the insulating gap to maintain impedance.  Too little gap and the impedance is too low, too much and it's too high.  We want as close to 50 ohms as possible so the device doesn't introduce distortions or reflections in the signal.

In version 2 (brass), there is an air gap between the resistor chain and the housing.  I sized the gap based on the dimensions of the resistor chain and the dielectric constant of the air insulator (ε = 1.0).  With air as the insulating medium, I needed a 0.0525" gap.

Version 2 uses Teflon (PTFE) as the insulator.  This is mainly to give it a high resistance to internal arcing, but also to better control the internal dimensions.  Because Teflon has a higher dielectric constant (ε ≈ 2.05), the insulation thickness needed to be increased to about 0.092".  For this, I cut layers of 0.015" thick Teflon sheet and stacked them in the cavity.

Let's look at some pictures!

 

Forming Threads
Forming The Threads
(click for a larger view)

The Machined Housing
The Machined Housing
(click for a larger view)
Reflow Soldering
Reflow Soldering
(click for a larger view)
Internals Installed
Internals Installed
(click for a larger view)
The Completed Device
The Completed Device
(click for a larger view)
In The Palm Of The Hand
In The Palm Of The Hand
(click for a larger view)

The middle-left photo shows the process of soldering the shunt resistors between the housing and the sampling port.  This method is far better than trying to heat the housing locally to the melting point of the solder.  Here, I'm using my reflow skillet, which is a modified electric cooking skillet with the thermostat defeated.  The aluminum block that you see has a thermocouple embedded in it, which goes to a digital PID controller that turns the skillet on & off through a solid state relay.  That allows me to very precisely control the temperature.  The aluminum block sits on a film of safflower oil to ensure good heat transfer.

The middle-right photo shows the resistor chain connecting the mainline to the sample port, the capacitive stub (purple wire) and the other parts.  Again, for a full description of the device's operation, check out my earlier article.

So how does it perform?  Well, the coupled frequency response is reasonably flat to at least 1.5 GHz.  That's the limit of my equipment's ability to measure.  I need to take some better (more careful) measurements, but here are the preliminary scans in the frequency domain:

Insertion Loss On The Mainline
Insertion Loss On The Mainline
(click for a larger view)
Coupling Over 1.5 GHz Span
Coupling Over 1.5 GHz Span
(click for a larger view)

So I'm pretty happy with it.  It's not perfect, but neither are these measurements.  I will have a new return loss fixture soon and will get some more careful measurements then.  Once its characteristics are fully known and I have calibration files put together for the spectrum analyzer, I'll be able to do some nice measurements using this device.

 

 

Add comment


Security code
Refresh