Calibrating phase shift of multiple RF circuits

Question

My situation: I have a 16x antenna phased array with each antenna connected to its own receiver pre-amplification circuit that I have custom made. Since the system is a phased array, the phase shift of each circuit (or electrical path length) must be as identical as possible to ensure we know what direction signals are arriving to the array. Since these circuits are hand made by me, and are fairly complex, there is a bit of fluctuation in the phase shift across all 16 circuits. My goal is to eliminate this phase shift between the individual circuits, so all circuits have identical electrical path length.

I have collected magnitude and phase measurements on my current batch of circuits using a VNA across our working frequencies (8-20 MHz), shown below.

• Plot 1: Amplification of each circuit (not applicable here)
• Plot 2: The phase shift of each circuit from input to output of the circuit
• Plot 3: The variation in magnitude and phase across all circuits

From the plot, I can see that the phase variation is pretty close (all within ~5 degrees of each other) - but can I do better?

My question: Is it possible to manually adjust the electrical path length of an RF circuit (i.e. with something like a potentiometer trimmer) so that I could adjust the phase and calibrate all of my circuits to have the exact same phase characteristics?

I'm relatively new to RF circuit design, so this may be a naive question. I just want to figure out if this is possible before I move ahead with my design.

Answer 1

What you want is a variable-length transmission line. I had a couple of these, mechanically variable probably with some sort of sliding contact inside. Expensive. We're not supposed to say where things can be bought, but it wasn't hard to find at my usual supplier of microwave stuff.

Such a solution would not be practical for your problem. Maybe look into more precise fabrication?

Answer 2

A peak to peak phase variation of 5º sounds tight enough to me, but if you want to play around with phases, there is a way to tweak them on your PCB, exactly as you mentioned.

The method involves changing the propagation delay of the waves on each trace by adding a dielectric of a different constant directly onto the trace.

This method only achieves higher delay, as air has a lower dielectric constant than the materials you're able to add (if you were working with a potted circuit, I suppose adding a bubble onto the trace would have the opposite effect).

Of course, by changing the characteristic impedance of the transmission line, you are also sacrificing transmission coefficient, and are at risk of adding reflected power. In theory, the capacitance added by adding the dielectric could also cause your line to resonate within your working bandwidth, but I've never encountered that problem.

I don't know how to predict these negative effects precisely, (I'm a technician, not an engineer,) but I can tell you what materials I've used and their general effects:

1. Loctite Hysol 1C. This epoxy is great for fine tuning, with minimal effect on transmission and reflexion. It also works well within a large range of environmental conditions.

2. Arlon. This material is useful for larger phase differences, at the cost of a higher loss in transmission and reflexion.

3. RF absorbant materials. These materials are usually used to ensure no radiated waves reflect off of the metal case of your module at higher frequencies. They can also be used to significantly affect phases, but cost tremendously in terms of transmission and reflexion.

For your application, I would reccomend you use Loctite Hysol 1C (also known as Torr-Seal), as your phase variation is already small. Use PPE though (gloves, respirator, eye protection). That stuff can be nasty.

I recommend you plug a network analyser in and play around with different materials. You can even use sticky tape, but that's hardly a permament solution. Normalise the data trace to the most delayed phase, and try to get the rest of them as close to zero. You can add any amount of any dielectric pretty much anywhere on your trace, as long as it is touching the trace, and not touching your ground.

Tell me how it goes !

-Patrick