Antenna tuning woes

[adamgerken]

Hey all, I'm having a bit of a headscratcher here. I've designed plenty of Bluetooth things and always went with recommended matches across the board and never had any issues. Recently I got a Rigol RSA5065N which has a VNA mode so I figured why not try to do an optimized tune. The issue is I just can't get it to work out.

Some background on the board: Simple OTS antenna with a pi match.



On the board it's fairly well laid out, although I probably would have done it a bit different:



There is a small snap in make/break connector there but that ended up being a disaster to use so I just soldered on a short SMA pig tail. I tried various different methods of calibration and nothing seemed to work so I made some little pig tail calibration leads that match the pig tail soldered to the board. I have an open that's just open (I know it should be shielded but, how much can that matter?) a short, and a load that's two 100ohm resistors soldered in 180 degrees off from each other.





I shorted out the series element of the filter and removed the shunts to get a baseline impedance of the antenna which seems to work. I get 1+j-6.2 so using a calculator (and double checking by hand I figured I needed 0.9nH in series to get up to the unity conductance circle:

This only got me up to 2+j1 (shown as red marker)



Ok, no problem, at least we're going in the right direction. I experimentally found a value of 2.2nH lands pretty close to where I want... 4+j11.75. So now I need to add a shunt cap to wrap around to the 50ohm point. I decided to try to sneak up on it so I used a value of 4.3pF which should have been in the ballpark at 33.5+j13 but for some reason the recorded value looks like it continues on the inductor path to a value of 3.5+j25 (shown as red marker.)



I've tried many different combinations of components now (the above is just the most recent) with many different cabling set ups, calibration points, etc. I can always experimentally get close with the first component, even though the values are usually many multiples off from what I expect but then no matter what any time I add in the second component the point does not move towards the center, it stays way around the edge of the Smith Chart.

After probably 50 attempts at tuning this thing over the past three days I'm at a complete loss of what I'm doing wrong. Is the Rigol garbage at being a VNA? is it my cal method? the test set up? Johanson got all my part kits mixed up?

EDIT: no idea why the attachments aren't working, sorry.

[adamgerken]

For some completeness, this is what I was going for in an ideal scenario:

[antenna]

One thing you might try is, with your shunt parts still removed, also remove the series jumper and adjust your port extension/delay such that the marker is at ∞Ω.  Then, replace the series jumper and measure the impedance again.  This way, you are measuring the impedance at the matching location, not where you connected the coax.  At 2.4GHz, a little transmission line can make a lot of rotation around the chart, which is exactly what I think is happening based on what the recommended matching components say about the impedance of that chip antenna.

[adamgerken]

One thing you might try is, with your shunt parts still removed, also remove the series jumper and adjust your port extension/delay such that the marker is at ∞Ω.  Then, replace the series jumper and measure the impedance again.  This way, you are measuring the impedance at the matching location, not where you connected the coax.  At 2.4GHz, a little transmission line can make a lot of rotation around the chart, which is exactly what I think is happening based on what the recommended matching components say about the impedance of that chip antenna.

Good point. I'll give that a go and report back. Thank you.

[adamgerken]

One thing you might try is, with your shunt parts still removed, also remove the series jumper and adjust your port extension/delay such that the marker is at ∞Ω.  Then, replace the series jumper and measure the impedance again.  This way, you are measuring the impedance at the matching location, not where you connected the coax.  At 2.4GHz, a little transmission line can make a lot of rotation around the chart, which is exactly what I think is happening based on what the recommended matching components say about the impedance of that chip antenna.

Good point. I'll give that a go and report back. Thank you.

With a username like 'antenna' I guess I'm not shocked but you were right on the money. The values were still a little hunt and peck but similar to what the calculator said. Got it tuned up and looks like it's working mint. Thank you so much!