I've assembled a MJW patch which was CNC milled to a very high accuracy using a 3mm back plate and a 1.2mm front plate set as close to 3mm above the back as possible.
Measuring it on my VNA and spectrum analyser (for return loss) shows a return loss of about 16dB at 2405MHz and the S11 is shifted up the chart (inductive).
I'll post some pictures / plots when I have chance.
Any tips on how to adjust the two resonances ?
I've tried bending the edges up and down a little - but could really do with being a bit more scientific as to what to tweak to make sure the resonances are in the right places.
Show us a plot. You want to see something like this.
That's the notched version but the square one should be similar. This is a simplification but there are two way's round the patch, one slightly longer than the other. The longer is the lower frequency the shorter the higher, so depending on which you want to move adjust the spacing slightly accordingly.
I might add, 16 dB return loss is not bad, though you can get better than that, -20dB of so. As long as the amplifier is happy it will be fine as it is. I will have to look at the design again and make some tweaks for a broader match if people are having trouble, but don't get fixated on return loss, this design is a compromise and match and bandwidth suffer to allow the LHCP.
here's a few plots - there's some issues with them though, for instance the VNA S11 reference plane isn't correct but the VNA RL shows to resonances but they are too far apart ?
The Rigol shows about a 20dB return loss but cannot resolve the two dips despite trying higher resolution RBW and VBW. Likewise the HP plot shows a 20dB return loss but cannot resolve the two dips - going to try a different directional coupler and see if that helps.
Only the MiniVNA seems able to resolve the two resonances. I've tried increasing / decreasing the hieght with limited success but also cracked the solder joint to the central tube - doh !
Using a 3mm drill shank I can confirm the spacing is 3mm all round but I still have the PTFE insulation in place round the feed pin.
I'm tempted to leave it alone and maybe figure some way of testing for CP.
- return_loss.jpg (871.53 KiB) Viewed 770 times
- RL_two_dips.jpg (903.91 KiB) Viewed 770 times
- s11.jpg (949.2 KiB) Viewed 770 times
From observation, the resonant frequencies are OK, just above and just below the band, so it is about the position and size of the probe, but of course changing that changes everything else a little. This would be easy if it was not for that big 22mm pipe going through the middle.
From the article, you will see the angle of the probe rotates around the smith chart and its distance from the centre changes the impedance. Your result shows quite a low impedance, but close to the real axis - are you sure the probe is in the right place? What is the connector - SMA? I am also wondering about the feed pin and the insulation. I simulated 1mm, 2mm and 3mm feed diameters and it makes a difference, but not as large as you see. My assumption is 2mm.
The patch was machined to within 0.01mm of the drawings and by adding a little bit of copper tape I've managed to get the tuning about as good as it's going to get. VSWR is 1.33 across the band.
The feed connector is an N type and the pin is 1.2mm in diameter. I added a bit of tape close to it in an attempt to tuned out its inductance - kind of worked.
Then added another small bit of tape to shift the resonances into place - so I figure that's about it. Sure is critical to get it bang on the nose, even with precision machining !
EDIT - When I measured the RL on my SA I found that without the tape tuning in place the RL was still acceptable at -16dB and resonances were about +30 and -20MHz. So all good exactly as built and the copper tape was just my OCD !
- VNA_190428_192441.jpg (223.28 KiB) Viewed 730 times
- PATCH_TOP.jpg (565 KiB) Viewed 730 times
- S11_tuned.jpg (1.04 MiB) Viewed 730 times
Mike's right, probably not worth messing about with further. A RL of -16dB will give about 2.5% reflected power, which will be reduced further by cable loss. If your PA is happy with this then ok.
I have made two of these patch feeds. Both had a RL of around -18dB, but took a lot of fiddling around with the corners of the patch. The question is, does it radiate properly ? Not that easy to assess, but what I did was to uplink a test signal (CW from FT817 + transverter) using a 1W amp and a 90cm dish to the NB transponder and look at the downlink on the BATC web SDR. Signal was S8-9. I then used an alternative design to compare with and the results were similar. So to me, performance was ok.
Construction-wise, yes follow the design but I used a steel rule, snips, hacksaw and filing. So not too critical, which for amateur use is great and is why I made a second one.
By varying the positioning of the copper I can get different results, such as two sharp resonance points or two points with a flatter response between them.
I don't have a VNA and the refection bridge on the Siglent analyser is only rated up to 2GHz, so don't know if the results are accurate.
Does any of the results I obtained look OK to use ?
If so I can make a final version using some copper tape soldered in place.