help for PCB antenna

[tomars]

I have added dipolo antenna to My PCB (Right)
I need 433 Mhz anetnna
There are also RFIS 13Mhz coil and Touch key on same PCB
Measuring with NanoVna I have the resonance at 300 MHz
I lost half a day to remove two turns at a time, and re measured, but the only thing that changed were decibel, the frequency changed a little.
Perhaps they affect other parts of the circuit too much ?
What can I try ?

[ Moderator action ] : Fixed typo on title

 

[FvM]

Without seeing your measurement setup, we can't know which resonance your are actually seeing. It might be just a feed cable sheet resonance.

I don't understand how you are matching the differential dipole to the apparently single ended transceiver output.

 

[vfone]

I think the spacing between turns in your meander antenna is too small.
If the spacing is too small, the cross-coupling between elements will affect antenna resonance, antenna efficiency, and antenna polarization.

 

[tomars]

--- Updated Apr 18, 2023 ---

I think the spacing between turns in your meander antenna is too small.
If the spacing is too small, the cross-coupling between elements will affect antenna resonance, antenna efficiency, and antenna polarization.

The length of the single track is 330 mm, width 0.25 mm
330 top + 330 bottom

I also tried two triangles, but the measurements are disastrous.
I would like to increase the bandwidth, to reduce the influence that external conditions have.

 

[FvM]

I suggest a simple test, does S11 curve change considerably if you touch the feed cable or the VNA? If yes, the measurement is mostly meaningless.

To eliminate the cable effect, place multiple ferrite tubes on the cable.

See https://www.edaboard.com/threads/el...nce-in-antenna-measurement.129541/post-567000

 

[tomars]

Not, do not change touching VNA or cable.
Only change moving antenna.

Any advice on making an antenna more immune to other parts of the PCB ?

 

[volker@muehlhaus]

As the others already mentioned, a symmetric antenna (dipole) for an asymmetric feed isn't a good idea, unless you use a balun. In addition, the nearby PCB ground edge will affect dipole impedance.

Have a look at monopole-style antennas like the F antenna, which use the PCB ground as antenna ground. This will also save space. Overview appnote by TI: https://www.ti.com/lit/an/swra161b/swra161b.pdf

 

[tomars]

But inverted F only i found GHZ
In this pdf and online.
Only i have 80mm

 

[volker@muehlhaus]

But inverted F only i found GHZ

I recommended the F antenna because that works well with the radiator parallel to PCB ground. But yes, it is not simple to design an efficient 433 MHz antenna for your small PCB.

You can try meandering the radiator, like you did for the dipole. But that is limited, not sure if it works.

From my experience with PCB antenna design: the dipole is not a useful solution here.

 

[FvM]

TI has some application notes about Sub-GHz antenna. Appended AN003 seems to be no longer available at ti.com

https://www.ti.com/lit/pdf/swra161

As a general consideration for electrical small antenna, the smaller the antenna compared to wavelength, the lower the radiation resistance and bandwidth and the higher the resonance Q. Respectively the stronger de-tuning by hand effect.

A symmetrical dipole doesn't utilize the large ground plane to increase the effective antenna size, thus a monople antenna is often preferred for sub-GHz. (Actually it's a kind of asymmetrical dipole with the ground plane acting as second leg).

 

[tomars]

Maybe by doubling the turns I can get to 433Mhz

 

[FvM]

Consider an unfolded wire length of least lambda/4 (173 mm).

 

[tomars]

I can also make it to 433mhz.
Perhaps doubling the coils.
But the problem is that the bandwidth is smaller than the instability due to environmental conditions.
All I have to do is bring my hand closer that the curve comes totally out of the marker.

An antenna with little gain but a large bandwidth would be much better.

--- Updated Apr 20, 2023 ---

Consider an unfolded wire length of least lambda/4 (173 mm).

Monopole /4 require grout bottom.
I have gronud laterally.

 

[FvM]

If your application involves radio transmission, a lossy wide band antenna is rarely a good choice, e.g. due to output power limits for certification.

You should also focus on (expectable) S21 rather than S11. S11 of -10 dB means that still 90% of transmitter power is radiated or absorbed by the antenna.

 

[volker@muehlhaus]

Such a small antenna is narrowband, that is physics. If tolerances are an issue, you can introduce loss into your antenna to make it more wideband, at lower gain.

One possibility is to keep the F structure and add R||C at the open end. C will tune frequency down and R adds loss.

I have attached an example of a small 868 MHz antenna (for handheld remote) where that was done: make it wideband and robust by adding the resistor, and accept the resulting low gain. You can try to adopt this concept for our 433 MHz design.

 

[FvM]

Do you have a comparison with and without 2k resistor? I wonder if the resistance value introduces relevant losses.

 

[volker@muehlhaus]

Do you have a comparison with and without 2k resistor? I wonder if the resistance value introduces relevant losses.

This is not my design, I only simulated that many years ago.

To answer your question, I now compared both bases, with and without 2kOhm resistor:

Gain excluding mismatch losses is ~ 11dB higher without the resistor. Adding the resistor reduces gain, but helps with matching.

 

[tomars]

Waooooo
2K Miracle
The only problem is that the thickness of the resistor won't fit, I'll have to hide it inside a downturn.

NO res

2K res

 

[volker@muehlhaus]

Great, glad this works for you!

It seems to me that your antenna test PCB is smaller than the actual device PCB. You need to be aware that antenna tuning also depends on PCB ground size, so you need to tweak with the actual PCB size.

 

[tomars]

Not, is same, it is 70 mm high
I'll do one more test with an extra loop and then make a complete PCB with the whole circuit.
It feels like the right way, though.



I am no longer as lost and bewildered as before.

13Mhz RFID coil work also if one layer is covered.