Dynamic antenna tuning by turning on/off inductor

[eriksen]

Hi,


EDIT: Frequency is 868MHz, and physical dimensions of PCB is approx 40mm*20mm.

I am trying to make a small wireless unit were I want to change the antenna matching based on wether the unit is placed in a hand or not.
My question is based on the attached picture.


I want to change the matching by turning on or off L2 when I sense a weaker signal.

My question is:
Will the trace with the L2 inductor affect the matching when the pin is left floating?
Does anyone have experience with something similar?

Thank you very much!

-Fredrik

 

[johnjoe]

It depends for example on your frequency. The trace and the input circuit of your IO-pin will have some influence. It's written floating but what about the high frequency behavior? It's better to use a rf-switch and to generate well defined conditions of your impedance network.

 

[biff44]

because of parasitic reactances and line lengths, switching an inductance is normally NOT done this way. What I have seen is some switching device, such as a FET or a PIN diode placed physically right next to the L2 inductor. THe control signal turns this local FET/Diode ON/OFF, and the parasitics are minimized.

 

[volker@muehlhaus]

My question is:
Will the trace with the L2 inductor affect the matching when the pin is left floating?
Does anyone have experience with something similar?

If we assume trace length much shorter than the wavelength, this will simply add parasitic capacitance to ground. You can approximate the capacitance from conductor area and substrate thickness and permittivity (plate capacitor equation).

 

[eriksen]

Thanks for alot of good input.

I have edited the start post to include frequency and board dimensions. 868MHz and approx 40mm*20mm.

The trace would need to be approx. 15mm.
And the reason for trying out a solution like this is mainly about BOM cost.
So that's why I wanted to hear if anyone has experience with this simple method.

Using an RF-switch is also something I am considering, but it seems unnecessary if the above mentioned solution is viable without to much noise.
Is there multiport RF-switches that could switch between several different inductors witout costing to much available?

 

[johnjoe]

The main problem of your idea is the high-impedance I/O. The trace is a minor problem because you can take him into account of your matching network. But honestly I don't believe you can estimate the behavior of the I/O at 866 MHz in both cases, shorten to ground and high-impedance. You don't know the internal circuit. That's why I believe it will not work, or work somehow.

 

[eriksen]

Thanks.

I can understand that.
So that it's called a high impedance port when floating is not enough to be sure about how it's going to react?

So I've tried to come up with a cheap alternative as possible and landed one something like this:
PE4239 from Peregrine.
**broken link removed**

My initial thought is to put this in front of the matching inductors and connect it like on the attached picture.
This would give me two different matchings based on signal to CTRL pin.
If i'm not missing something. (?)

Thanks.
-Fredrik

 

[chuckey]

why not use a varicap to tune the input inductor, then detune it by varying the bias on the varicap?
Frank

 

[volker@muehlhaus]

why not use a varicap to tune the input inductor, then detune it by varying the bias on the varicap?

The difficulty is that this is an analog solution, and usually requires higher voltages to tune the varactor.

A recent solution antenna matching approach used in mobile terminals is based on digitally controlled capacitor banks:
**broken link removed**

 

[eriksen]

Thanks for answers. The qfe15xx solution seems powerful but expensive.

Any comments on my suggestion above using the rf-switch switching in either inductance?
The two imagined use cases are either the unit is standalone or in a wristband, which will shift the impedance drastically.

Measurements done with an VNA shows that changing the inductor value today between 10 and 12 nH does the trick between the two cases.

But my concern is switching between two components, as I have never tried anything like it.

Thanks.

 

[D.A.(Tony)Stewart]

I recall reviewing Delphi's hundreds of patents for remote control garage openers for cars including auto-tuning antenna.

Here's an interesting one which tunes the length of the element (L & C)
https://www.google.ca/patents/US6061025

Maybe that can inspire you to read all their patents and make a better widget.

 

[mtwieg]

The difficulty is that this is an analog solution, and usually requires higher voltages to tune the varactor.

A recent solution antenna matching approach used in mobile terminals is based on digitally controlled capacitor banks:
**broken link removed**

From what I've seen in RF CMOS research, adjustable capacitance (usually with a switched bank of fixed caps) is always preferable to switching inductors for reasons of noise and size. Or you could use a variable gyrator instead, but that probably would be even noisier.

 

[tony_lth]

QFE15XX may have 2H issue, and needs tons of test. Too complex.

 

[vfone]

I've been playing with antenna tuners (and with tuners in general) for many years now, at various frequencies from VLF to microwave frequencies.

Using antenna tuners in mobile phones is an overrated approach and it is almost 100% a marketing business. I've been evaluating solutions from Qualcomm, Ericsson (ST), Infineon (Intel).
The overall improvement that you get is insignificant compared to the price of the development, hardware, and the processing power that the BB processor needs to accommodate all the variables.

In top of these issues, I've seen unpardonable mistakes made by designers when they use the same antenna tuner for both paths, TX and RX, when the split between TX and RX in some cellular bands could be up to few hundreds of MHz. So they tune for best performance in TX but lose sensitivity in RX, because most of the tuners are narrow-band by definition.

 

[volker@muehlhaus]

So they tune for best performance in TX but lose sensitivity in RX, because most of the tuners are narrow-band by definition.

Sure. The point with antenna tuners in mobile phones is to reduce transmitter power consumption -> increase battery life.
I've been supporting some of this work (EM simulation of front end modules and antenna tuners) ... these guys are working hard even for 0.1 dB better insertion loss.