Matching Network Design

[rvargo]

Hi,

I am new to antennas, so this is a fairly basic question. I am trying to design a matching network for a 2.4GHz chip antenna. I use a vector network analyzer to measure the antenna's impedance with no matching network. I then use software to design the matching network, and I lay out that circuit on the board. However, the impedance I measure with the matching network is nowhere near what the simulation says it should be.

We were reading some on this and saw that we should account for the electrical delay. Could somebody explain why this is? Also, if you have any other suggestions or hints on how to get accurate measurements?

Thanks for any help.

 

[Higgler]

look up "smith chart impedance match", it's all about matching an impedance.

 

[biff44]

When you said you measured the "impedance" of the antenna, at what physical position did you measure that impedance? At the coaxial connector at the input to an antenna, further down the line, right at where the antenna proper physically starts (ignoring any lead-in 50 ohm line)? Answer that first and we will take it from there.

 

[rvargo]

I have laid out pads for a matching network very close to the physical antenna. These pads are in a pi filter form. I short the series pads using a zero ohm resistor, and keep the parallel pads open. I measure the impedance at the series pad, right after the 0 ohm resistor. So it is close to the physical antenna.

 

[biff44]

Yep, that is the sort of thing I do when measuring an antenna. If you can do that, then move the electrical delay around until it looks like a smallish dot at the short circuit position on the smith chart display, when you unsolder the short you should be getting pretty accurate antenna measurements!

One problem I run into with antennas, especially the ones below 1 GHz, is that the mass of the network analyzer, and its cable, make a really big ground plane for the antenna. This will change the antenna characteristics somewhat. So the antenna impedance you measure will be a little off from what it really would be if you disconnected it from the network analyzer. Is your matching network a "little" off, or a lot off?

Also, if you are doing a very narrowband antenna, it might just need to be re-resonated slightly. Is it well matched just a little off-frequency?

http://www.MaguffinMicrowave.com

Microwave and wireless design consulting

 

[rvargo]

Sorry again for the basic questions, but this whole area is new to me. Could you explain exactly what the electrical delay is and why I need to account for it?

When do I need to account for the electrical delay? For example, I am getting demo boards of some antennas from the antenna makers and would like to measure them to see how well they are matched. Do I need to do the procedure (short to ground, then adjust electrical delay) for this case? Or is this something I only need to do when designing the matching network?

Thanks.

 

[biff44]

When you measure a 1-port device using a microwave network analyzer, what the hardware is actually measuring is reflection coefficient. What that means, in a nutshell, is that the network analyzer generates a stable signal (1 GHz for example), sends it out port 1, the signal travels along a transmission line, hits a load of some sort (your antenna), some of the energy goes into the antenna and is radiated, while some of the energy bounces back from the antenna. That bounce back energy travels back along the transmission line, goes back into Port 1 of the network analyzer, and the machine compares the bounceback amplitude and phase to the original amplitude and phase generated inside of the machine. That ratio is the reflection coefficient.

If you add a length of 50 ohm transmission line between port 1 and the antenna, the answer you get for reflection coefficient magnitude will be the same, but the phase answer will be different. This is because the signal travelling towards the load has to go thru this additional length, and the reflected signal has to go thru the additional length once again. You have to compensate for that additional length to get an accurate measurement.

One way is to use a short circuit at the exact location that the antenna would be, and vary the "time delay" on the network analyzer front panel. This just dials in a phase offset to the reflection coefficient displayed.

This is important, since the amplitude AND phase of the reflection coefficient are used to calculate the measured load impedance. If you get the phase wrong, your measured impedance will be wrong. And any matching network you design with that wrong measured data will not work well!

http://www.MaguffinMicrowave.com

Microwave and wireless design consulting

 

[rvargo]

Thanks, that clarified a lot for me!

 

[rvargo]

One last question...does the location where I am shorting to ground matter?

The board has an antenna, with the matching network pads nearby. There are pads for 2 parallel and 1 series connection. Does it matter which of the 2 parallel I short?

After the matching network is a transmission line that is about 50mm in length, then a connector. Do I want to short to ground near the connector instead?

Thanks again for all of your help!

 

[biff44]

It is entirely up to you. Where you use the temporary ground, and then use that electrical delay to correct the antenna data, that is the exact location where the network analyzer will calculate the impedance data at. If you decide you use your temporary ground 1 foot away from the actual antenna, then you will want later install your matching network 1 foot away from the antenna!

For other reasons (bandwidth), you usually want the matching network as close as possible to the load you are trying to match, so I would do it close to the actual antenna.

There are other, better means for doing all this. You can make a number of different standards (trl standards if you like agilent products) to do this measurement more accurately. The higher the frequency, the less likely you are actually making a "good" temporary short circuit, and therefore the less likely you are going to get accurate data. Certainly above 2.5 GHz, I would use a more accurate method than just dropping in a temporary short.

Also, a short cicuit must be physically small! No long lead lengths, use multiple vias to ground, etc. If it is not a short (ie. has inductance) you will get erroneous data too.

http://www.MaguffinMicrowave.com

Microwave and wireless design consulting