Narrowband Matching Network to Broadband Network

Narrowband Matching Network to Broadband Matching Network

I had a source that has a 60.95-j45.75 impedance. Through the help of some on this forum I got a matching network that provided great matching to 50 Ohm at 300 MHz.

The thing is that I need broadband matching as oppose to the narrowband matching. A BW of 100 MHz is desired. The image below is the matching network. Can someone help me get 100 MHz bandwidth out of this?

I figure that I need to create a Band Pass Filter as the matching network that has a 300 MHz center and bandwidth of 100MHz, i.e. cutoff frequencies at 200 & 400 MHz.

Or what about transformers?

An exact matching using only L and C components is most likely not possible. I guess, the best way is to accept a certain mismatch and roughly compensate the capacitive source impedance for the center frequency with an inductor. Using transformers and transmission line segments can offer additional options for wide band matching. I guess, you don't intend a lossy RLC matching?

You are doing something wrong in your analysis of this matching network. You should get a >20dB return loss bandwidth of nearly 200MHz with this network.

edit: see post #7 for a possible reason you are seeing a narrow bandwidth with this network.

I figure that I need to create a Band Pass Filter as the matching network that has a 300 MHz center and bandwidth of 100MHz, i.e. cutoff frequencies at 200 & 400 MHz.

Click to expand...

A 100MHz bw is 250 to 350MHz so are you really asking for a 200MHz bw?

Hello, infinite_gbps
As i can see you have used ideal L and C in you matching network - so you get very narrow bandwidth.
If you use real, for example, Murata's elements, i guess the plot will be different.

pavel_adameyko said:

Hello, infinite_gbps
As i can see you have used ideal L and C in you matching network - so you get very narrow bandwidth.
If you use real, for example, Murata's elements, i guess the plot will be different.

Click to expand...

Not so. This is a very low Q network and I would expect that the difference between ideal components and real components will be minor. With real (SMD?) components the S21 will be typically around 0.15dB (instead of 0dB at 300MHz) but the bandwidth of S21 and the return loss will be virtually identical.

This assumes the real circuit is built using good RF practice and typical SMD components.

The current network should give <0.25dB insertion loss over 200-400MHz with reasonable quality SMD parts. The loss will be slightly less with better quality parts.

I didn't expect a .25 dB insertion loss, but you are right, the shown return loss graphic doesn't fit the shown source impedance and matching network. It's showing a kind of high Q
resonator.

In reality I suspect the (60.95, -45.75) Ohm source itself will show a different impedance across 200-400MHz and this will complicate the matching network.

Perhaps IG should remeasure the source impedance to see how it changes across 200-400MHz and this will reduce the possibility of disappointment when testing the network on the real hardware. i.e. does the source on its own still look like 60.95, -45.75 across 200-400MHz. I suspect it will not.

Can you post up the s parameters of the source over 200-400MHz?

I suspect a more complex network will be required as the source may have a very different impedance across 200-400MHz.

G0HZU said:

You are doing something wrong in your analysis of this matching network. You should get a >20dB return loss bandwidth of nearly 200MHz with this network.

edit: see post #7 for a possible reason you are seeing a narrow bandwidth with this network.

A 100MHz bw is 250 to 350MHz so are you really asking for a 200MHz bw?

Click to expand...

Hi G0HZU, thanks for your help. Want to change the discussion to focus more on the BW. What equations support this >20dB RL at 200 MHz? Or the program that you showed me, how can my analysis change?

All I was doing is using the S-parameters of the source to match to a LNA with a 50 Ohm impedance across that band of interest.

G0HZU said:

The current network should give <0.25dB insertion loss over 200-400MHz with reasonable quality SMD parts. The loss will be slightly less with better quality parts.

Click to expand...

Any way to redo my analysis, either with equations or in ADS? Maybe I should just get the transfer function of the lumped element matching network to see the proper frequency response. What do you think?

G0HZU said:

In reality I suspect the (60.95, -45.75) Ohm source itself will show a different impedance across 200-400MHz and this will complicate the matching network.

Perhaps IG should remeasure the source impedance to see how it changes across 200-400MHz and this will reduce the possibility of disappointment when testing the network on the real hardware. i.e. does the source on its own still look like 60.95, -45.75 across 200-400MHz. I suspect it will not.

Can you post up the s parameters of the source over 200-400MHz?

I suspect a more complex network will be required as the source may have a very different impedance across 200-400MHz.

Click to expand...

S11 of source attached. You are right I need to get proper match from 200-400 MHz. Like you stated, will I be able to use the network above or have to do something more complicated.

FvM said:

I didn't expect a .25 dB insertion loss, but you are right, the shown return loss graphic doesn't fit the shown source impedance and matching network. It's showing a kind of high Q
resonator.

Click to expand...

Hi FvM, I appreciate your help in the past and you pitching in now.

Any way to alter the analysis above? Thought I was doing it correctly. I'm opening to doing the math to show what I should be doing but please let me know what analysis I should be doing to understand the Q I should be getting and proper response. Thanks.

The S11 file for your source shows the impedance changes a lot over the 200-400MHz range. This will make it much harder to match with a decent bandwidth.

Can I ask what the source consists of? The S11 file looks like the response from a narrowband filter. This is why you only see a narrow bandwidth match.

G0HZU said:

The S11 file for your source shows the impedance changes a lot over the 200-400MHz range. This will make it much harder to match with a decent bandwidth.

Can I ask what the source consists of? The S11 file looks like the response from a narrowband filter. This is why you only see a narrow bandwidth match.

Click to expand...

It is looking into a switch that switches a network in and out. The S11 is when the switch is closed.

Can you elaborate on the your last two sentences a little please? Thanks.

Any ideas in getting the required BW, i.e. lower Q? I was thinking about adding some resistance to get the Q down a little and in the mean time try to think of a way to match to the LNA that is 50 Ohms.

Is the network after the switch a narrow RF filter or duplexer? i.e. does your system look like antenna + duplexer/filter + switch + (new matching network) + LNA?

The reason I think your source looks like a filter is because the reflection coefficient swings around the smith chart a lot either side of 300MHz.
If you plot the return loss of the source you will see it is very poor immediately either side of 300MHz. This is a similar response to a narrowband 300MHz filter.

It will be difficult to match this without having the following problems:

The network produced to give a good broadband match will be quite complex and will probably be very sensitive to component tolerance.

The network will probably end up with an undesirable insertion loss at 300MHz (ahead of your LNA)

Are you expecting to receive signals over 200-400MHz with your LNA?

G0HZU said:

Is the network after the switch a narrow RF filter or duplexer? i.e. does your system look like antenna + duplexer/filter + switch + (new matching network) + LNA?

The reason I think your source looks like a filter is because the reflection coefficient swings around the smith chart a lot either side of 300MHz.
If you plot the return loss of the source you will see it is very poor immediately either side of 300MHz. This is a similar response to a narrowband 300MHz filter.

It will be difficult to match this without having the following problems:

The network produced to give a good broadband match will be quite complex and will probably be very sensitive to component tolerance.

The network will probably end up with an undesirable insertion loss at 300MHz (ahead of your LNA)

Are you expecting to receive signals over 200-400MHz with your LNA?

Click to expand...

So you bascially got it. There is a switch on the antenna that opens it and closes the loop. The antenna goes into an LNA. There is no rf filter or duplexer.

The Q coming out of the antenna is already high (relatively speaking). I just expect that the broadband matching can help.

I FOUND Bowick BTW!!!!! :-o 2nd Edition

It states that (in the "Low-Q or Wideband Matching Networks" on page 69) that one needs fairly broad range of frequencies matched that one can simply use two L sections. They would be series connected rather than using back to back configurations of Pi or T networks.

Agree with this? Not sure what he is talking about as be goes on.

Great book though!!!!

It looks like the antenna you are using is a very narrowband design and that is the reason your S11 file shows the response that it does.
Therefore, you might be better going for a wider bandwidth antenna rather than try and design a (crazy?) matching network to partly make up for the antenna. Is changing the antenna feasible?
Otherwise I don't see why you are bothered about a wide match if the basic antenna is only going to receive efficiently over a narrow bandwidth anyway...

I hope I'm not taking this thread off course...

It states that (in the "Low-Q or Wideband Matching Networks" on page 69) that one needs fairly broad range of frequencies matched that one can simply use two L sections. They would be series connected rather than using back to back configurations of Pi or T networks.

Agree with this? Not sure what he is talking about as be goes on.

Click to expand...

I think that Bowick is referring to fairly basic source and load matching tasks whereas you are trying to match over a wide bandwidth that shows a very diverse range of impedances. You won't be able to match it with a couple of L matches. Can you change the antenna?

I agree with what your saying and no your not taking this thread off course. I like the discussion. It helps me think.

My issue is that I did not design the antenna and it is known that the Q is already so high on the antenna. Your comment earlier about the S11 looking like a narrowband filter resulting in my matching looking like narrowband made me ask the question as to why is the antenna narrowband in the first place. But can't do anything about it at the moment.This is a good lesson learned for me.

Can you answer a few questions for me? This will help me make my point to some people.

How "crazy" will that matching network be? Is it next to impossible to get a Q below lets say 10? Even 15?

Also can you elaborate on what you meant by saying "fairly basic source and load matching task"? I thought broadband matching meant that you have frequencies that saw different impedances. Are the differences so huge that make it impractical?

Thanks so much for your help. My practical knowledge around this stuff has increase because of you.

I had a source that has a 60.95-j45.75 impedance.

Click to expand...

The statement has been effectively disproved by the tabulated S11. I don't expect, that a wideband impedance matching can be achieved with this higly resonant source impedance. As a simple example, consider a series LC circuit. It could be only compensated by a negative inductance and capacitance, which both doesn't exist.

Yes, and even if a (reactive) network was found that gave an improvement in match across a wider bandwidth it would only work well under ideal conditions.
eg if you moved the antenna or changed the way it was grounded or you walked it close to other objects then the matching network would no longer be effective because the antenna s11 would change.

The best way to get a stable match and good rx performance would be to consider a different antenna. A lot depends on what system specification you have been given for antenna and receiver performance over a given bandwidth.

What initially appeared to be a simple matching exercise has actually revealed itself to be a very unrealistic antenna matching task. I think maybe you need to discuss this with your colleagues to see what is really required.

---------- Post added at 01:52 ---------- Previous post was at 01:23 ----------

Also can you elaborate on what you meant by saying "fairly basic source and load matching task"? I thought broadband matching meant that you have frequencies that saw different impedances. Are the differences so huge that make it impractical?

Click to expand...

in this case yes they are. Basically, the further the load (at any givevn frequency) appears from the centre of the smith chart, the worse the (unmatched) match is and the harder it is to get good bandwidth if you try and match it with an LC network. If you plot your s11 data on a smith chart you will see it quickly sweeps close to the edge of the chart as soon as you look at frequencies more than 30MHz away from 300MHz.

This is why your task is so difficult with this narrowband antenna. This is because 250MHz and 350MHz are outside the tuned antenna's natural bandwidth and are therefore on very different parts of the smith chart and they are also close to the edge of the chart. So even if you wanted to just match this source at 250MHz you would have a very narrow bandwidth and a unique matching network. It will almost certainly be unsuitable for 350MHz.

To make things even harder, if you could 'see' the s11 plot on the smith chart as you moved the antenna or moved objects close to it you would see the S11 plot wobble on the smith chart like a jelly as the antenna impedance changed and this would degrade the performance of any matching network a lot. It wouldn't wobble so bad at 300MHz because the antenna is tuned to 300MHz and it will wobble in an area close to the centre of the smith chart (no big deal)

Usually, broadband impedance matching is much simpler than this so don't let this task put you off trying to match 'kinder' sources or loads in the future.

In addition, you should consider, that there is apparently a lower limit for the Q of electrically small antennas. I guess, it plays a role in your case. I quote the statement from Texas Instrument Application Note AN003 (SRD Antennas):

McLean [3] has described the fundamental theoretical limit for the minimum Q-value of a small antenna. If the antenna can be placed inside a sphere of radius a, the minimum Qvalue-value for a loss-less antenna is

Qmin = 1/(ka)³

where k = 2pi/λ

This expresses the absolute minimum Q value the antenna can take. Unfortunately, the theory does not tell us how to implement a minimum Q antenna. The antenna Q can of cause be reduced by introducing loss (a resistor) in addition to the radiation resistance, but this would reduce the antenna efficiency, see below.

J.S.McLean: A re-examination of the fundamental limits on the radiation Q of electrically small antennas, IEEE Trans Antennas Propagat, vol 44, pp672-675, May 1996

Click to expand...

FVM,

Thanks for the reference. Good read.

G0HZU, thanks for the continued help.

I have another question which plays off this fact that my impedance changes significant as I move away from 300MHz. Then I open the switch on the antenna and now looking into the LNA is high kOhms. Do I have to worry about stability issues? Link: https://www.edaboard.com/threads/192357/#post805304

Please respond at the link above. Don't want the moderators coming after me.