Multi element matching using the Smith chart

[deba_fire]

Multi element matching using Smith chart

Hi

Multi element matching is needed for separating optimizing the Q and bandwidth of the circuit. For multi element matching(>3) components, how does one go about this using Smith charts? As for multi element there are an infinite paths to reach the center of Smith chart. Is there a method which is followed?

Thanks

 

[vfone]

Re: Multi element matching using Smith chart

When try to reach the center of the Smith chart using multi-element matching components, important is to try staying on the same side of the Smith chart (capacitive or inductive) from where you start.
Depending how many matching components are used, not always is possible to do this.

 

[FvM]

Re: Multi element matching using Smith chart

You'll implement multi element matching for a reason, e.g. increase the matching bandwidth. Respectively the circuit topology and element values will be chosen to achieve this objective.

A smith chart is constructed for a single frequency, in so far it's not well suited to design wide band matching.

 

[vfone]

Re: Multi element matching using Smith chart

When designing a multi-element matching network using Smith chart (which is done very often in RF design), the wide band characteristic of the resulting circuit is a consequence of this fact.

 

[FvM]

Re: Multi element matching using Smith chart

When designing a multi-element matching network using Smith chart (which is done very often in RF design), the wide band characteristic of the resulting circuit is a consequence of this fact.

Thanks, I must confess, I have used Smith chart only for small band yet.

I can imagine that dividing the transformed impedance ratio into two or three sections of equal ratio will result in bandwidth improvement. But won't it be more straightforward to perform an equal impedance ripple fit by numerical methods?

 

[vfone]

Re: Multi element matching using Smith chart

Yes, the filter design approach of performing numerical methods to get an equal-ripple response could be applied to a multi-element impedance matching network also.
Smith chart approach might not give you directly an information about the circuit bandwidth but this can be analyzed after the right impedance matching is done.
An RF simulator can do easily a simultaneous optimization for impedance matching and for desired bandwidth.

 

[deba_fire]

Re: Multi element matching using Smith chart

Thanks for useful suggestions FvM and vfone. Two element matching is very straightforward using Smith charts. For multi-element matching, one needs to use the numerical method similar to any filter design.

You have mentioned RF simulator for for simultaneous optimization of impedance matching and bandwidth. I use Spectre circuit simulator from Cadence but haven't seen such optimization. Can you suggest one?

 

[vfone]

Re: Multi element matching using Smith chart

NI AWR, ADS Keysight can do such optimization. Important is what inputs and outputs you chose for optimization.
I don't know if those simulators use any of the published broadband impedance algorithms (Carlin, Fano-Youla, Cuthbert) but they get good results.
Biggest challenge in broadband matching is the frequency dependence of the load impedance. This is the reason that I mentioned to try staying (if is possible) on the same side of the Smith chart, because this would help somehow in this matter.

 

[volker@muehlhaus]

Re: Multi element matching using Smith chart

I am an ADS user, but for really challenging matching circuits, I'm impressed by Optennilab.
https://www.optenni.com/

 

[E Kafeman]

Re: Multi element matching using Smith chart

I am very much biased as I did develop a extremely fast wide band impedance tuning algorithm many years ago. It is implemented in AnTune software. That software is probably the only software that performs wide band optimized matching in real time, as fast as VNA data transfer to a PC allows. Resulting net impedance takes account for non ideal components and tries to minimize related losses.
Another important function, AnTune provides semi auto port forwarding and impedance calibration at location for the intended network at PCB. Great help if VNA not already have that feature. Failing to set correct time delay is else a common problem when implementing a network, especially at ++GHz frequencies and with some distance at PCB between antenna and network as a few mm wrong settings drastically can change resulting real network impedance compared to expected simulated network impedance.
Especially for handheld wireless units must often two (or more) impedance situations be taken in account, free space and handheld. AnTune handles that too. It must then be a compromise, but this problem can be reduced in an early antenna development stage so that impedance shift due to handheld and free space cost less as resulting impedances can be seen during antenna development in AnTune for antenna+optimized network before any real network is implemented.
Idea is that a slightly mistuned antenna and a optimized network gives better final tuning result for two different situations, then what a better tuned antenna for just one situation and its matching network allows.
It is a similar problem when optimizing for TIS&TRP which can be two different impedances and goal impedance can then also be very frequency depending, especially if it is a multiband radio.
Optimize an embedded wide band antenna for best efficiency result seldom in a pure 50 Ohm antenna. It is also complex to design the antenna dual band or very wideband and afterwards hope that an fitting impedance network can be found that correct impedance to fit actual radio.
Many of by me designed wideband antennas do not have any nice impedance curve, because I do only check resulting sum impedance for network and antenna, when developing the antenna. AnTune do then show me optimized matching result at same moment as I adjust antenna shape. This part is probably where I save most of the time when designing antennas. Final resulting impedance for just the antenna is of less interest.
Set any network topology or let AnTune auto select.
Works with most types of VNA. All software data formats and VNA settings are in open formats. Makes it easy to create your own component library or adjust VNA settings with aid of Excel or Notepad.
AnTune is very much focused on solving problem related to non 50 Ohm impedance goal for embedded wideband antennas, ~300MHz and upward, so it is maybe less useful for full size antennas in free space and less wide frequency requirements.
AnTune is not a simulation software, it is a real time software. Need for that reason some kind of connection to a VNA (GPIB/USB/LAN). It can of course import S-parameters from file and do same calculations as with live data but that method is a much less effective way to develop embedded antennas, impedance network and RF-filters.

 

[tony_lth]

Re: Multi element matching using Smith chart

Yes, AnTune software developed by E_Kafeman is very good, I used it when I worked in Nokia.
In the several bands matching, AnTune is even better than Optennilab.