matching network where Rs > Rl ?

[ssbothwell KJ6RSG]

hi guys. been really busy and havent had a chance to work on any of my projects lately. anyways, i was just messing around with some matching network calculators online and noticed they seem to get crazy values if i set the source impedance higher then the load impedance. i can use the same values but put the small one as Rs and the big one as Rl and the calculation seems to come out okay. this seems to be the case in all the calculators i have found.

why is this going on?

and, can a matching network be used in either direction? eg., can i setup the network so Rs is the smaller value and Rl is the bigger value but then use Rs as the load and Rl as the source?

[k4kyv]

I have used a simple L-network, with series rotary inductor, and variable cap to  ground.  To match a higher impedance than the  source, the cap  goes to the output side of the inductor.  To match a lower impedance, the cap goes to the input side.  I used a switch to change the cap over to the opposite side of the inductor.  I believe that's a standard method used with commercially built tuners.

[WD8BIL]

In GENERAL..... (anti-expert statement)

When feeding a low Z source, I assume you're talking antennas here, you'll want the inductive reactance toward the Low Z.

In your model flip the matching network as you flip the source/load Rs and see what happens.

[W4AMV]

I think you are confusing yourself about the definition of Rs and RL. The math does not care if Rs > RL or visa versa. To transform a resistive value UP you apply the series to parallel transform (see that pdf sent) and the shunt matching element will be on the HIGH Z side. Recall you will obtain a Q for the match (and the L network is a minimum Q network) and the Q can be positive or negative. So the match system could be high pass or low pass. So a real Z to real Z match will result in two solutions and the shunt element of the final L match will be on the HIGH Z side of the network. Complex to complex match a little more involved and a possibility of FOUR solutions result!  I'll see if I can right this all up with some examples and post.

[ssbothwell KJ6RSG]

thanks for the tips guys. i'll experiment with flipping the network around.

alan, i am going to go back over those notes you sent me. thanks again.


can you guys recommend some software for modeling matching networks?

[W4AMV]

This is well done. There is a dedicated tab for matching design and much more.

http://electroschematics.com/835/rfsim99-download/

[ssbothwell KJ6RSG]

alan, that program is very cool.

i just tried entering a a hypothetical 1000 Zin to 30 Zout match into RFSim99 and it gave me a really nice chart. however, it put the capacitor on the 1000r side (Zin) of the inductor. everyone in this thread is saying it should go on the Zout side of the inductor.

if i manually move the capacitor to the Zin side of the inductor then the simulation doesnt dip at the resonance point. i'm attaching images of the two simulations to clarify this.


WD8BIL, you are correct that i am trying to understanding matching networks and how they are used to connect low z loads (antennas) to high z sources.

can any of you guys recommend me any articles or textbooks that explain this stuff in relatively simple terms?

[W4AMV]

Again, for your case 30 ohm match to 1000 ohm. In the solution provided by RFSIM, 2 solutions occur. One low pass, second one high pass. Any other arrangement as you explored, shunt element on the low Z side, does not work for this case. The answer to your question is in understanding the equivalence of circuits. In this case, how a series circuit has an equivalent parallel circuit. Let me see if I can find a simple write up. Yes, there is some algebra involved. Or, the Smith Chart shows this very nicely as a graphical solution. It would be nice to show the solution to your problem on the Chart (visual is nice) and then show it via the algebra.

[ssbothwell KJ6RSG]

thanks alan. i look forward to any information you can provide.

what does it mean that a  matching network is low-pass or high-pass? i didn't realize matching networks operate like filters, although it makes sense considering the topologies are so similar.

if a having the shunt element on the low z side does not work in this situation, is it not possible to match those values (1000 to 30) with an L-Network?

[W4AMV]

what does it mean that a  matching network is low-pass or high-pass? i didn't realize matching networks operate like filters, although it makes sense considering the topologies are so similar.

lossless matching networks are filters! If the shunt element is C, series element L, then low pass. Flip these and high pass. 

if a having the shunt element on the low z side does not work in this situation, is it not possible to match those values (1000 to 30) with an L-Network?
[/quote]

YES. The program you used to create the match of 1000 ohm to 30 ohm gives you a solution. Indeed that is the correct solution to match 1000 to 30. Understand that in this solution the shunt reactance ( XL or XC ) is on the high impedance side (1000 ohm) of the circuit. YES, it will not be possible to place the shunt element on the low impedance side and match to 1000 for this simple matching circuit. However, for more complex matching networks it is possible and you may investigate those with the tool I sent, RFSIM.

[ssbothwell KJ6RSG]

so for the purposes of impedance matching, what is the difference between a high pass and a low pass network?

also, what is the significance of the shunt being on either the high or low impedance side?

[W4AMV]

so for the purposes of impedance matching, what is the difference between a high pass and a low pass network?

For the purpose of Z matching, no difference. However, matching networks are filter networks, therefore their frequency response will follow. Again, look at RFSIM frequency plots.

also, what is the significance of the shunt being on either the high or low impedance side?

No significance. Again, for the simple L, the shunt element will be on the HIGH Z SIDE.

[ssbothwell KJ6RSG]

ahh i see. so if i was building a matching network and i wanted to take advantage of the harmonics (perhaps in a VFO) i would use a high pass network?

out of curiosity, is there such a thing as a broadband matching network? what if you are working in the GHz range and want to operate at a bunch of different frequencies? do you have to use different matching networks for each band?

[W4AMV]

ahh i see. so if i was building a matching network and i wanted to take advantage of the harmonics (perhaps in a VFO) i would use a high pass network?

out of curiosity, is there such a thing as a broadband matching network? what if you are working in the GHz range and want to operate at a bunch of different frequencies? do you have to use different matching networks for each band?

YES, correct and if you wish to pass just the fundamental, then LP.

On the second question YES. This is where the notion of impedance matching and the concept of filters come together. In affect, a wideband filter is a broadband impedance matching network. Its type is BANDPASS. How to design these things is a whole new thread.

Important to note... For the simple matching system like the "L", the impedance match is exact only at a single frequency. Hence the frequency response as shown in the simulator. The plot reponse is also telling you the degree of MISMATCH with the change in frequency.

[ssbothwell KJ6RSG]

[quote
On the second question YES. This is where the notion of impedance matching and the concept of filters come together. In affect, a wideband filter is a broadband impedance matching network. Its type is BANDPASS. How to design these things is a whole new thread.
[/quote]

well i'll leave this for another thread but this is really interesting. the fact that filters and matching networks are the same thing is really making me curious. i need to go a lot deeper in the arrl handbook section on filter design.

Important to note... For the simple matching system like the "L", the impedance match is exact only at a single frequency. Hence the frequency response as shown in the simulator. The plot reponse is also telling you the degree of MISMATCH with the change in frequency.

so what is the bandwidth value in RFSim99 representing? i assumed that was the range around the center frequency where the impedances are matched, but you are saying that is not the case?

[W4AMV]

No, that is not what I am saying, YES, you are correct... around the frequency range where the match is perfect. For example, investigate if the low pass L is used and a match is desired at 1 MHz. Then at 1 MHz we indeed see an exact match. As we move above or below 1 MHz we experience a loss in power transfer which is directly related to the degree of mismatch. One metric to define the frequency bandwidth of the match is to investigate where the loss is 1 dB in excess of the perfect match. There are other definitions.

[ssbothwell KJ6RSG]

ahh i see. i just made a 1000r:30r L-Network around 1mhz. bandwidth is 654kHz and -1dB at about 1.36mHz and 420kHz.

can you send me any info on what the square plot chart is representing? there are two lines, red and blue. blue has a huge dip at the resonant point of the network.

i assume these two lines represent are the two ports and they represent forward and back reflecting waves, is that right?

what does the dip mean? is it indicating the reverse reflected waves being very low dB at the resonant frequency?

[W4AMV]

So sorry for delay. On the plots, as I recall for your network response when you use the match tool, you will see insertion loss and return loss plots. The plot with the DIP, is return loss. So, the perfect match at the design frequency implies NO RETURNED power. The other plot is insertion loss plot. So at the same design frequency, perfect match is ZERO LOSS. The program makes heavy use of S-parameters. Read all the literature... lots available and there you will find the info on these screwball return loss and insertion loss definitions. You should also now begin to see why filter networks are alos matching networks and visa-versa!

73.

[ssbothwell KJ6RSG]

thanks for explaining that. i've been reading relevant passages from the arrl handbook and am starting to understand this stuff.

am i right to say that in these models we are assuming that the impedance is purely resistive? do you have to use an antenna analyzer to figure out the actual reactance and to know how to totally tune-out the capacative and inductive reactance?