0
\$\begingroup\$

I am trying to design a distributed amplifier in which the parasitic capacitances of a small-signal FET model are linked to inductors to form a distributed line across the drains and gates and I'm frustrated because nothing is matching what I expect from theory.

I know the characteristic impedance of my two lines should be sqrt(L/C) and I've chosen my values as such.. however I am still getting a ringing in my input impedances even when I just simulate a simple LC distributed circuit cell for a transmission line, and there is no real part as can be seen in the graph.

So my questions are - does the characteristic impedance calculation not work in AWR for a distributed element model of a transmission line?

If I measure the S(5,5) or S(1,1) it is resonant at DC, which makes sense because an inductor is a short at DC frequencies, but then what does the resonant frequency wc=1/sqrt(LC) mean and what is its significance in the AWR simulations? enter image description here

Thanks for your time!

-- edit:

so the ultimate goal is the distributed amplifier as it appears below where this cell is repeated five times. The inductors add in series, so at the end there is another L/2 value inductor while between each cell the inductors are value L. The four ports are all terminated in a 50Ohm impedance. I simplified by taking out the fet small-signal model to inspect the input impedance.

By ringing I mean that the impedance value is purely imaginary and oscillates. I guess I'm not surprised that a few LC circuits are not giving an input impedance of sqrt(L/C) but I just can't justify to myself why they disobey this relationship because it should scale with length and number of cells.

enter image description here

I really apologize for the horrible screenshots, I have a cap on how many images I can post.

--edit So I realized I was plotting Z(1,1) instead of ZIN in AWR, which solves a lot of my issue understanding what I'm doing wrong. But thank you for your time and thoughts I really appreciate it!

\$\endgroup\$
8
  • 1
    \$\begingroup\$ What is "AWR Microwave Office"? \$\endgroup\$
    – Transistor
    Commented Feb 21, 2016 at 21:41
  • 1
    \$\begingroup\$ I don't think you can just "throw in" a couple of inductors and capacitors like that and expect them to work as a transmission line. I have a feeling you don't have a clue what you're doing ! The tool will work fine for sure but don't expect that you can just use it in a meaningfull way without following a tutorial or reading the manual. \$\endgroup\$
    – Bimpelrekkie
    Commented Feb 21, 2016 at 21:56
  • 1
    \$\begingroup\$ Unless you are going to add infinite L-C nodes, it's not going to behave like a transmission line. If you add maybe 100+ you may start to see an ok approximation (that's how things like LTSpice to it), but in no way will it be accurate. Use the built in transmission lines. \$\endgroup\$
    – Tom Carpenter
    Commented Feb 21, 2016 at 22:13
  • \$\begingroup\$ A single L/R will not adequately simulate a transmission line, nevertheless your first LCR circuit will produce a 2nd order TF with \$\small \zeta=0.5\$ so the resonance peak will only be 15%. Are you sure you're loading the C with \$\small 50\Omega\$? Is the connection to the output terminal of the port correct? \$\endgroup\$
    – Chu
    Commented Feb 22, 2016 at 1:13
  • \$\begingroup\$ @TomCarpenter, in my experience, 100's of sections are not required and even 4 or 5 LC sections is enough to start seeing behavior approximating a transmission line. \$\endgroup\$
    – The Photon
    Commented Feb 22, 2016 at 2:16

2 Answers 2

Reset to default
1
\$\begingroup\$

Distributed Amps the way you built should be ringing, in theory as well as in practice. To get them to work without ringing, you need an "m-derived network." old Tek manuals from the vacuum tube era show how.

For example, see here http://w140.com/tekwiki/images/3/36/545a_vert.png

the center-tapped inductors are key. Google "t coil network" or "m derived network" for the design equations.

i refer to any of the tapped inductors in the tektronix schematic. for an impedance of Z=1 and a fet capacitance of C=1, we need each half of the inductance to be 1/3 and the mutual inductance to be m=1/2. that gives a 3 db bandwidth of B=2.7 and rise time tr=.79. In theory, a capacitance of C/12 from end to end of the tapped inductor is understood.

so if, for example, Z=50R and C=1.5pF, we get an inductance of 1.25 nH, m=1.875 nH, B=5.7 GHz and tr= 59 ps. Overshoot is less than a percent.

\$\endgroup\$
0
\$\begingroup\$

You need your shunt capacitors at the end of your TL to be C/2 to make the unit cells symmetric c/2--L--C--L--C--L--C/2

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.