Hi, everyone
I need a balun with such performance: when the single-end port is floating or load 50 ohms, add a signal to one of the unbalanced-end, the other port will get a 180 degree inverse signal to the input, and the same p-p voltage.
I found that the balun model in the rfLib can't  reach my demand. When I add a signal to one of the unbalanced-end with the other two ports load 50 ohms, the single-end port have a signal, but the other unbalanced-end get nothing.  
So I modified the veriloga file as following, but I don't konw if it is correct. Could someone tell me? Thanks.

module balun(single, bal_p, bal_n);
 inout single, bal_p, bal_n;
 electrical single, bal_p, bal_n;
  parameter real rin = 50 from (0:inf);
  parameter real rout = 50 from (0:inf);
  parameter real loss = 0 from [0:inf);

  real      a1, a2, a3;
  real      s21, s31, s12, s13, s23, s32;

  electrical single_int, bal_p_int, bal_n_int;

  analog begin

     @(initial_step("static") or initial_step("pss") or
     initial_step("pdisto")) begin
      // unfold the S parameters from dB to real values
      s21 = pow(10,-loss/20);
      s21 = s21/sqrt(2.0);    // energy converation law!
      s31 = -s21;
      s12 = s21;
      s13 = s31;
      s23 = -1;      
      s32 = -1;      
     end

     // calculate the normalized incident waves: a1, a2
     a1 = V(single)/sqrt(rin) + I(single, single_int)*sqrt(rin);
     a2 = V(bal_p)/sqrt(rout) + I(bal_p, bal_p_int)*sqrt(rout);
     a3 = V(bal_n)/sqrt(rout) + I(bal_n, bal_n_int)*sqrt(rout);
     
     // input
     // be careful about the sign of the controlled source
     I(single_int) <+ V(single)/rin;
     I(single_int) <+ -a2*s12/sqrt(rin);
     I(single_int) <+ -a3*s13/sqrt(rin);
     
     // output positive node
     I(bal_p_int) <+ V(bal_p)/rout;
     I(bal_p_int) <+ -a1*s21/sqrt(rout);
     I(bal_p_int) <+ -a3*s23/sqrt(rout);
     
     // output negative node
     I(bal_n_int) <+ V(bal_n)/rout;
     I(bal_n_int) <+ -a1*s31/sqrt(rout);
     I(bal_n_int) <+ -a2*s32/sqrt(rout);

  end
endmodule // balun

Thanks for reply
This is my test-bench netlist:

Balun1 (single bal_p bal_n) balun rin=50 rout=50 loss=0
R0 (single 0) resistor r=50
R1 (bal_n 0) resistor r=50
PORT0 (bal_p 0) port r=50 num=1 type=sine dbm=-30 sinephase=0 freq=1G \\
       mag=1 phase=0 xfmag=1 pacmag=1 fundname=\"fre\"
simulatorOptions options reltol=1e-3 vabstol=1e-6 iabstol=1e-12 temp=27 \\
   tnom=27 scalem=1.0 scale=1.0 gmin=1e-12 rforce=1 maxnotes=5 maxwarns=5 \\
   digits=5 cols=80 pivrel=1e-3 ckptclock=1800 \\
   sensfile=\"../psf/sens.output\"
tran tran stop=2n errpreset=moderate write=\"spectre.ic\" \\
   writefinal=\"spectre.fc\" annotate=status maxiters=5

Sorry that I don\'t konw how to insert a picture.

Thanks. I think I got it.
But there is another question. You put several resistors in the model, so the input (and output) impedance is 50 ohms. But I am quite doubt that does the actual balun have this performance? That means, for a certain frequency (or a wide-band use), the balun can show a impedance of 50 ohms at all the three ports?
If so, what does the "50 ohms" comes? Doesn't a balun made of  inductor/transfermor?
While if not, why do we simulate our schematic use the unrealistic balun model?