Hi

I was trying simulate a Direct Conversion Mixer with Eldo-RF and ADMS but the sources do not work good and the results off simulation is not agree.
The result don't show the constellation both the input and the output.
I will describe the SPICE and Verliog AMS archives:
Verilog Ams
Quote:


Spice:

Quote:


Thanks for atention

Ps. I use the tutorial http://www.designers-guide.org/Modeling/modeling-rf-systems.pdf for develop this situation

Ps. Sorry for the English

First thanks for the answer

pancho_hideboo wrote on Jul 21^st, 2008, 5:12pm:

Sorry I wrote wrong in the last post. FUND2=1GHz.
pancho_hideboo wrote on Jul 21^st, 2008, 5:12pm:

If I use FUND1=FUND2=1GHz I obtain this constellation(Plot source with v(11).h(1,0)):
The plot off in and out has attach

IT's OK

But IF I will change the verilog source, add a second order intermodulation. I don't see the modification of constellation.
I'm thinking that the results of mixer h(1,-1) is different of the dc output h(0,0) but I would like to see in same time. What I think is correct?
pancho_hideboo wrote on Jul 21^st, 2008, 5:12pm:

Ok, I have used a EZWAVE for plotting the constellation, I have used a TCL archive for tune the sampling:
Quote:

pancho_hideboo wrote on Jul 21^st, 2008, 5:12pm:

it's same of ADS
constellationdiagram(complex(i_data,q_data, T_symbol,delay)


pancho_hideboo wrote on Jul 21^st, 2008, 5:12pm:

Yes, It's same to ADS envelope analysis

trumpeu wrote on Jul 21^st, 2008, 11:11am:

In the above, there is no specification of harmonic to observe.


And my problem is that I don't see the dc off-set of second order intermodulation

     
thank you for  attention

This is result of ganho_dB=72, iip3_dBm=-10 and iip2_dBm=10.

Different from IM3, there are many types in IM2 definition.
See the following.
http://www.designers-guide.org/Forum/YaBB.pl?num=1193566207

For example, self mixing of RF input, Half-IF, proximity of two tones in RF, etc.
Your Verilog-A is not proper for any type of IM2 which causes inband interference.

The following is modified Verilog-A code for RF self mixing or RF two tones in close proximity.
But strictly speaking, this implementation is not so proper, you have to scale and limit IM2 implementation.
Also you can't express Half-IF with this implementation.
Anyway you have to build your own mathematical formulation to model IM2 issue you concern.

"trumpeu_iq_down_mixer.va"
// trumpeu Demodulator IQ
`include "discipline.h"
`include "constants.h"

module trumpeu_iq_down_mixer(rf_in, lo_i_in, lo_q_in, if_i_out, if_q_out);
inout rf_in, lo_i_in, lo_q_in;
output if_i_out, if_q_out;
electrical rf_in, lo_i_in, lo_q_in;
electrical if_i_out, if_q_out;

parameter real rin = 50;       //input impedance
parameter real rout= 50;       //output impedance
parameter real ganho_dB = 10;  //Gain in dB
parameter real iip3_dBm = -10; //third order intercept point
parameter real iip2_dBm = 2;   //second order intercept point
   
real ga, gb, gc, Aiip3, Aiip2, vin, vin2, vout;

analog begin
  @(initial_step) begin //Parameters conversion
     ga = pow(10, ganho_dB / 20.0);  //Gain conversion
     Aiip3 = sqrt(pow(10, iip3_dBm / 10.0) * 2 * rout * 0.001);      
     gc = (4.0 / 3.0) * ga/(Aiip3*Aiip3);                  
     Aiip2 = sqrt(pow(10, iip2_dBm / 10.0) * 2 * rout * 0.001);
     gb = ga / Aiip2;
  end // initial_step

  I(rf_in) <+ V(rf_in) / rin;
  I(lo_i_in) <+ V(lo_i_in) / rin;
  I(lo_q_in) <+ V(lo_q_in) / rin;
         
  vin = V(rf_in);
  vin2 = vin * vin;
  vout = (ga - gc*vin2) * vin;
         
  V(if_i_out) <+ vout * V(lo_i_in) + gb*vin2;
  V(if_q_out) <+ vout * V(lo_q_in) + gb*vin2;
end // analog

endmodule