Learning Forum

[Marin Milovac]

Hello,

I have been trying to simulate TDR of the "Ideal" 50Ohm line (50-ohm source, line, and termination).

Schematic:++



PCB:


The thickness for 50R has been calculated via Altium's integrated 2D solver to be 0.325mm. The total delay of the trace is around 662ps:

Once I simulate TDR with SIwave's integrated TDRwizard I get this as a response:



I understand the first dip to ~9.5 ohm is due to the simulation starting and the current not instantaneously becoming the I=U/Z but rather "ramping up" to that value. But I don't understand why there is a ringing afterward. Shouldn't the response be a flat line if everything is matched correctly and not swinging +-3 ohms? Do you maybe know why this is the case? Is it typical for simulation or do you think that I am doing something wrong?

For any further clarification please don't hesitate to ask.
Marin 

 

[Dan Dvorscak]

What did you use for your frequency sweep in the TDR Wizard setup in SIwave? I think the default sweep setup only goes up to 5GHz. Just looking at the waveform it looks like that rining is ~6GHz or so. Given that the TDR source in the Circuit tool uses a 35ps rising edge by default, we have to be careful with out of band energy unless you have upped the bandwith in the TDR setup. Alternately, you can go into the Circuit setup and slow down the edge of the TDR source to see if that changes anything. 

[Marin Milovac]

Hello,

Thank you for your answer.

Bellow I am providing screenshots of TDR and voltage levels of various rise time simulations that I have simulated (1ps, 10ps, 100ps, and 1ns).

1ps:
Voltage levels

TDR

 10ps:
Voltage levels

TDR

 100ps:
Voltage levels

TDR

 1ns:
Voltage levels

TDR


From what I can see the main problem for TDR is the output voltage (red) because the ringing of the output voltage gets translated into noise in the TDR. The ringing is there for 1ps, 10ps, and 100ps. For 1ns there is no ringing but the edge is too slow to get any information about impedance discontinuities. 

Also, two very interesting things that I have noticed:

First is that the resistor input voltage also starts ringing BEFORE the signal propagates to the resistor (especially at fast rise times) seemingly breaking causality:



Second is that for fast rise times voltage rises proportionately to the set rise time for about half of its amplitude, but for the second part of its amplitude it does not rise according to the set rise time:



Any help would be much appreciated.

Marin

[Dan Dvorscak]

Hello Marin,

First off, have you tried upping the maxiumum frequency of the TDR simulation in SIwave? If you have left it with the defaul maximum of 5GHz, I'm concerned that this ringing is above what SIwave is providing Circuit in the s-parameters.

Yes, I also have some concern about causality here. Are you using the default state-space method in Circuit, or convolution. If you run the casuality check on these s-parameters in the Network Data Explorer, what are those results?