Parametric simulation with orcad capture- transient analysis unreliability

[israel_Y]

Please help,

I am simulating resistance to frequency converter using Orcad Capture 16.5. Exact same schematic from the IEEE paper
"A High-Resolution, Linear Resistance-to-Frequency Converter" Kouji Mochizuki and Kenzo Watanabe 1996. the schematic is attached here.

The part where i face problem is when i vary the time duration in the transient analysis, the output varies unpredictably. (the Rval values that create oscillation vary): For example.

1. for 800ms long simulation, Rval = 1 and Rval=66 ohm ONLY create oscillation while the rest are flat
2. for 2sec long simulation Rval = 1 and Rval=51 ohm create oscillation while the rest are flat

I simulated many times to see if i am making some mistake but unfortunately i couldnt detect it.

Please suggest why this problem occurs, is there initialization setting for analog simulation (i know there is digital initialization for flip flops).

I appreciate any idea/hunch/guess in advance. thank you

 

[keith1200rs]

Skip the initial bias point calculation (UIC option on transient analysis command line). Also, set the maximum time step to something sensible (a small fraction of the expected oscillation period.

Keith

 

[israel_Y]

Thank you so much, you were spot-on.. now it is working perfectly, and giving sensible output.
I have struggled with it for the past couple of days.

 

[LvW]

Do you really intend to use resistors as low as 1ohm only - in a divider with kohm resistors? The voltage difference in the bridge is very low and may cause problems for real amplifiers.
I don`t think that the UIC option will help - however, in any case follow the recommendation from Keith (sufficient time resolution, small time steps).

 

[israel_Y]

Do you really intend to use resistors as low as 1ohm only - in a divider with kohm resistors? The voltage difference in the bridge is very low and may cause problems for real amplifiers.
I don`t think that the UIC option will help - however, in any case follow the recommendation from Keith (sufficient time resolution, small time steps).

Dear LvW,

the sensor has very high R resolution (deltaR) even less than 1 ohm thats why . you might be right for actual implementation, the opamp might give me some offset and mismatch problem... but for simulation sake only, the UIC option helped perfectly.. if u have any suggestion to read high resolution please help. (apart from decreasing the bridge R values, because i want to keep the bridge current low).
Thank you