Problems of HFSS optimization - Bandwidth evaluation and step of geometry variables
时间：03-30
整理：3721RD
点击：
There are two questions here:
1) We know that it is easy to determine the 10-dB bandwidth by using the XWidthAtYVal(-10) function in the simulated return-loss plot. Thus when I try to achieve a 10-dB bandwidth larger than 50 MHz, I set an optimization goal like XWidthAtYVal(dB(S(1,1)),-10)>=50e6. Everything goes OK until the second iteration where errors pop up and the optimization process terminates. The error messages are
Failed to evaluate calculation 'XWidthAtYVal(dB(S(1,1)),-10)'.
Engine: External solver for experiments failed!
Optimization Analysis failed - Solver returned error code. More information may be available in XXX.
Does anybody meet the same problem? Why the optimization process terminates after just one iteration?
2) For geometrical variables under optimization, they may vary with a very small amount between adjacent iterations. For example, in one iteration, the dipole length may be 80 mm, while it becomes 80.004 in the next iteration. The change from 80 mm to 80.004 mm is just 0.004 mm, which is easily overwhelmed by the manufacture tolerance. Therefore a bigger step like 0.5 mm is desired. How to specify the sweep step of geometrical variables when setting the optimization goal?
Thx in advance.
1) We know that it is easy to determine the 10-dB bandwidth by using the XWidthAtYVal(-10) function in the simulated return-loss plot. Thus when I try to achieve a 10-dB bandwidth larger than 50 MHz, I set an optimization goal like XWidthAtYVal(dB(S(1,1)),-10)>=50e6. Everything goes OK until the second iteration where errors pop up and the optimization process terminates. The error messages are
Failed to evaluate calculation 'XWidthAtYVal(dB(S(1,1)),-10)'.
Engine: External solver for experiments failed!
Optimization Analysis failed - Solver returned error code. More information may be available in XXX.
Does anybody meet the same problem? Why the optimization process terminates after just one iteration?
2) For geometrical variables under optimization, they may vary with a very small amount between adjacent iterations. For example, in one iteration, the dipole length may be 80 mm, while it becomes 80.004 in the next iteration. The change from 80 mm to 80.004 mm is just 0.004 mm, which is easily overwhelmed by the manufacture tolerance. Therefore a bigger step like 0.5 mm is desired. How to specify the sweep step of geometrical variables when setting the optimization goal?
Thx in advance.