0

I tried to simulate a Laplace function of a PID controller with LTspice in the time domain, but an error message appears "The laplacian is singular at DC". enter image description here

cs1 is the controller, so I just have a unit jump (Ugen) + controller (cs1)

with KP=0.9995594504821; KI=0.0000772197554; KD=0.00540381443834925; Tp=KD/100

ast30277
  • 11
  • 3
  • Without seeing the subcircuit one can only guess, but I'd guess that the integral part uses a bare 1/s, which indeed can cause problems. If so, the solution is to add a pole way up, somethig like 1/(s+1p). However, this means that Laplace expressions are used, and they are known to be problematic in .TRAN. – a concerned citizen Jul 02 '20 at 10:29
  • thanks, but if I use the formula with 1 / (s + 1p), my results are no longer correct. Can you also build a PID controller like this? with 1 / (s + 1p) in the I-part instead of (1 / s) – ast30277 Jul 02 '20 at 10:49
  • p should be a very small value. Lower than the lowest frequency (in rad/s) you are interested in simulating correctly. – AJN Jul 02 '20 at 11:51
  • @ast30277 There must be a misunderstanding, 1p refers to 1 pico, as in 1e-12, which means you'll have a pole in the hundreds of GHz region. I can't see how that could affect your results in a meaningful way. And, yes, for Laplace expressions this trick is common. – a concerned citizen Jul 02 '20 at 12:42
  • thank you very much, it was very helpful. My results were not affected – ast30277 Jul 02 '20 at 12:50

0 Answers0