Maker Pro
Maker Pro

LTSPICE and 3 phase w/ transient analysis - SLOW!!!!!!!!!!!!!!

jackorocko

Apr 4, 2010
1,284
Joined
Apr 4, 2010
Messages
1,284
I been messing around with LTSPICE lately in an attempt to play around with some 3 phase wiring ideas I had been trying to understand. All was working out well till I tried to simulate a step down isolation transformer according to this pdf file here page 44 http://www.ieca-inc.com/images/Spice-Simulation_Using_LTspice_Part_1.pdf

In the above link he uses a 50Hz 3 phase sine wave. I noticed that he used a time delay to setup properly the 3 phase. I was using prior to this the PHI[deg] to set 120 and 240 degrees per phases 2 and 3. But when trying to do a transient analysis over even a microsecond takes forever once you add in the isolation transformers. The key is using a time delay to setup the 3 phase, but what exactly is the time difference between 3 phase on 60Hz like in the states and why so long for spice to simulate with the phase shift set in PHI?

I came to some numbers, but I know I am wrong with my mathematical understanding.

I used for phases

1 - 0 ms/delay
2 - 5.66666ms
3 - 11.33333ms

Now, how I came up with those answers, 60Hz cycle means the wave changes 120 cycles per second. so 1000ms/120 = ~8.3ms * 66% = ~5.6ms * 2 = ~11.3ms. Which I know is utter crap, but how are those values really derived?

A final summary of the above - Best way to do transient analysis with mutual inductance using 3 phase that won't slow the simulation down to a crawl?
 

Harald Kapp

Moderator
Moderator
Nov 17, 2011
13,700
Joined
Nov 17, 2011
Messages
13,700
but what exactly is the time difference between 3 phase on 60Hz
60 Hz is equivalent to a period of 16.67 ms (rounded). One period is eqivalent to a rotation of 360 ° in phase.
Therefore t = phi * 16.67 ms / 360 °
Your numbers are off a bit but in the right direction.
120 ° -> 5.557 ms
240 ° -> 11,113 ms

Here are some tips for working with LTSPICE. "Ground in multiphase systems" may (or may not) have an answer to the issue of slow simulation.

I can only guess that is it miuch easier for LTSPICE to delay a signal (sine) using a time delay function as LTSPICE operates in discrete time steps and a delay in time simply means to shift operations to a later step in time. Hereas using a phase delay LTSPICE will have to calculate the shift in time from the phase - and probably does so for every timestep - which may teke longer.
 

jackorocko

Apr 4, 2010
1,284
Joined
Apr 4, 2010
Messages
1,284
I was right in my thinking. I actually came up with 5.5ms but knew that wasn't exactly right because of rounding errors so I took and educated guess and added in the decimal places from the 50Hz numbers posted in that pdf. It was close ;)

Anyway, thanks for the clarification/formula and the guess/validation that spice is working as I suspected.
 
Top