1
\$\begingroup\$

I'm working on a AC PWM controller with 2 mosfets. It took me a long time to get this to work in LTspice but i still can't get it to work. I send pulses to the mosfets to generate the PWM modulation on the AC side. Everything works until i want to implement an optical galvanization. For this i would need to define 2 grounds in LTspice and this is the point where things start to get weird. I've already looked on the internet for more information about this topic. In LTspice you can only define 1 ground node, this is the reference voltage from where LTspice starts to calculate. If i want to use a second ground i have to use the COM node. I tried to do this but it still doesn't work. I know my mosfet part works and i know my opto coupler part works. When i try to connect both LTspice starts to behave strange. I've connected the COM node and GND node with a 99M resistor( this was a solution i found online) but this doesn't work at all. How can i get this simulation to run? Attached are 2 print screens. The first is a working mosfet switch, the second is with the opto coupler. I hope somebody here can help me to get this to work.

PWMok PWMNok Thank you all!

\$\endgroup\$
5
  • 1
    \$\begingroup\$ you need to simulate everything with one GND... there is no galvanic isolation in spice. \$\endgroup\$
    – schnedan
    Sep 30, 2020 at 14:33
  • \$\begingroup\$ What do you hope to achieve with the 2nd schematic? Do you really intend to drive two power MOS with an optocoupler in series wth a 100k resistor (connected to mains, by the look of it, even if rectified)? \$\endgroup\$ Sep 30, 2020 at 14:42
  • \$\begingroup\$ @aconcernedcitizen yes, the 100K and 10K resistor act as a voltage divider so my opto doesn't blow up. It's not going to be a product that i'm going to sell but i don't see a reason why this won't work? I might need to choose other voltage divider resistors that can deliver more current. \$\endgroup\$ Sep 30, 2020 at 14:47
  • \$\begingroup\$ @MichielReyntjens Not only the optocoupler is strongly dependent on the current collector, but the time constants -- even without the transistor -- will be in the tens of microseconds, if not more. Think of what that will do to the power dissipation. Try reading this and other similar documents about this, because your words "i don't see a reason why this won't work" are on the worrying side. \$\endgroup\$ Sep 30, 2020 at 15:18
  • \$\begingroup\$ @aconcernedcitizen i didn't thought about this. Rise time according to spice is 50ns so that shouldn't be a problem. Fall time is 250micro seconds so i should find a solution for that. \$\endgroup\$ Sep 30, 2020 at 15:34

1 Answer 1

1
\$\begingroup\$

You need to use only one GND in LTSpice.

This makes it sound like a limitation, but it isn't. In the real world the grounds will be connected, just via a very high impedance link.

I have often done LTSpice simulations for isolated DC-DC converters, where you have the same issue: two isolated grounds.

The fix for this is to remember that LTSpice simulates the real world with models. In the real world the two side are linked via a very high resistance, and a very small capacitance. So, for the lack of any concrete numbers to use, I use a 1Meg resistor in parallel with a 1nF capacitor to link the two grounds (number are open to some debate, as I've been given some, I've updated these numbers used here). Then I call one ground "iso_gnd" or something so that I can use that as the ground reference for my probes.

So I have a V_IN and GND_IN, then a V_OUT and GND_OUT. I connect the proper ground to GND_IN, connect GND_OUT to GND_IN via a 1Meg resistor in parallel with a 1nF capacitor. And simulate. To get the output voltage, I plot the voltage of V_OUT minus GND_OUT.

\$\endgroup\$
2
  • 1
    \$\begingroup\$ Just a note, this is the behaviour of SPICE since 40+ years ago: everything needs to be referenced to the global node zero (0). Also, no need for such extreme values (1G || 1p), when even in real SMPS you find connections of 1Meg || 1n (or similar) between primary and secondary. \$\endgroup\$ Sep 30, 2020 at 14:40
  • 1
    \$\begingroup\$ @aconcernedcitizen I used those figures as a finger in the air, as you've given something with a bit of backing, I'll update it. \$\endgroup\$
    – Puffafish
    Sep 30, 2020 at 14:47

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.