0
\$\begingroup\$

I'm trying to simulate a circuit for a design I'm making. In a few parts of what I'm interfacing with uses relays that when closed will short a few power supplies to the chassis ground and in LTSpice and Circuit lab there is no "chassis ground" or "earth ground" just one ground. It doesn't short the rest of the circuits with power supplies tied to their own grounds just a few spots these relays are used. I've been trying to find the answer on here and elsewhere and found Need help modelling the circuit with parasitic capacitance of a power supply in this scenario in SPICE where he just hand drew it but is there a way to make it with subskt or is there a proper way to simulate this that won't skew my sim in weird ways?

\$\endgroup\$
  • \$\begingroup\$ In your third sentence, what is "It"...LTspice? CircuitLab? one ground? How is the chassis ground of your supplies related to "earth" ground? Are they truly floating? \$\endgroup\$ – Elliot Alderson Oct 19 '18 at 20:03
  • \$\begingroup\$ @ElliotAlderson LTSpice is just a spice freeware. Circuitlab.com is what pulls up when you click the little schematic button when you're asking/answering questions here so a lot of people use it. The chassis ground and earth ground aren't related, I was just emphasizing there is only one global ground and can't separate. \$\endgroup\$ – BrownKuma Oct 19 '18 at 20:26
  • \$\begingroup\$ I know very well what LTspice and Circuitlab are. What is confusing is what "It" refers to at the beginning of your third sentence. \$\endgroup\$ – Elliot Alderson Oct 19 '18 at 21:19
  • \$\begingroup\$ If chassis ground and earth ground aren't "related" does that mean they are galvanically isolated? Yes, SPICE requires that you have a ground node but it's not clear yet why that is a problem for you. \$\endgroup\$ – Elliot Alderson Oct 19 '18 at 21:21
  • \$\begingroup\$ @ElliotAlderson Oh I'm sorry I saw that as "LT" not "it" with the font used on this site. I'm building it in LTSpice but when I came across not being able to add chassis ground I looked at circuitlab just to compare. This is for a senior design project for school aimed at making a device for an aircraft so there is no true "earth ground." the closest we can plan for is "chassis ground" \$\endgroup\$ – BrownKuma Oct 19 '18 at 21:41
0
\$\begingroup\$

Due to the nature of the SPICE (ever since 40+ years), the ground (0, GND) node is global, but just a net, like all nets. And all nets can be renamed to whatever name/label/etc is needed. In this case, that chassis you're looking for, in whichever SPICE software, is just a cosmetic appearance for a net/node, labeled chassis, or earth, or whatever. In LTspice there is only one cosmetic like that, called COM, but which has the same functionality of a common node (like chassis, earth, etc, unless mentioned otherwise). It, too, needs to be referenced to ground, so some connection is needed, if you so choose to use it. This is true for whichever SPICE program that uses those nodes as floating nodes. Here's a minor example of usage:

bla

Note the R2 that connects the triangular shape ground (the ground, 0, GND) to the cosmetic COM. Its value is 1g to simulate a separate, floating ground, but really it's there with the sole purpose of providing a reference to ground without which the solver cannot calculate. Also, the plot shows V(2,COM), which is just a local reference for probing; can be anything, really.

\$\endgroup\$
  • \$\begingroup\$ So, just to confirm, in this model the COM is chassis ground, correct? \$\endgroup\$ – BrownKuma Oct 26 '18 at 15:07
  • \$\begingroup\$ @BrownKuma If that's what you want it to be, sure. It really doesn't matter even if you're using an unnamed net, all that matters is that you keep the reference to that node (if that's your purpose). For visibility purposes, you can name that net as CHASSIS, and reference everything you need to be referenced to it. \$\endgroup\$ – a concerned citizen Oct 26 '18 at 16:58
  • \$\begingroup\$ I'm trying to conceptualize it correctly. I did it the other way and the simulation didn't seem right so after some research I realized there shouldn't be hardly any current going to the chassis so that should be shunted with the 1g resistor, am I correct? \$\endgroup\$ – BrownKuma Oct 28 '18 at 16:24
  • \$\begingroup\$ @BrownKuma You have to separate the real life case to the SPICE world. In SPICE, everything needs to be calculated with a reference. That ref. is the ground, or the 0 (GND) node. And while in real life you can have a separate, floating ground, in SPICE you can't, because that separate ground needs to be referenced to 0. The solution is to add a large enough resistor so that the currents are small to not influence the schematic, while allowing referencing to 0. Still, even in real life, you have the finite impedance of air, various parasitics, etc. \$\endgroup\$ – a concerned citizen Oct 28 '18 at 17:01

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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