Error message: LTspice says that there is 'more than one sub-circuit definition with this name, which is ill-formed.'

Hi Tulipa ,

The absence of an error is not proof that something worked.

Remember, a netlist does not have hierarchy or scope other than line by line; a SPICE netlist is completely flat and non-ordered. LTspice, on the other hand, gives you scope and hierarchy in its schematic editor, a very useful and popular feature. LTspice's syntax checker and netlister essentially must evaluate your schematic and hierarchy and flatten all into a simulateable (flat) netlist. If LTspice must guess your intentions without clear instruction, the determinism of LTspice is lost. There's no error, but at what cost? Older versions of LTspice actually did this, giving you the impression that all is fine, when in fact it was flipping a coin. Now, you get an error.

That said:

Your request is essentially asking LTspice to netlist thus:

BUCK_REGULATOR.LOCAL_FUNCTION

and 

LDO_REGULATOR.LOCAL_FUNCTION

What if you also have 

LOCAL_FUNCTION

defined on your schematic? Is that:

GLOBAL.LOCAL_FUNCTION

Since scope must be determined, each call to those subcircuits is now:

X1 1 2 BUCK_REGULATOR.LOCAL_FUNCTION

X2 3 4 LDO_REGULATOR.LOCAL_FUNCTION

X3 5 6 GLOBAL...

The netlister would have to analyze a schematic for all instances of the sub and determine if scope is necessary, then add text to make each instance unique, and then produce a netlist with subcircuit names (which do not match the original schematic). Once done, the syntax checker must go through again and see that it all worked out. It would do what you are doing, changing the names, but less cleanly.

For models, you can use:

.options allow_ambiguous_models

Which enables coin-flipping behavior in LTspice re models.

mike

Hi Tulipa ,

The absence of an error is not proof that something worked.

Remember, a netlist does not have hierarchy or scope other than line by line; a SPICE netlist is completely flat and non-ordered. LTspice, on the other hand, gives you scope and hierarchy in its schematic editor, a very useful and popular feature. LTspice's syntax checker and netlister essentially must evaluate your schematic and hierarchy and flatten all into a simulateable (flat) netlist. If LTspice must guess your intentions without clear instruction, the determinism of LTspice is lost. There's no error, but at what cost? Older versions of LTspice actually did this, giving you the impression that all is fine, when in fact it was flipping a coin. Now, you get an error.

That said:

Your request is essentially asking LTspice to netlist thus:

BUCK_REGULATOR.LOCAL_FUNCTION

and 

LDO_REGULATOR.LOCAL_FUNCTION

What if you also have 

LOCAL_FUNCTION

defined on your schematic? Is that:

GLOBAL.LOCAL_FUNCTION

Since scope must be determined, each call to those subcircuits is now:

X1 1 2 BUCK_REGULATOR.LOCAL_FUNCTION

X2 3 4 LDO_REGULATOR.LOCAL_FUNCTION

X3 5 6 GLOBAL...

The netlister would have to analyze a schematic for all instances of the sub and determine if scope is necessary, then add text to make each instance unique, and then produce a netlist with subcircuit names (which do not match the original schematic). Once done, the syntax checker must go through again and see that it all worked out. It would do what you are doing, changing the names, but less cleanly.

For models, you can use:

.options allow_ambiguous_models

Which enables coin-flipping behavior in LTspice re models.

mike