qucs#992 corrections in NMOSFETs PMOSFETs Transistors Z-Diodes libs

[oxmon-2500]

Value out of range - deactivated by commenting it out
NMOSFETs IRF3709ZCS_L
NMOSFETs IRF3709Z_S_L
NMOSFETs IRF3805S-7P
NMOSFETs IRF7821
PMOSFETs IRF7220
PMOSFETs IRF7233
PMOSFETs IRF7410
Transistors MMST4401

Typos - corrected
NMOSFETs IRF7465 missing <Model> tag
NMOSFETs IRFB4229PBF Component Name and .Def Name equalized
NMOSFETs IRFR024N Component Name and .Def Name equalized
NMOSFETs IRFR110 Component Name and .Def Name equalized
NMOSFETs IRFR1205 Component Name and .Def Name equalized
NMOSFETs IRFR120N Component Name and .Def Name equalized
NMOSFETs IRFR220 Component Name and .Def Name equalized
NMOSFETs IRFR320 Component Name and .Def Name equalized
NMOSFETs IRFR3910 Component Name and .Def Name equalized
NMOSFETs IRFR4105 Component Name and .Def Name equalized
Z-Diodes 1N5342B 'Vj' occurred 2x
Z-Diodes 1N5344B 'Vj' occurred 2x
Z-Diodes 1N5346B 'Vj' occurred 2x
Z-Diodes 1N5377B .Def:Z-Diodes_1N5377B -> .Def:Z_Diodes_1N5377B

[felix-salfelder]

On Mon, Mar 23, 2020 at 02:17:20AM -0700, Szymon Blachuta wrote: Value out of range - deactivated by commenting it out [..]

I can't see the issue with these subcircuit models. If a value is out of range, there are three possibilities - the value is not supported by the device and there is a warning. - the range check is wrong, and it is actually within range. - the value is a typo, should be a different one, possibly within range. what will commenting out fix?

Typos - corrected [..]

changes the name of a several components. on what basis? what happens if somebody uses these components?

[oxmon-2500]

what will commenting out fix?

Using this components results in aborting the simulation:

checking netlist...
Errors occurred during simulation on Mon 23. Mar 2020 at 15:49:49:190
Aborted.
Errors and Warnings:
checker error, value of 'Vj' (11.8882) is out of range ']0,10]' in 'Diode:D2'

If the component is unusable then it is better not to offer it to build it into a circuit.
Maybe the range check is wrong, thus it remains unusable. The range should be other issue.
It seems to me it is a typo, because there are many more Vj values within the range.
The components are still in the library as comments. Hopefully someone will take a look at it, correct the value and uncomment it

Typos - corrected [..]
changes the name of a several components. on what basis? what happens if somebody uses these components?

In all component definitions:
<Component BSP125>
...
<Model>
.Def:NMOSFETs_BSP125 _net1 _net3 _net2

the component name and the name in .Def (after underscore) are the same. Only in those erroneous the second name slightly differs and looks like typo. I wanted to ask the author Gunther Kraut, but his eMail <gn.kraut@online.de> is no more up to date.

[felix-salfelder]

On Mon, Mar 23, 2020 at 09:23:18AM -0700, Szymon Blachuta wrote: >what will commenting out fix? Using this components results in aborting the simulation: Errors and Warnings: checker error, value of 'Vj' (11.8882) is out of range ']0,10]' in 'Diode:D2'

Is this not correct behaviour?

If the component is unusable then it is better not to offer it to build it into a circuit.

It is unusable with qucsator. Does not imply that the subcircuit model is wrong, and there are other simulators out there. If the Warning/Error is not good enough (I think it is), maybe a comment should explain the issue, until the devices are implemented or changed...? (the latter is a qucsator issue, not qucslib).

In all component definitions: <Component **BSP125**> ... \<Model\> .Def:NMOSFETs_**BSP125** _net1 _net3 _net2 the component name and the name in .Def (after underscore) are the same. Only in those erroneous the second name slightly differs and looks like typo. I wanted to ask the author Gunther Kraut, but his eMail ***@***.***\> is no more up to date.

OK i see. Was not obvious from the patch. (There was something strange going on in how these subcircuit definitions are used, but I'm afraid I do not remember the details). Maybe the name after .Def is the one used in the netlist, ideally unique, but otherwise arbitrary. (Can you confirm this?). With this, I think your change is an improvement. No need to ask the author in that case. thanks felix

[oxmon-2500]

there are other simulators out there

I see, I will try this components with qucs-s

[felix-salfelder]

On Mon, Mar 23, 2020 at 10:50:19AM -0700, Szymon Blachuta wrote: >there are other simulators out there I see, I will try this components with qucs-s

Qucs runs any (suitable) simulator you choose, (set the QUCSATOR environment variable.) With gnucsator, you may choose from about half a dozen Diode implementations, all with potentially different ranges (have not checked). Add your own if needed. More simulators or diodes may come or not. The point is, that you can't possibly know if a modelname (+ parameters) is understood and supported before you run the simulator. The current error handler implementation reflects this. ("asking" the simulator without actually running it is not implemented. that would be overkill, ymmv). (qucs-s is a different project, don't know what they do, but they might also import your patches in the end.)