Spice netlist doubled device names

[buleks]

When i generate spice netlist using eeschema i get for each device something like this:
rr4 vg2 net-_r4-pad2_ 10
rr5 net-_d3-pad1_ vg2 1k

Instead print r5 it repeat first letter.

It not occur in stable release.
I compile code form branch master.
Source revision:commit 4bdac4de705849777f5a59e0f6f7ee1dd6274a8c

[Maciej Suminski (orsonmmz)]

Just to be sure, does it cause any simulation errors?

[buleks (buleks)]

Simulation works without errors because for spice this is not any problem.

[jean-pierre charras (jp-charras)]

Hi Orson,
AFAIK, we have already discussed this feature, added because Spice simulator has some constraints (the first letter of a reference is constrained).
The current behavior is very acceptable, and I am not sure it should be modified.

[buleks (buleks)]

I have example why it should be modified.
When i want to build transformer, on schematic i must put two inductors L1,L2.
Then i must specify coupling coefficient like this "K1 L1 L2 0.98" .
With this issue i must put this directive: K1 LL1 LL2 0.98

[jean-pierre charras (jp-charras)]

I am not convinced: This is not a proof.

Your 2 inductor could be named I1 and I2 (I for inductor) or T1 and T2 (T for transformer).
Each country has its habits.

Be sure in some (many?) countries, resistors are not labeled Rn and capacitors not labeled Cn.

[buleks (buleks)]

This is clear for me now.
I agree that it shoudnt be modified.

[JKB (joel-bertrand)]

There are some side effects. For example, instead of writing
    .save @r7[i]
you have now to write
    .save @rr7[i]
to save R7 current or
    .save i(vv9)
if V9 is a 0V source to measure current.

[PCB Wiz (1-pcb-wiz)]

I can see what KiCad spice is trying to do, but the current 'solution' is rather blunt, and it breaks spice netlist portability, as well as making ID less natural.

There is a lot of WEB info out there, around spice netlists.
http://www.spiceopus.si/getstarted.html
http://denethor.wlu.ca/ltspice/

My suggested fix would be to ADD A PREFIX ONLY AS NEEDED.
That avoids most RR,CC,VV,QQ,DD cases, but can still patch someones incorrect prefix.

eg I tried a simple sim

.title KiCad schematic
.include "ad8009.lib"
.include "fzt1049a.lib"
.include "laser.lib"
VV1 /in GND pulse(0 3 100n 1n 1n 20n 100n )
XU1 Net-_R2-Pad1_ Net-_N1-Pad1_ VDD VSS Net-_Q1-Pad2_ ad8009
VV2 VDD GND DC 10
VV3 GND VSS DC 10
CC2 /out Net-_C2-Pad2_ 1p
RR5 /out Net-_C2-Pad2_ 2.5
RR1 Net-_N1-Pad1_ GND 220
RR3 Net-_C2-Pad2_ Net-_N1-Pad1_ 220
CN1 Net-_N1-Pad1_ Net-_C2-Pad2_ 1p
DD1 /out GND laser
RR4 /out Net-_R2-Pad1_ 220
RR2 Net-_R2-Pad1_ /in 160
QQ1 VDD Net-_Q1-Pad2_ Net-_C2-Pad2_ fzt1049a
.tran 10p 150n
.end

Here we see XU1 patch is needed, as U1 is a subcircuit.
I manually changed C1 to N1, whence it promptly lost the spice model info (so it _does_ already expect you to follow conventions)
I can manually remap to Passive:Capacitor:1p, and now C prefix correctly overrides the poor choice of N

ONLY U1 and N1 here, actually need a prefix fixup applied all others are ok.

The suggested smarter (no simple repeats) fixed NETLIST would look like this :

.title KiCad schematic
.include "ad8009.lib"
.include "fzt1049a.lib"
.include "laser.lib"
V1 /in GND pulse(0 3 100n 1n 1n 20n 100n )
XU1 Net-_R2-Pad1_ Net-_N1-Pad1_ VDD VSS Net-_Q1-Pad2_ ad8009
V2 VDD GND DC 10
V3 GND VSS DC 10
C2 /out Net-_C2-Pad2_ 1p
R5 /out Net-_C2-Pad2_ 2.5
R1 Net-_N1-Pad1_ GND 220
R3 Net-_C2-Pad2_ Net-_N1-Pad1_ 220
CN1 Net-_N1-Pad1_ Net-_C2-Pad2_ 1p
D1 /out GND laser
R4 /out Net-_R2-Pad1_ 220
R2 Net-_R2-Pad1_ /in 160
Q1 VDD Net-_Q1-Pad2_ Net-_C2-Pad2_ fzt1049a
.tran 10p 150n
.end

[Maciej Suminski (orsonmmz)]

Let's think about it when v6 development cycle begins.

[Maciej Suminski (orsonmmz)]

I have pushed a fix for the bug, as it turned out to be a simple and low risk change in exchange for much greater benefit.

[KiCad Janitor (kicad-janitor)]

Fixed in revision bc67d1ad8cde85fd0d88b95b7b5c259614d08bb4
https://git.launchpad.net/kicad/patch/?id=bc67d1ad8cde85fd0d88b95b7b5c259614d08bb4