LTspice: .SAVEBIAS doesn't

[Mike Rocket J. Squirrel E]

Strange. The .SAVEBIAS <filename> statement was working just fine. Now,
however, when I run a .TRAN, it's not writing the date -- there is no
such <filename>. Is there something that can cause it to ignore this card?

--
Mike "Rocket J Squirrel" Elliott

 

[Mike Engelhardt]

Mike,

Strange. The .SAVEBIAS <filename> statement was working just fine. Now,
however, when I run a .TRAN, it's not writing the date -- there is no
such <filename>. Is there something that can cause it to ignore this card?

The command's operation is documented in the help
in the section LTspice=>Dot Commands=>.savebias.
If you stipulate a time at which to save the bias and
the simulation doesn't run beyond that time, then the
file isn't written. There's any number of stipulations
you can make for the bias point which if never met
will mean the file is never written. There's also
an implicit one regarding .tran analysis: If the
..op point fails and that's the one you wanted to
save, the .tran analysis will run as if you'd specified
skipbp(uic), but the .savebias file is never written.

--Mike

 

[Mike Rocket J. Squirrel E]

Mike Engelhardt wrote:

Mike,


Strange. The .SAVEBIAS <filename> statement was working just fine. Now,
however, when I run a .TRAN, it's not writing the date -- there is no
such <filename>. Is there something that can cause it to ignore this card?


The command's operation is documented in the help
in the section LTspice=>Dot Commands=>.savebias.
If you stipulate a time at which to save the bias and
the simulation doesn't run beyond that time, then the
file isn't written.

I assumed that unless time is specified, .savebias writes the end-of-run
operating points to file.

There's any number of stipulations
you can make for the bias point which if never met
will mean the file is never written. There's also
an implicit one regarding .tran analysis: If the
.op point fails and that's the one you wanted to
save, the .tran analysis will run as if you'd specified
skipbp(uic), but the .savebias file is never written.

Does that also mean that if you tell .tran to uic, then .savebias does
not happen?

What about when the error file reports,

"Direct Newton iteration failed to find .op point. (Use ".option
noopiter" to skip.)", or,

"Direct Newton iteration for .op point skipped," or,

"Trouble finding operating point....skipping operating point for
Transient analysis.", or

Regardless of those warnings, the .tran does run, so there are op's that
can be saved.

I'll send my SPICE file to the Help e-mail address.

I'm trying to iterate the circuit so it starts up with all the long time
constant parts in their settled condition. I'd like to .savebias at the
end of the first run, then .loadbias those conditions at the start of
the second run, then .savebias those op's at the end of *that* run, then
start a third .tran with .loadbias the second run's op's, etc., until
the circuit is starting up with everything settled so I can then do some
short runs and look at other short time constant transients.

--
Mike "Rocket J Squirrel" Elliott

 

[Mike Engelhardt]

Mike,

The command's operation is documented in the help
in the section LTspice=>Dot Commands=>.savebias.
If you stipulate a time at which to save the bias and
the simulation doesn't run beyond that time, then the
file isn't written.

I assumed that unless time is specified, .savebias
writes the end-of-run operating points to file.

Of course not. The .savebias/.loadbias intent to speed
up the .op solution of large simulations, e.g., a
transistor-level full-chip simulation, though it has other
uses if you edit the file from a .nodeset to a .ic statement.
If you want a specific time step saved, then you specify that
time in the manner described in the above mentioned page
of the documentation.

There's any number of stipulations
you can make for the bias point which if never met
will mean the file is never written. There's also
an implicit one regarding .tran analysis: If the
.op point fails and that's the one you wanted to
save, the .tran analysis will run as if you'd specified
skipbp(uic), but the .savebias file is never written.

Does that also mean that if you tell .tran to uic, then
.savebias does not happen?

Depends how you write the .savebias command. If stipulate
a time, then that will be obeyed, but if you want the .op
solution and you don't do it, then the solution it didn't
do won't be written.

What about when the error file reports,

"Direct Newton iteration failed to find .op point. (Use
".option noopiter" to skip.)", or,

"Direct Newton iteration for .op point skipped," or,

"Trouble finding operating point....skipping operating
point for Transient analysis.", or

Regardless of those warnings, the .tran does run, so
there are op's that can be saved.

Ah! Here we have why the .savebias didn't write anything.
It didn't find the .op solution(notice the part "skipping
operating point for Transient analysis") so it didn't
write it. From my first post:

...There's also an implicit one regarding .tran
analysis: If the .op point fails and that's the
one you wanted to save, the .tran analysis will
run as if you'd specified skipbp(uic), but the
.savebias file is never written...

I'll send my SPICE file to the Help e-mail address.

Thanks. Saw that first, of course, and have already
responded before seeing your post here with more
detailed questions than were sent with the file. BTW,
it's better to send such files as attachments then
included text. Line wrap makes it difficult to
figure out what you're doing.

Best Regards,

--Mike

 

Is failure to write a .savebias file noted in the error log?

On 10 Nov 2003 05:25:25 GMT, "Mike Engelhardt" <pmte@concentric.net>
wrote:

Strange. The .SAVEBIAS <filename> statement was working just fine. Now,
however, when I run a .TRAN, it's not writing the date -- there is no
such <filename>. Is there something that can cause it to ignore this card?

The command's operation is documented in the help
in the section LTspice=>Dot Commands=>.savebias.
If you stipulate a time at which to save the bias and
the simulation doesn't run beyond that time, then the
file isn't written. There's any number of stipulations
you can make for the bias point which if never met
will mean the file is never written. There's also
an implicit one regarding .tran analysis: If the
.op point fails and that's the one you wanted to
save, the .tran analysis will run as if you'd specified
skipbp(uic), but the .savebias file is never written.

 

[Mike Engelhardt]

Pat,

Is failure to write a .savebias file noted in the error log?

In this case, there was the message that the biaspoint calc
failed. It didn't mention that therefor the solution that
was never found wasn't written to disk, just as there's
no separate explanation that no waveform data was written
if none could be calculated.

But no, as there's so many programmable conditions for the
..savebias file to be written, it takes a somewhat broad-
visioned forensic analysis to determine why it wasn't
written -- whether because it never reached the time,
temperature, or source voltage values over the run of
the simulation.

The user here just has the burden to notice that the file
is missing or from an old run.

--Mike

 

[Mike Rocket J. Squirrel E]

pat.lawler wrote:

Is failure to write a .savebias file noted in the error log?

[snip]

No it's not. I guess, though, that it is strictly not a failure, just
the result of there being no operating point found, which *is* noted in
the error log.

Mike (I won't call him Mikey as other have in this group, being a "Mike"
myself and having gotten seriously tired of the "Mikey likes it!" jokes)
is the most helpful guy in the world, but I find his helpfile writing to
be less than clear -- to me! (It's not likely his fault: I find the
writing style of most SPICE documentation to be slightly odd. I can
generally work my way through just about any other app's documentation,
but SPICE writing conventions and I just don't get on well together.
It's a personality thing.)

In this particular case, I would have been able to suss what I was doing
wrong if I knew that:

1. Since the file will be written with the results of the initial
solution to the operating point, if there is no solution to the
operating point found (either because it's too darn hard to do or
LTspice has been told not to bother, see the error file) then no file
will be written.

2. You can, however, save your node voltages by specifying a time step,
regardless of whether an initial operating point solution was found or
not. The .savebias command will write out the node voltages present at
the time specified. This can be handy if you wish to use the node
voltages found at the end of a .tran run, rather than the ones at the start.

This last bit of information, the first sentence of Step 2., was the bit
that I needed help understanding. Now things are peachy, life is good,
and my design work can continue apace. Thanks to all on this group who
have been patient with me.

--
Mike "Rocket J Squirrel" Elliott