I’ve been trudging along happily with LTspice 17.1.14. I’m sure that the newer versions offer some improvements, but I’m leery of the incompatibilities and bugs that are being reported. ? With this in mind, would anyone like to offer a suggestion regarding upgrading?
toggle quoted message
Show quoted text
From: [email protected] <[email protected]> On Behalf Of Andy I via groups.io
Sent: Thursday, May 22, 2025 7:30 PM
To: [email protected]
Subject: Re: [LTspice] 24.1,x update woes? On Thu, May 22, 2025 at 07:04 PM, Hawker wrote: Am I correct in assuming in 24.1 you cannot have a .include to a model you are not using?
I have not tried that, but I greatly doubt that you cannot do that.? It has never been a restriction before (in LTspice or anywhere, as far as I know).? IMHO it would serve no useful purpose to have such a restriction. Did you read that somewhere, that it was not allowed? C:\Users\{user}\Documents\LTspice\user.jft(1): Syntax error (unexpected input).
.MODEL 2N5457 NJF(VTO=-1.8 BETA=0.00135 LAMBDA=0.001 RD=35 RS=31.5 + CGS=2.25E-12 CGD=6E-12 KF=6.5E-17 AF=0.5 )
? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?? ^^^^^^^^^
Was there more to that error message than that?? Usually error messages have more information. Could it be the stray plus sign between the RS and CGS parameters?? If you see a floating misplaced plus sign in your user.jft, try editing it out.? If it is a 2-line .MODEL statement, that could be an important clue. OTOH, there were problems with many (most?) of the JFET models that came in LTspice's standard.jft, for several years.? They may have finally whipped them into shape.
|
Re: Modelling magnetic core of EP10 with T38 material of TDK micronoas
I wrote:
Not everyone uses the same turns.
That should be "terms", not "turns".
?
Andy
|
Re: Modelling magnetic core of EP10 with T38 material of TDK micronoas
On Fri, May 23, 2025 at 10:55 AM, <edhooruper@...> wrote:
Is it possible to model a ferrite magnetic core in Ltspice ?
LTspice does not have a built-in magnetics modeler.? You need to turn to another program (such as the previously recommended FEMM) for that.? Once you get the circuit parameters, then you can use them in LTspice to simulate the circuit.
?
As you read in LTspice's Help, it does include some tools in the form of the Chan model for nonlinear magnetics.? They might include some of the model parameters you are trying to use, such as the cross-sectional area for which LTspice has a parameter also named "A".? But note that there are MANY parameters with similar names, such as "A" (or A-sub-something) and you need to fully understand which parameters you are dealing with in each situation.? Not everyone uses the same turns terms.? Most inductor modelers are familiar with AL, usually written as A-sub-L.? It is not the same as the A in the Chan model, nor the a, Ac, Ap, At, Aw, Aw(B), Aw(I), Awp, Aws, or A-T in the book you referred to.? Interestingly, the book apparently also uses a parameter named A (which you quoted in your reply), yet it does not list it in its table of Symbols.? What does that suggest?
?
Andy
?
?
|
Re: Modelling magnetic core of EP10 with T38 material of TDK micronoas
I also hve the circuit in Ltspice ,?
Is it possible to model a ferrite magnetic core in Ltspice ? Is there something similar pspice model editor for Ltspice ?
|
Re: Modelling magnetic core of EP10 with T38 material of TDK micronoas
On 23/05/2025 16:35, edhooruper via
groups.io wrote:
So I am modelling
the magnetic core in Pspice model editor
I got the B-H curve values from the TDK mcironas ferrite design
magnetic tool and uploaded that into pspice model editor
I also referred to the
for modelling the magnetic core
The values mentioned in the book matches with the ?
But the issue is the Pspice model editor uses Jiles Atherton model
to calculate the parameters Ms , A , C and K
When I give the B-H curve values into the Pspice model editor ,
the model editor internally somehow calculates the above mentioned
parameters which is way too far away from the ones mentioned in
the book (there is no values of MS , A , C and K mentioned in the
datasheet)
And when I give the parameters from the book into the Pspice model
editor , the curve does not fit and I am not getting the expected
results
Any idea on how to resolve this ?
Below table shows both the book values (2nd column) and spice
model editor values (3rd column)
|
Parameters
|
Values
from the book (MS is calculated)
|
?Values extracted
from curve by Pspice
Model
|
|
MS
|
342110
|
360595.7907737
|
|
A
|
1.26
|
10.185
|
|
C
|
2.60
|
0.39337
|
|
K
|
4.194e3
|
8.246
|
Please note, this group's purpose is to support LTspice, please
don't expect PSpice support.
--
Regards,
Tony
|
Re: Modelling magnetic core of EP10 with T38 material of TDK micronoas
On Fri, May 23, 2025 at 10:35 AM, <edhooruper@...> wrote:
But the issue is the Pspice model editor uses Jiles Atherton model to calculate the parameters Ms , A , C and K
This is not a group about PSpice.? There are other resources where you can ask questions directly about PSpice-unique problems.? This group is for LTspice.
When I give the B-H curve values into the Pspice model editor , the model editor internally somehow calculates the above mentioned parameters which is way too far away from the ones mentioned in the book (there is no values of MS , A , C and K mentioned in the datasheet)
And when I give the parameters from the book into the Pspice model editor , the curve does not fit and I am not getting the expected results
Any idea on how to resolve this ?
Is it possible you used the wrong units, or omitted units-multipliers?? That's only a guess.
?
Andy
?
|
Re: Modelling magnetic core of EP10 with T38 material of TDK micronoas
So I am modelling the magnetic core in Pspice model editor I got the B-H curve values from the TDK mcironas ferrite design magnetic tool and uploaded that into pspice model editor I also referred to the for modelling the magnetic core The values mentioned in the book matches with the ? But the issue is the Pspice model editor uses Jiles Atherton model to calculate the parameters Ms , A , C and K When I give the B-H curve values into the Pspice model editor , the model editor internally somehow calculates the above mentioned parameters which is way too far away from the ones mentioned in the book (there is no values of MS , A , C and K mentioned in the datasheet) And when I give the parameters from the book into the Pspice model editor , the curve does not fit and I am not getting the expected results Any idea on how to resolve this ? Below table shows both the book values (2nd column) and spice model editor values (3rd column)
|
Parameters
|
Values from the book (MS is calculated)
|
?Values extracted from curve by Pspice Model
|
|
MS
|
342110
|
360595.7907737
|
|
A
|
1.26
|
10.185
|
|
C
|
2.60
|
0.39337
|
|
K
|
4.194e3
|
8.246
|
|
Hawker,
?
If you only check in here once daily (after receiving your daily Digest), this is going to be an awfully tedious problem to solve.
?
Please check back more often!
?
Andy
?
|
On Fri, May 23, 2025 at 10:04 AM, Hawker wrote:
Indeed I have 24.1.8. Is this a known problem install?
No.
Should/can I go back to a previous one?
You can, if you want to.? If you do go back, go back to version 24.0.12 or earlier.? Of the 24.1.x versions, the latest (24.1.8) should have the fewest problems and is preferred.? Do not try 24.1.1 or? 24.1.2 or 24.1.3 or ....
Is there a a way to go back before 24.1.8 if indeed this version has issues?
Of course.? Back up all your own library files first.? If you kept the install file for the older version, use it; or download it, and start over.? The group's main webpage has a link to the installation file.
I have been unable to get most of my simulations that worked fine before to work now.
That suggests you may have problems elsewhere in your installation, not with LTspice 24 itself.
I have this message for example for simulations that have worked fine in 24.0.x
C:\Users\{user}\Documents\LTspice\user.jft(1): Syntax error (unexpected input).
.MODEL 2N5457 NJF(VTO=-1.8 BETA=0.00135 LAMBDA=0.001 RD=35 RS=31.5 + CGS=2.25E-12 CGD=6E-12 KF=6.5E-17 AF=0.5 )
? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ^^^^^^^^^
C:\Users\{user}\Documents\LTspiceXVII\lib\cmp\TL074.301(43): Expected device instantiation or directive here.
If I comment out the unused TL074.301 include I get past that but my 2N5457 model still gives me the syntax error.
Did you try the other suggestion, about that plus sign in the 2N5457 that should not be there?? Did you find the cause of that extra plus sign?? What exactly does your .INC command look like that tries to load? TL074.301?
?
We can offer you suggestions, but they don't help if you do not try them.
?
Andy
?
?
|
On Thu, May 22, 2025 at 07:18 PM, John Woodgate wrote:
There are reverse-compatibility issues with 24.1.8, but exactly which 24.1.x are you using? Issues before x.x.8 are suspect. Are you getting errors or just warnings?
Indeed I have 24.1.8. Is this a known problem install? Should/can I go back to a previous one? Is there a a way to go back before 24.1.8 if indeed this version has issues?
I have been unable to get most of my simulations that worked fine before to work now.
I have this message for example for simulations that have worked fine in 24.0.x
C:\Users\{user}\Documents\LTspice\user.jft(1): Syntax error (unexpected input).
.MODEL 2N5457 NJF(VTO=-1.8 BETA=0.00135 LAMBDA=0.001 RD=35 RS=31.5 + CGS=2.25E-12 CGD=6E-12 KF=6.5E-17 AF=0.5 )
? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ^^^^^^^^^
C:\Users\{user}\Documents\LTspiceXVII\lib\cmp\TL074.301(43): Expected device instantiation or directive here.
If I comment out the unused TL074.301 include I get past that but my 2N5457 model still gives me the syntax error.
^
|
On Thu, May 22, 2025 at 07:04 PM, Hawker wrote:
Am I correct in assuming in 24.1 you cannot have a .include to a model you are not using?
I have not tried that, but I greatly doubt that you cannot do that.? It has never been a restriction before (in LTspice or anywhere, as far as I know).? IMHO it would serve no useful purpose to have such a restriction.
?
Did you read that somewhere, that it was not allowed?
?
C:\Users\{user}\Documents\LTspice\user.jft(1): Syntax error (unexpected input).
.MODEL 2N5457 NJF(VTO=-1.8 BETA=0.00135 LAMBDA=0.001 RD=35 RS=31.5 + CGS=2.25E-12 CGD=6E-12 KF=6.5E-17 AF=0.5 )
? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?? ^^^^^^^^^
Was there more to that error message than that?? Usually error messages have more information.
?
Could it be the stray plus sign between the RS and CGS parameters?? If you see a floating misplaced plus sign in your user.jft, try editing it out.? If it is a 2-line .MODEL statement, that could be an important clue.
?
OTOH, there were problems with many (most?) of the JFET models that came in LTspice's standard.jft, for several years.? They may have finally whipped them into shape.
?
|
There are reverse-compatibility issues with
24.1.8, but exactly which 24.1.x are you using? Issues before
x.x.8 are suspect. Are you getting errors or just warnings?
On 2025-05-23 00:04, Hawker wrote:
Howdy
I finally got around to updating from 24.0.x to 24.1.x and a lot
seems to have broken for me.
Am I correct in assuming in 24.1 you cannot have a .include to a
model you are not using?
I tend to have a "template" for a type of work where I may
define a pile of parts, but not always use them all. Up to now
that has not been an issue but now I get an error like
?
C:\Users\{user}\Documents\LTspiceXVII\lib\cmp\TL074.301(43):
Expected device instantiation or directive here.
I can change it to a comment and it works.
I am also finding some of my legacy models that came from LT
Spice no longer work. For example how can I fix this one?
It's been working fine for nearly 20 years.
C:\Users\{user}\Documents\LTspice\user.jft(1): Syntax error
(unexpected input).
.MODEL 2N5457 NJF(VTO=-1.8 BETA=0.00135 LAMBDA=0.001 RD=35
RS=31.5 + CGS=2.25E-12 CGD=6E-12 KF=6.5E-17 AF=0.5 )
? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?
? ? ? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?
?? ^^^^^^^^^
--
Best wishes John Woodgate RAYLEIGH Essex OOO-Own Opinions Only
If something is true: * as far as we know - it's science *for
certain - it's mathematics *unquestionably - it's religion
|
Howdy
I finally got around to updating from 24.0.x to 24.1.x and a lot seems to have broken for me.
Am I correct in assuming in 24.1 you cannot have a .include to a model you are not using?
I tend to have a "template" for a type of work where I may define a pile of parts, but not always use them all. Up to now that has not been an issue but now I get an error like
?
C:\Users\{user}\Documents\LTspiceXVII\lib\cmp\TL074.301(43): Expected device instantiation or directive here.
I can change it to a comment and it works.
I am also finding some of my legacy models that came from LT Spice no longer work. For example how can I fix this one?
It's been working fine for nearly 20 years.
C:\Users\{user}\Documents\LTspice\user.jft(1): Syntax error (unexpected input).
.MODEL 2N5457 NJF(VTO=-1.8 BETA=0.00135 LAMBDA=0.001 RD=35 RS=31.5 + CGS=2.25E-12 CGD=6E-12 KF=6.5E-17 AF=0.5 )
? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?? ^^^^^^^^^
|
Re: Directory for custom symbols
The most important aspect of ensuring that all models, symbols, and subcircuits are checked in to the version control system is that it also ensures that everyone can always know that the specific version used for any simulation is known, can be replicated, and any variations understood and tracked. That knowledge is invaluable.
I would argue that there is value in having the central sim-server only run simulations on files that it checks out of the master repo. Works very well for software projects. Every piece of the work is exactly known. No tweaks ever sneak into a project.
Donald.
Sent from Proton Mail Android
-------- Original Message -------- On 5/22/25 09:54, Jeff Kayzerman wrote: Thanks eewiz that clears things up!
?
Thanks Tony, we have really good redundancy and resiliency so it's unlikely the share drives would go down but I get your point and that's a good idea about using SVN. SVN complicates things a little bit because what I'm trying to accomplish is less about syncing between EE's machines and is actually about syncing to a remote server. I built them a system where they have a web app that they can drag and drop a zip file that contains the .net files and all symbols, models etc... which gets sent to a server cluster to be executed. This process works well and some EE's are using it and just accept that they need to include any dependencies not in the standard library into the folder that they zip. But of course others complain about this... so if they can get a centralized location that LTspice on each machine reads from it makes it easier (theoretically). Doing this in svn would mean they have to commit any and all new changes to symobls/models and then when the server that picks up the job needs to do an svn update prior to running the job. It's not a significant code change just requires some additional setup.
?
There will also need to be training on svn and there will be organizational push-back around adopting it I'm sure.
?
Thanks though, will discuss with the team and see if the benefits of version control is worth the extra effort.
|
Re: Deactivate on Circuit or part of the circuit in LTspice
On Thu, May 22, 2025 at 10:35 AM, <oerni@...> wrote:
It would be sufficient if subcircuit hierarchical blocks could be avoided from loading and inclusion into netlist per switch which can be controlled by a floating point parameter....(but keeping them as schematics to be opened still by double-click)
I solved it workaround wise in a way that I created a subckt with names 0.999 for used and 0.001 for unused....for example I can use .PARAM FILTERS=0.001 or .PARAM FILTERS=0.999 to turn them ON+OFF (also via .STEP)
Out of nothing but my curiosity - why does it need to be a floating point parameter?? Would integers (0 and 1) have worked?
?
If their meanings were ON and OFF, then values of 1 and 0 would seem to be more appropriate, since there is precedent already for those values to equate with the binary values TRUE and FALSE.
Andy
?
|
Re: Deactivate on Circuit or part of the circuit in LTspice
In a reply to a rather old set of messages thread, <oerni@...> wrote:
I got them working after one week of work even without any reply on my request here about a week or two ago...Many thanks for all your help.
Where was that request for help?
?
Was it the question on the 6th of May about Vlad's special Filters?? Vlad himself is the best and might be be the only expert on them.? I have not seen them discussed here much, which suggests there are not many others who have used his Filters and understand how they work.
?
Unfortunately, Vlad has made himself scarce in this group for a couple of years.? (We hope he will return soon.)? He tells people a way to contact him directly, in his Filter docs.? Did you try contacting him, and did he respond?
?
Andy
?
|
Re: Deactivate on Circuit or part of the circuit in LTspice
On 22/05/2025 16:35, oerni via
groups.io wrote:
Searching the
internet for this feature brings many wishes but not one good
solution. Now I go to matlab which has this feature available.
Sorry guys.
If you're desperate for an application more like LTspice where is
possible, check out Qucs. This has been a feature for a long time.
Most of us will find various other ways to work around this. The
easiest being to simply delete the unneeded section(s), run the
analysis, and then "Undo".
If you want to ignore a block of components, temporarily short all
internal nodes to 0, then they will be excluded from the netlist
without throwing an error.
If you just want to reduce the size of the .raw file, use the .SAVE
directive to be selective about the analysis results that you are
interested in, and disregard the rest. The .raw file can be
drastically reduced by this action.
--
Regards,
Tony
|
Re: Deactivate on Circuit or part of the circuit in LTspice
It would be sufficient if subcircuit hierarchical blocks could be avoided from loading and inclusion into netlist per switch which can be controlled by a floating point parameter....(but keeping them as schematics to be opened still by double-click)
I solved it workaround wise in a way that I created a subckt with names 0.999 for used and 0.001 for unused....for example I can use .PARAM FILTERS=0.001 or .PARAM FILTERS=0.999 to turn them ON+OFF (also via .STEP)
In the subcircuit file .SUB I just placed the netlist output from LTSPICE for my schematic which needs to be turned ON/OFF.
I keep the original schematic with all its wires and ports as a way to still edit the schematic in LTSPICE....but every time I edit it, I have to copy the netlist again into the subcircuit which contain 0.001 and 0.999.
I did not manage to use strings, not even for the latest LTspice version.....maybe someone knows how this would work.
On my .ASY symbol I make the "spicemodel" visible and I put {FILTERS} for it....depending on the value of FILTERS, either the complete model or a nearly empty dummy is loaded.
But this does not allow to double click on the symbol to edit the underlayed schematic. I have to open and edit it manually, copy manually the new netlist into the subcircuit, but then I can turn ON+OFF parts even in higher hierarchical blocks and downpass the FILTERS parameter down to the subcircuit....and I can step my simulations with and without the huge digital filters. I got them working after one week of work even without any reply on my request here about a week or two ago...Many thanks for all your help.
Unfortunately I have 108 inputs, each with two huge digital filters out of a filter chain....and there are usages which don't need the filters and can simulate in seconds while everything with filters takes 15min. till it delivers plots and automated measurements but most of the times I end the LTspice task afterwards as it does not react on inputs anymore after simulation with about 10GB .raw file. But it works and I can see details in the created logfile.
Searching the internet for this feature? brings many wishes but not one good solution. Now I go to matlab which has this feature available. Sorry guys.
|
Re: Directory for custom symbols
Thanks eewiz that clears things up!
?
Thanks Tony, we have really good redundancy and resiliency so it's unlikely the share drives would go down but I get your point and that's a good idea about using SVN. SVN complicates things a little bit because what I'm trying to accomplish is less about syncing between EE's machines and is actually about syncing to a remote server. I built them a system where they have a web app that they can drag and drop a zip file that contains the .net files and all symbols, models etc... which gets sent to a server cluster to be executed. This process works well and some EE's are using it and just accept that they need to include any dependencies not in the standard library into the folder that they zip. But of course others complain about this... so if they can get a centralized location that LTspice on each machine reads from it makes it easier (theoretically). Doing this in svn would mean they have to commit any and all new changes to symobls/models and then when the server that picks up the job needs to do an svn update prior to running the job. It's not a significant code change just requires some additional setup.
?
There will also need to be training on svn and there will be organizational push-back around adopting it I'm sure.
?
Thanks though, will discuss with the team and see if the benefits of version control is worth the extra effort.
|
Re: Directory for custom symbols
On 22/05/2025 03:02, Jeff Kayzerman via
groups.io wrote:
Thanks Tony, I am indeed talking about a network drive that
is mapped to all engineers computers as the "M:" drive. I am not
an EE, I am a software dev helping them out with some automation
so I don't fully understand how all of this works.
?
Does it make sense to have a top level directory in the M:
drive, "LTspice Common" (or something to that extent). Then
under that have some folder structure where there is one folder
for "models", "symbols", "libraries" etc... and those
directories are flat with just files.
?
Then which paths go into which boxes in control panel?
Obviously symbols would go into "Symbol Search Path[*]".
Libraries goes in "Library Search Path[*]". Models, .dio,
anything else all needs to go into the top "User Libraries
Directory"? Just guessing here.
?
Since you're an "Software Dev", you'll understand the complications
involved when users are creating, sharing, using and updating remote
files. For that reason, I suggest you don't use the simple mapped M:
drive on all users' machines.
I (mostly) don't have multiple users sharing LTspice libraries etc.
and other documents, but I do have multiple machines. For this and
other reasons, I use a change control system (Subversion) on my
central server, and all other machines have the Subversion client.
In this scenario, you don't require mapped network drives, you keep
synced copies of the shared resources on each machine. This has the
advantage that all (well, nearly all) files are still available if
(when) the server isn't available, which otherwise wouldn't be the
case, and everyone's work would stop. A perfect case of single point
failure.
In that situation, each machine has local copies of "user files" in
the locations previously discussed in this thread, and listed in
Settings (formerly Control Panel) > Search Paths. So, no wheels
need inventing. Machines could be set up to e.g.: "svn update" in
the morning and "svn commit" in the evening. (Other change control
options are available.)
The other advantage with a change control system, is the built-in
capability to easily revert to a previous version if someone messes
up any particular file(s), which then causes issues.
I'm sure as a software dev, you're used working this way, but
hardware engineers not so much...
--
Regards,
Tony
|
Tony Casey
John Woodgate