Re: Tabs in Waveform window
On 21/05/2025 15:56, Andy I via
groups.io wrote:
Sorry, I meant to write "File
> Save Plot Settings As ...".? It is there but not
obvious at first.
?
But the other one to open a new Plot Settings file is
missing.
Yes, you're right. It's just the "Open Plot Settings File..." that
is missing from the Right-click File menu. However, the Ctrl-O
keyboard shortcut does work, provided it hasn't been changed. I
can't remember what all the installation default V24 keyboard
shortcuts are, but I remember I didn't like some of them, and
reverted to my old settings.
--
Regards,
Tony
|
Re: Tabs in Waveform window
Sorry, I meant to write "File > Save Plot Settings As ...".? It is there but not obvious at first.
?
But the other one to open a new Plot Settings file is missing.
?
Andy
?
|
Re: Tabs in Waveform window
On Wed, May 21, 2025 at 08:08 AM, Tony Casey wrote:
However, I just realised there is a problem with floating plot windows: when in use, there is no way to access the "Save Plot Settings As..." and the "Open Plot Settings File..." menu options, you can only reload and save one plot settings file. This needs reporting.
That might be partially taken care of already.
?
With the floated plot window, you can right-click in the plot window, then from the pop-up menu, choose File > Save Plot Settings. or File > Save Plot Settings As, or File > Reload Plot Settings.? But there is no choice to Open a different Plot Settings File, as one had from the Plot Settings main menu tab.
?
Andy
?
|
Re: Tabs in Waveform window
On 21/05/2025 13:18, J.C.A. Dekkers via
groups.io wrote:
Thank you too and I
am using that feature right now. From one of my previous lives I
used a simulator where you can tabulate the output panes, which is
quite useful since you can bundle results per subject.
In case you haven't come across it before, you can save any number
of plot settings in separate .plt files, that define different plot
configurations.
However, I just realised there is a problem with floating plot
windows: when in use, there is no way to access the "Save Plot
Settings As..." and the "Open Plot Settings File..." menu options,
you can only reload and save one plot settings file. This needs
reporting.
--
Regards,
Tony
|
Re: Tabs in Waveform window
Hi Andy,
?
Thank you too and I am using that feature right now. From one of my previous lives I used a simulator where you can tabulate the output panes, which is quite useful since you can bundle results per subject.
?
Regards,
Jacco
|
Re: Directory for custom symbols
On Wed, May 21, 2025 at 06:49 AM, eewiz wrote:
In addition to any symbol/library base folders there are also the standard.??? files.
For example, the standard.bjt file for transistors and standard.dio, .mos, .ind, .cap, .res and .bead files for those other parts.
All of those standard.??? files are just text files.
No, some of them are binary files.? Only the .bjt, .dio, .jft, and .mos ones are text, but those are the ones you are most likely to want to customize.? The others are binary files but can be edited inside of LTspice.
?
Andy
?
|
Re: Directory for custom symbols
Jeff
?
In reference to number "2. There is one top level to browse a location called "User libraries directory"" below.
?
There is a "User libraries directory" path entry but that item does not point to those symbol/library team directories asked about.
In addition to any symbol/library base folders there are also the standard.??? files.
For example, the standard.bjt file for transistors and standard.dio, .mos, .ind, .cap, .res and .bead files for those other parts.
All of those standard.??? files are just text files.
You may create empty text files named user.???.
Such as user.bjt, user.dio, user.mos, etc...
Put all of those user.??? files in any one directory, possibly your "TEAM" directory or somewhere deeper.
?
Make that "User libraries directory" path entry point to those user.??? files.
Now, whenever any schematic may call Q1 as a 2NOddBall, LTspice will search it's standard.bjt file first and if 2NOddBall is not found, LTspice will then search your "User libraries directory"\user.bjt file for the 2NOddBall transistor.
?
Sent:?Wednesday, May 21, 2025 at 3:43 AM
From:?"Mathias Born via groups.io" <mathias.born@...>
To:[email protected]
Subject:?Re: [LTspice] Directory for custom symbols
Yes, it is recursive. This was changed precisely to address your use case. Simply update to LTspice 24.1.8. You'll find the help also improved accordingly.
?
Best Regards,
Mathias
?
On Tue, May 20, 2025 at 10:10 PM, Jeff Kayzerman wrote:
My team wants to have a shared directory where all Electrical Engineers can place their symbols and libraries so others can re-use them instead of re-creating it.
?
This is what the help doc says
Search Paths
This panel allows you to enter additional paths than the default to find symbols and libraries. When entering symbol and library search paths, list each directory on it's own line.
This panel also allows you to edit the directory where user-defined discrete component libraries are located, in the field labeled "Location of user.* component library files."
?
Questions:
1. Is the search recursive? I would imagine since the text says "list each directory on its own line" that it is not recursive. However one of my engineers says he has used a nested directory structure for his symbols for years
2. There is one top level to browse a location called "User libraries directory". Is this what I should be using to point to our shared location? If so, why would I need the "Symbol Search Path" and "Library Search Path" boxes that require one directory per line.
3. If someone edited a standard component will it take precedence if it is defined in one of these paths (presumably the "User libraries directory"?
?
Thank you
|
Re: Tabs in Waveform window
Jacco,
?
I am guessing that you already know about changing LTspice's "Default Window Tile Pattern" from the default (Horizontal) to Vertical.? That gives you a full-height plot window, which can be more comfortable for using multiple plot panes - although it does change to half-width (but does not need to stay that way).? ?You might use that as an alternative to "Floating" the plot window as Tony suggested.
?
Floating the plot window is useful even if you have only one monitor, but it is especially handy when you have two.
?
Andy
?
|
Re: plotting group delay multiplied by frequency
On Wed, May 14, 2025 at 10:23 AM, Tony Casey wrote: Unfortunately, you can't plot gd(V(out)) as you can - say ph(V(out)).
Neither can you plot d(ph(V(out))) - an irritating little box pops up
telling you: "Ph() differentiation is not supported. Use the waveform
viewer group delay display feature." I worked around this a long time ago by using complex math, see /g/LTspice/message/10802
|
Re: Tabs in Waveform window
Thank you Tony.
?
Regards,
Jacco
|
Re: Tabs in Waveform window
On 21/05/2025 09:38, J.C.A. Dekkers via
groups.io wrote:
Maybe this question has already been asked, but I couldn't
find it. Sorry for that.
?
I am looking for multiple tabs/pages in the Waveform
(results) window. One can easily add a plot pane above or below
but it would be handy if one can add a tab en keep the number of
panes limited to a small number. Is such thing possible?
No, this isn't possible. The only other thing you can do is to float
the Waveform Viewer window: Right-click > Float Window. This is
useful if you have multiple monitors.
--
Regards,
Tony
|
On 21/05/2025 02:48, Andy I via
groups.io wrote:
I believe that your UniversalComp has a bug.? The
subcircuit's code includes this fragment:
if(V(out)>0,-Hys,Hys)/2
But that compares V(out) against global Ground.? That's not
OK.
?
It sort of works (the end result is OK) as long as the two
supply voltages straddle 0V = Ground.? But for example, if one
were to use V+ = +25V and V- = +20V, then V(out) is always >0
and the if() statement above always chooses -Hys on both rising
and falling edges.? And then there is no hysteresis.
?
It even fails in your test schematic just by changing V2 to
-1mV (V- = +1mV).
Well spotted! Thanks for looking. I did note this yesterday too,
while checking a whole bunch of other things. I have uploaded a fixed
model that addresses this.
I also amended the test
circuit to address the issue brought to light yesterday about
the naming syntax of .MEAS directives in 24.1.x (only alpha-numeric
characters allowed). Also fixed .MEAS so propagation delay measurements are consistent regardless whether inverting or non-inverting configuration is used.
I also did some benchmarking of other comparator models, for comparison. To get more reliable results, all analyses were .STEPed 100 times and repeated 5 times and averaged.
Circuit???????????????????????????? Time/100 ?? Hysteresis??? Trise??? Tfall??? Tpd+??? Tpd-
DiffSchmitt (A-device)??????????????? 2.82s?????? 10.0mV????? 22.0ns?? 22.0ns?? 17.3ns? 17.4ns
Universal Comparator (BI version)???? 3.12s?????? 10.0mV????? 31.7ns?? 31.7ns?? 11.2ns? 11.2ns
LT1720 (LT binary model)????????????? 3.75s?????? 3.5mV?????? 5.0ns??? 4.8ns??? 24.6ns? 11.6ns
LM139ATC (Transistor level) RL=39k ?? 4.24s ?? ?? 0.0mV?? ? ? 23.0ns?? 26.0ns?? 10.9ns? 75.0ns
LMV7219 (TI transistor level)???????? 5.79s?????? 15.0mV????? 2.1ns??? 2.1ns??? 9.1ns?? 15.8ns
Notes:
- With the DiffSchmitt device, the Vh parameter needs to be set to 1/2 of the desired total hysteresis, and Vt must be set to zero for asymmetrical power rails.
- For the DiffSchmitt, Tau was set to the programmed Delay of the BI source, for comparison.
- None of the other models have any programmable parameters.
--
Regards,
Tony
|
Re: Performance comparison: Win 11 vs Linux/Wine
On 21/05/2025 04:24, vbifyz via
groups.io wrote:
I had an impression already that Linux/Wine is a faster
platform for LTspice than Windows. Now I have a data point to
demonstrate it.
The file is uploaded, belka.asc .? It is a
simple RF mixer with a BJT audio amplifier. It runs 5ms of
transient analysis. The simulated RF frequency is 4MHz.
The LTSpice version is 24.0.12 on all 3 computers I compared.
Windows computer is Asus Vero laptop, i5-1235U, 16GB RAM,
Windows 11 with all latest updates.
Linux computer 1 is Dell Precision 5530 laptop, Xeon E-2176M
CPU (8th gen core),? 32GB RAM
Linux computer 2 is Asus ZenBook 14, i7-8565U, 16GB RAM
The Linux OS on both is Mint 22.1, recently updated. Wine is
the latest from the repository, no tweaks.
Now to the numbers. I pushed the start button and timed it to
the button turning green again. No traces displayed, no previous
runs (fresh window).
Windows : 1min 39s
Linux 1 : 0min 21s
Linux 2 : 0min 30s
I tried to add a security exception for LTspice in Windows,
so that the real-time antivirus is not in the way. It is known
to slow down many applications. There was no noticeable
difference.
If you look at the single thread benchmarks for these CPUs,
i5-1235U is expected to be about 50% faster than both of the 8th
gen CPUs, although Xeon has similar multithreaded score.
?
I invite other people to reproduce and to try different
files.
Benchmarking tests are pointless unless you change only one thing at
a time. Performance varies hugely with different hardware, often
unpredictably. Also, because operating systems are schedulers, you
should always average over a number of separate repetitions -
preferably, at least 10.
I previously conducted LTspice performance comparisons on my
dual-boot Ryzen 9 desktop PC, between Windows 10 and Wine running on
Arch Linux. Despite my overwhelming preference for using Linux, I
had to concede that Linux is slower for some core operations in
LTspice This is hardly surprising, as there is an extra layer. If
there were a native version of LTspice, the comparison would be
fairer and more useful
For these tests, I added ".options nomarch", to minimise graphics
operations, which can be quite a performance drag if you have
multiple .STEPs, which you always should do for benchmarking. I also
used a 16GB RAMDISK to eliminate writing to the hard disk.
--
Regards,
tony
|
Hi there,
?
Maybe this question has already been asked, but I couldn't find it. Sorry for that.
?
I am looking for multiple tabs/pages in the Waveform (results) window. One can easily add a plot pane above or below but it would be handy if one can add a tab en keep the number of panes limited to a small number. Is such thing possible?
?
Regards,
Jacco
|
Re: Directory for custom symbols
Yes, it is recursive. This was changed precisely to address your use case. Simply update to LTspice 24.1.8. You'll find the help also improved accordingly.
?
Best Regards,
Mathias
?
On Tue, May 20, 2025 at 10:10 PM, Jeff Kayzerman wrote:
My team wants to have a shared directory where all Electrical Engineers can place their symbols and libraries so others can re-use them instead of re-creating it.
?
This is what the help doc says
Search Paths
This panel allows you to enter additional paths than the default to find symbols and libraries. When entering symbol and library search paths, list each directory on it's own line.
This panel also allows you to edit the directory where user-defined discrete component libraries are located, in the field labeled "Location of user.* component library files."
?
Questions:
1. Is the search recursive? I would imagine since the text says "list each directory on its own line" that it is not recursive. However one of my engineers says he has used a nested directory structure for his symbols for years
2. There is one top level to browse a location called "User libraries directory". Is this what I should be using to point to our shared location? If so, why would I need the "Symbol Search Path" and "Library Search Path" boxes that require one directory per line.
3. If someone edited a standard component will it take precedence if it is defined in one of these paths (presumably the "User libraries directory"?
?
Thank you
|
Re: Performance comparison: Win 11 vs Linux/Wine
Found the reason for the Windows slowness. The .asc file was on the flash drive I am using to move the projects between computers. The run leaves a 130MB .raw file in the directory where the .asc file is. I can see the drive becoming 100% busy in the Task Manager during the run.
Once I copied the file to the local SSD and ran from it, it is way faster, about 16s. Haven't tried this on Linux laptops yet, but it is obviously an issue of Windows file access being very slow to the USB flash drive compared to Linux. The stick is USB 3.0 Lexar 64GB, fairly old. I am happy with its speed otherwise. Big file writes are 20-30MB/s, reads 80-100MB/s under both systems.
?
Mike
|
Re: Performance comparison: Win 11 vs Linux/Wine
On Tue, May 20, 2025 at 10:46 PM, vbifyz wrote:
I had an impression already that Linux/Wine is a faster platform for LTspice than Windows. Now I have a data point to demonstrate it.
I'll point out the obvious, which is that it is an apples-to-oranges comparison.
?
For it to be a fair comparison, it should have been made on the same exact hardware, and with the same LTspice*.ini file.? Neither of those is true.? How that skews the results, I can't say but it might be very significantly.
Windows : 1min 39s
Linux 1 : 0min 21s
Linux 2 : 0min 30s
For kicks, I tried this on an old Windows laptop here, with much less memory and a different CPU.? So it is difficult to compare expectations, except? that I would have expected mine to be slower than yours.??But somehow mine was about 6 to 9 times faster than yours!? And two to three times faster than your Linux tests.? And I had marching waveforms enabled and even plotting a waveform, which made mine run slower than it could have done.
?
That has me wondering.? What were your LTspice settings that made it so slow?? What else might have been running on your computers?
?
I tried to add a security exception for LTspice in Windows, so that the real-time antivirus is not in the way. It is known to slow down many applications. There was no noticeable difference.
Maybe it was still in the way and holding things up.? That significantly?? Perhaps.
?
I am not currently set up with Linux.
?
Andy
?
|
Re: Performance comparison: Win 11 vs Linux/Wine
Try running windows 11 & wine on the same system and running the sim for each os.
|
Performance comparison: Win 11 vs Linux/Wine
I had an impression already that Linux/Wine is a faster platform for LTspice than Windows. Now I have a data point to demonstrate it.
The file is uploaded, belka.asc .? It is a simple RF mixer with a BJT audio amplifier. It runs 5ms of transient analysis. The simulated RF frequency is 4MHz.
The LTSpice version is 24.0.12 on all 3 computers I compared.
Windows computer is Asus Vero laptop, i5-1235U, 16GB RAM, Windows 11 with all latest updates.
Linux computer 1 is Dell Precision 5530 laptop, Xeon E-2176M CPU (8th gen core),? 32GB RAM
Linux computer 2 is Asus ZenBook 14, i7-8565U, 16GB RAM
The Linux OS on both is Mint 22.1, recently updated. Wine is the latest from the repository, no tweaks.
Now to the numbers. I pushed the start button and timed it to the button turning green again. No traces displayed, no previous runs (fresh window).
Windows : 1min 39s
Linux 1 : 0min 21s
Linux 2 : 0min 30s
I tried to add a security exception for LTspice in Windows, so that the real-time antivirus is not in the way. It is known to slow down many applications. There was no noticeable difference.
If you look at the single thread benchmarks for these CPUs, i5-1235U is expected to be about 50% faster than both of the 8th gen CPUs, although Xeon has similar multithreaded score.
?
I invite other people to reproduce and to try different files.
?
Mike
|
Tony,
?
I believe that your UniversalComp has a bug.? The subcircuit's code includes this fragment:
if(V(out)>0,-Hys,Hys)/2
But that compares V(out) against global Ground.? That's not OK.
?
It sort of works (the end result is OK) as long as the two supply voltages straddle 0V = Ground.? But for example, if one were to use V+ = +25V and V- = +20V, then V(out) is always >0 and the if() statement above always chooses -Hys on both rising and falling edges.? And then there is no hysteresis.
?
It even fails in your test schematic just by changing V2 to -1mV (V- = +1mV).
?
Andy
?
?
|
Tony Casey
eewiz
Frank Wiedmann