How to get uniform export data from LTspice?

Sergius

unread,

Jan 8, 2012, 4:01:52 PM1/8/12

to

I;m trying to get transient data for future Matlab - calculating SNR
of
ADC output, but time scale of .txt exported file is not uniform- time
steps
seem to different from point to point. What I must to linearize or
resample?

Robert Macy

unread,

Jan 9, 2012, 10:50:41 AM1/9/12

to

this has been covered several times with the LTspicegroup. And, yes,
Iforgot how to do it.

There is some form of the way to export the data file that includes
the step size, or number of steps, ??

I wrote a function to linearnize the steps for LTpsice and used it
until I found that out .

Did you oin the LTspice group, it's easy.

Jim Thompson

unread,

Jan 9, 2012, 11:09:21 AM1/9/12

to

Robert, Is there a way to get uniform time-step data?

This is all I know of, from Helmut [Digest # 4440]:

[begin quote from LTspice group]

--- In LTs...@yahoogroups.com, "jtanalog" <yahoo@...> wrote:
>
> Is there any way in LTspice to generate a .OUT file such as are found in other Spice variants?
>
> Or is there some alternate way to "print" data points in a list?
>
> - Jim Thompson
>

Hello Jim,

Make the waveform window the active window.
From the menu click File and then Export.

Best regards,
Helmut

[end quote from LTspice group]

In PSpice you just insert a .print statement into your .cir file with
the stepsize you want. The data appears in the .out file.

...Jim Thompson
--
| James E.Thompson, CTO | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| Phoenix, Arizona 85048 Skype: Contacts Only | |
| Voice:(480)460-2350 Fax: Available upon request | Brass Rat |
| E-mail Icon at http://www.analog-innovations.com | 1962 |

I love to cook with wine. Sometimes I even put it in the food.

Robert Macy

unread,

Jan 9, 2012, 6:47:39 PM1/9/12

to

On Jan 9, 9:09 am, Jim Thompson <To-Email-Use-The-Envelope-I...@On-My-

Web-Site.com> wrote:
> On Mon, 9 Jan 2012 07:50:41 -0800 (PST), Robert Macy
>
>
>
>
>
> <robert.a.m...@gmail.com> wrote:
> >On Jan 8, 2:01 pm, Sergius <sergey...@gmail.com> wrote:
> >> I;m trying to get transient data for future Matlab - calculating SNR
> >> of
> >> ADC output, but time scale of .txt exported file is not uniform- time
> >> steps
> >> seem to different from point to point. What I must to linearize or
> >> resample?
>
> >this has been covered several times with the LTspicegroup. And, yes,
> >Iforgot how to do it.
>
> >There is some form of the way to export the data file that includes
> >the step size, or number of steps, ??
>
> >I wrote a function to linearnize the steps for LTpsice and used it
> >until I found that out .
>
> >Did you oin the LTspice group, it's easy.
>
> Robert, Is there a way to get uniform time-step data?
>
> This is all I know of, from Helmut [Digest # 4440]:
>
> [begin quote from LTspice group]
>

> --- In LTsp...@yahoogroups.com, "jtanalog" <yahoo@...> wrote:
>
>
>
> > Is there any way in LTspice to generate a .OUT file such as are found in other Spice variants?
>
> > Or is there some alternate way to "print" data points in a list?
>
> > - Jim Thompson
>
> Hello Jim,
>
> Make the waveform window the active window.
> From the menu click File and then Export.
>
> Best regards,
> Helmut
>
> [end quote from LTspice group]
>
> In PSpice you just insert a .print statement into your .cir file with
> the stepsize you want. The data appears in the .out file.
>
> ...Jim Thompson
> --
> | James E.Thompson, CTO | mens |
> | Analog Innovations, Inc. | et |
> | Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
> | Phoenix, Arizona 85048 Skype: Contacts Only | |
> | Voice:(480)460-2350 Fax: Available upon request | Brass Rat |

> | E-mail Icon athttp://www.analog-innovations.com| 1962 |

>
> I love to cook with wine. Sometimes I even put it in the food.

Jim,

I played with the form to try it out and now have NO idea where it is,
or what I had done. I'll look around to see if I can find it.

Jim Thompson

unread,

Jan 9, 2012, 8:23:12 PM1/9/12

to

Thanks!

It sure seems that Mikey would have an equivalent to every other
Spice, the .print statement ???

| E-mail Icon at http://www.analog-innovations.com | 1962 |

Robert Macy

unread,

Jan 10, 2012, 10:03:11 AM1/10/12

to

On Jan 9, 6:23 pm, Jim Thompson <To-Email-Use-The-Envelope-I...@On-My-

Here's the function I wrote to use in free Matlab clone, octave
2.1.50a which should work in later versions:

function [t,v]=fillinsteps(x,y,n);
% FILLINSTEPS to smooth an ordinate varying waveform
% use the form [t,v,n]=fillinsteps(x,y,n);
% where x is the ordinate
% y is the abcissa
% n is the number of steps
if (nargin==0)
help fillinsteps;
return;
endif

if (nargin ~=3)
help fillinsteps;
return;
endif

[ry,cy]=size(y);
t=max(x)*(1:n)/n;
for i=1:n
[xx,ix]=min(abs(x-t(i)));
if (t(i)>=x(ix))
if (ix==cy)
v(i)=y(cy);
else
v(i)=y(ix)+(y(ix+1)-y(ix))*(t(i)-x(ix))/(x(ix+1)-x(ix));
endif
else
if (ix==1)
v(i)=y(1);
else
v(i)=y(ix-1)+(y(ix)-y(ix-1))*(t(i)-x(ix-1))/(x(ix)-x(ix-1));
endif
endif
endfor

Note, it IS a linear interpolation, so you will see the noise floor
come up in the FFTs.

I think there is some finagling to get the x and y into this function,
probably export, or print, then past onto Excel, maybe, then scoop up
and put into the array, or...

Maybe it was simply strip off the header from LTspice and 'read' the
text file as a variable, then strip out x and y. Yeah that sounds
better.

Jim Thompson

unread,

Jan 10, 2012, 10:38:49 AM1/10/12

to

What do you get with LTspice if you choose "ASCII data files" (CSDF)?

| E-mail Icon at http://www.analog-innovations.com | 1962 |

Robert Macy

unread,

Jan 10, 2012, 7:02:20 PM1/10/12

to

On Jan 10, 8:38 am, Jim Thompson <To-Email-Use-The-Envelope-I...@On-My-

Web-Site.com> wrote:
> On Tue, 10 Jan 2012 07:03:11 -0800 (PST), Robert Macy
>
>
>
>
>

I don't know.

But someone finally posted the complete method:

Use LTspice2Matlab.m function to read and convert the .raw file to
uniform steps.

Before I knew that existed, I would go to FILE, select Export, and the
file automatically comes out as a two column non-uniform text list of
calculations.
I would remove the Header with a text editor, and simply 'read' the
file into octave, then I used my custom fillinsteps.m [not knowing
someone had already written interp1.m and was a built-in function]
then I would save the data in ascii and get a text list of uniform
steps.

A bit of a workaround, but when you don't do this often, workarounds
take less time. Probably would go through all these steps if I did it
again, because I tend to reinvent the wheel each time and that's the
logical sequence

Jim Thompson

unread,

Jan 10, 2012, 7:19:10 PM1/10/12

to

It occurs to me that one could simply import the text list into PSpice
as a PWL, run it as a source, but call a .PRINT statement with the
timestep I choose to convert it to evenly spaced.

Is LTspice2Matlab.m a Matlab routine, or where do you run it?

| E-mail Icon at http://www.analog-innovations.com | 1962 |

Robert Macy

unread,

Jan 11, 2012, 11:32:27 AM1/11/12

to

On Jan 10, 5:19 pm, Jim Thompson <To-Email-Use-The-Envelope-I...@On-My-

If you have Matlab, run as a normal function putting in all the
arguments and requesting all your outputs.

In octave, you can place it in a script file to run, or run from the
console line entry.

I only used it to verify and found it did work. But, every time I
'uniformed' the steps, I always reverted to my old technique since I
could count on it.

Do you have Matlab, or free octave?

Did you try the waveform source and PRINT function to see if that
worked?

Jim Thompson

unread,

Jan 11, 2012, 11:50:01 AM1/11/12

to

Neither. But with the requests for Spice to IBIS coming in I do need
a convenient numeric processor.

| E-mail Icon at http://www.analog-innovations.com | 1962 |

Robert Macy

unread,

Jan 11, 2012, 1:46:10 PM1/11/12

to

On Jan 11, 9:50 am, Jim Thompson <To-Email-Use-The-Envelope-I...@On-My-

Web-Site.com> wrote:
> On Wed, 11 Jan 2012 08:32:27 -0800 (PST), Robert Macy
>
>
>
>
>

Stay away from trying to use Excel ;)

I recommend octave 2.1.50a for its plotting! However, if you're doing
high power arrays, latest version is better, accepts 3 indices.

More advantageous, the users group will usually post help response
within a few hours 24/7. If you're like me, you work anytime and don't
want to interface with 'customer service' only during business hours.

I use octave to test DSP algorithms BEFORE writing the C, C++ code.

First time Directions:
Get a good text editor, like SC1 from Scilab. Set it up to open .m
files.
Install octave.
place all your working .txt files and .m files in the octave_files
folder.
Run octave.
Either command line prompt, step by step, or
write a TestScript.m file using the editor
At the prompt type in the name of the script file, in this case
TestScript [don't add the .m It's implied] and octave will run your
script.
You can open files, read files either text or binary, save files
either text or binary. etc.

Fastest way to learn is try to do the first thing you want done?
LTspice analyze the problem and export results in text format. Keep
simple. only one variable. Edit the file to remove header, only have
numbers, save as out.txt.

At the command prompt type
> read("out.txt");
then type
> t=out(:,1)'; v=out(",2)';
that will put the thing into 'normal' forms

you now have time as t in nonuniform steps and you have v at each time
point.

How many points do you want? 1000?
type in
> [T,V]=fillinsteps(t,v,1000);
[or use the built in function interp1.m]
and voila! even number of steps

save as a text file by typing
> save -ascii "StepsUniform.txt" T V;
and then look at that file. You'll find that the arrays are across and
not down
if you want down like the original format change the variables:
> T=T'; V=V'; save "StepsUniform.txt" T V;
or if you want the 'down' array to be in the similar format as the
original:
> OUT=[T,V]; save -ascii "OUT.txt" OUT;
then you can use your text editor to change the header to the original
header and you're done.

Jim Thompson

unread,

Jan 11, 2012, 2:06:45 PM1/11/12

to

Thanks, Robert, I'll give it a whirl.

| E-mail Icon at http://www.analog-innovations.com | 1962 |

Jim Thompson

unread,

Jan 11, 2012, 2:53:07 PM1/11/12

to

On Wed, 11 Jan 2012 10:46:10 -0800 (PST), Robert Macy
<robert...@gmail.com> wrote:

[snip]

Robert, For those of us using only Windows... which installer should I
download?

| E-mail Icon at http://www.analog-innovations.com | 1962 |

Robert Macy

unread,

Jan 11, 2012, 5:58:35 PM1/11/12

to

On Jan 11, 12:53 pm, Jim Thompson <To-Email-Use-The-Envelope-I...@On-

My-Web-Site.com> wrote:
> On Wed, 11 Jan 2012 10:46:10 -0800 (PST), Robert Macy
>

> | E-mail Icon athttp://www.analog-innovations.com| 1962 |

>
> I love to cook with wine. Sometimes I even put it in the food.

windows binary?

I opt for going to that website which carries old programs and getting
octave 2.1.50a At least, you'll be COMPLETELY compatible with my
comments, and this is incredible, but when they changed to the 3
version, they broke all my plot functions and what they replaced it
with is SLOW!!! All to completely maintain that octave is an EXACT
Matlab clone. The group still has flurries of questions on how to
speed up plotting.

You can't believe the difference in the plot routines/commands/speed.
For example, I can plot a 3D mesh plot that is 250 by 250 data points
instantly and rotate it around with mouse control to view anything I
want. The NEW plot program starts bogging down at only 25 by 25 with
the same speed of screen updates!!! And believe me, there is more
information in a rotated plot than you can believe.

I wrote so many custom plot routines I could NOT abandon them, for
example I can plot a complex function as a line, which you can view
down its end and see real-imaginary or rotate slightly and see along
the time line to see either real, or imaginary. Plus, with the
rotation capability I can view everything, real and imaginary along
the whole time line. Very handy for phase shifting those FFTs to find
the EXACT delay even when the delay is between sample points. Like
time line resolution is in steps of 50 uS, but you can easily resolve
the delay down below 1 uS

The versioon I stopped upgrading at is small, at around 7.577MB:
octave-2.1.50a-inst.exe
self installs, just keep hitting default answers [enter] to every
question, until it's done

talk about bloating code, when they went to ver 2.9 that is 35MB !!!
I have no idea what it's at now. Probably set up to run on WinXP or
such.

The good part is that whatever you learn using octave 2.1.50a will be
applicable to any updated version, ...except the plot routines.

Jim Thompson

unread,

Jan 11, 2012, 6:02:45 PM1/11/12

to

Got a link that takes me to that 2.1.50a installer?

| E-mail Icon at http://www.analog-innovations.com | 1962 |

Robert Macy

unread,

Jan 12, 2012, 8:56:42 AM1/12/12

to

On Jan 11, 4:02 pm, Jim Thompson <To-Email-Use-The-Envelope-I...@On-My-

http://www.femm.info/Archives/bin/octave-2.1.50a-inst.exe?download

odd place?

Jim Thompson

unread,

Jan 12, 2012, 9:38:30 AM1/12/12

to

Thanks!

| E-mail Icon at http://www.analog-innovations.com | 1962 |

Robert Macy

unread,

Jan 12, 2012, 10:11:06 AM1/12/12

to

On Jan 12, 7:38 am, Jim Thompson <To-Email-Use-The-Envelope-I...@On-My-

Web-Site.com> wrote:
> On Thu, 12 Jan 2012 05:56:42 -0800 (PST), Robert Macy
>
>
>
>
>

While you're there, you should download a copy of femm 4.2
femm42bin.exe
only 1.7MB and self installs

Totally awesome finite element analyses for magnetics, electrostatics,
and heat flow.