% imports binary PSpice data from *.dat files (PSpice 6.0, 9.1 and 10.3.0)
%
% Examples
% readdat
% data = readdat
% [data text] = readdat
% [data text] = readdat(filename)
% data = readdat(filename)
% readdat(filename)
%
% If you type in just "readdat" and hit enter, the tool will ask you for a
% filename and plot the imported data, e.g. for fast review.
%
% Alternatively you can define a filename for any combination of output.
%
% If you assign the output of this function to only one variable, you'll
% receive a struct that might look as follows:
%
% data =
%
% Head: [1x1 struct] -- all the header information
% Time: [60243x1 double] -- time (not linearly spaced!)
% Data: [60243x29 single] -- all data traces
% Name: {1x29 cell} -- all node names
%
% If you assign two variables to the output, you'll get an array with all
% traces (including time) and also a cell with corresponding node names.
%
% data =
%
% 0 0 -0.0000 -0.0000 -0.0304 ...
% 0.0000 0 -0.0000 -0.0000 -0.0304 ...
% 0.0000 0 -0.0000 -0.0000 -0.0304 ...
% 0.0000 0 -0.0000 -0.0000 -0.0304 ...
% 0.0000 0 -0.0000 -0.0000 -0.0304 ...
% 0.0000 0 -0.0000 -0.0000 -0.0304 ...
% 0.0000 0 -0.0000 -0.0000 -0.0304 ...
% ... ... ... ... ...
%
% text =
%
% 'Time'
% 'V(R_free:2)'
% 'V(OUT)'
% 'V(FET)'
% 'V(CELL)'
% 'V(PATCH)'
% ...
%
% Note, that this tool is an early "beta version" and I could not check it
% with all the different versions of PSpice. At least you'll get an idea
% about the general binary *.dat-file structure to add you code or
% modifications.
%
% Suggestions, comments and discussions are welcome...
%
% by Frank Sommerhage
% (matlab@sommerhage.com)
function [data text] = readdat(filename)
%
% pSpice *.dat files are hybrids (text and binary). They can be interpret
% by a MicroSim application called PROBE. Using programs like PSpice A/D to
% open the files and export data to text files and subsequently import it
% into Matlab takes much time...
%
% This m-file imports (at least analog) data from *.dat files into Matlab.
% Also, this code is lightning-fast and returns the complete data (30 data
% traces with about 60000 points each) from a 7.5 MByte file in less than
% one second from an external HDD (USB 2.0).
%
% Also, for some reason the imported data has one data point more (the very
% first one) than the usual exported data from PSpice. I guess there is a
% tiny bug in PSpice that forgets this very first point. The rest is 100%
% identical with copy-pasted data.
%
% I worked with transient analysis only (mostly PSpice 10.3.0). If you need
% digital data or measurement results from a pSpice *.dat file, you might
% need to add code to this m-file or change it...
%
% REVISION HISTORY
% -----------------------------------------------------------------------
% DATE | NAME | JOB/TASK/MODIFICATION
% =======================================================================
% 2008-08-05 | Frank Sommerhage | initial revision (PSpice 10.3.0)
% 2008-08-07 | Frank Sommerhage | changes in readData() function to
% | | increase the import speed
% 2008-08-16 | Frank Sommerhage | added compatibility for
% | | PSpice 6.0 and PSpice 9.1 Student
% 2008-08-17 | Frank Sommerhage | minor changed in readInventory()
% 2008-09-17 | Frank Sommerhage | improved code (M-Lind)
% 2008-09-21 | Frank Sommerhage | added fclose(fid) - stupid me
% -----------------------------------------------------------------------
%
global oldpath
if nargin < 1 % if missing file-information ask for filename
% this keeps the recently used path - very useful!
if ~isvarname('oldpath') || ~isempty(oldpath)
oldpath = '';
end
% open file dialog
[file, path] = uigetfile({'*.dat', 'PSpice binary files'}, ...
'Open PSpice data file', oldpath);
% check if a file was selcted
if file == 0
return
else % save path for future use and form filename string
oldpath = path;
filename = [path file];
end
end
% open the file
f.id = fopen(filename);
% read a couple of mostly useless flags
f.flags = fread(f.id, 16, 'ubit4')';
% extract file version (I think this is some kind of file version?!?)
f.version = f.flags(15);
% files with different file versions are formated differently:
%
% *** file version 7 ***
% I've got PSpice 6.0 (for DOS), which uses version 7. That is my start. It
% is that old, it has a year 2000-bug. Version 7 is my earliest reference.
%
% *** file version 8 ***
% is used by PSpice 9.1 Student. There, they increased the precision of
% time data from single to double.
%
% *** file version 9 ***
% present in PSpice 10.3.0. The flag (coding for the file size in the
% inventory or the 'distance' to the next container with more S or A data)
% has a higher precision now (single to double). The file inventory has
% longer hex-strings to code for container positions. They also added a new
% variable to the header: "CurTime".
%
% I haven't worked with more recent versions - Sorry. Provide me a file and
% I'll try to make this m-code compatible to it...
%
% next comes a container with an inventory for data in the file
I = readInventory(f);
f.size = I.flag;
% % add inventory and file info to output struct
% data.Inventory = I;
% data.File = f;
% % (if you want to see it, remove the %-sign
% go ahead with import of header container
data.Head = readHeader(f, I.H);
% move into node name import
data.Name = readNodes(f, I.N);
% import the time line (very important - time is not spaced linearly!!!)
data.Time = readData(f, I.S, 'S', ...
data.Head.AnaRows);
% continue on data import
data.Data = readData(f, I.A, 'A', ...
data.Head.AnaRows, ...
data.Head.AnaCols-1);
% close file
f.id = fclose(f.id);
% the following code checks what you want and formats the data accordingly
switch nargout
case 2 % return array with traces and a cell with node names
text = data.Name';
data = [data.Time, data.Data];
case 1 % return 1 struct with full file information
data.Name(1) = []; % time is extra, therefore delete its node name
otherwise % plot data and return nothing to workspace
plot(data.Time, data.Data);
legend(data.Name(2:end));
clear data
end
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% SUBFUNCTIONS %
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% =========================================================================
function I = readInventory(f)
%
% If you open *.dat files with a (hex-)editor, you'll see something like
% this (e.g. PSpice 10.3; older PSpice versions have shorter hex-strings):
%
% i=0x0000000000000000 -I've got no idea?!?
% H=0x00000000000000DF position of the header information
% h=0x0000000000000000 -IDK?!?
% N=0x00000000000002A1 position of node information
% n=0x0000000000000000 -no idea?!?
% S=0x00000000000002FB position of start parameters
% s=0x0000000000000000 -nope?!?
% A=0x0000000000000358 position of analog data
% D=0x0000000000000000 position of digital data (I suppose)
% M=0x0000000000000000 position of measurement results (?)
%
% This function reads the inventory and converts container positions (given
% as hex) into decimal numbers.
%
I = readContainerHead(f, 0, 'I'); % read inventory header
chs = 8 + 4*(f.version>8); % adjust container head size
temp = fread(f.id, I.size-chs-1, '*char')'; % read inventory information
temp = reshape(temp, [], 10)'; % reformat inventory string
for i = 1:10 % 10 entries (see above)
I.(temp(i,1)) = hex2dec(temp(i,5:end)); % convert hex, add to struct
end
% =========================================================================
function H = readHeader(f, pos)
%
% This is an example for header information. It is stored in *.dat files
% as a list of strings:
%
% AllCol=0
% AnaCols=2
% Analysis=Transient Analysis
% AnaRows=0x00009974
% CirName=* Profile name *
% CirSub=
% Complex=1
% DataType=0
% Date=06/18/07
% DigCols=0
% DigRows=0x00000000
% EndSweep=000000000000000000000000000000000000000000000.89
% Param1=Time
% Param2=
% ProgID=<your serial number>
% RevNo=10.3
% StartSweep=0
% SweepMode=1
% Temp=27
% Time=16:53:11
% CurTime=000000000000000000000000000000000000000000000000
%
% Some of the data is stored as hex, some as dec, there are strings and
% dates. I tried to parse everything. If you have additional information
% in your files, you might want to add some code here to parse the header
% accordingly.
%
head = readContainerHead(f, pos, 'H'); % read container info
chs = 8 + 4*(f.version>8); % container head size
temp = fread(f.id, head.size-chs-2, '*char')'; % read header information
while ~isempty(temp) % until end of container
[token temp] = strtok(temp, char(0)); % read a pstring
spos = strfind(token, '='); % find the = sign in string
name = token(1:spos-1); % name is afore '='
value = token(spos+1:end); % value comes past '='
if isempty(value)
value = []; % delete empty strings ('')
else
number = str2num(value); % try to convert to number
if isempty(number)
if length(value) > 1 && strcmp(value(1:2), '0x') % check if hex data
value = hex2dec(value(3:end)); % convert hex to decimal
end
elseif strfind(name, 'Time') % do not convert to number
elseif strfind(name, 'Date') % do not convert to number
if length(value) > 8
value(7) = []; % correct year 2000 error
end
elseif strfind(name, 'ProgID') % do not convert to number
elseif strfind(name, 'RevNo') % do not convert to number
else
value = number; % or use number as value
end
end
H.(name) = value; % add value to struct
end
% =========================================================================
function N = readNodes(f, pos)
%
% Reads the text information with node names for each data trace. I didn't
% need the additional node information. Here some examples:
%
% Time
% V(OUT);V1(R_lowpass);V2(C_lowpass);V(OUT);V(C_lowpass:2);V(R_lowpass:1)
% V(CELL);V(R_series:2);V(R_free:1);V(R_attached:2);V(C_att ...
% I(V_stm);I(V_stm:+)
% I(R_lowpass);I(R_lowpass:1)
% ...
%
% Each line represents one data trace and the additional information is
% useless at this point - at least to me. That is why I cut the names to
% their short form - as you would see them in PSpice. If you need this
% further information, just remove the part between the stars below.
%
head = readContainerHead(f, pos, 'N'); % read container info
chs = 8 + 4*(f.version>8); % container head size
temp = fread(f.id, head.size-chs-2, '*char')'; % read header information
N = {}; % start names as empty cell
while ~isempty(temp) % until end of container
[token temp] = strtok(temp, char(0)); % read a pstring
spos = strfind(token, ';'); % find a ';' (see above)
if isempty(spos) % take it (e.g. 'Time')
name = token;
else % ************* % delete the code between
name = token(1:spos(1)-1); % the stars to get the full
end % ************* % name information...
N{end+1} = name; % add name to cell
end
% =========================================================================
function data = slowreadData(f, pos, type, r, c)
%
% This is an older version of the data import tool. It uses a while loop to
% read its way through the file. You might want to use it, if you need to
% implement reading for digital data or measurement results. Actually, I am
% not sure, whether it is still working...?!?
%
if nargin < 5, c = 1; end % see next function below...
head = readContainerHead(f, pos, type);
switch type
case 'A', precision = 'single'; psize = 4;
case 'S', precision = 'double'; psize = 8;
otherwise, return
end
data = zeros(r, c); % create data array
p = 0; % start at pos 0 in array
z = 4 + 4*(f.version>8) + 4, % container head size
k = (f.version>8) % file version corrections
fseek(f.id, -(z+k), 0); % file-pntr to this container
% fread(f.id, 30, '*char')', return
tic % get current time
while head.flag % until no more containers
head = readContainerHead(f, ftell(f.id)-8-1, type); % current flag info
n = (head.size-z)/psize; % # of data in each container
% read and reshape data in container, and store in data array
data(p+1:p+n/c,:) = reshape(fread(f.id, n, precision)', c, [])';
p = p+n/c; % increase array pointer
fseek(f.id, head.flag(1)-head.size-2*~k, 0); % jump to next container
if toc > 3 && mod(p, 2500) % display waitbar if ...
waitbar(p/r); % process takes more than 3
end % seconds
% fread(f.id, 30, '*char')', return
end
data = data(1:r,1:c); % shorten to rows & columns
% =========================================================================
%
% Data is written in various ways. The data is not continuous and appears
% in containers (luckily with constant size - usually 10 points/trace).
% Time is not linear: check yourself with diff(time) or cumsum(time) !!!
% I don't know, what digital data or measurement result would look like in
% the file, since I never used it. Therefore D and M containers cannot be
% imported with this code (you would have to add the necessary code below).
%
% This part of the m-code makes this tool so fast. The skip-option in fread
% makes much slower while- or for-loops obsolete. Thank you Mathworks!
%
function data = readData(f, pos, type, r, c)
if nargin < 5, c = 1; end % no column? --> make it 1
head = readContainerHead(f, pos, type); % read container head
switch type
case 'A', precision = 'single'; psize = 4;% analog data: always single
case 'S'
if f.version > 7
precision = 'double'; psize = 8; % since ver 8, time is double
else
precision = 'single'; psize = 4; % older versens are single
end
% case 'D', precision = ''; psize = NaN; % I don't know the parameters
% case 'M', precision = ''; psize = NaN; % I don't know the parameters
otherwise, return % do not import
end
h = 8 + 4*(f.version>8); % correct for file version
skip = head.flag(1)-head.size+h; % skipsize between containers
n = (head.size-h)/psize; % (usually n=10/c) % # of data in each container
precision = [num2str(n) '*' precision '=>' precision]; % precision string
data = fread(f.id, r*c, precision, skip)'; % read, skip, read, skip, ...
data = reshape(data, c, [])'; % reorganize data for columns
% =========================================================================
function head = readContainerHead(f, pos, verify)
%
% Some examples for container heads in a *.dat file. They start with a
% character to determine the type of the container. Here's an example of
% a few heads in my files (in a hex/text-editor)
%
% 'I ' <-- I for inventory
% 'H ' <-- H for header
% 'N Z ' <-- N for node names
% 'S / ' <-- S for time (in my case!)
% 'A 4 ' <-- A for analog data
%
% The char (e.g. 'H') is followed by a int64 flag. In the inventory, this
% number represents the total filesize in I and in A and S containers it
% represents the distance to the next container with A and S information in
% it (what a waist of disk space). The last part codes for the current
% container's size as int32.
%
if nargin < 3, verify = []; end % no verify? --> turn it off
fseek(f.id, 8+pos, -1); % move file pointer to pos+8(!)
char = fread(f.id, 1, '*char')'; % verification character
head.pos = pos; % keep position for later
if f.version > 8
head.flag = fread(f.id, 1, 'int64')'; % filesize / next container pos
else
head.flag = fread(f.id, 1, 'int32')'; % filesize / next container pos
end
head.size = fread(f.id, 1, 'int32')'; % container size
if char ~= verify % stop if wrong container
error(['Wanted ' verify ' got ' char ' at ' num2str(pos+8) '!']);
end
