But getting things like corner definitions in there is difficult. You’d have to hand-enter them. Luckily, Cadence allows you to load these definitions from a CSV file or an XML format they call SDB.

The CSV file is pretty useless. The reason is that each corner entry is another column in the file, not another row. This is the transpose of the way everyone else does CSV files—where a row is another record, and the columns are the fields in that record.

But, with some use of Python’s etree XML capabilities, you can get the SDB file to work pretty easily.

Here’s a sample script that takes a sample SDB file (which you can export from ADE-XL) and then adds multiple corners.

import xml.etree.ElementTree as ET
import itertools
import os.path

tree = ET.parse('adexl-sample-corners.sdb')
root = tree.getroot()
corners = root.find('active').find('corners')

vdds = [3.0, 3.3, 3.5]
temps = [-40, 25, 85]

model_comb = itertools.product(['slowNMOS', 'fastNMOS'], ['slowPMOS', 'fastPMOS'], ['lowResistor', 'highResistor'])

model_corners = [''.join(k) for k in model_comb]
all_combs_keys = ['modelsection:/path/to/modelfile.scs', 'var:VDD', 'var:temperature']
all_combs = itertools.product(model_corners, vdds, temps)
corner_list = [ dict( (all_combs_keys[i], v) for i,v in enumerate(c) ) for c in all_combs ]

for i,corner in enumerate(corner_list):
  if 'corner name' in corner:
    c_name = c['corner name']
  else:
    c_name = 'C{0}'.format(i)
  c = ET.SubElement(corners, 'corner', enabled="1")
  c.text = c_name
  print(c.text)

for k,v in corner.iteritems():
  if k.startswith('modelsection'):
    modelfile = k.split(':')[1]
    models = c.find('models')
  if not models:
    models = ET.SubElement(c, 'models')
    mm = ET.SubElement(models, 'model', enabled="1")
    mm.text = os.path.split(modelfile)[-1]
    mt = ET.SubElement(mm, 'modeltest')
    mt.text = 'All'
  mb = ET.SubElement(mm, 'modelblock')
  mb.text = 'Global'
  mf = ET.SubElement(mm, 'modelfile')
  mf.text = modelfile
  ms = ET.SubElement(mm, 'modelsection')
  ms.text = v
  if k.startswith('var'):
    var_name = k.split(':')[1]
    vars = c.find('vars')
    if not vars:
      vars = ET.SubElement(c, 'vars')
      var = ET.SubElement(vars, 'var')
      var.text = var_name
    val = ET.SubElement(var, 'value')
  val.text = str(v)

tree.write('generated.sdb')

The first section makes all the different corner combinations of process, voltage and temperature. As a small tangent, I also constructed all the process corners themselves (slow/fast NMOS, slow/fast PMOS, etc).

function [ S ] = csv_struct_read( file_name )
  %csv_strct_read: S = csv_struct_read( file_name )
  % Read CSV file, with headers, as a struct array
  % Header fields are struct fields. Dimension of
  % struct array is NR x 1 where NR is number of
  % non-header lines (rows) in CSV.
  A = importdata(file_name, ',', 1);
  if size(A.textdata, 1) > 1
    D = A.textdata(2:end, :);
    D(strcmp(D, '')) = num2cell(A.data);
    S = cell2struct(D, A.textdata(1,:), 2);
  else
    S = cell2struct(num2cell(A.data), A.colheaders, 2);
  end
end

Watch movie online The Transporter Refueled (2015)

Also, we got another contribution from Iou Bingyong: svnMenu.il. He says:

I’m in China and I have tried your script today . It works fine in IC615, great works!
I wrote a skill script to add the “Subversion” menu in schematic/layout editor. In this way the “menus” sub-directory is not needed and I think this should be a little easier to setup.

This is useful for debugging multiple supply domains, supply/ground bounce, and differential circuits.

procedure( relVT(netname)
  VT(netname) - VTGND
)

procedure( selVT()
  wid=hiGetCurrentWindow()
  path = geGetAdjustedPath(wid strcat(geGetInstHier(wid) "/" car(geGetSelSet())~>net~>name))
  relVT(path)
)

procedure( setGnd()
  wid=hiGetCurrentWindow()
  path = geGetAdjustedPath(wid strcat(geGetInstHier(wid) "/" car(geGetSelSet())~>net~>name))
  VTGND = VT(path)
)

procedure( plotSelVT()
  let( ((sVT selVT()))
    plot(sVT)
  )
)

hiSetBindKey("Schematics" "CtrlR" "plotSelVT()")
hiSetBindKey("Schematics" "CtrlG" "setGnd()")

Unfortunately, when you do a logarithmic sweep, you can’t cover from negative to positive frequencies–which is what you want when you generate a Nyquist plot. With the code below, you can reflect the positive frequencies onto negative to get the full Nyquist plot:


; load “~/cadence/skill/abConcatWaveforms.il”
lg = -getData(“loopGain” ?result “stb”)
; abConcatWaveforms externally defined
lg2 = abConcatWaveforms(flip(conjugate(lg)) lg)
w = newWindow()
ocnYvsYplot(?wavex real(lg2) ?wavey imag(lg2))
stb_x = xmin(abs(lg2-complex(-1.0,0.0))**2)
stb_margin = value(lg2, stb_x)
addTitle(sprintf(nil “stb_margin = %f dB” -dB20(stb_margin)))
xLimit( list(-1.6 0.2) )
yLimit( list(-0.5 0.5) )


You’ll need Andrew Beckett’s abConcatWaveforms function defined and loaded.

Incrementing index:

for(i 1 10
print(i)
)


Run through a list:

myThings = list(1 2 3 4 “a” “B” “c”)
foreach(t myThings
print(t)
)


Open a File


h = outfile("new_file.txt", "w")
h = infile("existing_file.txt")
h = outfile("partial_file.txt", "a")

Opens new file for writing, existing file for reading, and partial file for appending, respectively.

Write to a file


write("Hello World!\n", h)
fprintf(h, "gm = %f", OP("/M23" "gm"))
print(OP("/M23" "gm"), h)
pprint(OP("/M23" "gm"), h)


Write “Hello World!” to a file (newlines are explicit); construct formatted string and write to file; print using native format to a file; pretty-print (useful for nested lists etc) to a file. File handle is h.

Read from a file


str = lineread(h)


Reads a single line as a string

• Into the file “cdsLibMgr.lib” was created the menu item “SVN Unlock Cell”
• Into the file “svnLockCell.il” were created the procedures “svnUnlockCell” and “svnUnlockCellFormCB”
• Inside the procedures “svnUnlockCellFormCB” and “svnUnlockCVFormCB” were made the following changes:

• where you have: if( !(rexMatchp(“.*.svn$”, file~>readPath) || rexMatchp(“.*%”, file~>readPath))
• was changed to: if( !(rexMatchp(“.*.svn$”, file~>readPath) || rexMatchp(” “, file~>readPath))

Behaviors observed:

• We don´t want the files *.cd% under version control, but if for some reason they are under version control that exist the possibility to lock them or unlock them.
• Copying a “cell” or a “view” by “Library Manager” doesn´t copy files *.cd% which is not a problem because we don´t want these files under version control.
• The command “[user@host CDSVN]$ svn lock *” or “[user@host CDSVN/bin]$ lock directory/” locks the files *.cd%, however, the “CDSVN Unlock Cell” or “CDSVN Unlock View” does not unlock.
• The “CDSVN Lock Cell” or “CDSVN Lock View” does not unlock the files *.cd%, however, the commands “[user@host CDSVN]$ svn unlock *” or “[user@host CDSVN/bin]$ unlock directory/” are able to unlock.
• Copying by the command “[user@host CDSVN]$ svn copy” copies the files *.cd% only if they are already under version control. The same behavior occurs for “CDSVN Copy Cell” or “CDSVN Copy View”. This is not a problem.
• The command “[user@host CDSVN]$ svn add”, “CDSVN Add Cell” or “CDSVN Add View” does not add files *.cd% even because this type of file is in the list of ignored files as showed below:

[user@host /home/user/.subversion] vi config
global-ignores = *.cd% *.cd- *.cdslck *.Cat% *.abstract.status *.abstract.messages *.inca* *inca* .cdsvmod expand.cfg% transcript_ms

[user@host CDSVN/]vi cadence_ignores.txt
*.cd%
*.cd-
*.cdslck
*.Cat%
*.abstract.status
*.abstract.messages
*.inca*
*inca*
….cdsvmod
expand.cfg%
transcript_ms

Their updated files are attached. Special thanks to PCS for coordinating this. Remember, the CDSVN scripts are licensed under GPL.

cdsLibMgr.il
svnLockCell.il
cdsLibMgr.il

Also, you’ll notice that Google Ads are prominently displayed on this blog. I wanted to recoup the hosting costs of running this blog. Unfortunately, that didn’t end up being as profitable as I thought. So, as soon as I hit $100 of ad revenue (which is the minimum to cash out of my Google AdSense account), I’ll remove the ads.

Unless I change my mind. Which is unlikely.

Also, I might try out some more social features on the web site. If I get the time. Which is also unlikely.