ngspice / Bugs / #292 nint(x) works in independent sources and expressions but not recognised in B sources.

marcel hendrix - 2014-09-18

The ngspice-26 manual does not list nint(x) as a supported function, and
mentions that nint(x) is allowed in param statements. Therefore the
signaled behavior is formally not a B-source bug.

I think numparm is to blame, as:

::html B1 NINT_FUNCTION1 GND V=NINT(2) B2 NINT_FUNCTION2 GND V=NINT(boo) B3 NINT_FUNCTION3 GND V=NINT(V(in))

should have worked for B1 and B2 ( B1 <- 2, B2 <- 2.5).
Reason: numparm does not substitute the NINT(<x>) strings. </x>
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
- Andy Fierman - 2014-09-18
  
  nint(x) is defined in the table on page 57 of:
  
  http://ngspice.sourceforge.net/docs/ngspice-manual.pdf
  
  However, it is not clear to me exactly what this table is defining: it has
  no title and is not clearly referenced in the text.
  
  This is the only place that nint(x) is described.
  
  Does this table refer specifically to functions that can be used only in
  .param statements?
  
  If so, can they also be used in expressions for independent sources and
  other component values like:
  
  R1 net1 gnd {nint(5.6)}
  
  V1 net2 gnd sin(0 1 {nint(2.5k)})
  
  Coming from using LTspice, I'd expected to find that if a function is
  defined then it would work in any context.
  
  It would certainly make the user's life much simpler (but not the
  developers!).
  
  :)
  
  Andy.
  
  signality.co.uk
  
  On 18 September 2014 18:31, marcel hendrix mhx_at_sf@users.sf.net wrote:
  
  The ngspice-26 manual does not list nint(x) as a supported function, and
  mentions that nint(x) is allowed in param statements. Therefore the
  signaled behavior is formally not a B-source bug.
  
  I think numparm is to blame, as:
  
  B1 NINT_FUNCTION1 GND V=NINT(2)
  B2 NINT_FUNCTION2 GND V=NINT(boo)
  B3 NINT_FUNCTION3 GND V=NINT(V(in))
  
  should have worked for B1 and B2 ( B1 <- 2, B2 <- 2.5).
  Reason: numparm does not substitute the NINT(<x>) strings.</x>
  
  [bugs:#292] http://sourceforge.net/p/ngspice/bugs/292 nint(x) works in
  independent sources and expressions but not recognised in B sources.*
  
  Status: open
  Group: v1.0 (example)
  Created: Thu Sep 18, 2014 12:38 PM UTC by Andy Fierman
  Last Updated: Thu Sep 18, 2014 12:38 PM UTC
  Owner: nobody
  
  In Ngspice 26 from git, it looks like the NINT(x) function (Nearest
  integer, half integers rounded towards even) function is not recognised
  when used in dependent (B) sources.
  
  The netlist below illustrates the problem:
  
  ngspice nint(x) test jig
  
  .param boo=-2.5
  + ref={NINT(2.5)}
  
  V1 IN GND {0.5} SIN(0 4.1 1k)
  V2 REF1 GND {ref}
  V3 REF2 GND {NINT(boo)}
  
  Uncommenting any of the 3 lines below generates an error report:
  
  B1 NINT_FUNCTION1 GND V=NINT(2) B2 NINT_FUNCTION2 GND V=NINT(boo)
  *B3 NINT_FUNCTION3 GND V=NINT(V(in))
  
  .TRAN 1u 1m
  *.OP
  
  probe V(in) V(ref1) V(ref2) ; V(nint_function1) V(nint_function2)
  V(nint_function3)
  
  .END
  
  The netlist above runs without errors if the B1 - B3 lines are commented
  out. Running the netlist with B1, B2 and B3 lines uncommented produces an
  error report containing this information:
  
  ==============================
  :
  Error: no such function 'nint'
  Error: no such function 'nint'
  Error: no such function 'nint'
  Error on line 24 : b1 nint_function1 0 v= nint ( 2.0000000000e+00 )
  parameter value out of range or the wrong type
  Error on line 25 : b2 nint_function2 0 v= nint ( -2.50000000000000000e+00 )
  parameter value out of range or the wrong type
  Error on line 26 : b3 nint_function3 0 v= nint ( v(in) )
  parameter value out of range or the wrong type
  :
  ===============================
  
  Many thanks for your attention,
  
  Andy.
  
  signality.co.uk
  
  Sent from sourceforge.net because you indicated interest in
  https://sourceforge.net/p/ngspice/bugs/292/
  
  To unsubscribe from further messages, please visit
  https://sourceforge.net/auth/subscriptions/
  
  Related
  
  Bugs: ~~#292~~
  
  If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
  - Andy Fierman - 2014-09-19
    
    "The ngspice-26 manual does not list nint(x) as a supported function,"
    
    Not sure that's true now: it was announced as a new feature in the ngspice
    release 26, January 12th, 2014 and is in thesame table on page 57 in the
    stable released manual:
    
    http://ngspice.sourceforge.net/docs/ngspice26-manual.pdf
    
    Cheers,
    
    Andy.
    
    signality.co.uk
    
    On 18 September 2014 22:33, Andy Fierman andyfierman@signality.co.uk
    wrote:
    
    nint(x) is defined in the table on page 57 of:
    
    http://ngspice.sourceforge.net/docs/ngspice-manual.pdf
    
    However, it is not clear to me exactly what this table is defining: it has
    no title and is not clearly referenced in the text.
    
    This is the only place that nint(x) is described.
    
    Does this table refer specifically to functions that can be used only in
    .param statements?
    
    If so, can they also be used in expressions for independent sources and
    other component values like:
    
    R1 net1 gnd {nint(5.6)}
    
    V1 net2 gnd sin(0 1 {nint(2.5k)})
    
    Coming from using LTspice, I'd expected to find that if a function is
    defined then it would work in any context.
    
    It would certainly make the user's life much simpler (but not the
    developers!).
    
    :)
    
    Andy.
    
    signality.co.uk
    
    On 18 September 2014 18:31, marcel hendrix mhx_at_sf@users.sf.net wrote:
    
    The ngspice-26 manual does not list nint(x) as a supported function, and
    mentions that nint(x) is allowed in param statements. Therefore the
    signaled behavior is formally not a B-source bug.
    
    I think numparm is to blame, as:
    
    B1 NINT_FUNCTION1 GND V=NINT(2)
    B2 NINT_FUNCTION2 GND V=NINT(boo)
    B3 NINT_FUNCTION3 GND V=NINT(V(in))
    
    should have worked for B1 and B2 ( B1 <- 2, B2 <- 2.5).
    Reason: numparm does not substitute the NINT(<x>) strings.</x>
    
    [bugs:#292] http://sourceforge.net/p/ngspice/bugs/292 nint(x) works
    in independent sources and expressions but not recognised in B sources.*
    
    Status: open
    Group: v1.0 (example)
    Created: Thu Sep 18, 2014 12:38 PM UTC by Andy Fierman
    Last Updated: Thu Sep 18, 2014 12:38 PM UTC
    Owner: nobody
    
    In Ngspice 26 from git, it looks like the NINT(x) function (Nearest
    integer, half integers rounded towards even) function is not recognised
    when used in dependent (B) sources.
    
    The netlist below illustrates the problem:
    
    ngspice nint(x) test jig
    
    .param boo=-2.5
    + ref={NINT(2.5)}
    
    V1 IN GND {0.5} SIN(0 4.1 1k)
    V2 REF1 GND {ref}
    V3 REF2 GND {NINT(boo)}
    
    Uncommenting any of the 3 lines below generates an error report:
    
    B1 NINT_FUNCTION1 GND V=NINT(2) B2 NINT_FUNCTION2 GND V=NINT(boo)
    *B3 NINT_FUNCTION3 GND V=NINT(V(in))
    
    .TRAN 1u 1m
    *.OP
    
    probe V(in) V(ref1) V(ref2) ; V(nint_function1) V(nint_function2)
    V(nint_function3)
    
    .END
    
    The netlist above runs without errors if the B1 - B3 lines are commented
    out. Running the netlist with B1, B2 and B3 lines uncommented produces an
    error report containing this information:
    
    ==============================
    :
    Error: no such function 'nint'
    Error: no such function 'nint'
    Error: no such function 'nint'
    Error on line 24 : b1 nint_function1 0 v= nint ( 2.0000000000e+00 )
    parameter value out of range or the wrong type
    Error on line 25 : b2 nint_function2 0 v= nint ( -2.50000000000000000e+00
    )
    parameter value out of range or the wrong type
    Error on line 26 : b3 nint_function3 0 v= nint ( v(in) )
    parameter value out of range or the wrong type
    :
    ===============================
    
    Many thanks for your attention,
    
    Andy.
    
    signality.co.uk
    
    Sent from sourceforge.net because you indicated interest in
    https://sourceforge.net/p/ngspice/bugs/292/
    
    To unsubscribe from further messages, please visit
    https://sourceforge.net/auth/subscriptions/
    
    Related
    
    Bugs: ~~#292~~
    
    If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
    - marcel hendrix - 2014-09-19
      
      "The ngspice-26 manual does not list nint(x) as a supported function,"
      
      Not sure that's true now: it was announced as a new feature in the ngspice
      release 26, January 12th, 2014 and is in the same table on page 57 in the
      stable released manual:
      
      http://ngspice.sourceforge.net/docs/ngspice26-manual.pdf
      
      I looked at the release notes, and nint is not for B-sources (note the
      different formulation for pwr). That this is clear to me is probably
      because I am (by now) well aware of the 3 different ngspice parsers.
      (Yes, I believe that they should be transparent.)
      
      I recommend to retract the bug report for the B-source parser and
      file one for the bug in numparam instead.
      
      -marcel
      
      If you would like to refer to this comment somewhere else in this project, copy and paste the following link:

Robert Larice - 2014-09-19

Hello Andy, Marcel,

Its not a numparm bug, but a missing implementation in the B parser.

Some time ago, I've put some effort into the numparm parser
to make some details compatibel to another spice.
Thats somewhat expressed in the regression test
tests/regression/parser/xpressn-1.cir
For example nint() was missing, got implemented, and was then tested there.

Concerning this (numparm) section, I currently know of the various
pwr pwrs ^ and ** operators, which need to be polished too.

I've not cared about the B world at that time.

Thus, indeed its a bug/missing feature
and a similiar regression .cir file needs to be written for B too
(and for .control of course)

Robert

Last edit: Robert Larice 2014-09-19

If you would like to refer to this comment somewhere else in this project, copy and paste the following link:

Robert Larice - 2014-09-20

Hello Andy,

I've pushed commits to implement nint for 'B' and for '.control'
A regression test is pushed as well.

Thanks for the report.
Robert

If you would like to refer to this comment somewhere else in this project, copy and paste the following link:

Robert Larice - 2014-09-20

status: open --> closed-fixed
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
- Andy Fierman - 2014-09-20
  
  Robert,
  
  Many thanks!
  
  Deep joy.
  
  Cheers,
  
  Andy.
  
  signality.co.uk
  
  On 20 September 2014 16:34, Robert Larice rlar@users.sf.net wrote:
  
  status: open --> closed-fixed
  
  [bugs:#292] http://sourceforge.net/p/ngspice/bugs/292 nint(x) works in
  independent sources and expressions but not recognised in B sources.*
  
  Status: closed-fixed
  Group: v1.0 (example)
  Created: Thu Sep 18, 2014 12:38 PM UTC by Andy Fierman
  Last Updated: Sat Sep 20, 2014 03:33 PM UTC
  Owner: nobody
  
  In Ngspice 26 from git, it looks like the NINT(x) function (Nearest
  integer, half integers rounded towards even) function is not recognised
  when used in dependent (B) sources.
  
  The netlist below illustrates the problem:
  
  ngspice nint(x) test jig
  
  .param boo=-2.5
  + ref={NINT(2.5)}
  
  V1 IN GND {0.5} SIN(0 4.1 1k)
  V2 REF1 GND {ref}
  V3 REF2 GND {NINT(boo)}
  
  Uncommenting any of the 3 lines below generates an error report:
  
  B1 NINT_FUNCTION1 GND V=NINT(2) B2 NINT_FUNCTION2 GND V=NINT(boo)
  *B3 NINT_FUNCTION3 GND V=NINT(V(in))
  
  .TRAN 1u 1m
  *.OP
  
  probe V(in) V(ref1) V(ref2) ; V(nint_function1) V(nint_function2)
  V(nint_function3)
  
  .END
  
  The netlist above runs without errors if the B1 - B3 lines are commented
  out. Running the netlist with B1, B2 and B3 lines uncommented produces an
  error report containing this information:
  
  ==============================
  :
  Error: no such function 'nint'
  Error: no such function 'nint'
  Error: no such function 'nint'
  Error on line 24 : b1 nint_function1 0 v= nint ( 2.0000000000e+00 )
  parameter value out of range or the wrong type
  Error on line 25 : b2 nint_function2 0 v= nint ( -2.50000000000000000e+00 )
  parameter value out of range or the wrong type
  Error on line 26 : b3 nint_function3 0 v= nint ( v(in) )
  parameter value out of range or the wrong type
  :
  ===============================
  
  Many thanks for your attention,
  
  Andy.
  
  signality.co.uk
  
  Sent from sourceforge.net because you indicated interest in
  https://sourceforge.net/p/ngspice/bugs/292/
  
  To unsubscribe from further messages, please visit
  https://sourceforge.net/auth/subscriptions/
  
  Related
  
  Bugs: ~~#292~~
  
  If you would like to refer to this comment somewhere else in this project, copy and paste the following link:

Andy Fierman - 2014-12-19

Hi,

If I run the netlist below in ngpsice.com (release 26) then the tilde in:

ainv1 ~ind outd buff

inverts the input OK.

If I remove the tilde so that line becomes:

ainv1 ind outd buff

and insert one into:

aoutbridge1 [~outd] [out] dac

then that inverts the input OK too.

However, if I run this in EasyEDA using ngspice from git then the tilde is
ignored in either location.

Is this a bug in ngspice from git?

This is the example shared on ngspice.com:

http://www.ngspice.com/index.php?public_circuit=CFHOrs

and on EasyEDA:

http://easyeda.com/editor#id=QQ20mH31Z

Cheers,

Andy.

signality.co.uk

tilde test
.tran 1.2u 1.2m
.param

tdlo = 20n

tdhi = 50n
**

in_low = 0.1

in_high = 0.9

out_low = 0.1

out_high = 0.9

out_undef = 0.5
**
VCC1 vcc 0 1
V1 in 0 pulse(0 1 0 10n 10n 500u 1m)
*
ainbridge1 [in] [ind] adc1
.model adc1 adc_bridge(in_low = {in_low} in_high = {in_high})
*
ainv1 ~ind outd buff ; inversion using [~ind] works OK
*
.model buff d_buffer(rise_delay = {tdhi} fall_delay = {tdlo}

input_load = 0.5e-12)
*
aoutbridge1 [outd] [out] dac ; inversion using [~outd] works OK here too
*
.model dac dac_bridge(out_low = {out_low} out_high = {out_high} out_undef =
{out_undef}

input_load = 5.0e-12 t_rise = 50e-9

t_fall = 20e-9)
*
*.probe v(in) v(out)
*
.end
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:

nint(x) works in independent sources and expressions but not recognised in B...

Group

Searches

Help

#292 nint(x) works in independent sources and expressions but not recognised in B sources.

Related

Discussion

The netlist below illustrates the problem:

.END

signality.co.uk

Related

The netlist below illustrates the problem:

.END

signality.co.uk

Related

The netlist below illustrates the problem:

.END

signality.co.uk

Related