Meaning of "User error: PWL source's repeat value (R) must be >= 0 and < last value in Time-Voltage list"

174 views
Skip to first unread message

Roshni Uppala

unread,
May 6, 2015, 10:23:02 AM5/6/15
to xyce-...@googlegroups.com
What is a Time-Voltage list? I am using Xyce 6.2 on a cluster of computing nodes.

Thanks,
Roshni

xyce-users

unread,
May 6, 2015, 2:42:50 PM5/6/15
to xyce-...@googlegroups.com
Roshni,

The Xyce 6.2 (and soon to be released Xyce 6.3) Reference Guides are indeed terse about the syntax for the Piecewise Linear (PWL) source.  The PWL source syntax is mentioned in the Xyce Reference Guide sections on the V and I sources, but the explanations could be better.

An example for this group would be as follows. The time-voltage list in source VPWL is simply the pairs within the ( )'s.  So, the entire waveform is played out once between t=1 and t=11 seconds, since TD=1 and 10S is the last value in the time-voltage list.  This specification then starts repeating the waveform definition given between t=7 (which is set by R=7) and t=10 (which is the last point in the time-voltage list) until the end of the simulation.

Repeat whole waveform, with time delay of one seconds and R=7 seconds
VPWL 1 0 PWL(0S 0V  2S 3V  3S 2V  4S 2V  4.01V 5V  4.5S 5V
+ 4.51S -2V  7S 1V  9S -1V  9.01S 4V  10S 3V) TD=1 R=7

The error message you cite ("User error: PWL source's repeat value (R) must be >= 0 and < last value in  Time-Voltage list") is an attempt to catch a PWL definition that won't work.  That error message is triggered if either the repeat value (R) is < 0 or if R is >= the time in the last time-voltage pair.  So, for the VPWL example above, R must be >=0 and < 10, since 10s is the last time-value in the source definition for VPWL.   Another note is that if R is omitted on the PWL instance line then R defaults to 0.

These requirements on the Repeat (R) parameter are discussed in the text after the Table on the Piecewise Linear Parameters in the V and I device sections of the Xyce Reference Guide.  However, no examples are given there. Some working examples for the Xyce PWL source can be found in the Xyce Regression Test suite under Xyce_Regression/Netlists/VPWL 

--The Xyce Team

Roshni Uppala

unread,
May 7, 2015, 8:17:08 AM5/7/15
to xyce-users
Thanks.
I am using a PWL file instead in this way, Vrowt1 Nvrowt1 0 PWL FILE "write_row1.txt" where
write_row1.txt is as follows:

0.000000000000  0.0000
0.000000010000  0.0000
0.000000015000  1.0000
0.000000065000  1.0000
0.000000070000  0.0000
0.000000075000  0.0000

Best regards,
Roshni

Roshni Uppala
Graduate Student- Electrical & Computer Engineering
University of Dayton

--
You received this message because you are subscribed to a topic in the Google Groups "xyce-users" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/xyce-users/tyv_xi2fXVw/unsubscribe.
To unsubscribe from this group and all its topics, send an email to xyce-users+...@googlegroups.com.
To post to this group, send email to xyce-...@googlegroups.com.
Visit this group at http://groups.google.com/group/xyce-users.
For more options, visit https://groups.google.com/d/optout.

xyce-users

unread,
May 7, 2015, 6:24:38 PM5/7/15
to xyce-...@googlegroups.com, roshni...@gmail.com
Roshni,

We're not sure what this issue is then.  Do you have a simple netlist that produces the error message you cited ("User error: PWL source's repeat value (R) must be >= 0 and < last value in Time-Voltage list")?  What tool did you use to make write_row1.txt?  There are sometimes issues with files made on Windows (and especially with ones made in MSWord), but I thought you used Linux. 

This simple netlist works if the text you cite for write_row1.txt is placed in that file.  Does it work for you?

   Simple VPWL Test Circuit

   .tran 1e-9 1e-6

   Vrowt1 Nvrowt1 0 PWL FILE "write_row1.txt"

   R1 Nvrowt1 0 1K
  .PRINT TRAN V(Nvrowt1)
  .end

--The Xyce Team


On Thursday, May 7, 2015 at 6:17:08 AM UTC-6, Roshni Uppala wrote:
Thanks.
I am using a PWL file instead in this way, Vrowt1 Nvrowt1 0 PWL FILE "write_row1.txt" where
write_row1.txt is as follows:

0.000000000000  0.0000
0.000000010000  0.0000
0.000000015000  1.0000
0.000000065000  1.0000
0.000000070000  0.0000
0.000000075000  0.0000

Best regards,
Roshni

Roshni Uppala
Graduate Student- Electrical & Computer Engineering
University of Dayton

Roshni Uppala

unread,
May 11, 2015, 6:46:54 PM5/11/15
to xyce-...@googlegroups.com, xyce-...@googlegroups.com
Yes. Your test script runs on the cluster.

 I am using linux system and had transferred files to the cluster using the FileZilla. Many times during that transfer, files got converted to a non-unix version. So yes you are right.

However, I tried to convert all the files and tried to run my script; it gives the same error again. Maybe the netlist has some error.

xyce-users

unread,
May 12, 2015, 4:49:33 PM5/12/15
to xyce-...@googlegroups.com, roshni...@gmail.com
Roshni,

Your netlist may indeed have an error in it, or maybe your netlist triggered some latent bug in how the Xyce parser handles PWL sources.  It's hard to say without more details.  So, our questions are:

  Can you share your netlist?  (Our memory says "no", since they tend to be for large cross-bar networks related to your research.) 
  Can you make the error occur in a smaller netlist? 
  Does your netlist run in some other SPICE variant, which supports a file-based PWL syntax?

--The Xyce Team

Roshni Uppala

unread,
May 12, 2015, 6:20:39 PM5/12/15
to xyce-users
Hi,

The netlist actually ran successfully in my laptop using Xyce. Also, a small test script (the one you shared regarding PWL) runs fine on the cluster that I am using right now.
I am actually trying to break the netlist down into a simpler one. Probably we can go from there? I will let you know soon.

Thanks for your help.

-Roshni

Roshni Uppala
Graduate Student- Electrical & Computer Engineering
University of Dayton

xyce-users

unread,
May 14, 2015, 5:49:18 PM5/14/15
to xyce-...@googlegroups.com, roshni...@gmail.com
Roshni,

If the netlist runs fine on your laptop but not on the cluster then it may be a difference in the file formatting on those two systems, or caused by characters added during the file transfer between the two systems.  What OS is each system using?  Also have you verified that the netlist files and the files specifying the PWL source data are exactly identical on your laptop and the cluster?

We have seen cases where (for example), a netlist made in a Microsoft editor will cause errors during Xyce netlist parsing.  However, it typically causes a LOT of errors on multiple lines in a large netlist.  This reported bug is "interesting" because those file-formatting differences don't typically cause such a specific error in the parsing of a single device instance.

--The Xyce Team


On Tuesday, May 12, 2015 at 4:20:39 PM UTC-6, Roshni Uppala wrote:
Hi,

The netlist actually ran successfully in my laptop using Xyce. Also, a small test script (the one you shared regarding PWL) runs fine on the cluster that I am using right now.
I am actually trying to break the netlist down into a simpler one. Probably we can go from there? I will let you know soon.

Thanks for your help.

-Roshni

Roshni Uppala
Graduate Student- Electrical & Computer Engineering
University of Dayton

To unsubscribe from this group and all its topics, send an email to xyce-users+unsubscribe@googlegroups.com.

Roshni Uppala

unread,
May 15, 2015, 10:48:48 AM5/15/15
to xyce-users
Hi,

The netlist runs fine on my laptop, however I use Xyce 6.1 version on my laptop. The version of Xyce that I installed on cluster is 6.2. I am going to see if the version of xyce is creating an issue by installing Xyce 6.2 on my laptop.

Yes, I am aware of the file formatting issues. For that reason, I make sure to convert all txt files or pwl files to unix version before running on the cluster. Moreover, no microsoft editors are used. Also, yes I have checked the files; they are identical.

Also, the OS on my laptop is CentOS 6 and the cluster is red hat linux 6.5. I was able to install Xyce 6.2 successfully on the cluster and it passed all the tests from the regression suite.

Actually, I will get back to you with a smaller sized netlist code. I am checking if that runs correctly on my laptop first. Hope this might help?

Thanks,
Roshni

Roshni Uppala
Graduate Student- Electrical & Computer Engineering
University of Dayton

To unsubscribe from this group and all its topics, send an email to xyce-users+...@googlegroups.com.

To post to this group, send email to xyce-...@googlegroups.com.
Visit this group at http://groups.google.com/group/xyce-users.
For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to a topic in the Google Groups "xyce-users" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/xyce-users/tyv_xi2fXVw/unsubscribe.
To unsubscribe from this group and all its topics, send an email to xyce-users+...@googlegroups.com.

xyce-users

unread,
May 18, 2015, 12:21:11 PM5/18/15
to xyce-...@googlegroups.com, roshni...@gmail.com
After off-line communication with Roshni, we have determined that the issue is that Xyce 6.2 (which is the first version to support the "repeat" feature on PWL sources) requires more than one point in the piecewise linear time-voltage list.  In the cases where Roshni was seeing this error message, she had been using a file of time-voltage pairs that contained only a single pair.  This breaks the repeat function's validation logic, which is implicitly assuming at least two time-voltage pairs (i.e. something that defines at least one linear segment in time/voltage space).

If one wants to create a "piecewise linear" source that is actually just a constant value, one needs to have at least two time-voltage pairs --- the constant value is assured by making both pairs have the same voltage, but different times.

Thank you for pointing out that there is now an issue with this feature.
To unsubscribe from this group and all its topics, send an email to xyce-users+unsubscribe@googlegroups.com.

To post to this group, send email to xyce-...@googlegroups.com.
Visit this group at http://groups.google.com/group/xyce-users.
For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to a topic in the Google Groups "xyce-users" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/xyce-users/tyv_xi2fXVw/unsubscribe.
To unsubscribe from this group and all its topics, send an email to xyce-users+unsubscribe@googlegroups.com.
Reply all
Reply to author
Forward
0 new messages