Rise & fall time of a periodic signal

Dear side,

Yes. However, I have no idea what Cadence tool you are using to vary the capacitance nor what version of the Cadence tool  you are using. Hence, I am unable to provide any specific suggestions.

Shawn 

thanks for the reply , I am using IC 6.1.8 and sweeping the capacitance using the Tools -> Parametric sweep in ADE-L

ADE XL has also reached end-of-sale (well, you have another 10 days) so is in the same status as ADE L. You should use Explorer or Assembler as Shawn said.

The flow is the same - you create an expression and then perform the sweep - it's just the mechanism for defining the sweep is different - the sweep is part of the design or global variables in the Data View assistant in Explorer or Assembler. You can do it with ADE XL too by sweeping global variables (you cannot sweep per-test design variables with ADE XL though). Since you're using IC6.1.8 there's no good reason to start using ADE XL now, you should use Assembler instead.

Andrew. 

thank you very much for the reply Andrew  & Shawn. I will try to see Assembler and Explorer.

In the meantime can you please let suggest how can I ensure whether the parametric sweep using ADE-L is going to converge. Asking this because out of the 5 linear steps the first 4 got completed soon and the parametric sweep is now stuck at the last step for last few hours (perhaps 4hrs or more).

The progress bar shows 80% but I am not sure if it is actually able to simulating the last step. I don't see any errors in the CIW as well.

I am not sure how much more time will it take so any suggestions will be helpful.

The first step would be to look at the spectre log file rather than just the summary of parametric analysis sweep. Simulation->Output Log would show that.

There's no hope of anyone here being able to diagnose what your problem is from so little information... we don't know what the circuit does, how it is (or even if it is) failing, and so on.

You might also want to consider turning on the "Enable Diagnostic Mode" on Setup->Environment so that the spectre log file gives more information as to what is going on and why progress is slow.

Regards,

Andrew.

thanks for the reply Andrew.

The circuit involves simulating an extracted view under different capacitance loads as per my previous snapshot. The first 4 sweeps went well and after looking hard on the ADE-L window I saw the status bar read "status ready" which was earlier something like "simulating 5e-12..."

The expressions inside ADE-L also seem to change in value to reflect that the final step has been simulated but the progress bar didn't seem to go beyond 80%.

Dear sidm,

sidm said:

The expressions inside ADE-L also seem to change in value to reflect that the final step has been simulated but the progress bar didn't seem to go beyond 80%.

Andrew is correct, the spectre.out file contains far more information regarding issues encountered in simulation than the "progress bar" you note and observe. That file is far more useful in debugging issues than the latter. Another item to consider is if you are submitting this job to an LSF farm or running it interactively.

I will note, however, that I used the Parametric Tool extensively in its day and found it would occasionally not complete the parametric run when running interactively. There are a multitude of reasons this occurred, but without the spectre.out file, I can't provide any specific guidance to you.

Shawn

thanks for the reply Shawn,

in the simulation log file i see "spectre completes with 0 errors, 61 warnings, and 27 notices."

But the progress bar of the parametric sweep is still stuck at 80% as per my previous snapshot.

I've not seen reports of this happening, but anyway you should be moving to using Explorer instead which has a different sweep mechanism,

Andrew

Dear sidm,

I agree with Andrew regarding your use of Explorer or Assembler.

However, you are making debugging a bit difficult as you haven't uploaded the spectre.out file nor answered my question regarding whether you submitted this parametric analysis to your interactive machine or whether you submitted it to an LSF farm (i.e. a distributed analysis). I am wondering if you did the latter there was a communication issue between your session and the machine on which the last iteration was performed. In that case, the spectre.out file would show the simulation completed, but the job completion notice may not have been communicated properly to the Parametric tool.

Shawn 

thanks Shawn, I am not using LSF.

The log ends something like the following :

During simulation, the CPU load for active processors is :
        Spectre  0 (84.3 %)      1 (85.9 %)      2 (85.1 %)      3 (85.6 %)
                 4 (85.2 %)      5 (85.7 %)      6 (86.1 %)      7 (85.3 %)
        Other   
Initial condition solution time: CPU = 253.087 s, elapsed = 42.5773 s.
Intrinsic tran analysis time:    CPU = 16.8409 ks, elapsed = 2.82258 ks.

************ AHDL Lint Summary ************
Number of accepted steps = 14548
#Steps Type         Instance:File:Line No.
 14397 contribution I0.mipitx.R4.r1:Internal (bsource):0
Total time required for tran analysis `tran': CPU = 17.0943 ks (4h 44m 54s), elapsed = 2.86543 ks (47m  45.4s).
Time accumulated: CPU = 17.1164 ks (4h 45m 16s), elapsed = 2.89933 ks (48m  19.3s).
Peak resident memory used = 2.44 Gbytes.


Notice from spectre.
    1116 notices suppressed.
    4005637 warnings suppressed.

finalTimeOP: writing operating point information to rawfile.

Opening the PSF file ../psf/finalTimeOP.info ...

Warning from spectre during info `finalTimeOP'.
    WARNING (ASL-6233): "Internal (bsource)" 0:  I0.mipitx.R4.r1:  The value 0.0004757 has been clamped to 0.001. Set parameter 'rclamp' to change the clamped value if the value is within the permissible limit.
        

modelParameter: writing model parameter values to rawfile.

Opening the PSF file ../psf/modelParameter.info ...
element: writing instance parameter values to rawfile.

Opening the PSF file ../psf/element.info ...
outputParameter: writing output parameter values to rawfile.

Opening the PSF file ../psf/outputParameter.info ...
designParamVals: writing netlist parameters to rawfile.

Opening the PSFASCII file ../psf/designParamVals.info ...
primitives: writing primitives to rawfile.

Opening the PSFASCII file ../psf/primitives.info.primitives ...
subckts: writing subcircuits to rawfile.

Opening the PSFASCII file ../psf/subckts.info.subckts ...

~~~~~~~~~~~~~~~~~~~~~~~
Post-Simulation Summary
~~~~~~~~~~~~~~~~~~~~~~~
                Analysis        Analysis ID        CPU (s)      Elapsed (s)
         1          tran              tran        1.71e+04        2.87e+03
         2          info        finalTimeOP       1.20e+00        1.20e+00
         3          info        modelParameter    1.23e-01        1.24e-01
         4          info           element        1.36e+00        1.36e+00
         5          info        outputParameter   1.16e+00        1.16e+00
         6          info        designParamVals   1.50e-01        1.50e-01
         7          info        primitives        6.07e-02        6.18e-02
         8          info           subckts        7.87e-02        8.01e-02
------------------------------------------------------------------------------------
     Total                                        1.71e+04        2.87e+03
   Average                                        2.14e+03        3.59e+02
Elapsed Min         info        primitives        6.07e-02        6.18e-02
Elapsed Max         tran              tran        1.71e+04        2.87e+03

~~~~~~~~~~~~~~~~~~~~~~~


Aggregate audit (12:47:56 AM, Thur Dec 24, 2020):
Time used: CPU = 17.1 ks (4h 45m 23s), elapsed = 2.91 ks (48m  27.0s), util. = 589%.
Time spent in licensing: elapsed = 66.8 ms.
Peak memory used = 2.44 Gbytes.
Simulation started at: 11:59:29 PM, Wed Dec 23, 2020, ended at: 12:47:56 AM, Thur Dec 24, 2020, with elapsed time (wall clock): 2.91 ks (48m  27.0s).
spectre completes with 0 errors, 61 warnings, and 27 notices.

Dear sidm,

sidm said:

I am not using LSF.

The log ends something like the following

Thank you sidm!

From your comments, it appears my hypothesis concerning a communication issue between the Parametric tool and LSF is not correct. Based on your uploaded spectre.out file, my thought is now that the "progress bar" is not complete as ADE-L is attempting to compute your added ADE-L output expressions for transition times.

1. If you are sweeping the capacitance, I am assuming your last analysis is performed with the capacitance at its maximum value. Is this correct?

2. If it is at the maximum capacitance, are you sure the output waveform is crossing the minimum and maximum thresholds you set for computing the transition times under this maximum capacitance value?

3. Perhaps if you:

a. Disable the added ADE-L expressions for transition times and then run the Parametric analysis, does the simulation complete and the progress bar indicate 100%?

b. Examine the waveforms in ViVA to validate that all cross the minimum and maximum thresholds you set in the rise/fall time expressions.

c. Enable the transition time ADE-L expressions for plotting

d. Plot (do not re-simulate) the expressions after selecting the Parametric results by using the ADE-L plot button (beneath the "Stop" simulation button).

Shawn

Dear sidm,

sidm said:

I am not using LSF.

The log ends something like the following

Thank you sidm!

From your comments, it appears my hypothesis concerning a communication issue between the Parametric tool and LSF is not correct. Based on your uploaded spectre.out file, my thought is now that the "progress bar" is not complete as ADE-L is attempting to compute your added ADE-L output expressions for transition times.

1. If you are sweeping the capacitance, I am assuming your last analysis is performed with the capacitance at its maximum value. Is this correct?

2. If it is at the maximum capacitance, are you sure the output waveform is crossing the minimum and maximum thresholds you set for computing the transition times under this maximum capacitance value?

3. Perhaps if you:

a. Disable the added ADE-L expressions for transition times and then run the Parametric analysis, does the simulation complete and the progress bar indicate 100%?

b. Examine the waveforms in ViVA to validate that all cross the minimum and maximum thresholds you set in the rise/fall time expressions.

c. Enable the transition time ADE-L expressions for plotting

d. Plot (do not re-simulate) the expressions after selecting the Parametric results by using the ADE-L plot button (beneath the "Stop" simulation button).

Shawn

Hi Shawn,

Yes the last analysis is performed with the capacitance at its maximum value.

Yes at the maximum capacitance the output waveform is crossing the minimum and maximum thresholds.

I will try to Disable the added ADE-L expressions for transition times and then run the Parametric analysis,

Please upload the entire spectre.out file (don't just paste it, use Insert->Image/Video/File from the insert link below; you might need to rename the file with a .txt suffix), ideally running with diagnose enabled. I'd like to get an idea about how long each point in the sweep takes, and if there are any issues with the last point. The excerpt you posted above wasn't enough to determine much, and it may be that this is an update problem with ADE L, and the simulation itself ran fine. Or it might be a genuine simulation problem, and the enhanced log file might give a clue.

Thanks.

Andrew

thanks for checking this Andrew.  I am checking internally so that I can upload the .out file.

One more thing that I observe is that the operating system is CentOS , not sure if that can be a cause of the problem.

I doubt that CentOS would have anything to do with this. I can’t really see how it could, at least. 

thanks for the reply Andrew.  I tried a simpler testcase and observed that parametric simulation went through for it with the progress bar going to 100%.