PWM resolves #4089 and #4090 and #4839 by AHaumer · Pull Request #4812 · modelica/ModelicaStandardLibrary

AHaumer · 2026-09-11T16:21:56Z

Fixes the bugs adressed in #4089 and #4090 and refines the example for testing.

christiankral

I do understand the changes in line 30-31 and 42-43 of Electrical/PowerConverters/DCAC/Control/IntersectivePWM.mo‎

Unfortunately I cannot understand why the quantities and the units of the some variables have been changed, as this is a non backwards compatible change. Shouldn't there be a solution the same units ...

AHaumer · 2026-11-01T11:31:49Z

Thanks a lot for your thorough review @christiankral

I have to admit I don't understand your remark "why the quantities and the units of the some variables have been changed".
Which quantities and units have been changed?

Regarding the changes amplitude=uMax -> 2 and offset=-uMax/2 -> -1: There is a new gain block diving the signal by uMax.
This is not equivalent! I have to rethink this whether the new or the old version is correct.

Regarding the comparions SVPWM and IntersectivePWM: I have to dig into this.

into PWM

AHaumer · 2026-11-02T10:42:33Z

After enhancing the documentation including clarification of the voltages, I'll check and adapt:

Electrical.PowerConverters.DCAC.Control.IntersectivePWM
Electrical.PowerConverters.DCAC.Control.SVPWM
take care of Bugfix Electrical.PowerConverters.DCAC.Control.SVPWM #4840
as well as these two examples
Electrical.PowerConverters.Examples.DCAC.PolyphaseTwoLevel..PolyphaseTwoLevel_PWM
Electrical.Machines.Examples.InductionMachines.IMC_InverterDrive

Since it is not possible to separate PR #4840 from this one in order to resolve the issues addressed here properly, these rather limited changes in SVPWM done in #4840 (issue #4839) have been incorporated here.
#4840 will be closed, and the issue addressed there is resolved with this PR.

into PWM

MSL-RegressionTesting-CI · 2026-11-10T16:46:09Z

Failure: Some checks of the regression testing failed.
Details:
https://serv.ltx.de/prs//4812/Modelica/PR_comparison_report.html
https://serv.ltx.de/prs//4812/ModelicaTest/PR_comparison_report.html

MSL-RegressionTesting-CI · 2026-11-10T18:33:55Z

Failure: Some checks of the regression testing failed.
Details:
https://serv.ltx.de/prs//4812/Modelica/PR_comparison_report.html
https://serv.ltx.de/prs//4812/ModelicaTest/PR_comparison_report.html

AHaumer · 2026-11-11T10:11:54Z

@casella @GallLeo @MatthiasBSchaefer
As far as I see, "Some checks of the regression testing failed." doesn't make sense if we work on a bugfix.
Of course a bugfix can change the results, and new refercne results have to be generated.
How do we handle this properly?

MatthiasBSchaefer · 2026-11-11T10:23:16Z

@casella @GallLeo @MatthiasBSchaefer As far as I see, "Some checks of the regression testing failed." doesn't make sense if we work on a bugfix. Of course a bugfix can change the results, and new refercne results have to be generated. How do we handle this properly?

Obviously the regression testing will fail, as long as the reference files are not updated. This absolutely makes sense.

You can download - potentially - new reference files from the report:
1.) click the links given in the comment for Modelica or ModelicaTest
2.) then choose the box, where the examples for the pull request were tested with Dymola (box in the lower right corner of the Dymola- section).
3.) There you can choose the relevant example
4.) In the report of the relevant example you can find a download-Button in the mSL-Reference section
5.) check the trajectories (they are only potentially new reference files)
6.) If you agree with the results, you can create a new commit adding these files as new references

AHaumer · 2026-11-11T14:37:39Z

@MatthiasBSchaefer I don't get it.
2) lower right corner? 0 failed?

MatthiasBSchaefer · 2026-11-11T15:11:56Z

@AHaumer Yes, the box you marked in the picture.
This is where all the examples were simulated.
This is successful (0 failed), even if the trajectories diverge.

New reference results were created based on the simulation, not based on the trajectory-comparison. Such, the box in the lower right corner of the Dymola-pattern is the place of new reference files.

new reference results

MSL-RegressionTesting-CI · 2026-11-11T18:32:43Z

Failure: Some checks of the regression testing failed.
Details:
https://serv.ltx.de/prs//4812/Modelica/PR_comparison_report.html
https://serv.ltx.de/prs//4812/ModelicaTest/PR_comparison_report.html

AHaumer · 2026-11-11T19:47:22Z

@MatthiasBSchaefer how do I see what failed?

Failure: Some checks of the regression testing failed. Details:
https://serv.ltx.de/prs//4812/Modelica/PR_comparison_report.html
https://serv.ltx.de/prs//4812/ModelicaTest/PR_comparison_report.html

I tried to follow your advice and committed new reference results (Modelica.Electrical.Machines.Examples.InductionMachines.IMC_InverterDrive and Modelica.Electrical.PowerConverters.Examples.DCAC.PolyphaseTwoLevel.ThreePhaseTwoLevel_PWM) to the PR itself (is this correct?).
Why I was wondering:
For ThreePhaseTwoLevel_PWM I expected the new comparisonSignals with 4 signals but found an old one, whereas the csv contained the 4 signals.

The whole procedure seems to be complicated and confusing.

into PWM

MSL-RegressionTesting-CI · 2026-11-13T15:54:38Z

Failure: Some checks of the regression testing failed.
Details:
https://serv.ltx.de/prs//4812/Modelica/PR_comparison_report.html
https://serv.ltx.de/prs//4812/ModelicaTest/PR_comparison_report.html

MSL-RegressionTesting-CI · 2026-11-13T17:15:14Z

Failure: Some checks of the regression testing failed.
Details:
https://serv.ltx.de/prs//4812/Modelica/PR_comparison_report.html
https://serv.ltx.de/prs//4812/ModelicaTest/PR_comparison_report.html

christiankral

I investigated Modelica.Electrical.PowerConverters.Examples.DCAC.PolyphaseTwoLevel.ThreePhaseTwoLevel_PWM, changing VNom from VNom=Vmax/sqrt(2) to half of this value: VNom=Vmax/sqrt(2)/2. Then I compared VNom with harmonic.y_rms for

SVPWM
Intersective
and found the following result:

The ratio of the two results is 115.3V / 100V = 1.153 = $2 / \sqrt 3$

christiankral · 2026-11-13T21:54:30Z

@AHaumer I do understand, that there is a difference for VNom=Vmax/sqrt(2) as the intersective PWM "saturates". But why is do the two results differ for half of this value.

christiankral · 2026-11-13T22:28:42Z

I also compared uMax of in the intersective PWM model for the attached Modelica example ThreePhaseTwoLevel_PWM.mo:

MSL 4.1.0

uMax for the SVPWM and the intersective PWM are equal.

PR #4812

uMax for the SVPWM is now different from the intersective PWM case. This is a non backwards compatible change. Therefore, we have to put this into the release notes.

AHaumer · 2026-11-14T09:25:34Z

I investigated Modelica.Electrical.PowerConverters.Examples.DCAC.PolyphaseTwoLevel.ThreePhaseTwoLevel_PWM, changing VNom from VNom=Vmax/sqrt(2) to half of this value: VNom=Vmax/sqrt(2)/2. Then I compared VNom with harmonic.y_rms for
* SVPWM

* Intersective
  and found the following result:
The ratio of the two results is 115.3V / 100V = 1.153 = 2 / sqrt(3)

@christiankral This is correct. Both PWM methods are in the linear mode without overmodulation.
How would you increase the base harmonic of the carrier based modulation without being nonlinear?
If we increase artificially the amplitude of the reference signal by 2/sqrt(3), we would achieve a higher base harmonic rms
until we reach saturation at sqrt(3)/2 of the maximum voltage and achieve a contiguous on-block at the maximum of the reference voltage.
This is not the desired behaviour. If we want to achieve higher output, we have to increase the base harmonic
without increasing the maximum of the reference voltage by injecting a zero system which doesn't affect the machine
(neglecting the potential of the starpoint).
I could add (in a backwards compatible manner) such a block that adds a zero system. Shall I?

AHaumer · 2026-11-14T09:39:27Z

I also compared uMax of in the intersective PWM model for the attached Modelica example ThreePhaseTwoLevel_PWM.mo:

MSL 4.1.0

uMax for the SVPWM and the intersective PWM are equal.

PR #4812

uMax for the SVPWM is now different from the intersective PWM case. This is a non backwards compatible change. Therefore, we have to put this into the release notes.

@christiankral you are right. This is a bugfix, the handling of maximum voltage was wrong. Here's a comparison:

MSL 4.1.0

SVPWM
parameter Real uMax "Maximum length of space vector = half diagonal of hexagon";
IntersectivePWM
parameter Real uMax "Maximum amplitude of signal";
PWM
parameter Real uMax "Maximum amplitude of signal";
PowerConverters.DCAC.Control.SVPWM svPWM(uMax=uMax, …)…;
PowerConverters.DCAC.Control.IntersectivePWM intersectivePWM(uMax=uMax, …)…;

PR4812

SVPWM
parameter Real uMax "Maximum space phasor = half diagonal of hexagon";
IntersectivePWM
parameter Real uMax "Maximum amplitude of signal";
PWM
parameter Real uMax "Maximum amplitude of signal";
PowerConverters.DCAC.Control.SVPWM svPWM(uMax=uMax/(sqrt(3)/2),…)…;
PowerConverters.DCAC.Control.IntersectivePWM intersectivePWM(uMax=uMax,…)…;

There are 3 bugfixes, and - you are rtíght - all of them have to be documented in the release notes:

correcting the timing of IntersectivePWM
correcting the sequence of switching patterns in SVPWM
correcting calculation and usage of maximum reachable voltage

Should I edit the release notes in this PR?

MSL-RegressionTesting-CI · 2026-11-14T12:25:44Z

Failure: Some checks of the regression testing failed.
Details:
https://serv.ltx.de/prs//4812/Modelica/PR_comparison_report.html
https://serv.ltx.de/prs//4812/ModelicaTest/PR_comparison_report.html

resolves modelica#4089 and modelica#4090

1889f92

AHaumer added this to the MSL4.1.1 milestone Sep 11, 2026

AHaumer requested review from casella and christiankral September 11, 2026 16:21

AHaumer self-assigned this Sep 11, 2026

AHaumer added bug Critical/severe issue L: Electrical.PowerConverters Issue addresses Modelica.Electrical.PowerConverters needs backporting Needs back-porting from the master branch create new reference results create new reference results labels Sep 11, 2026

AHaumer mentioned this pull request Sep 11, 2026

Fix propagation of start time in IntersectivePWM #4090

Closed

AHaumer linked an issue Sep 11, 2026 that may be closed by this pull request

Faulty parameter propagation in IntersectivePWM #4089

Open

resolves modelica#4089 and modelica#4090

26d9679

beutlich force-pushed the PWM branch from 6245c0c to 26d9679 Compare October 20, 2026 20:47

This was referenced Oct 22, 2026

PowerConverters.DCAC.Control.SVPWM: non-standard vector sequence in odd sectors (double commutations, extra harmonics) #4839

Open

Bugfix Electrical.PowerConverters.DCAC.Control.SVPWM #4840

Closed

christiankral reviewed Oct 31, 2026

View reviewed changes

Comment thread Modelica/Electrical/PowerConverters/DCAC/Control/IntersectivePWM.mo Resolved

Comment thread Modelica/Electrical/PowerConverters/DCAC/Control/IntersectivePWM.mo Resolved

Comment thread Modelica/Electrical/PowerConverters/DCAC/Control/IntersectivePWM.mo Resolved

AHaumer marked this pull request as draft November 1, 2026 11:32

AHaumer and others added 6 commits November 1, 2026 14:34

Merge branch 'PWM' of https://github.com/AHaumer/ModelicaStandardLibrary

0cf419b

into PWM

Documentation of Electrical/PowerConverters/UsersGuide/DCACConcept

1eb5888

improved documentation DCAC

1090704

corrected html

7f7dd2b

enhanced documentation

44e1dbb

Merge branch 'master' into PWM

16266ea

AHaumer added 2 commits November 2, 2026 13:22

corrected PWM

547a91b

Incorporated modelica#4840

c518bc1

AHaumer closed this Nov 2, 2026

AHaumer reopened this Nov 2, 2026

AHaumer changed the title ~~PWM resolves #4089 and #4090~~ PWM resolves #4089 and #4090 and #4839 Nov 2, 2026

AHaumer requested a review from MSL-RegressionTesting-CI November 10, 2026 15:43

AHaumer added 2 commits November 10, 2026 17:06

2 PowerConverter examples were misplaced in ModelicaTest

e03b977

Merge branch 'PWM' of https://github.com/AHaumer/ModelicaStandardLibrary

b6fb890

into PWM

AHaumer added 2 commits November 10, 2026 18:50

take care that zero vector is excluded in example Test_SVPWM

be06cda

improved TestExample

0be9a5e

AHaumer requested a review from casella November 11, 2026 10:07

AHaumer and others added 2 commits November 11, 2026 18:53

new ReferenceResults for 2 Examples due to BugFix

1a1f392

Merge branch 'master' into PWM

837dd18

AHaumer and others added 3 commits November 13, 2026 15:35

corrected comparisonSignals and created new ReferenceResults

e473b87

Merge branch 'PWM' of https://github.com/AHaumer/ModelicaStandardLibrary

1d2704b

into PWM

Merge branch 'master' into PWM

daf33cc

update ReferenceResults

43ca683

christiankral reviewed Nov 13, 2026

View reviewed changes

AHaumer requested a review from christiankral November 14, 2026 09:52

Conversation

AHaumer commented Sep 11, 2026

Uh oh!

christiankral left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

AHaumer commented Nov 1, 2026

Uh oh!

AHaumer commented Nov 2, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

MSL-RegressionTesting-CI commented Nov 10, 2026

Uh oh!

MSL-RegressionTesting-CI commented Nov 10, 2026

Uh oh!

AHaumer commented Nov 11, 2026

Uh oh!

MatthiasBSchaefer commented Nov 11, 2026

Uh oh!

AHaumer commented Nov 11, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

MatthiasBSchaefer commented Nov 11, 2026

Uh oh!

MSL-RegressionTesting-CI commented Nov 11, 2026

Uh oh!

AHaumer commented Nov 11, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

MSL-RegressionTesting-CI commented Nov 13, 2026

Uh oh!

MSL-RegressionTesting-CI commented Nov 13, 2026

Uh oh!

christiankral left a comment

Choose a reason for hiding this comment

Uh oh!

christiankral commented Nov 13, 2026

Uh oh!

christiankral commented Nov 13, 2026

MSL 4.1.0

PR #4812

Uh oh!

AHaumer commented Nov 14, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

AHaumer commented Nov 14, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

MSL 4.1.0

PR #4812

MSL 4.1.0

PR4812

Uh oh!

MSL-RegressionTesting-CI commented Nov 14, 2026

Uh oh!

This branch has not been deployed

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

christiankral left a comment •

edited

Loading

AHaumer commented Nov 2, 2026 •

edited

Loading

AHaumer commented Nov 11, 2026 •

edited

Loading

AHaumer commented Nov 11, 2026 •

edited

Loading

AHaumer commented Nov 14, 2026 •

edited

Loading

AHaumer commented Nov 14, 2026 •

edited

Loading