SIMULINK BUG, or Am I missing something in DATA STORE MEMORY block?

16 vues (au cours des 30 derniers jours)
Eleazar
Eleazar le 22 Mai 2013
I have a Simulink model with the following problem:
At the top-level I have one data store memory block A, with one respective read and write block. And even though after the reading, the value travels thru a S-Function at the same level, 3 different Switch-Case Action Subsystems each one with different S-Functions, returns to top-level, and passes thru a Merge block, the writing of the new value is correct.
Now, inside just 1 of the 3 Switch-Case Action Subsystems I have another data store memory block B, also with one respective read and write block inside the subsystem too, same level. After the reading, the value travels thru the S-Function, together with the value from memory block A.
The value for block A is always correct. However, there is no reading from block B.
I already tried:
1-Changing the data type, in the memory block and the S-Function.
2-Double check in sample time.
3-Different initial value.
4-Set ports connected to Merge block as SS_REUSABLE_AND_GLOBAL.
5-Diffent computers (x32 and x64).
6-Put B at the same top-level with A.
7-The same problem still exists in a smaller model too.
But still not reading from block B. Is there something missing?
Anyone please. I am stuck with this for a week.
Thanks.
-Ele

Réponses (2)

Nirmal Gunaseelan
Nirmal Gunaseelan le 22 Mai 2013
Looks to me like an access issue. How about you make B as a global data store? This is just to make sure that your conditionally executed subsystem has and provides access to B outside itself.
  4 commentaires
Eleazar
Eleazar le 30 Mai 2013
Well, I realized the problem is not about block execution order. Funny is that, in the initial run of the model, there is no reading from block B. But if I run the model a second time, it works perfectly.
I think there is an issue with how many S-fcn's can open same .txt file at the same time. 2 S-fcn's writing in the file work fine, but one more even just reading it maybe cannot if the file is empty. But as I said, in the second run, with some data in the file, the model runs fine.
I think that is why the data store memory block B is not updated from the output of the s-fcn. A third pointer to the same .txt file is not working. Btw, the s-fcn's are in C language.
This is just my guess. Maybe I am wrong.
Kaustubha Govind
Kaustubha Govind le 30 Mai 2013
Eleazar: Would it be simpler if you try to debug this by modifying your S-function to not perform file read/write access (because that may be a C-language related issue)? This might help you determine if it is really a file I/O issue or a Simulink issue.

Connectez-vous pour commenter.


Muruganandham Subramanian
Muruganandham Subramanian le 30 Mai 2013
How you are reading the value for DSR B (data store read)?? from workspace?? Is DSR B depends on DSR A output?, If it depends, check DSR memory used inside the subsytem, passing right value to DSR B ?
  1 commentaire
Eleazar
Eleazar le 31 Mai 2013
Values of DSR B are only in the Simulink model, and there is nothing from or to the workspace.
The original model is big, so I created a much reduced toy-model out of it, just to reproduce the problem. Just a few block and 2 DSM blocks. There is 1 .txt file accessed by the 3 S-fnc's (in C lang.), and ech S-fnc in a Switch-Case Action Subsystem. The S-fnc's are used to read and update a pointer to the .txt file, and i/o to the .txt file too. The pointer is stored in the DSM B.

Connectez-vous pour commenter.

Catégories

En savoir plus sur Simulink Functions dans Help Center et File Exchange

Produits

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by