Not knowing how your data looks like it's hard to give any valid hints.

Yes, it has changed. A lot. In general S7 architecture has changed. All internal registers known from S7-300/400, including DB and DI, do not exist anymore. However they can be emulated. But only on S7-1500 and for blocks in STL only.

You may find this helpful.

What is asked, is about optimization for the S7-1200 and S7-1500. TIA Portal, or more accurately said - Step7, has (almost) nothing to do with your query. Coding the S7-300 in Step7 (TIA Portal) should deliver the same functionality as you had in Step7 Classic.

Optimization of a process is very subjective thing, bound to what you are aiming to achieve. Difficult to say what your case needs - as already said. For a fact: Holding on to the traditional indirect addressing techniques are going to come with penalties.

Using structured UDTs, using arrays and coding in SCL helped for my own projects. Our S7-1200 projects looks very different than the S7-300 ones we did years ago. Yet our projects are not as time critical as it seems yours are.

I do not know that the S7-1200 range is marketed as being better than the S7-300 range when it comes to processing. Convincing Siemens to deliver a better processing cycle from the S7-1200 might be a big ask, but you are welcome to confront them directly. Use the Support Request or contact your local representative. Give them a sample S7-300 program to suggest a conversion to S7-1200 that equates the cycle time. You may need to remind the Siemens guys they are loosing market share to brands who deliver faster processing.

Hi Mauro,

I recommend that you take a look at the Programming Guideline for S7-1200/1500. It contains a lot of information about the optimal programming from a technical point of view.

Since these PLCs come with optimized access, you should always prefer to use this method. Since optimized access is only available for blocks, variables inside optimized data blocks will be more performant compared to using bit memory, timers or counters.

Kind regards
Stefan

Your experience differs from my own.  I even have good performance with UDTs containing UDTs and multi-instance FBs using UDTs as parameters as well as static and temp tags.

Mauro.Bottizzo

Thanks for the link, I will take a look, of course.
But, I would like to say that I have made many and many test with these CPU comparing and analizing the results: using many different programming techniques: only direct access for every tags, using strutures, variant pointers, references and more.

What I understend as well is that: using a small amount of structured data as in/out variables of the function bloks don't affect very much the CPU performance, but if the structure is bigger than only some tags, even when used ad in/out of the FBs (in fact pointers), the CPU will die. The same program can cycle from 10 ms to 300 ms.

So, now, before to continue di discussion, I will take a look of your link.