How well does TIA v15 compare to Step7 5.6?

“S120 doesn't TECHNICALLY need IRT, but you lose some of the good features & tight control if you don't use it. Short version is, the product may not absolutely require IRT, but the typical application usually does.”


What’s the typical application?

mk42 said:

Definitely true, if possible. If not, the multiuser server is a (paid for) alternative to the (free, but iffy) synchronize feature. It isn't perfect, but it is way less bad.

Click to expand...

I'm now on my first project with Allen-Bradley PLCs and I really like how you go online and just stay there, and you modify everything while online. And other guy immediately picks up your changes.
Bad thing about this is that you're not really aware of what other guy is doing, which is also a problem of its own.


I think TIA showing you where are the changes by the other guy is good, but Siemens should work on making it smoother and faster when synchronizing. A lot of times, from my experience, it would show online/offline differences, and when I'd check it would say no differences. This was caused by timestamp difference. And you cannot 'upload' timestamp and downloading your, same, software would cause timestamp issue for the other guy. This was quite a big timewaste, and we resolved by simply asking each other 'can I download' etc.

JRW said:

But what is your application?
Also remember, 1500H doesn’t support TO’s.

Click to expand...

The application is a high precision servo system. What are TO's?

doomsword said:

Anyways, with Trace you can have limited samples, while with iba basically you can have unlimited.

Click to expand...

Is this a different issue than the one JRW advised to use the "Automatically repeat recording" setting for?

mk42 said:

Yes. You can change the actual values of tags in DBs while monitoring, via a watch table, or possibly from the DB editor. Or obviously from an HMI. Watch table is often the easiest way.

Click to expand...

So then you have to make the watch table first. Just a little bit more mouse and keyboard work than ideal.

mk42 said:

There is a new snapshot feature that allows you to take the current values in a DB and save them to be the offline start value in your program.

Click to expand...

New? Is it any different from saving DB actual values from online, to the offline project in Classic?

mk42 said:

I've only ever heard of MRPD used in conjunction with IRT, so I can't imagine it would support it before IRT support is available. I'd be shocked if it IRT support arrived without MRPD.

Click to expand...

I haven't worked with either before, but from what I've been able to research, MRPD is just for the bumpless redundancy, by sending all telegrams in both directions on the ring. That way, even in case of a cable break, there are no lost telegrams. And MRPD is required by IRT. So MRPD = cool, and should work by itself, but perhaps there's no need for it for other things than isochronous real-time applications...

mk42 said:

Siemens says they have customer references that say development was 33% faster in Portal than Simatic Manager. I dunno how much I think that is typical, but I definitely prefer Portal, once I got used to it.

Click to expand...

I'm sure they do. And they sell Portal very hard. Whenever someone tries to sell me something really hard, I wonder how bad it really is. Kinda like when someone repeatedly demands that you trust them.....

mk42 said:

There's definitely a hump for the 1st project, learning how the 1500s are different from the 300/400s. IF you adapt to the new way of using them, then you'll be fine. If you try to use them like they were still a 300/400, then you'll struggle.

FWIW, it is strongly recommended to avoid STL for the 1500 in any case. It is way slower now, and is pretty much only around for legacy migrations. The recommendation is to use SCL instead, but most things you used to need STL for can be done in LAD now.

Click to expand...

Yeah, I read the programming guidelines for 1200/1500. My two key take-home points were: only use optimized DB's and symbolic addressing. I will add "avoid STL" to the list.

(bummer. I really liked pushing and popping, getting the code down to the least number of lines. Anyone wanting easy to read code must have hated me for it )

Anyway, your advice is worth a lot. And makes sense, since they built the 1200/1500's on CPU's without the registers that STL is made for. I can't stand LAD btw, but I guess the compiler handles FBD just as well.

And what I really liked with STL was how you knew exactly what was happening at each line of code. Can't imagine SCL gives you the same control?

A TO is an a acronym for Technology Object.
Sounds like your not really familiar with 1500 and S120.
You should hire an experienced integrator

mk42 said:

2) Absolutely definitely not. The whole project is compiled, and then any changes are downloaded. However, there is a feature called "download without re-initialization" that allows you to add new tags to a DB without affecting the data already in place.

Click to expand...

What is your experience with this carrying over when changing TIA version ?
My main gripe is that when you change TIA version, you have to compile and download again before you can go online. If the download without reinitialization works in this case as well, it could take out one of the biggest pains,
Havent done much testing myself with this scenario.

JesperMP said:

What is your experience with this carrying over when changing TIA version ?
My main gripe is that when you change TIA version, you have to compile and download again before you can go online. If the download without reinitialization works in this case as well, it could take out one of the biggest pains,
Havent done much testing myself with this scenario.

Click to expand...

Never tried them in combination, and I haven't had to upgrade project versions in the field very often. I think you can hypothetically "upgrade" a CPU that was downloaded via the prior version (say, v15 -> v16) after you migrate the project without stopping it. I've usually had changes that needed to get downloaded along with the new version (like updating FW) or it was a workbench CPU where I didn't really pay attention to the process because I didn't care.

JRW said:

“S120 doesn't TECHNICALLY need IRT, but you lose some of the good features & tight control if you don't use it. Short version is, the product may not absolutely require IRT, but the typical application usually does.”


What’s the typical application?

Click to expand...

OK, I guess I said that with blinders on. For some reason I always think of coordinated motion when I think of S120, which typically needs IRT.


I guess there are a million other use cases where it isn't necessary. EPOS gets you pretty far without needing IRT.

superted said:

So then you have to make the watch table first. Just a little bit more mouse and keyboard work than ideal.

Click to expand...

Honestly, I always used VAT tables for that in Simatic Manager, so I'm not sure what you're doing instead.

superted said:

New? Is it any different from saving DB actual values from online, to the offline project in Classic?

Click to expand...

I'm not sure how you were doing that in Classic, but it sounds like it ends up getting you to the same place. The new feature has tool bar buttons for it (or in the right click menu of project tree), which presumably either streamlines the process or makes it more flexible, depending on what you're doing.



Most of my Classic experience was offline, I never messed around with the online blocks much.

superted said:

I haven't worked with either before, but from what I've been able to research, MRPD is just for the bumpless redundancy, by sending all telegrams in both directions on the ring. That way, even in case of a cable break, there are no lost telegrams. And MRPD is required by IRT. So MRPD = cool, and should work by itself, but perhaps there's no need for it for other things than isochronous real-time applications...

Click to expand...

I guess MRPD may not actually REQUIRE IRT, but the two are linked in the Siemens implementation. I've never seen a Profinet device that supported MRPD that didn't also support at least passing IRT traffic. Not sure how the PN spec lays it out.

superted said:

Yeah, I read the programming guidelines for 1200/1500. My two key take-home points were: only use optimized DB's and symbolic addressing. I will add "avoid STL" to the list.

(bummer. I really liked pushing and popping, getting the code down to the least number of lines. Anyone wanting easy to read code must have hated me for it )

Anyway, your advice is worth a lot. And makes sense, since they built the 1200/1500's on CPU's without the registers that STL is made for. I can't stand LAD btw, but I guess the compiler handles FBD just as well.

And what I really liked with STL was how you knew exactly what was happening at each line of code. Can't imagine SCL gives you the same control?

Click to expand...

You can't switch STL to LAD or FBD anymore, but you can still swap FBD and LAD back and forth freely. So at least you still get that if you hate LAD....

SCL gives you the status a bit differently, but I don't think it is less comprehensive than STL. When you monitor SCL it gives you can see the values of every tag in every line, and I think it gives you how different functions/instructions evaluate via hovertext.



A few other general keys that may go unemphasized in the guidelines document:

• Default to FBs instead of FCs. There are a lot of new features coming out that only work with FBs, because they are tied to the instance. FCs still have a place sometimes, though.
• Don't use S5 timers/counters (T/C memory). Use the newer IEC style timers counters instead. This was actually a best practice in Simatic Manager as well, but the blocks were hidden in a library and therefore often unused.
• Along with symbolic addressing, data types become very important. Building UDTs to use all over your code simplifies so much.
• You generally want to use DBs for tags whenever possible, instead of M memory. Part of the reason for this is easy duplication, but you also can't use custom data types in M memory. I typically see the recommendation to only use M memory for scratch tags that are just for temporary testing. As I'm cleaning up code, any tags I see in M memory besides the clock byte I instantly know were supposed to be temporary, and therefore should be replaced.

superted said:

So then you have to make the watch table first. Just a little bit more mouse and keyboard work than ideal.

Click to expand...

No, like mk42 said, you can change it in function monitoring, in watch table and db itself, with no declaration view/monitoring view stuff..

New? Is it any different from saving DB actual values from online, to the offline project in Classic?

Click to expand...

For getting a snapshot, not really. But then you can set the actual values as initial, for all or only for setpoints.. While always seeing clearly what you have and what you haven't changed.

I'm sure they do. And they sell Portal very hard. Whenever someone tries to sell me something really hard, I wonder how bad it really is. Kinda like when someone repeatedly demands that you trust them.....

Click to expand...

Maybe they are just scared they can't support you anymore with you still being on WinXP

Yeah, I read the programming guidelines for 1200/1500. My two key take-home points were: only use optimized DB's and symbolic addressing. I will add "avoid STL" to the list.

(bummer. I really liked pushing and popping, getting the code down to the least number of lines. Anyone wanting easy to read code must have hated me for it )

Anyway, your advice is worth a lot. And makes sense, since they built the 1200/1500's on CPU's without the registers that STL is made for. I can't stand LAD btw, but I guess the compiler handles FBD just as well.

And what I really liked with STL was how you knew exactly what was happening at each line of code. Can't imagine SCL gives you the same control?

Click to expand...

Yeah.. You can still do everything "the old way", non-optimised and STL. (if you get stuck). But you can also do everything better by using optimised and SCL..

And of course you know what is happening at every line of code...