PLC with 64bit floating point and integer arithmetic?

Does anyone know of any such PLCs?

Usually I'm lazy and don't want to think about precision and overflows.

Simatic S7-1500

Even with Long Reals and Long Integers you can create an overflow. Quite easily actually.
So a bit of thinking will still be required.

That's true Jesper, but I find I am getting close to the limit too many times when I 'think' about it with 32bit math.

These S7-1500's look pretty good. Thank you, Steffen.

I'm curious where anyone has found a statement that S7-1500 supports 64 bit floating point arithmetic.

In the 5 documents I've found so far I've discovered that it can execute at 64nS, and that TIA supports 64 bit windows, but I can not find a statement about supporting 64 bit arithmetic.

Omron - even the shoe box CP1L has 64 bit floating point. 2 year warranty as well.

Danw: from what I can find, they have a type LREAL and LINT, and a separate library of "long functions", eg LREAL_ABS. Most regular instructions don't accept them.

BobB: Wow, nice one Omron. Any reason not to use an Omron?

have look at www.oscat.de or a nice collection of functions, also double precision

That is all I use these days AustralIan - I have been known to knock back work if I have had to use another brand at times. Even the shoe box has a very extensive list of instructions and you have a choice of using either binary or BCD timers and counters. BCD for legacy conversion. The CJ2 series you can use a mix of binary and BCD if you wish.
A couple of my favourite instructions are BSET (block set), BCNT (bit count - great for alarms) and APR (linear approximation - put together a table and the instruction does a linear conversion).
Different processors have slightly different instruction sets for motion and the like but the basics are identical.
Have used my last Schneider PLC for example - painful software and way too slow to program and set up.
There are others that I will not use either.
I will not use the new Omron NJ either - IEC stuff - auto allocation and tag based - YUCK! The way it is going though. I program using I/O numbers directly - way quicker. Just type in the instructions in a line as well such as TIMX 100 &300 is timer 100 30.0 seconds - so quick and easy.
Lot of stuff is symbol as well > D1000 &2034 is D1000 is greater than 2034 - just easy.

AustralIan said:

Danw: from what I can find, they have a type LREAL and LINT, and a separate library of "long functions", eg LREAL_ABS. Most regular instructions don't accept them.

Click to expand...

When I try using LReal/Lint tags in a 1500 with instructions like add/==/abs I don't get any compile errors. I was just playing around though, I've never needed them in a real project.

Hmm, sounds promising.

I generally don't recommend them except for high-volume applications with limited engineering input, but the Rockwell Micro 820/830/850 does support both 64 bit LINT and 64 bit LREAL datatypes, with specific math instructions for both.

If you needed to do some very specific math in an otherwise low-end control system, those might suit your needs.

If you want 64 bit math because you want *more* flexibility, functionality, ease-of-use and broad application, look to something else.

S7-1200 also support 64bits

Hi,

The basic S7-1200 also support 64 bits integers and floats.


Best regards,
Kelkoon

I would think most major brands support the LREAL, LINT, and ULINT standard IEC datatypes. I know for sure all B&R, Beckhoff, and PackDrive (formerly ELAU, now Schneider-Electric) do. I would be shocked if a VxWorks platform didn't, so between those, Beckhoff, and Siemens that's probably 95% of all PLCs.