New Ethernet CLICKs are out ... New Do-mores are ???

[ADC Product Engineer]

Honestly, how big does a program have to be before a PLC runs into a scan time issue?

It's not really how big rung wise that the program is, but what you have in it.  For example with the DirectLogic PLC's, if you want to go slow, use a lot of BCD or Real math.  These instructions are incredibly slower than using signed or unsigned decimal which is a native instruction to the processor.

(Yes, I know BCD is the default Timer and Counter data type for DirectLogic.  There is a reason for this buried in antiquity.  Decimal math is still incredibly faster.)

[deep6ixed]

Our boat had her keel laid in '82 with technology that put a man on the moon, remember the unit in Apollo 13 that Tom Hanks was looking at to steer the command module.  Yeah thats a VERDAN unit, ours was MARDAN its cousin...

Our computers were loaded from paper tape in 2004...

So a Do-More could easily run Navigation Control.

[Dean]

Let's not forget Latvia's greatest export to date:

[Frontier]

Out of curiosity, what are people getting for their average scan times on a Do-More?  What min and max ranges are to be considered normal?

[BobO]

Can vary considerably. Minimum generally a couple hundred microseconds, depending on I/O. Max depends greatly on the type of application. Simple logic is very fast, string processing can be pretty slow.

What are you doing and what kind of performance do you need?

[Controls Guy]

I managed to bog one down to 13ms.  Only 200 rungs but TONS of MATH boxes.  Probably an average of better than one per rung.  I pretty much can't energize a coil without resorting to a MATH box.

[BobO]

You must have a lot of floating point or complex functions. Basic integer math is blistering fast.

[Frontier]

I'm running an average of 6ms, with a max of 9ms. The application is a brewhouse, and the most critical thing is a PID loop with a 20 second lag time. I have a CTRIO card for reading two pulse flow meters, a fairly hefty RS485 bus, and 8 RTD inputs. I was mostly curious how my scan compared so I would know if I should optimize some of the program.

[Controls Guy]

You must have a lot of floating point or complex functions. Basic integer math is blistering fast.

Tons of boolean math (XORing DINT's with each other and with masks, etc.), and some looping, though not extreme.  If I get a chance I'll profile it and see if there are any bottlenecks.

[BobO]

I might have cut some of those from the copro due to resource issues, but they were part at one point. I'll check when back at the office. Should definitely be in the new hardware.

[Controls Guy]

The 13ms wasn't a problem, but more power is always a good thing; I won't turn it down!

Yeah, I guess the lesson is that some weirdos like myself do a LOT of boolean math (and a lot of math in general), so the better it's handled the happier we'll be.   

[BobO]

Gate array was snug, so I made the bool math box stuff optional to save space. Didn't think they would be used heavily. Sorry. Should be space in the new box. Will be about a 20x performance boost for those instructions.