# plc vs computer control

V

#### Vishal A Bhatt

Hi, I am from india and here p4 pcs cost about 1/5 of the plc cost

also now as we have excellent GUI with help of vb and vc++ plcs dont stand a chance in market where the look and feel of the product is given higher
preference. Due to ics with integrated adc dac and rs 232 communication integration is childs play. I think only cost factor can help plcs in
comming days.

J

#### James Ingraham

Jeez, Vishal. Why not just throw a Molatov cocktail in here?

I will confess that about five years ago I predicted that the PLC would be gone in 20 years. I was wrong. Lots of people are about to respond with specifics about why, but the bottom line is this: the PLC companies will fight back. GE Fanuc has already embraced the "PAC" concept. National Instruments actually took their pure PC-based stuff and made a PLC out of it. They, too, call it a "PAC". Even Allen-Bradley (which doesn't use the PAC term) minimizes the references to "PLC" in its literature, and stresses the multiple capabilities of its Logix "platform".

Is this PC/PLC convergence? Mabye. But rest assured that PLCs are not going anywhere. Certainly not in favor of Windows and VB / VC++. Heck, I can't even get my customers to accept Sequential Function Charts or Function Block Diagrams. It's Ladder or nothing.

-James Ingraham
Sage Automation, Inc.

A

#### ANA

PC based control systems might be cheaper. But has anyone used this PC based control system before, running it 24/7 without rebooting the PCs? Unless your OS is Unix based, I doubt that it can operate 24/7 without rebooting. I think that's the advantage of PLCs over PCs.

ANA

C

#### Curt Wuollet

That's a very interesting observation. I'm of the opinion that Windows and VB/VC++ is a large part of the reason there isn't more PC control and convergence. MS has sort of poisoned the well even though there are reliable alternatives. Perhaps that's why the big automation vendors cowtow to Bill.

Regards

cww

V

Hello,
PCs kill not only PLC at all (as a kind of specialized solution vs mass-production), but they kill so called RT OSes also.

The multi-core architecture is very good for control.

--
Best regards.

W

#### William Sturm

I would agree that the defacto Microsoft standard is the reason why more PC's aren't used for control applications. Windows is just too complicated. PC control was a great budding industry in the DOS days, and even in the early windows days.

It would be nice if there was a simple graphical OS tailored for factory automation. As long as it could "seamlessly" network with Windows boxes, it just might work. It also has to be simple and fit on a flash drive. Maybe even boot from a CD. It would be impossible to kibosh the OS if it was on some kind of ROM.

Bill Sturm

M

#### Michael Griffin

I think a major factor has been that PC and PLC hardware and software were not really ready for this. This is changing though. Low cost compact fanless PCs, flash memory, Ethernet I/O, commodity software and increasing expectations for linking the factory floor to the rest of the company are changing the fundamental conditions.

The PLC market could shake out into commodity hardware assemblers, software packagers, and systems companies who bundle the hardware and software together and sell them with support contracts. The traditional PLC logic solving software could become just another commodity software component like databases or web servers. The major differentiator between brands of PLC would become how well the commodity components were put together to link the
factory floor to the rest of the company and exchange meaningful data easily.

If the major automation vendors stayed in the PLC business, it would be as systems companies. There would be plenty of opportunity during a disruptive change like this though for smaller companies to take a major share of the market.

I wouldn't be surprised to see the situation in automation be played out much the same as it has in the IT world. At first proprietary software on proprietary hardware was the norm. Then hardware became a commodity, and now software is as well. The companies that are expected to survive outside of niche markets are either the ones that become the lowest cost providers, or else the ones that find out how to make money by adding value on top of the commodities they sell. The ones that persist in trying to milk fat revenue streams out of locked-in customers are expected to go to the wall.

M

#### Michael Griffin

> I would agree that the defacto Microsoft standard is the reason why more
> PC's aren't used for control applications. Windows is just too
> complicated. PC control was a great budding industry in the DOS days, and
> even in the early windows days.
>
> It would be nice if there was a simple graphical OS tailored for factory
> automation. <

There is. It's called Linux. There are trimmed down GUIs if you don't want all the desktop features of KDE or Gnome. It's a very modular OS.

> As long as it could "seamlessly" network with Windows boxes,
> it just might work. <

The Samba network software can act as a client or a server on a Windows SMB network. Samba has been doing "Windows" networking longer than Windows has.

> It also has to be simple and fit on a flash drive. <

This is done all the time.

> Maybe even boot from a CD. <

The Knoppix distribution of Linux is the most popular for this, but there are numerous others.

> It would be impossible to kibosh the OS if it
> was on some kind of ROM. <

You can also boot off a network if your network card supports it.

J

#### Jiri Baum

Hello,

Simple graphical OS for factory automation... have you considered linux?

Apparently the handheld people can get a reasonable graphical base system [1] in 12MB (compressed), which should fit on a flash drive just fine with enough space left for the actual application; or it can boot from CD. Seamless networking depends on what you want to do with it, but certainly at least files and printers can be shared, and it can serve web pages on the intranet.

It's not specifically tailored for factory automation, but it has been used for various embedded tasks in the past, so it's probably more a question of collecting the information that's already out there than actually making
anything new.

Jiri

[1] That's linux, X windows and GTK+; binary-compatible with Debian packages.
--
Jiri Baum <[email protected]> http://www.csse.monash.edu.au/~jirib
MAT LinuxPLC project --- http://mat.sf.net --- Machine Automation Tools

A

#### Automation Linse

Your "prediction" has been given for nearly 20 years now ... and so far seems far from true.

What will kill your chances of market success is that "P4 PC" which costs 20% the price of PLC has parts within which cannot be purchased 6 months from now. I suppose your could REPLACE the full PC every time it fails, but I'd certainly not want to support such a system. Even GM and other big users who attempted to bias in favor of PC-control & not PLC have backed off. The on-going day-to-day support/repair of the PC have been nightmarish.

PLCs may be expensive, but today one can still by 20-year old parts for repair (either from the vendor or eBay).

- LynnL

C

#### Curt Wuollet

Hi Lynn

That is probably true for machines aimed at the disposable desktop PC market, but there is absolutely no difference between them and any "modern" PLC that uses custom silicon and hybrid ICs as, for example AB, does in this respect. In fact, I humbly submit that you can't buy the AB chips _now_. The primary reason that you can't get parts for a
5 year old PC is that no one is interested in doing so. And the whole shebang stays afloat on a much shorter replacement cycle. But if you manufacture purpose built boxes and manage spares in the same way, I simply can't see where there would be any difference. True, this would
add to the cost, but it's certainly doable. In fact these P4 PCs are pretty close to what's inside the newest and most powerful PLCs already. And there are several racks that will accept either "traditional" PLC CPUs or CPUs with the resources to run minimal but recognizable operating systems. So from a technology and hardware standpoint it's a non-problem. It's entirely possible to simply manage the problem away. The whole issue is getting enough people to buy the same gear to make the numbers work. The least popular PC MB probably sells 2 orders of
magnitude more units than the most popular PLC. If AB sold a PC CPU for their SLC racks, they could deliver a lot of bang per buck and get numbers up to commodity levels. I expect one of the big vendors to catch on in a few years.

Regards

cww

M

#### Michael Griffin

In reply to Jiri Baum's comments - With regards to flash drives there are essentially three different kinds. There are IDE flash drives which emulate conventional drives, and may include "wear leveling". The "wear leveling" spreads out the writes over the drive to avoid wearing out the flash too quickly. These tend to be expensive, but work with software and operating systems that don't otherwise support using flash.

There are also small DIP style flash packages (e.g. M-Systems "Disk On Chip"). These are common in the small size you are referring to (12MB). These are good for small embedded systems, as you can buy PC/104 CPU boards with
sockets for these devices. They require software support for them though.

There are also adaptors for flash cards which allow them to be used as drives. These are larger than a "Disk on Chip", but are a good deal cheaper for an equivalent size. Some of the mini-ITX boards support these directly. For
other boards there are adaptors available which allow them to be used via an IDE port.

In terms of cost versus capacity, the flash cards are the most attractive option. Since they don't have wear leveling though, you can't use them unless you can avoid writing to them all the time. To use them with Linux, I believe that you would simply mount them read only, just like booting from a CD-ROM (e.g. as with Knoppix). Scratch-pad temporary directories would be mapped to a RAM disk.

Given the cost and capacity of flash cards, there is little reason to try to pare a system down too much unless you have severe power, size, or cost constraints. An economical installation can still include software with Gnome or KDE dependencies.

As for X and GTK+ in a 12MB compressed image, I would be interested in a reference for that. I was under the impression that the 8 - 12MB installations were for console mode only. The "small" general purpose distros
including 'X' that I have seen are generally about 50MB (although this does include a selection of application software).

C

#### Curt Wuollet

Hi Michael

I think the primary reason the hardware hasn't been ready is simply that it's such a small market that no one has really made the effort. Examples of appropriate hardware are certainly out there, but they are extremely low volume and hence, not cost effective.

Agree on the software, primitive runtime executives are wonderfully robust and reliable, but the results are really bad when people try to add on general networking. And fitting comms into a relay mindset has produced the totally inflexible, "one act ponies" that are scarcely
better than nothing. It should be obvious to the most casual observer that the results obtained from adding the simple PLC functions to a robust, full featured OS will be far more desireable than attempting to add sophisticated networking to a glorified endless loop.

But, in the end, if you take a cell phone or DVD player in one hand, and a PLC in the other and showing the guts, ask any reasonable, uninterested observer which should cost more, the PLC position is pretty inexplicable. This will be corrected eventually. Commodities are a wonderful thing for consumers. This whole business is services and knowledge based anyways.

Regards

cww

J

#### Jiri Baum

Hello,

In reply to Michael Griffin - the X and GTK+ in 12MB compressed came from a talk about Familiar Linux at linux.conf.au 2005 - I hope I haven't misheard it (or misread my notes, for that matter), because I can't find the information on their site. However, the download sizes (with applications) are 25.2MB for the GTK-based one and 14.5MB for the Qt-based, so it's probably something like that, anyway.

Jiri
--
Jiri Baum <[email protected]> http://www.csse.monash.edu.au/~jirib
MAT LinuxPLC project --- http://mat.sf.net --- Machine Automation Tools

B

#### Brian E Boothe

Linux Isn't Something I'd Throw At my Operators and say hey Use this, uhh, Ya sure. half them Don't even Know what a desktop is in windows... And Linux isn't something u can use straight outta the box.

J

#### Jiri Baum

Hello Brian,

(a) This thread's target audience is integrators (in-house or out-sourced), not operators.

(b) You're mistaken as far as the "out of box" usability of Linux is concerned. Perhaps you should try one of the modern Live CD distributions? (Ubuntu, Knoppix, etc.) No installation required - just pop the CD in and go.

Jiri
--
Jiri Baum <[email protected]> http://www.csse.monash.edu.au/~jirib
MAT LinuxPLC project --- http://mat.sf.net --- Machine Automation Tools

C

#### Curt Wuollet

Perhaps, of course many operators would express bewilderment at any change. But that's not the context. If you throw a PLC at them, they would have no idea what to do with it either. But if you build a reliable inexpensive control system no one would know unless they opened the cabinet. The fact that Linux can run headless, that is, with no keyboard or monitor, its reliability, features and power make it ideal in this role. On even a modest PC, you can drive very high IO counts and do math and complex operations that would be difficult or impossible with typical PLCs. And the resources at your disposal are worlds beyond the capabilities of nearly all PLCs.

How you do your HMI is another matter. But I submit that even here, in most industrial applications, operators point and poke or perhaps click at fixed screens that could be running on almost anything. In a good job oriented design, they are working with the application and tools, not the operating system. In what I've been doing lately, I'm using Optimate operator panels. The user would never know whether there was a SLC, a 90/30, or a Linux box behind the scenes. Nor would or should it matter to them as long as it works. And the fact that the Linux box could also run a graphical HMI should also be invisible to the user as they are working with a program that could be running, at least in theory, on any platform. So how would they even know if you threw Linux at them? Surely they aren't programming or doing setup or administration on the machines. Our printing company has tons of "window" type graphical interfaces, none of which run on Windows and they haven't even noticed.

Regards

cww

M

#### Michael Griffin

With regards to suitable PC hardware, the early stages of a volume market are developing for consumer applications, particularly for entertainment and vehicle systems. People want small, low powered, fanless systems for low
noise so they can be located living rooms and play music or video.

Prices for a Mini-ITX MB start at CND$120, with CND$240 being typical (if you get one with features like built in compact flash slot or dual LAN). A power supply is about CND$50 (60W). Cases start at CND$100, but these are high quality metal work, which are intended to look like stereo components. Compact flash cards are less than CND\$50 for 256MB. This sounds fairly cost
effective to me.

The main "lack" at the moment is an industrial style case.

C

#### Curt Wuollet

Those would be pretty close on some applications already as there is a trend towards no local I/O in conveyors for example and some high I/O count machines.

And Industrial Style housings is a curious term as many PLCs come in pretty shoddy plastic wrappers these days. A terminal strip for power and a DIN rail mount would go a long ways.

But I was encouraged today as I had to go to work to "train" on a large capital system installed recently by Heidelberg (big in printing) it's a
"Perfect Binder" a bookmaking machine. Lots of Siemens and Lenze, etc. strung together with ProfitBus. But the Flex console, what we would term SCADA, uses TimeSys Linux to control the machine. Of course they run the display and db part on NT as they still need to interface to all the monopoly servents, but we have a Linux box controlling a multi million dollar machine directly over ProfitBus I/O in a mission critical
application done by an industry leader. That sounds like PC control to me. And I'd love to hear why they didn't just use the NT:^)

Regards
cww

M

#### Mark Bayern

Linux? Maybe...

I'm probably 4-5 years out of date, but Heidlelberg's 'Sunday 3000' presses are (were?) run by multiple QN/X systems networked together.

The printing plants I've been in _never_ ran any equipment based on MSWindows.

> That sounds like PC control to me. And I'd love to hear why they
didn't just use the NT:^) <