PLC Wars... which one?

P

Thread Starter

Paul Tolsma

Hi all. We are presently going thru a "What PLC should we use" exercise. The application is high-speed assembly machinery (around 100 pieces per minute) with primarily discreet I/O that runs from 100-300 in and half that out. Typical
machines also have 3-6 servos that are usually used in step and direction mode, sometimes with encoders. We have some installed GE Fanuc 90/30 PLCs (340 CPUs) running state logic but we are looking to change that to ladder. HMI will be via
Ethernet and we would also like to investigate moving to distributed I/O (ie, DeviceNet type stuff) to cut down on wiring if we can do that without sacrificing speed. I would think the contenders would be AB and Siemens in addition to GE. Anyone have any experience with these PLCs, or others, in this type of application?

Thanks... paul t
 
Why limit yourself to a PLC? I have been continually frustrated by the
inability of PLC manufacturers to provide a fast, open approach to
PLC/HMI/Motion. My current solution to this dilemna? SoftPLC from
Teledenken. It looks, smells, and tastes like an Allen-Bradley PLC5. The
difference is it supports fast non-proprietary I/O of your choice (I use
interbus-S), TCP/IP communication for HMI using a standard PC ethernet card,
and direct no-hassle interface to Indramat's SERCOS servo controller (20+
axes of coordinated motion control over a fiber). Realtime, real fast, real
cheap.

If I absolutely HAD to use a PLC, Simatic S7/Profibus I/O and masterdrive
servos would be my choice. You can do PLC/HMI/Motion in one programming
environment and have a one-stop-shop solution. You also have to figure out
how to make it all work...documentation is not exactly a Siemens strong
point.

Just my $.02...time to crawl back into my bunker for the rest of the PLC
wars.

-------------------------------------------------

Jerry Holzer
R&D Electrical Engineer
Curt G. Joa, Inc - Boynton Beach, FL

--------------------------------------------------
DISCLAIMER: The contents of this message may not be consistent with the
views or policies of my employer,
Curt G. Joa, Inc.
 
P

Paul Anderson

Here's another thought if processor speed is a problem, as in Paul's type
of application. The marketplace currently has many specialty "modules" that
are not only produced by the PLC manufacturers, but their "partnering"
companies as well. In the case of high speed applications, a module that
has its own processor (independent of the PLC CPU) and I/O may be a better
answer. High speed counter cards, encoder / resolver modules or
programmable limit switch modules allow you to continue to use the PLCs (or
at least the same manufacturer) that you now have.
I realize that we aren't here to "hawk our wares", but feel free to
contact me direct if I can be of any further assistance.

Paul Anderson
Application Engineer
Electro Cam Corp.
 
M

Michael Griffin

At 09:49 02/12/99 -0500, Paul Tolsma wrote:
<clip>
>Hi all. We are presently going thru a "What PLC should we use" exercise.
>The application is high-speed assembly machinery (around 100 pieces per
>minute)
If the machines are one-up, then 100 pieces per minute is 1.67
pieces per second. This is certainly high speed. I usually feel pretty happy
about 4 to 8 seconds, but I suppose your machines are somewhat different. I
don't know what sort of motions your equipment performs, but does the scan
rate have any significant effect on the cycle time of your machines? Also,
what about the I/O connections? We have some machines where we are losing
about 10% of potential cycle time because of the limited I/O update rate
from some older PLCs to remote I/O racks. These machines perform a rapid
series of repetative short cylinder strokes. Its not the PLC scan rate
that's hurting us, its the rate at which the I/O racks update over the
remote I/O network.

>with primarily discreet I/O that runs from 100-300 in and half that
>out. Typical machines also have 3-6 servos that are usually used in
>step and direction mode, sometimes with encoders.
If the digital I/O are 16 bit cards, then you have between 10 and 30
I/O cards, plus also stepper and servo cards. This seems to rule out the
smallest PLCs.

>We have some installed GE Fanuc 90/30 PLCs (340 CPUs)
>running state logic but we are looking to change that to ladder.
I'm not sure what you mean by this. Does this mean you want to
perform your sequencing by a series of step latches or other simple ladder
techniques, or are you planning to use more advanced techniques in
conjuction with ladder? Using pure ladder for anything but the very simplest
jobs is rather like building a sailboat out of toothpicks. I would suggest
that one of your major criteria should be that the PLC and programming
software offers good support for techniques which help you to organise the
program in a logical manner.

>HMI will be via
>Ethernet and we would also like to investigate moving to distributed I/O (ie,
>DeviceNet type stuff) to cut down on wiring if we can do that without
>sacrificing speed.
To return to the point about speed mentioned above, you should
investigate in detail how distributed I/O will affect your response time.
You need to take into account the fact that the I/O update is not
necessarily synchronised with the PLC scan rate. Also the "headline" specs
may tell you the speed of the network, but they don't necessarily tell you
how fast it can get data in and out of the CPU to the end devices, or what
the network response time of these end devices is.
You could also consider that there are also drives which you can
control by remote I/O, rather than using cards directly in the rack. I
haven't used any of these myself, so I can't tell you how well these might
work for you.
One thing that I suggest though is that you make sure the device
networks are designed with electricians in mind - that is simple, easy, and
quick to troubleshoot and repair. Nobody is going to thank you for
simplifying wiring if the overall system is more difficult to maintain.

>I would think the contenders would be AB and Siemens in
>addition to GE. Anyone have any experience with these PLCs, or others, in this
>type of application?
<clip>
I don't want to recommend any particular brand, but I would like to
mention another point which has recently been occupying my attention. I
suggest that you also consider how well any particular PLC will network to
plant wide data systems for monitoring purposes, and to make sure your
design remains open to this. I think that this is an area that will see a
lot of activity over the next few years.


**********************
Michael Griffin
London, Ont. Canada
[email protected]
 
D

David Lawton Mars

Just out of interest, what OS does your SoftPLC use? I guess it's
Win95/98/NT? If so, how robust do you find it?
Maybe I'm a traditionalist but I'm still convinced the tried and trusted PLC
is the best option at this moment in time. I can see SoftPLC's becoming very
popular in the future but I'd prefer to wait until I can be confident they're
reliable. I can just about handle the odd SCADA crash but the platform
operating the actual plant needs to be (as good as) bullet proof. I'm not
convinced SoftPLC's are there yet.
What do others think?
You mentioned the Siemens one-stop-shop, but what about AB ControlLogix too?
The motion stuff now looks very interesting and the overall architecture
looks extremely powerful - it also supports the comms protocols you'd expect
i.e. DeviceNet, ControlNet, Ethernet. I believe SERCOS may also be released
but I'm not sure when.
Failing that, AB SLC -> 1394 works well. I've also used Trio and Quin for
motion and they're both very good.
Regards
David
 
D
Paul,

I have been thru this routine more than once! All PLCs do a good job,
so it normally comes down to where are my factories located and who can
support these locations the best. If you are an OEM that builds for the general
public, I would consider using both AB for the US and Siemens if you supply
to Europe. These two companies are firmly rooted in Industrial controls and
probably aren't going to be pushed out of the business and have almost
equally good support globally. The problem with US vs. Europe is acceptance
of the control system you offer. We don't like theirs and they don't like ours,
it is just a matter of comfort or experience with the equipment.

Dale
 
T
Check out the PLC Direct (www.automationdirect.com) series of PLCs. Having used both them and AB, I prefer the PLC direct. Cheaper, as powerful, easier to program, better software, more durable, etc., etc., etc.

tom
 
B

Bill Sherman

I have built control systems with similar constraints, namely 200in/100out and 80 parts/min (750msec/machine cycle) and achieved good results with an A-B SLC 5/04.

By putting all i/o in local racks and writing the code with addresses in the beginning of the plc memory scan rates of 5 msec (for a 200 line program) were achieved. At one time a remote rack was added but that bumped the scan rate so high that the machine slowed down remarkable and we had to eliminate it .

It is important to note that with a state logic type program the machine cycle rate will be somewhat dependent on the scan time.

i.e. (scan time)*(number of steps)=time the machine is waiting for the plc to process inputs.

e.g.for a 30 step program with a scan of 5 msec the machine ran at 80/min but at a scan of 15 msec it ran at 57/min.

(5 msec)*(30 steps)=150msec+machine motions of
600msec=750msec/machine cycle. i.e. 80/min.

(15 msec)*(30 steps)=450msec+machine motions of
600msec=1050msec/machine cycle. i.e. 57/min.

Of course there are many other factors but the A-B slc 5/04 is a fast processor and has done a good job.
 
T

Tanweer Ahmed

In my opinion you should aviod using distributed I/O. For high speed motion
control we prefer using high speed modules rather than ordinary discrete
inputs.

Tanweer Ahmed
[email protected]
 
P
Hi Jerry. Thanks for your response. I hadn't really thought that PC control was
ready for prime time yet, altho I'm willing to be proved wrong <g>. The history
here, though, is that we were using small state-logic controllers (from CTC,
ironically enough). For maybe 6-7 years that sufficed. Two years ago, a new EE
(now moved on to the manufacturing side of the company) came in and started
using GE Fanuc PLCs with the Adatek state logic CPU. He wanted to migrate the
company to ladder and GE Fanuc as a platform. I showed up two months ago <g> as
the new controls engineer and am stuck with this mishmash. The stated reason
internally for the move to a new platform was that ladder is more familar to
more people than state, altho if that's the case then why this 2-yr hiatus with
Fanuc and state control rather than just moving to RLL? And, yeah, I agree that
Siemens S7 would be a good platform... I also just saw the new AB ControlLogix
HW and SW and man, it is tremendously impressive. I have never used Step 7,
altho I have my local Siemens guys coming in. I think that either Siemens or AB
is a winner; I am not especially happy with GE in this application, even though
in my previous job... which was not high-speed machine control... I built maybe
30 systems over three years using Fanuc controllers.

In sum, I don't think that we're ready for PC-based control yet... we are not
even doing PLC with RLL <g>. There's also the off-shore issue; some of our stuff
gets shipped, sooner or later, to subsidiaries and that raises yet another
support issue. The biggest problem will be going thru the corporate processes to
create standards for internal and external use. We just began working on
programming standards this week; HW is a ways away. Speaking of which, do you
have any company standards on control architecture, RLL and HMI's? And where is
Boynton Beach, anyway?

Thanks... Paul T
 
softPLC uses a proprietary realtime kernel that loads via msDOS. Once the
realtime kernel loads, DOS goes bye-bye and you are left with a very robust
system that behaves a lot like a PLC. The difference is that you can get a
plethora of add-in drivers for things like fieldbus cards and custom
functions that are raaly a pain using standard PLC's. No driver available?
Write your own (assuming you have the chops).

I'm not conviced that this is a "bullet proof" as a PLC, but in today's
increasingly defensive control environment I felt that there isn't any
choice. The big automation vendors are walling themselves into armed camps,
actually making it harder for interoperability. To quote Princess Leia form
Star Wars..."the tighter you close your fist the more of us will slip
through your fingers".

The problem with controllogix motion is that too much is done at the PLC
level...not at the drive level. The SERCOS card for controllogix is
reportedly limited to 8 axes. Phbbth. SERCOS can easily handle 40 if you let
the drive take care of positioning. They MAY actually have something to sell
in...oh...say 5 years.

When's the last time you tried to find anything that actually TALKS to
controlnet? AB can't even deliver a controlnet interface for the 1394 before
the end of the year. Devicenet is too slow to be of any practical use for
most people in the machine building biz. I'm sure I could find a profibus
lawnmower if I needed to though. And if I couldn't, the probably would be
one for interbus-S.

I also use 1394-based systems when forced to. Aside from the flaky power
supply systems, a new motion control program every 5th drive, and crappy
feedback (yes I know how to get absolute feedback... and it ain't pretty)
the GMC products are wonderful. Creonics laughed all the way to the bank on
that deal.

Siemen's SIMOLINK (fiber-optic drive sync similar to SERCOS) can supposedly
handle better than 100 axes.

Anyways, there's a lot of ways to skin a cat and your mileage may vary. I'm
still trying to find the best way to build my machines, but there's always
some kind of comporomise. Thanks for letting me rave!
 
Boynton Beach is between Boca Raton and West Palm Beach. There's a big Motorola plant here...and us. I can understand not wanting to jump into PC
control. We don't want to do that, either...but the market is forcing us to. My advice to you is to look before you leap with regards to controllogix.
It's a PC control in PLC's clothing.
We generally use Eaton PanelMate for HMI, but AB PanelView is getting spec'd more and more often these days. I am a RLL programmer, so I obviously prefer that environment. That's the only reason to use AB PLC's in my opinion...they have the best programming software (RSlogix). Otherwise, we
make custom machines...so what the customer sez goes.
 
R
In a message dated 12/13/99 11:23:03 AM Pacific Standard Time,
[email protected] writes:

<< That's the only reason to use AB PLC's in my opinion...they have the best programming software (RSlogix). Otherwise, we make custom machines ... so what the customer sez goes. >>

You obviously have never used the AI series then. :)

My opinion is the RSLogix stuff is a step backwards from the AI product.
 
E

Enrico Guasco

Hi Paul,

in Italy we have several applications running SoftPLC with Profibus RIO and EXOR terminals (as ProfiBus nodes) we've used in place of a number of GE90/30 PLCs. These applications are for production control in textile mills, (and be sure you can't afford lose a single data per week, so reliability is a basic request!!!), we use a SoftPLC to read 5200 multiplexed IOs from eight spinning machines, calculate the breaking, display the current nr of breakings to the
machine operators with a BCD display, we use about 1Mb data registers to do this. The previous plant used a 90/30 CPU for each machine and update time for BCD display was 3,5", program
scan per spinning mill was 13 to 15ms. With SoftPLC we are updating the plant(s) every 1,5" and program scan (EIGHT (8) spinning mills), and
some more calculations we couldn't do previously, is about 12 to 13 ms, CPU is a Pentium 166Mhz. Data table size for this application is >20K words.


Enrico Guasco
 
M
My experience tells me that either one of these PLCs is suitable, but so would many other PLCs eg. Mitsubishi or Omron (see elsewhere in the list). Most PLCs these days can support the majority of 'non-proprietary' I/O.

Look at SLC-5/05 with Ethernet, DeviceNet and/or AS-Interface and a 1746-HSRV or 1394 servo-controller (or any servo controller with DeviceNet). ControlLogix would be the high tech. solution here.

The Omron is partnered with Trio Servo drives, or any servo controller with DeviceNet.

The main points are (no implied order of importance):-

What specialty modules are likely to be required (i.e Servo Amplifiers/Controllers, Positioning, Encoder Modules, Stepper modules, High Speed Counter, etc.)? Are these supported by your candidate platform ?

What networking/fieldbus/IO do you want to use ? Proprietary is all right if you are unlikely to change the PLC (CCNet from Mitsubishi is good), an
'open' or standard bus would be better if it is possible that you may change PLCs (some OEMs design equipment around a single or number of PLCs and allow their customers the choice); In which case choose the network that is common to the most likely group of PLCs. AS-interface is a good choice for discrete
only IO.

Cost, size, speed. Most PLCs are much of a muchness these days; often the price depends on buying power. Don't dismiss PLCDirect/AutomationDirect, particularly if your buying power is not great.

Programming tools. It is no good getting the fastest, cheapest PLC there is if the cost advantage is eaten up by the development process. Also the machine has to be maintained once it is installed, have a look at the available programming tools and the PLCs instruction set before you make the final decision.
 
E

Enrico Guasco

Hi Paul,

in Italy we have several applications running SoftPLC with Profibus RIO and EXOR terminals (as ProfiBus nodes) we've used in place of a number of GE90/30 PLCs. These applications are for production control in textile mills, (and be sure you can't afford lose a single data per week, so reliability is a basic request!!!), we use a SoftPLC to read 5200 multiplexed IOs from eight spinning machines, calculate the breaking, display the current nr of breakings to the
machine operators with a BCD display, we use about 1Mb data registers to do this. The previous plant used a 90/30 CPU for each machine and update time for BCD display was 3,5", program
scan per spinning mill was 13 to 15ms. With SoftPLC we are updating the plant(s) every 1,5" and program scan (EIGHT (8) spinning mills), and
some more calculations we couldn't do previously, is about 12 to 13 ms, CPU is a Pentium 166Mhz. Data table size for this application is >20K words.

Enrico Guasco
 
A

Anthony Kerstens

Yes. A very good example, and old standby, is Modicon's High Speed Logic Solver. It co-exists in either primary or remote I/O racks, has high speed inputs, and has logic programming in addition to the main processor.

Anthony Kerstens P.Eng.
 
A

Archie G. Mullins

My Opinion? Siemens S7!! High Speed Profibus built-in (12 Mb) I have sucessfully Installed a "Siemens Solution" for a plant for 100 PPM with "Cut to Length" tolerences of +/- 0.020" using a SE70 - CUMC Servo System with the code loaded on the Drive via F01 Option. I was able to reduce waste by 100% and increase production by 65%. Step 7, however, is not the friendliest software in the world. It is VERY Powerful!!! It is avalible in 7 different programing languages (STL,Ladder,FBD,Graph,HiGraph,SCL,C++) As you know, with flexibility comes Complexity!!! I hated Step 7 at first, but now, I wish every project I did was Step 7. If you get familiar with STL (Statement List) you'll throw rocks at the other stuff. Its up to you!!!

Good Luck!!
 
Top