back in March Cameron Dorrough posted some interesting
information on his PLCs.
PLC 1 (tank farm)- CPU and 2 I/O racks with 128 digital inputs , 120
digital outputs , 20 analog inputs ( 4-20mA) , 0 analog outputs.
PLC 2 ( intermediate storage ) - CPU rack and 2 I/O racks, 96 digital
inputs , 64 digital outputs , 20 analog inputs ( 4-20mA ) , 0 analog
PLC 3 ( Refinery head ) - CPU and 2 I/O racks with 256 digital inputs
, 128 digital outputs , 32 analog inputs ( 4-20mA ) , 0 analog
My question :- are these proportions of analog to digital I/O typical?
On 19 Aug 2003 02:33:47 -0700, email@example.com (tom) wrote:
I vote no. I see 25-30 different PLC's a week, and don't see any real
"pattern". Of course, Cameron's application is likely in a much larger
plant than I usually deal with. In the automotive sector, I see plc's
with no alalog, near-all analog, and all in between.
it you had to average over all the systems you see in a year
what percentage of the inputs would be analog ( ie
4/20mA,+/-10V,thermocouple or RTD) vs digital on/off.
I dont consider MODBUS/Fieldbus as digital signals because they are
bringing connections from remote modules which ultimately have to have
an analog or a digital input.
firstname.lastname@example.org (Steve Cothran) wrote in message
Just to clarify a few things here:
defines "Digital" to mean "The production of a discrete signal based on a
change in state" - not terribly helpful, but certainly nothing to do with
Modbus or Fieldbus (or any other kind of bus ;-)
talks about "Discrete I/O" and "Discrete Logic" in the sense that Walter has
Neither of these sites has any definition of "communications" or "network"
or anything of that nature. The ISA do not seem to publish a list of
definitions on-line and the IICA have removed theirs as well (used to be
The bottom line: Until equipment manufacturers (eg. Siemens, AB and the
like) start using "Discrete" in their documentation, I for one intend to
keep using the term "Digital I/O". I have never seen the word "digital"
used to describe industrial communication networks or protocols.
Walter, if you have a definition of "Digital" in the controls arena that
refers to communications networks and not I/O could you please provide it.
Cameron (the Dinosaur :-)
Digital: "A description of which is stored or transmitted as a
sequence of discrete symbols from a finite set, most commonly this means
data represented using electronic or electromagnetic signals. The
opposite is ".
Digital Communication: "Electronic transmission of information that has been
Note that Websters defines "digital" to describe the *data* - not the method
of transmitting it. And "discrete" is not restricted to "on/off"! Maybe I
should have checked there first...
I'll reply at the top because this is getting rather long. There are
several points of note:
1 - Even the word 'discrete' is ambiguous since it can be used in the sense
of separate, individual or unique. It is correct, by the dictionary, to
say, "There were several discrete analog signals." Controls and
communications engineers avoid such a phrase because it would far too easily
2 - You used a phrase, "...a sequence of discrete symbols ...". That is
exactly the meaning I had in mind. In fact both the word 'digital' and
'discrete' are used here in the way I mean them.
3 - In industry three types of signalling are in use:
i) Signals in which one physical quantity varies in proportion to another.
These could be pneumatic, electric, hydraulic or anything else.
ii) Signals in which a physical quantity switches on or off according to
whether another physical quantity is above or below a certain threshold, or
exists in one of two distinct states. These could be pneumatic, electric,
hydraulic or anything else.
iii) Signals in which information, either continuously varying, on/off, or
some form of encoded text or other, is coded in a complex sequence of binary
states. These are always electronic or optical.
These three need names. The first is commonly called analog. The third is
commonly called digital. Consider the question, "Is that cell phone analog
or digital?" Witness such terms as "Personal Digital Assistant".. This
thing has more than just an on/off switch. Digital computers, digital
communications, and so on for 60,000,000 hits on Google. None of them deal
with on/off switches. So what do you call signal type ii)? We are already
using all three of these signal types and two of them have the same name. I
don't know who was first in calling on/off signals digital. The first I saw
it was Honeywell. I was annoyed about it then, since I could see a future
of digital communications coming. Even then, Honeywell told me that the
signals of their Data Hiway were 'digital'. I asked if that meant they were
the same as what came from a pressure switch and they looked at ;me as if I
was some kind of idiot. Obviously it was not the same digital.
Another question to consider: If someone asks, "Is that a digital or an
analog control system?" do you assume the question refers to a relay system
or to an analog controller implemented using digital technology? What about
an A/D converter? It converts an analog signal to a digital signal. A
milliamp relay that converts 4-20 to a simple on/off signal is not called an
A/D converter. It seems we are, in general very clear on what is meant by
'digital' unless we are talking of discrete signals. In that case it means
In case there is any doubt, the yet unreleased update of ISA 5.1 contains
the following paragraphs taken from various places within the document.:
"3.1.14 Controller - a device having an output that varies to regulate a
controlled variable in a specified manner. A controller may be a
self-contained analog or digital instrument, or it may be the equivalent of
such an instrument in a shared-control system."
3.1.19 Digital - a signal or device that generates or uses binary digit
signals to represent continuous values or discrete states.
3.1.20 Discrete - a signal or device that generates or uses on-off signals
to represent discrete states
(8) Readout/Passive Function I (indicate) normally applies to the
analog or digital readout of an actual measurement or input signal to a
discrete instrument or a distributed control system's video display unit.
a) Analog, digital, and/or discrete control devices and/or functions for
continuous control, indication, calculation, etc. that are micro-processor
based, configurable, communicate with each other, and share control and/or
display functions, such as distributed control and programmable logic
a) Process plant computer implemented regulatory and/or advanced control
analog/digital/discrete control and indication functions that are mini- or
main frame computer based.
In all of these quotes ISA is making it very clear than we are dealing with
three different types of signals here and has given them unique names.
(Totally off topic -- What ever is happening to ISA 5.1? I think they bit
off more than they could chew.)
Actually there is at least one vendor who has been using the term 'discrete'
for switched inputs but I did not make note of the name.
Agreed. This is why, IMHO, the word "discrete" should never be used to
describe I/O. "Discrete" can be applied to multi-state.. but read on.
The "sequence of discrete symbols" could (and in my books should have) just
as easily said "sequence of digital symbols" - I guess that particular site
is well aware of ISA 5.1...
True - but they *do* all deal with two-state (ie. "1/0 on/off") or "binary"
signals. This is really what is behind the term "digital". In the same
sense, "digital I/O" deals with "1/0 on/off" (otherwise known as "binary")
"PDA" and "Digital Computer" describes the internal workings or logic
(remember that Analog Computers do exist). The "analog or digital cell"
phone describes the method of transmission modulation and is a bit curly,
but technically, Analog Communications exist too.
"Binary"?? Technically, there is no reason why they should not be (and
mostly are) called "digital".
Maybe not the same digital, but digital all the same. If you allow the
concept of your pressure switch as "digital data" (off/on) to be separate
from "digital processing" (logic operations) and "digital communications"
(ie. the encoding for transmission of aforesaid digital data - eg. Data
Hiway) you can get around most parts of it.
All are digital because the information exists in only one of two states
(binary). If you think about it, the word "digital" doesn't really tell you
much. It needs to be suffixed by something (eg. "I/O") before it means
anything. This is why entertainment system manufacturers happily plaster
the word "digital" over anything they make - it doesn't actually *mean*
anything and hence they're not lying.
An analog control system does not have two-state I/O. The description tells
you what it *does* not how it works.
Yep. Quite correct. ;-)
No - but there is, technically, no reason why it shouldn't be - it's
actually a "1-bit A/D Converter".
I suspect that there is still argument over the terms. As we've discussed,
"discrete" is definitely barking up the wrong tree..
Were they US-made by any chance??
definitions on-line... << from an earlier post by Cameron.
Definitions are available on line from ISA for members. Go to
www.isa.org and login. Then on the left hand navigation area select
"Members & Leaders", then "Dictionary" then "Definitions".
The dictionary is not a standard; it refers to standards as sources.
Walt has the latest info, however.
In my industrial experience analog is typically used for measurements
of temperature, flow, etc., even if the signal is brought into the
control system by way of communications over a bus. Both discrete and,
more often, digital, are used for on/off inputs such as motor status,
switch status, etc. that have two or more discrete states.
For example, a valve may have two limit switches, and therefore four
possible states: Open, Closed, Mid, and Fail (both open and closed,
according to the switches) This would be two digital inputs, when
looked at from a hardware I/O standpoint, or, depending upon the
software, one or two digital I/O software blocks to handle the valve.
In other words, like everything else, "it depends".
Simple reference to "xx number of digital I/O, yy number of analog I/O
is not precise enough. To be meaningful, depending upon the context,
there should be more definition. ("xx number of valves, each with two
contacts representing open and closed, yy aux. contacts from motor
I'll bet that it is typical for this particular application at this time.
Go back a few years, and I'd expect to see a lot more analog signals, and
fewer digital signals. Go ahead a few years and you'll have few or no
analog signals. Of particular interest is that all the control actions are
digital. Could he have used analog legacy equipment to reduce costs?
Just to follow up my earlier complaint -- how did you interpret the terms
'digital input' and 'digital output'? Reading your answer I get the
impression you understood that Tom had a large number of digital, as in
Modbus / Fieldbus, etc. signals. There will certainly be more of these in
If, on the other hand, you understood discrete switches such as pressure
switches, there will be fewer of them in the future. In the process
industries the change over from F/L/P/T switches to transmitters is almost
complete worldwide for new installations.
But the changeover from transmitters and discrete devices to digital devices
is just gathering speed. I here that Shell International has already
switched to Fieldbus. At this point there is no longer much of a
distinction between analog and discrete or between inputs and outputs.
Hop on the bus, Gus. Turn a new leaf, chief. 'The times they are a
changing," says Dylan in the background.
Walter (the Dinosaur)
As for the ratio of discrete to analog, it can depend not only on the
industry and type of plant, but on individual plants. Most of the
discrete I/O I see on DCSs are for motor starters and valves. Usually
motor starters have one digital out to control and one digital input
for status. However, some people prefer two momentary outputs, one to
start and one to stop. Some have additional status inputs (contact in
MCC bucket indicating status, contact on field disconnect, etc.).
For valves, some people only have an output to open and close the
valve, others have an output and one input for valve position, others
have two inputs, one for valve open, other for valve closed.
Two identical plants, designed by diffent companies or different
people, can have very different I/O counts.
To answer your question: No. It's horses for courses, really. (See below)
There will always be a requirement for F/L/P/T switches way into the distant
future - since for simple applications (eg. high/high level shutdown,
overpressure shutdown, etc.) a switch is a cheap, uncomplicated way of
safely shutting the process down when process parameters are exceeded.
Probably what you really meant is "the change from F/L/P/T switches *as
primary process control*..."
Interesting you should write that - which part of Shell International are
you referring to? This project is for Shell Aviation.. :-)
For this application (and many others like it that I come across) there is
no practical (read cheap) way to do this using device networks.
Since the valves are installed in seriously inaccessible locations
(aboveground and belowground), the operators require *hardwired* remote
control of the valve actuators so they can still control the pipeline if the
PLC system craps itself. There are two ways this can be done:
Dual-redundant everything or hardwired control. We chose hardwired control,
since that is what they are already using on-site and since there are not
enough valves to warrant a Pakscan system or similar.
No! that is NOT what I really mean. F/L/P/T switches are practially banned
in many installations. The long term cost of ownershbip is considered much
lower for transmitters because they are simply more reliable, provide more
infomation and are tamper resistant. Their quality is much higher. I think
all the brains left the switch department many years ago and all development
and progress is in smart transmitters. This has been going on for more than
ten years now. A while ago I sarted a thread in this NG asking if this had
been a common observbation and indeed it was. It has been for every one of
my projects for the last ten years.
I meant Shell refineries, petro-chemical, gas plants and other process
facilites world wide.
I'm not sure what you mean by pipeline. Out here a pipeline runs from ten
to several thousand miles. Hardwired is totally out of the question. I
always find it odd that we can run the Trans Canada Pipeline spanning
thousands of miles, with all its jet engine turbo-compressors, metering
stations, etc. by microwave link from a control room in a shack on Black
foot Trail in Calgary (I expect the terrorists don't monitor this NG.) but
someone else can't trust a PLC to run a filling station.
I suppose it is typical of a tank farm. Especially the small number of AIs
(tank level temp and pres) and no AOs.. The relationship of DI to DO
(about 1:1) tells me he has a lot of switching valves. Motor controls tend
to have more outputs than inputs. Systems with a lot if interlocks tend to
have more DI than DO. A car wash has virtually all DO and little else. The
concept 'typical for a PLC' has no meaning as they have such a great variety
of different applications. The phrase 'typical of a given industry or type
of process' has meaning.
A small bitch of mine: A Modbus signal is digital, as is Fieldbus, Ether
Net, Profibus, etc. An on/off switch has a discrete signal unless you are
using ASi in which case it does have a digital signal.. This is now
official ISA terminology.
Tom, Walter is quite correct about the application. It was indeed a tank
farm - three of them to be exact. The I/O quantities I listed included
around 20% spare capacity rounded up to the nearest 16-point or 32-point
(digital) and 4-point (analog) interface.
The digital I/O is controlling mostly motor-operated valves and only a
couple of transfer pumps. The valves we are using require 3 PLC inputs
(open/close/fault) and 2 outputs (open/close). The pumps require 2 PLC
inputs (running/fault) and 1 output (run). The analog inputs are for
flowmeters and pressure transmitters (tank levels being a separate Enraf
On Walter's bitch: Perhaps I'm wrong, but my definition of digital is
"on/off". Digital *communications* (eg. Fieldbus/Ethernet) has nothing
whatever to do with process digital *I/O*.
Even using ASi (a commmunications system), I cannot see why there should be
any problem with referring to a switch as Digital I/O and a pressure
transmitter as Analog I/O - there is no such thing as Analog comms now is
there?? <end whinge.. :->
Polytechforum.com is a website by engineers for engineers. It is not affiliated with any of manufacturers or vendors discussed here.
All logos and trade names are the property of their respective owners.