Closing Control Loops

[CBFalconer]

Fred Bloggs wrote:
.... snip ...
If it was me it was supposed to be off the record.

My Thesis advisor once told me that the first time he visited WPI
he spent about an hour on the interstate -- he was looking for
Wu'sta. He passed by Wor-ches-ter several times before it sunk in...

What happened when he needed fish from Gloss-tuh?

--
"If you want to post a followup via groups.google.com, don't use
the broken "Reply" link at the bottom of the article. Click on
"show options" at the top of the article, then click on the
"Reply" at the bottom of the article headers." - Keith Thompson
More details at: <http://cfaj.freeshell.org/google/>
Also see <http://www.safalra.com/special/googlegroupsreply/>

 

[Tony Williams]

My Thesis advisor once told me that the first time he visited WPI
he spent about an hour on the interstate -- he was looking for
Wu'sta. He passed by Wor-ches-ter several times before it sunk
in...

The 'cester' bit in Worcester says that the original
town had been a Roman camp (an obscure one, probably
only there because the River Severn was navigable up
to there from the sea). Your friend was lucky though,
he could have been driving up and down, looking for
the original Olde English name of Weogornaceaster.

 

[Jim Thompson]

[snip]

My Thesis advisor once told me that the first time he visited WPI he
spent about an hour on the interstate -- he was looking for Wu'sta. He
passed by Wor-ches-ter several times before it sunk in...

Pronounced just like the sauce ;-)

And sort of like Gloucester.

Then there's Wooster, OH ;-)

...Jim Thompson

 

[Jerry Avins]

... snip ...



What happened when he needed fish from Gloss-tuh?

Does he slaver it with war-sester-shire saucr?

Jerry

 

[Tim Wescott]

On Saturday, in article
<[email protected]>
[email protected] "Tim Wescott" wrote:
--snip--



I realise that there has to be simplification, but a lot of theory does
not USUALLY cover the applied like what to do when an external sensor
or feedback in the loop says there is a fault and how to make sure the
loop (and the system) return to the correct state (whatever that may be).

An example would be printing shop guillotines that usually require two
spaced out buttons to be pressed at the same time for the whole of the cut
process. On the basis that if both hands are on the buttons then the
operator's hands are not in the machinery! Now dpending on how the system
operates there are basically three safety stop states - stop, move back
a little or return to start position.

I have seen some control systems really screwed by an 'abort' function
that need complete restarts or worse after that, basically because the
control loop hardware/software could not effectively reset correctly.

If you were to write an article about this (beyond "think hard about
your problem and deal with sensor failures") I'd be delighted to read it.

I haven't really addressed this kind of thing in detail, because I've
mostly worked with loops that lack redundant sensors and which were
inherently mechanically safe. I _do_ have an abhorrence of controllers
with modes, because the mode change always seems to be awkward at best.
I suspect that a modeless controller is going to be inherently better
positioned to deal with sensors coming and going (not perfectly, just
better, often).

Quite often I have seen customers who do not understand the limitations
and assumptions in their systems let alone any model they think they have
created.

Some of the major control loop problems I have seen is applying open loop
blocks to an overall system that is closed loop. Mainly because they
did not understand the limits of the blocks they were using, most
notably delays, determinicity and sampling restraints.

I don't think it's just that people don't understand the limits of their
models. I think that people are so used to academic problems that they
forget that their models might possibly not be sufficiently accurate.
There's at least a few places in the book where the phrases "if your
model is accurate enough" or "if your linear model applies" appear. I
also made a point in the chapter on dealing with nonlinearities of
showing what happens when you blithely use a linear model to a design a
controller for a nonlinear system (a big scary/embarrassing oscillation,
in the example).

--

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

Posting from Google? See http://cfaj.freeshell.org/google/

 

[Tim Wescott]

The 'cester' bit in Worcester says that the original
town had been a Roman camp (an obscure one, probably
only there because the River Severn was navigable up
to there from the sea). Your friend was lucky though,
he could have been driving up and down, looking for
the original Olde English name of Weogornaceaster.

In this case if there was a Roman camp it was _really_ obscure: it's
Worcester Massachusetts, no doubt named after the 'real' Worcester. Any
'original' name would have been from the local Native American tribes.

--

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

Posting from Google? See http://cfaj.freeshell.org/google/

 

[Fred Bloggs]

In this case if there was a Roman camp it was _really_ obscure: it's
Worcester Massachusetts, no doubt named after the 'real' Worcester. Any
'original' name would have been from the local Native American tribes.

Who were slaughtered in their sleep by the English settlers...

 

[Fred Bloggs]

In this case if there was a Roman camp it was _really_ obscure: it's
Worcester Massachusetts, no doubt named after the 'real' Worcester. Any
'original' name would have been from the local Native American tribes.

The English stole it from the Nipmuck Indians with some funny money and
cloth. The Indians had no concept of exclusive ownership, they thought
the English were buying the right to share, and killed more than a few
of them when they found what they were about.

 

[Jim Thompson]

Paul Carpenter wrote: [snip]

Some of the major control loop problems I have seen is applying open loop
blocks to an overall system that is closed loop. Mainly because they
did not understand the limits of the blocks they were using, most
notably delays, determinicity and sampling restraints.

I don't think it's just that people don't understand the limits of their
models. I think that people are so used to academic problems that they
forget that their models might possibly not be sufficiently accurate.
There's at least a few places in the book where the phrases "if your
model is accurate enough" or "if your linear model applies" appear. I
also made a point in the chapter on dealing with nonlinearities of
showing what happens when you blithely use a linear model to a design a
controller for a nonlinear system (a big scary/embarrassing oscillation,
in the example).

Indeed! Though I don't use the material on any regular basis (*), I
had four semesters on non-linear control theory in grad school. Great
fun, the real world!

(*) Though I recently was involved in a laser beam "wobulator"... a
mirror driven by a power amplifier... great fun tuning it up for
stability ;-)

...Jim Thompson

 

[Fred Bloggs]

I don't think it's just that people don't understand the limits of their
models. I think that people are so used to academic problems that they
forget that their models might possibly not be sufficiently accurate.
There's at least a few places in the book where the phrases "if your
model is accurate enough" or "if your linear model applies" appear. I
also made a point in the chapter on dealing with nonlinearities of
showing what happens when you blithely use a linear model to a design a
controller for a nonlinear system (a big scary/embarrassing oscillation,
in the example).

You should apply for this one next go-round. There might be big bucks in
it for you. The deal is to stabilize the internal mirror pointing system
against acoustic wave interference induced into the turret by platform
motion through the air.
http://www.dodsbir.net/sitis/archives_display_topic.asp?Bookmark=19430

 

[John E. Hadstate]

Fred Bloggs wrote:

-snip-

If it was me it was supposed to be off the record.

My Thesis advisor once told me that the first time he
visited WPI he spent about an hour on the interstate -- he
was looking for Wu'sta. He passed by Wor-ches-ter several
times before it sunk in...

When I went to Woopie Tech, it was "guys only". The nearest
"gulls" were at some sort of prep school just down the
street that ran through the center of campus.

And now I have to add my favorite rant: I hope you
discussed at some length the fact that every control loop
needs an auditor, a second, independent measurement device
that is not involved in closing the loop. The reason for
this is that once you have closed the loop, you cannot use
that measurement to know anything about what's really going
on in the process over the long term. If the controller is
functioning properly, the measurement (at the controller)
will be forced to track the setpoint no matter what is
actually happening in the real world. Providing such
auditors used to be standard practice in the process control
industry until about 15 years ago when engineers lost
control of their plants and bean counters started
"cost-reducing" everything. In fact, we used to have RTDs
designed with that in mind: one RTD to control and one to
audit in the same probe.

Also, of course, we had the clever E&I technicians.
Operators would complain that the controller measurements
weren't the same as the auditor measurements, so some dork
with a 12-inch screwdriver would "re-calibrate" the
RTD-to-current-loop converters to make the auditors "read
right."

 

[Jerry Avins]

Also, of course, we had the clever E&I technicians.
Operators would complain that the controller measurements
weren't the same as the auditor measurements, so some dork
with a 12-inch screwdriver would "re-calibrate" the
RTD-to-current-loop converters to make the auditors "read
right."

If sensors are in the same place, they ought to read the same value.
Audit probes did me the most good when they were located somewhat apart
from the control probe.

Jerry

 

[John E. Hadstate]

If sensors are in the same place, they ought to read the
same value. Audit probes did me the most good when they
were located somewhat apart from the control probe.

No!! Between the sensors and the computer is usually a
device ("converter" or "amplifier") to produce a 4-20 mA (or
10-50 mA) signal from whatever the sensor produces
(resistance, millivolts, etc.) Back in the computer room,
there is often another converter to change the 4-20 mA
signal to something like 0-10 Volts or 1-5 Volts. These
devices almost always have "Span" and "Zero" adjustments
accessible to "the dork with the 12-inch screwdriver".
Consequently, without a serious threat to "break fingers",
these converters are often used to make the computer say
whatever the operators want it to say.

 

[BFoelsch]

On Sun, 26 Feb 2006 13:16:39 -0500, "John E. Hadstate"

BIG SNIP

Also, of course, we had the clever E&I technicians.
Operators would complain that the controller measurements
weren't the same as the auditor measurements, so some dork
with a 12-inch screwdriver would "re-calibrate" the
RTD-to-current-loop converters to make the auditors "read
right."

Oh yes. 30 years ago, when I was in Control Engineering at a large,
nearly dead company that makes photographic film, we had a constant
battle wuth the Instrument Techs, who on average knew absolutely ZERO.

Standard procedure was to slip the pens on the shaft until the pen and
pointer agreed. What they agreed ON was irrelevant. Calibrations were
all "end-to end;" there was no telling anything from the 4-20 mA loop
signal. 50 percent of PV would indicate something near 50 on the
instrument, but the loop might be at 16mA!

We had an area full of high-end environmental chambers, which were
capable of great precision when properly set up. USUALLY, however, you
would look in the chamber and find a $1.99 desktop
thermometer/hygrometer that the operator put in as a check on the
$10,000 instruments whish were ALWAYS completely fouled up. Head of
the temperature calibration lab was a kindergarten teacher with no
technical education or credentials, who hired technicians in her
image.

Strange thing, that company still can't understand why their costs are
so high. Lost big money for 5 quarters in a row now.....

 

[Genome]

No!! Between the sensors and the computer is usually a device
("converter" or "amplifier") to produce a 4-20 mA (or 10-50 mA) signal
from whatever the sensor produces (resistance, millivolts, etc.) Back in
the computer room, there is often another converter to change the 4-20 mA
signal to something like 0-10 Volts or 1-5 Volts. These devices almost
always have "Span" and "Zero" adjustments accessible to "the dork with the
12-inch screwdriver". Consequently, without a serious threat to "break
fingers", these converters are often used to make the computer say
whatever the operators want it to say.

Yes, I am use LabView so I am have no problims. Evin cumputer get burn up
coz ov over monkey mistak. ....

Also hav 'honest' backup what no-wun am figur owt. Bonus, am also
iso9005.02.03

Nise Coffeeee plus Doner Kebab...... Easy.

DNA

 

[Fred Marshall]

[snip]

My Thesis advisor once told me that the first time he visited WPI he
spent about an hour on the interstate -- he was looking for Wu'sta. He
passed by Wor-ches-ter several times before it sunk in...

Pronounced just like the sauce ;-)

And sort of like Gloucester.

Then there's Wooster, OH ;-)

and Warrik (RI) ...

 

[Jim Thompson]

[snip]

My Thesis advisor once told me that the first time he visited WPI he
spent about an hour on the interstate -- he was looking for Wu'sta. He
passed by Wor-ches-ter several times before it sunk in...

Pronounced just like the sauce ;-)

And sort of like Gloucester.

Then there's Wooster, OH ;-)

and Warrik (RI) ...

I used to take the New Haven... RR from Boston to DC, then switch to
the C&O on into Huntington. Trying to figure out those conductor
calls was quite a challenge.

...Jim Thompson

 

[John E. Hadstate]

Yes, I am use LabView so I am have no problims. Evin
cumputer get burn up coz ov over monkey mistak. ....

Also hav 'honest' backup what no-wun am figur owt. Bonus,
am also iso9005.02.03

Nise Coffeeee plus Doner Kebab...... Easy.

DNA

Yep, monkeys with screwdrivers were a big problem in my
plant (as was a raccoon who decided to have a barbecue in
the business area of one of the 230 KV switching substations
;-)

Seriously though, there is a subtlety buried here that
should be brought out. We didn't really become focused on
the problems I described until we started using various Data
Historians in a big way (actually about the same time that
people were getting interesting in things like ISO-9000 and
Statistical Process Control.) After a lot of thinking, a
lot more talking, and several demonstrations where our
historical data showed us that what was happening was not
what we thought was happening, we began to get a handle on
some of the problems.

I can't emphasize enough how important is the collection and
long-term storage of process data to the health of a plant
that has a lot of closed-loop control, but don't rely on the
controller measurements to tell you anything about what's
happening in the process. This principle should be pasted
on every control engineer's forehead: "If a process variable
is important enough that it needs to be automatically
controlled it is important enough that it needs to be
independently audited."

 

[Spehro Pefhany]

[snip]

My Thesis advisor once told me that the first time he visited WPI he
spent about an hour on the interstate -- he was looking for Wu'sta. He
passed by Wor-ches-ter several times before it sunk in...

Pronounced just like the sauce ;-)

And sort of like Gloucester.

Then there's Wooster, OH ;-)

and Warrik (RI) ...

The train wicket guy in the UK who sold me the ticket to 'Lie-stir'
looked at me funny, but sold the ticket to Leicester nonetheless.



Best regards,
Spehro Pefhany

 

[Jerry Avins]

No!! Between the sensors and the computer is usually a
device ("converter" or "amplifier") to produce a 4-20 mA (or
10-50 mA) signal from whatever the sensor produces
(resistance, millivolts, etc.) Back in the computer room,
there is often another converter to change the 4-20 mA
signal to something like 0-10 Volts or 1-5 Volts. These
devices almost always have "Span" and "Zero" adjustments
accessible to "the dork with the 12-inch screwdriver".
Consequently, without a serious threat to "break fingers",
these converters are often used to make the computer say
whatever the operators want it to say.

I disn't say they /would/ read the same, but that thay /ought/ to. When
two thermometers in the same stirred pot of soup indicate different
temperatures, at least one of them is wrong. Loop or no loop.

Jerry