Variable Speed drive with ControlNet interface

Having had several nasty experiences with ControlNet, I, personally, would
not recommend using it for drive operation (not if you want reliable
operation anyway!).  Maybe this is why other manufacturers don't offer it?
DeviceNet has issues also, although I agree with you that it is more than
adequate for almost all drive applications and there are plenty of
manufacturers happy to provide the interface.

With all of these networks, you'll find that bit-speed means nothing - it's
the actual data transfer rate that's important.  I suspect that, for a servo
drive, a dedicated motion card using something like SSI would be faster than
ControlNet simply because it's a point-to-point link with no need to allow
overheads for other network traffic right in the middle of a critical move.

If you must use comms to a VS Drive, my suggestion would be Profibus.

JM2cW,
Cameron:-)

Folks,
Our Client (The End User) has specified networked drives as site standard
but with E3 Plus relays (A-B) you get only DeviceNet interface, whilst the
VS drives are offered with ControlNet interface.  Hence the reason for my
question this morning.

I have since then found out that ABB's ACS800 drives are available with
ControlNet interface.  One of the explanations provided by Rockwell person
is that a 1788CN2DN acts as a single node on the ControlNet network (with
over say 30 E3 Plus relays hanging off at the DeviceNet end).  Since the end
user has accepted and specified ControlNet, I will go along with it (My
uneasiness has been some what reduced by the knowledge there is an
alternative to Rockwell drives).

Cameron & Jake,
Thanks for your comments.
Regards,
Raj

Raj Sreenevasan wrote:

Raj, how long do you envision this drive system lasting?  Five years?
Ten years?  20 years?  Think about the state of the art back then.  Do
you really think that an industrial net has enough staying power to
leave such a thing in place?  Even if it did, it would morph and
probably present some incompatibilities.  We're already seeing that with
Profibus.

I would use a 4-20 mA current loop to control it.  It is as close to
being a fundamental standard as anything gets.

Of course, if you don't envision this drive having that kind of
longevity, then by all means put someone's network in it.

As for Cameron's suggestions regarding Control Net, your experience may
vary.  Our company uses both Control Net and Profibus.  We don't seem to
be having too much trouble.  Yes, Rockwell has a version control problem
going on here.  It is worse than some, but not the worst I've seen.  And
as far as I'm concerned, it's no worse than Profibus.  Industrial
networks take some tweaking to get them to work.  Welcome to reality.

Jake Brodsky

<rant> A version control problem??  Now, that's an understatement!!  Why
should a modern industrial network be brought to it's knees by a "version
control problem"?  Even Rockwell don't know what's going on! :-(  ...oh, and
try bumping one of the network connections on the scanner (as one of our
commissioning engineers did recently) and see how long it takes to restore
comms... several *minutes* perhaps?? </rant>

After finding this out the hard way, it is now company policy that we do NOT
recommend ControlNet to customers.  To be fair, Devicenet had similar issues
when it was first released (EDS file issues mainly), but this sort of
carry-on is not acceptable at any of the plants we work in.


Begs the question:  What's the worst you've seen? :-)

Cameron:-)

Cameron Dorrough wrote:

How long ago was this?  The earlier major releases were a problem.  We
still occasionally encounter some version hate with some firmware more
than two or three years old.  However, the new stuff seems to have this
under control.  As for rejoining the network, that's not usually a
problem for us.  We try to ensure the physical media is hard enough
(fiber optic) and protected with battery backed 24 volt power systems so
that this won't be an issue.  And in any case, we're talking about large
water utility and waste water utility work here.  Nothing moves that
quickly in this application.

In fairness to AB, most networks are far less complex than what they
attempted with ControlNet.  I think they bit off more than they could
chew, and the result wasn't pretty.


Most of my hateful problems were media related.

Some of the earlier TI-500 I/O networks which used a cable TV-ish
interface were pretty difficult to work with.  The problem was trying to
explain to the folks who were designing and installing it all about the
ways and techniques with RF (I have a radio background too).  The
manuals left everyone gasping in efforts to understand what they were
supposed to do.  They should have just said what it is and what it was.

I can recall intermittent RS-485 networks which didn't work at full
speed (a blistering 56 kBPS!).  The problem was that occasionally the
duct bank would fill with water.  Although the cable was installed with
no splices through that duct bank, it had a polyethylene jacket.  I
spoke to Belden and they patiently explained that polyethylene can
absorb up to two percent of its weight in water --and that's enough to
change the capacitance per foot quite drastically.

But the worst is/was Modbus.  How many flavors are there?  Thirty one at
last count?  And yet we can't seem to get past explaining to people that
Modbus is more a state of mind instead of a protocol.  My favorite rant
about Modbus is that it's the Cockroach of Protocols.  It was first.
There are many species of it.  It will always be lurking in the corners
somewhere, but it will never evolve in to something better than what it
currently does.   The only reason it continues to get perpetrated is
because there are so many tinkerers who think they own it and understand
it.

Jake Brodsky

"version

Six months ago.  After wrestling with the original setup (comms between two
processor racks) for several days "resetting the keeper" and so forth and
finally getting it going, out of the blue Rockwell gave us some obscure
warning that the comms would crash after thirty-something days if we didn't
update the firmware - right now!  The lastest release supposedly fixed the
problem (taking Rockwell 2 man/days to get it going, though - ie. one entire
day of lost production) and I haven't heard anything from site since.  As
long as no-one touches it, it should work fine... touch wood.  But never
again.


Agreed.  Actually, I haven't had too many problems with Modbus - using only
Modicon equipment of course and avoiding all other flavours like the plague!
;-)  The best I've come across lately is Modbus/TCP for high-level comms
between Citect and Schneider PLC offerings - apart from short startup
delays, no problems at all (once you get your head around the addressing)
and best of all... NO stupid "version control problems" to worry about - as
it should be! :-)

Cameron:-)