September/October 2010

Your Letters

Widening the focus

I really enjoyed Jim Pinto's article about re-engineering of engineers (March/April 2010 InTech).

When I was a young professional, I used to focus in technical issues; after many years of pure technical work, I could widen my sight. In fact, I got a management position.

Anyway, my business approach is still very technical, which I think limits my growing perspectives.

My challenge today is to get more knowledge related to business and combine it with my background, which for sure will make a big difference.

Claudio Ginobili, Engineering Manager, Siemens Argentina



Thank you for your feedback on my article about engineers re-engineering themselves.

In an engineering business, more engineers are needed in sales, marketing, and other disciplines.

I wish you continued success.


Jim Pinto

Industrial Ethernet all the rage

First of all, I think that all articles of this type (referring to the InTech March/April feature, "Industrial Ethernet all the rage") make it a point to explain that they are talking about Ethernet as used in an industrial setting versus the term "Industrial Ethernet," since Rockwell has managed to successfully use that term to refer specifically to EnetIP. Unfortunately, many people in the control industry do not know the difference and make that assumption.

The author attempts to infer that the standard TCP/IP Ethernet is not acceptable in the industrial network. This is simply not true, as UDP used with a vendor's own means of acknowledgement is not only very fast, but quite reliable and can provide all the failure modes he mentions. All the special, grossly complex protocols in the world will not make up for a poorly designed Ethernet network.

However, as an example, using EnetIP can double or triple the cost of a network and at the same time still require the use of good network design. I recently had a client who did not need a network of micro second timing yet he was convinced that EnetIP was the way to go. I could not convince him of the problems associated with using EnetIP as well as the cost.

He has, on his own, come to the conclusion that EnetIP is really a farce, and that the cost of implementation, the number of vendors who have successfully implemented it, and all other factors made using ModbusTCP way more attractive as well as Opto22's implementation of TCP and UDP. What he did not know is, any use of EnetIP equipment is automatically going to require the use of IGMP snooping and fully-managed switches throughout the network.

Normally, a simple network design does not require anything more than "quality" unmanaged switches much less those that have IGMP snooping. In addition, this industry has the propensity to infer that everything must happen in "real time," which of course is a misnomer in itself. A-B's default polling time of 10 ms is also a joke for several reasons. Number 1 is that it can bring a network to its knees instantly if IGMP snooping is not used to prevent it from doing so. Number 2 is that most likely, the A-B PLC does not even have the capability to respond to replies at that rate with a normal load of programming, and certainly not if all the polls are at that speed. Number 3: how many applications actually require that level of updates?

Granted a high-speed bottle line may need it at a local level, so I suggest a separate local network segmented and on a separate switch from the overall network or better yet, use a local device network or simply a high-speed serial network. If you really want determinate networking, it was invented years ago-it was called ARCNET.

Barrett Davis

Managing risk

In "Deepwater Horizon oil disaster lessons" article (July/August InTech), you pointed to the problem of management that moves quickly through the ranks to a senior position without learning anything. I think this is partially true, but only a part of the problem-and maybe not the most important part.

The management fast track is a problem for people who are arrogant and assume they know everything. It can get completely out of hand in a command and control type of management structure. It is not so much a problem for managers who are really smart and understand the need to get expert opinion before making a big blunder. Consensus is always slower and more expensive, but it can avoid unmanaged risk that can put a whole organization in a bad spot.

A bigger problem, in my opinion, is our current system of corporate incentives. Managers at all levels are paid based on short-term results. Based on incentive programs, managers often choose very risky business options or options with known bad long-term consequences. Managers (especially at the top) can make a lot of money by doing bad things to a corporation and cashing in short-term gains. If executive compensation was tied to long-term gains (say 10 years), many decisions would have a whole new context and outcome.

There are well-known ways to manage risk in a professional manner. We don't make use of these techniques, not because of ignorance, but because they are not financially rewarding to investors and managers who can make a quick profit from doing the opposite of good risk management. When financial institutions and corporate compensation systems both reward bad behavior, we get bad behavior. This type of behavior inevitably leads to big events instead of operational excellence.

Herman Storey, Herman Storey Consulting, LLC, Kingwood, Texas