Responding to the Response

Four main points of contention were raised by Jim Cahill in his blog responding to our white paper “PROFIBUS PA vs. Foundation Fieldbus, a cost comparison”:



   Devices on a segment


Pricing.  Jim points out that other factors beside cost determine the selling price of automation devices.  That’s true.  FF devices need more expensive parts as a result of the larger stack and associated processing power, but vendors do not necessarily have to charge more.  They do though in the comparison that the article cites – by about $100 per device more for an FF device than a PA device.

Linking.  Jim points out that at least some controllers can have an FF segment attach directly to them, so linking devices are not needed.  Also true.  However, if the instruments are far removed from the controller or segment loading requires multiple segments, linking devices are needed.  The article’s example required multiple segments for either FF or PA, so linking devices were needed – linking to HSE for FF and to PROFIBUS DP for PA.

Devices on a segment.  Jim mixes a discussion of determinism with the number of devices on a segment. The article does talk about devices and segment and update times in two different sections. When just talking about devices per segment in a non-IS environment, the FF Engineering Guidelines clearly state that no more than 12 devices should be on a network. For PROFIBUS, the recommendation is no more than 24. For the case of determinism, the arguments on either side are not simple. PROFIBUS’ position is that PROIFBUS is deterministic within a fail safe timer. Yes, there is a variance, but there is also a variance in FF as well. Therefore you do not need to oversample, which again leads to the fact that you can put more devices on a PA segment than an FF segment.

Easier.  Jim points out that FF and PA use the exact same physical layer.  That’s right.  They also both have a transducer block and Analog Input and Output blocks. The difference is in the overhead caused by “Control in the Field.”  In FF you have to set up the macro-cycle times, and the block linkages. There is simply more to set up.  This observation has been supported by end users who have used both.  Another thing that makes PA easier is that you can access it from DP – the protocol is the same.  And you need DP anyway for the discrete IO.  The software tools you need all work from the DP side so there’s no additional tool needed for that.  You need separate hardware tools for the process cabling side, but that’s not additional – in either the PA or FF case you need hardware tools for a discrete bus and a process bus.  Since both systems in the example require linking devices, no complexity is added in the PA scenario over the FF scenario.

Many of the benefits Jim cites in opening and concluding are about operational benefits and they apply to either fieldbus.  Unfortunately, operational benefits are harder to quantify, leaving the decision to be made solely on the basis of installation costs far too often.

Jim and I agree that you should use a fieldbus to achieve those operational (and installation and commissioning) benefits.  But we have to agree to disagree on which fieldbus to choose.