Analog VS Digital Signal

4-20mA is used because it is simple, rugged, easy to set-up in the field, independent of cable length and type, doesn't hurt anyone that touches bare wires, doesn't blow up sensitive electronics if the two wires are shorted, doesn't blow fuses, and is not prone to interference by cross-talk from other signals. A simple variable resistor can be used as a signal transmitter. It is also historically compatible with virtually everything in the industrial process instrumentation field and entirely compatible with EEx i methods of explosion prevention in the neighbourhood of flammable materials.

Beat that.

4-20ma was the improvement to pneumatic control. The current crop of fieldbus systems are the improvement to 4-20ma. Give them time to supplant the norm. Changes in industry don't happen over night. Heck, they don't happen in 15 years.

I am a controls engineer for a process skid builder and I would rather run a good bus system than 4-20ma. So much easier wiring it. Unfortunately, one draw back is there is no perfect system to handle both analog and discrete, and it is quite common at present to run two bus systems (i.e. Foundation Fieldbus for analog and Profibus for discrete signals). This is part of the hinderence to full acceptance of pure digital. Another is the mistaken notion that the digitization of the analog signal leads to signal error. Yes it does, but even in the analog 4-20ma system, we get digitization at the processor, so you aren't adding any additional error going digital all the way, but this tends to get ignored.

Don't know how many times I've looked at a P&ID with 40 or so valves with limit switches, a dozen or so temperature transmitters and another half dozen pressure transmitters and asked the customer why don't we go to a bus systems, at least with the discrete signal to eliminate the Swiss cheese effect on the control cabinet from 60 or so cable entries and got a resounding "no".

It takes time to change.

I use both.

Once you add communication to a control system, you multiply the setup and programming time. A communication bus should be simpler but it is not. Wiring is the only simplification but you loose the "redundance". If your bus cable is unplug, everything stops. There are ways around but it increase the complexity to the point where you negate the gains.

Communication makes everything more complicated. The added information also increase the logic to monitor and display it. It might be useful at times but it is not necessary for the process.

In today's complex systems, the "keep it simple" approach is often forgotten.

Now, a winning strategy is often to start with a basic system with 4-20mA and then eventually, when resources are available, add the Hart functions to get more info from the sensors. This allows to start simple, fills the low work periods and eventually produces a complete system.

One reason for the continued use of analog signals in industry is "natural standardization": there just aren't that many different ways to make an analog current signal. The sheer simplicity of the concept makes interoperability natural. Even field transmitters generating a different current range (e.g. the obsolete 10-50 mA standard) can be made to work with equipment designed for 4-20 mA: just replace the 250 ohm resistor with a 100 ohm resistor and upgrade the DC power supply. Digital communication, on the other hand, can be done in so many different ways that it takes a concerted effort for manufacturers to make their digital equipment interoperable. Just look at how long it took the Fieldbus Foundation members to finally agree on the FF standard! It seems to me that the most successful digital bus standards in the business often start with one company's good idea, which then becomes a de factor standard as other companies recognize the benefits and adopt it (e.g. Modbus from Modicon, HART from Rosemount, Profibus from Siemens).

Another important reason for the continued use of 4-20 mA analog signaling is familiarity among maintenance technicians. Troubleshooting a 4-20 mA loop is quite a different matter than troubleshooting a malfunctioning digital bus. The tools are much simpler for troubleshooting analog and the problems are always rooted in the fundamental laws of electric circuits rather than the idiosyncrasies of digital data formats. When digital busses do fail from wiring problems such as poor grounding, improper cable termination, insufficient cable bend radius, etc., the problems can be a nightmare to pinpoint. This is why I cringe whenever I hear someone extoll the virtue of low installation cost with digital busses. Sure, there is a lot less wire and fewer connections with a digital bus, and commissioning time is (theoretically) faster, but how does the total lifecycle cost compare to analog? Extra wire and conduit can be a lot cheaper than down-time on a critical process unit.

Reliability is another factor where analog usually wins over digital. A comparison between analog versus digital instrument reliability provides some insight here. The Rosemount "Alphaline" analog model 1151 pressure transmitter at 226 years demonstrated MTBF versus the Rosemount "smart" model 3051C pressure transmitter at 136 years demonstrated MTBF is a good example. Admittedly both of these instruments have 4-20 mA output signals, but the latter has digital internals while the former is 100% analog. Foxboro's legendary SPEC 200 analog control system mentioned by Bigg in a previous post is an excellent example of the generally high reliability you can get from a well-designed analog system. The reason for this is very straightforward: analog systems have fewer components than digital, and consequently exhibit fewer failure modes.

Another issue with digital in general is loop update speed. Analog instruments and 4-20 mA signaling exhibit negligible dead time, whereas digital instruments necessarily introduce dead time into a control loop due to their internal scan rate. As Phys pointed out in his post, digitization anywhere in a system (even within the controller) introduces this kind of artifact into the signal, but some digital systems are much slower than others. HART networks, for instance, poll far too slowly for most process control applications which is why it's typically used as a maintenance communication protocol rather than for critical process data. Even FOUNDATION Fieldbus is limited to about a dozen devices per H1 segment in order to keep macrocycle times less than 1 second. For many process control applications a delay of a second or so is not a big deal, but for others (e.g. liquid flow control, liquid pressure control, compressor surge protection) it can be prohibitive. I've actually worked on systems where we had to install a purely analog transmitter (instead of a "smart" digital transmitter) and a digital controller with a tight scan time in order to achieve the fast response times necessary for stable feedback control. Certainly there are proven examples of fast-responding digital protocols for critical control applications out there (CANbus for automotive applications comes to mind), but I've yet to see this manifest in the industrial control world where physical distances are large, field devices are complex, wiring is custom-installed, and incremental upgrade capability is important.