Encoder vs 4-20mA

The position of a valve is not the thing that is normally controlled. The thing that is controlled is the process variable: temperature, pressure, flowrate, etc..

Any issues with the valve position can be taken up through the correct tuning of the controller by adding integral action.

4-20mA position feedback is fine for nearly all applications, and links directly and simply into most BMS and PLC inputs. It is also interference-free and can detect a loose wire condition on the position circuit without affecting control of the process variable.

What is your application?

measuring water level thru a float and pulley

Why not use an ultrasonic level device instead?

What about waves and foam on the surface?

What about the change in vessel shape with temperature?

There are several things to consider here - the thing you mustn't loose sight of is the end result you want to achieve.

Positional information is derived from a quadrature encoder output by using an up/down counter to track the number of pulses, with the relative phase giving directional information. Using a fairly economical model, giving say 1000 pulses per rev, and a pulley sized to give 20 turns for your full range of movement (say 10 metres), you would have a total of 20,000 pulses over the 10m, so your smallest unit would correspond to 0.5mm (i.e. an accuracy of ±0.25mm). This would be 100% repeatable (neglecting mechanical imperfections). As the measurement is based on pulse counting, there would be no temperature variation.

The big problem with this approach is that the counter is incremental; position information and tracking is lost when the power is off, and the float would have to pass some known datum position to reset the counter.

A "4-20mA" analog device would generally be based on a multi-turn potentiometer. I checked out the datasheet for one we have used recently, and the accuracy is given as ±0.1% with a repeatability of ±0.02%. Resolution is described as "essentially infinite" - it depends on your analog-to-digital conversion electronics. With a 10m span you would get an accuracy of ±10mm. (again neglecting mechanical imperfections). The analog-to-digital conversion could also be subject to errors due to temperature effects.

The major advantage is that the positional data is absolute - after cycling the power off and on, the position is known without any recalibration etc.

There is a third, much more expensive, route which gives the advantages of both approaches, which is to use an absolute encoder and a mechanical turns counter. This could give similar accuracy and repeatability to the incremental encoder, with the retentive advantages of the analog approach.

I could go on, but that's a summary of the two methods you've mentioned. There are other aspects to be considered, such as the potential of count errors due to noise (in the case of the incremental encoder), and what you want to do with your measurements. Thing is, what output do you actually need, and how deep is your pocket?

"The big problem with this approach is that the counter is incremental; position information and tracking is lost when the power is off, and the float would have to pass some known datum position to reset the counter."

John, what if he used an absolute encoder?

Aside from that possibility, an encoder for level control has a bigger inherent problem in my opinion, that of signal dampening. With an analog signal it's easy to ignore transients, i.e. waves, with a simple time buffer. If you were to use an encoder the position signal would be bouncing around and required a de-bounce routine in software to interpolate an actual level value. Seems to me that can be complex.

But where did the OP ever say this was for level feedback, or even valve position control feedback? If we are referring to linear position of a rotating element, I would choose an encoder any day over a tachometer with a 4-20ma output.

"John, what if he used an absolute encoder?" ... Please see 6^th paragraph in #5.

"With an analog signal it's easy to ignore transients, i.e. waves, with a simple time buffer..." - if by 'time buffer' you mean time-averaging the (digitized version of) the analog signal, there's no difference between to two approaches. If you're talking hardware analog filters, this is (IMHO) just as simple to do in software using an encoder signal (I've done it).

I don't see any practical difference between twiddling a knob to set an analog filter parameter, and moving a slider on an HMI panel to get the same effect with a digital filter.

"But where did the OP ever say this was for level feedback ...?" - in his #3, replying to my #2.

"... linear position of a rotating element ..." - sorry, not meaning to be contentious, but you lost me there.

Veeder-Root

Take the free motion control Webinair from Automation Direct on Sept 1:

Go to www.automationtalk.com

Or call them up and order what they tell you to buy. It won't be easy, but Automation Directs internet and product documentation is good enough to figure out how to do what you want on some level.

As already said, the sophistication you seek sort of depends on how deep your pockets are - but you can learn a lot without spending too much, and maybe come up with something usable.