pid control, 4-20 or 1-10 vdc? 4

[PNA]

we have an application that has 1 temp controller and 2 modulating valves.

1 valve is 6 feet from the controller, the 2nd is 25 feet from the controller

not having been involved in this design, i was asked to see why 1 valve was behaving proper, but the one further away was not working at all

they were using 4-20ma with the valves wired in parallel

i suggested they use a 1-10 vdc output to achieve this control (both valves opening and closing the same amount)

it works much better now
however when another tech called the controller manufacturer, they suggested using 4-20ma, wire it in series and use an amplifier to boost the signal to the furthest valve,

can someone explain why this would be better?
and does pid "work" better using 4-20 versus voltage?

 

[ScottyUK]

In a properly designed system it should make no difference to the PID controller whether you use voltage or current output. Current output has some definite advantages in the real world because it has better noise immunity and performs better over long distances.

Regarding parallel and series: if you wire two 4-20mA positioners in parallel the current will split between the valves, in theory giving 2-10mA to each but in reality they are unlikely to divide equally. Wired in series both valves see the full 4-20mA signal.

The chances are the 'amplifier' is a repeater or a booster which only introduces a small burden on the driving circuit and provides an output to drive the second valve. With two positioners in series it is possible that the current source - PLC, DCS, etc - will run out of voltage to drive the current through the higher resistance; using the repeater ensures that both loops have adequate voltage to drive 4-20mA around the loop.


----------------------------------

If we learn from our mistakes I'm getting a great education!

 

[dpc]

Yes. The 4-20 mA signal is much less susceptible to RFI, induced noise, ground loops, and other troublesome things that we must deal with on an industrial environment. Twisted shielded pair wiring should still be used. It's OK to use 10V levels within a control panel, but once the signal leaves the panel, I'd strongly recommend 4-20 mA signals.

The original installation would probably have worked fine if they had just wired it up correctly.

 

[PNA]

not sure i understand how a current signal wired in series would work, even with an amplifier

how would you compensate for kirchoff's law with a single current source and 2 paths of variable impedance?

just curious

 

[dpc]

It acts as a current source - the output voltage produced will vary as required up to the compliance limit of the signal source.

There is a maximum ohmic burden that can put on the signal, but up to that point, you can put as many load devices in series as you want.

It's done all the time.

Check this little guy out:

http://www.datel.com/data/meters/dms-an20.pdf

 

[waross]

I think by "amplifier" someone meant a power supply. This installation needs someone who understands 4-20 ma loops to look at it for a few minutes.
PNA, I don't mean to belittle you. There was a time when I did not know anything about 4-20 ma loops, but this is a very simple installation. It is not any more difficult than using ohms law the other way around from what we are used to to calculate voltages in a series street lighting circuit.
The controller varies its internal impedance (resistance) to control the current.
Somewhere in the circuit is a power supply (often 24 VDC) to supply the voltage.
The power supply may be stand alone or it may be hidden in the controller.
The valves and the controller (if there is a stand alone power supply) will all have either a voltage rating or a resistance rating.
If the rating is in Ohms, multiply by 20 ma to get volts.
Add the volts of the valves (and the controller if there is a stand alone power supply). The summ of the volts should be less than the voltage supplied by the power supply.
If you still have trouble post the name and cat number of the controller and valves, and someone may be kid enough to look up the ratings for you.
Please try yourself first, we would rather check your work than do it for you.

Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[roydm]

Yes, instruments are designed to work on mA not voltage.
You shouldn't need any additional components, just wire the two valves in series as DPC says. The distance you mention is insignificant, 4-20 mA signals will go for hundreds of meters without having to worry about voltage drop.
The reason that your valves won't work in parallel - I suspect they have different input impedances. The one that works is much lower so of course it gets all the current whereas the other is high so it only gets a fraction. Even if they were identical then each valve would only get 2-10 mA each, thats not enough!
Take your multimeter on Ohms and tell us what each valve input reads. While you are at it also make sure the input is isolated from ground.
Unlike a lot of other instruments valves don't usually have a 1-5V input, it could be any odd voltage. Two wire valves in particular have a higher input impedance because they need to draw more power out of the loop to operate.
You also need to study up on the controller as DPC pointed out and make sure the burden (sometimes called maximum load resistance) is greater than the sum of the two valves.
If the sum of the valves is greater than the burden or one or both grounded then you will need to purchase a signal isolator.
Instruments put out current, not voltage, they will continue to put out the same mA no matter what the load until they run out of voltage.
Let us know what you find out, If you can't find the specs on the controller tell us the make & model
Regards
Roy

 

[Skogsgurra]

The superiority of 4-20 mA signal vs 0-10 V has been debated for decades. The common wisdom is that the current loop is superior.

My experience is the opposite. The voltage signal is easier to use in field installations (my experience is mostly with paper machines and steel mills where distances are in the 10 - 300 m range) and it is less prone to noise pick up. Of the around ten "problem installations" I have dealt with over the decades, there has not been a single 0-10 V installation, but many 0-20 or 4-20 mA installations.

The reason is simply that you are not at all dependent on wire resistance any more. That was the prime reason why mA signals were introduced long time ago. Having mechanical instrumentation, where the receiver was a d'Arsonval coil, called for well calibrated and compensated cabling in order to avoid wire resistance influence on the transmission. RF and transients pick up was not at all a problem at that time. It was simply a question of voltage drop in wires.

Introducing the 0-20 mA (or 0-60 mA, as it were in the very beginning) overcame that problem. The transmitters were designed with a Thevenin equivalent with internal resistance high enough to "laugh" at loop resistances in the 100 - 1000 ohms range and that worked very well. Insulation was not a problem since all receivers were floating (coils are inherently isolated) and could easily be connected in series.

The 0-10 V receivers today have input impedance in the Mohms range. Current in the caables is therefore seldom more than about ten microamperes. That makes voltage drop a non-problem. Even 100 ohms, which is a rather long cable does not introduce an error larger than 1 mV, which is 0.1 % of the signal range. The common installation has a much lower resistance and, hence, lower error.

What about RF pick up, then? No problem. Including a low pass filter in the transmitter output (typically 47 ohms, a 10 nF to GND and another 47 ohm just before the otput terminal takes care of "back feed" from field wiring to last base/emitter in the driver stage and input filtering, which is easy, since the impedance is high and band-width in process signals seldom is more than a few ten Hz.

So, the bad reputation the 0-10 V signal has is not because it is bad, but because its predecessors used current levels that were about 10 000 times higher and, therefore, caused voltage drops in the installation. Today's systems do not have that problem.

The current loop system, on the other hand, actually has problems today. A very recent case is a test stand for large diesel engines. There, the vibration transducers are outputting 4-20 mA signals that are sent over around 20 m long cables to the instrumentation, which is in the same space as the VFDs (around 1000 kW) that regenerate braking power. As soon as the VFDs are started, the vibration level (or, rather, the transducer's output) goes to nil - not exactly what one would expect.

The reason is that HF RFI from the generator cables introduce interference in the vibration transmitter outputs which takes the signal below 4 mA, which the system, of course, evaluates as wire break. This system has been heavily filtered but there are still occasions where "wire breaks" occur. And, the calibration of these transducers is something that must be questioned all the time. The truck manufacturer is now contemplating to move over to the 0-10 V signal system that has been tried for some time in parallel to the 4-20 mA system.

I think it is about time to abandon that "knee-jerk" automatic answer that current loop is less sensitive to HF and transients interference. It simply isn't true. And it never was, actually.

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[Skogsgurra]

Error: "Even 100 ohms, which is a rather long cable does not introduce an error larger than 1 mV, which is 0.1 % of the signal range" shall read "Even 100 ohms, which is a rather long cable does not introduce an error larger than 1 mV, which is 0.01 % of the signal range".

It doesn't matter, really, since transducer error usually is in the percent range. But if error is important, I think that 0.01 % (at the extreme end) is not a big problem.

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[PNA]

i thank everyone for responding to my question!

1) one issue is we have 2 valves that are supposed to have the same impedance as seen by the circuit, but are actually different enough to make it a pain. if this is true on a lot of components, (variations of impendance), would this not be a pain to try and match up in the field?

2) aside from using 2 separate dedicated outputs or a more intelligent control solution to minimize debugging, the voltage scenario was recommended by our supplier. and yes we tried the 4-20 as indicated in series, and for whatever reason, they both behaved differently

again, i appreciate everyone's responses.

paul

 

[roydm]

Skogs,
You may be correct in your preference for voltage signals around a generator but for the majority of instrument instalations 4-20 mA is superior. Some people seem to be under the illusion that mA signal is all about noise, it's not, the mA signal powers up the instrument. Take a pressure transmitter for example, it needs power for the straingauge or whatever cell it uses and the amplifier. to run it on voltage you would need a minimum of 3 wires and some form of isolation between each loop.
Control valves are another issue. In the past the positioner or I/P operated with a magnetic transducer similar to a speaker coil, so you need current to drive that. Sure you could send it a voltage but any resistance in the cable at all would give a voltage drop thus effecting the calibration. 4-20 avoids that because the driving device just puts out more voltage until the mA reaches it's level.
I don't think you will ever convince an Instrument tech that voltage signals are superior.
Regards
Roy

 

[dpc]

Gunnar,

Thanks for your thoughts on this - I'm sure you have a lot more real world experience troubleshooting RFI and interference issues than I do. You've given us some things to think about.

I think there can be problems with either approach if proper wiring, shielding and grounding methods are not used.

In my limited experience, I've never had a significant interference or noise issue with a 4-20 mA loop. I've had a issues with 0-10 V signals, even inside of a control panel. But I'm far from an expert.

Cheers,

Dave

 

[Compositepro]

It does not matter if the valve electrical impedance is equal or not - 4-20 ma transducers should never be wired in parallel. The transducers are designed to work on 4-20 not 2-10. A 10 ma signal would only open the valve half-way. A 2 ma signal would represent a negative number.

The big advantage of 4-20 ma compared to 0-20 ma, 0-10 v, 0r 1-10 v is that the instrument power is (or can be) delivered on the signal lines, so there is lower wire cost.

But, Gunnar's points about current versus voltage signals are of course correct.

 

[Skogsgurra]

Thanks for the comments. The loop powered advantage is, of course, unbeatable.

And, yes, any voltage signal that has not been conditioned/protected/filtered correctly can pick up interference. In that respect, all solutions are equally bad - more or less. A badly designed voltage channel and a badly designed current channel both pick up noise and react to it. Since mA loops are usually better engineered than ad hoc volt signal solutions and that often explains experiences like dpc's.

One distinct advantage with voltage signals is that simple resistors or potentiometers can be used without any electronics at the sending end.

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[itsmoked]

PNA; You should never put two 4-20mA devices in parallel. They will never work reliably. They could be matched to a tee in impedance only to have the sun hit one or the controlled product change temperature and the impedance of just one would change.

Friends don't let friends parallel 4-20..

Keith Cress
kcress -

http://www.flaminsystems.com

 

[PNA]

Again, thankyou everybody!

Paul