Main Index

App. Notes Index

The 4-20mA current loop signalling protocol has been with us for many years, and despite all the digital advances remains popular.  It has one particular characteristic that makes it very suitable for hostile environments, and (within sensible limits) it is immune to the distance from the transmitter or sender and the receiver.  Cable can be added or removed without affecting the accuracy - a unique feature for analogue interfaces.  It only needs 2 wires to work, because the power and signal use the same pair of wires.

Digital interfaces can be used of course, but how many can operate cheerfully with hundreds of metres of cable? The cable itself may be anything from a telephone wire twisted pair through to a length of twin mains cable that "just happened to be handy".  Because the minimum current is 4mA (representing zero input), the system is self-monitoring.  Should a fault develop, it is immediately apparent because the current will be out-of-range ... zero for a break, or exceeding 20mA if there is a short to some other voltage source.

The interface can be tested with nothing more sophisticated than a multimeter (analogue is fine!).  There are specialised ICs available to convert just about any sensor imaginable to the 4-20mA standard, but there is one caveat ... no sensor may draw more than 4mA at idle.  Since the IC or other interface circuit will need some power, there is usually less than 4mA available.  This eliminates many gas sensors and the like, because they commonly have a heated sensing element that draws more current than the 4-20mA standard allows.  Accordingly, some 4-20mA applications require local power for the sensor and transmitter circuits, but this only applies to a few specialised sensors.  A third wire may be used in some cases, providing power to the sensor electronics.

4-20mA is the standard current loop, but there are also others that have been used over the years.  Various manufacturers have come up with their own variants, but none of these can be considered standard.

There are also measurement microphones that use a 4mA current loop, but this is a completely different arrangement.  These microphones are supplied with a fixed current of 4mA, and it does not change with variations of signal level.  If there is enough interest, this is covered in some detail in Project 134 - 4mA Current Loop Microphone.

The general arrangement of a 4-20mA interface is shown below.  The receiver is simply any device that can measure the voltage across a resistor, and may be analogue or digital.  The sense resistor is typically between 100 and 500 ohms, but in many cases 250 ohms is considered 'standard'.  125 ohms is also a common sense resistor value, but it really depends on the receiver electronics.

The transmitter circuit is the interface between the sensor and the 4-20mA loop.  Whether discrete or using a specialised IC, the transmitter takes a signal (typically a voltage from a sensor, but resistance is also common) and converts it to a current.  The current is directly proportional to the input.  With no signal or at the lower limit of the sensor, the current is 4mA, rising to the maximum of 20mA at the upper limit of the sensor.  The 4mA standing current is an offset that allows the transmitter to function, and provides a confidence signal to show that the loop is operational.

The final part is the power supply.  This may be within the receiver unit in some cases, but it's not at all uncommon for it to be completely separate.  The voltage supply needs only to be capable of supplying a maximum of 20mA for as many sensors that it powers.  The voltage can be anywhere between 12V and 36V, although you may see 48V used on occasion.  24V is the most common and is well suited to most applications.  It is important to understand that the voltage doesn't actually matter, provided it is enough to overcome the loss across the sense resistor and wiring resistance, and still leave enough voltage at the transmitter to allow it to function.

Figure 1 - Typical 4-20mA Block Diagram

If the sense resistor is 250 ohms, the voltage across it will be 5V at 20mA and 1V at idle (4mA).  The cable resistance might be 400 ohms, so 8V will be 'lost' across the cable itself.  The sensor and sender may need a minimum of 12V to function, so we add the voltages ...

V_total = V_sense + V_cable + V_sender

V_total = 5V + 8V + 12V = 25V

In this case, you would choose a 36V supply, as this provides a good safety margin and allows for the cabling to be extended if needs be.  While this might seem like a strange thing to do, this signalling scheme has been used in thousands of industrial applications (including mining) where things are changed regularly.  The last thing anyone needs is a requirement that the system be recalibrated just because someone extended or shortened a cable!

Herein lies the real advantage of using a current loop.  If a good current sink is used in the sender unit, the cable resistance and power supply voltage don't change the calibration at all, provided they remain within the designated range.  The above system with its 36V supply will work perfectly with as much as 950 ohms of cable, or as little as 1 ohm.  If someone were to replace the power supply with a 48V unit, even more cable could be used, and none of these radical changes affects the calibration.  No other analogue system can compete, and very few digital signalling schemes can be used either.  Because the signal is analogue, it is often possible to operate happily even with high noise levels on the signal pair, because the noise can be filtered out without affecting the DC voltage.

In some cases, it is useful to connect the transmitter between the +ve and -ve terminals of a bridge rectifier as shown in the above block diagram.  This means that the transmitter is not polarity sensitive, so if the wires are swapped around inadvertently the system keeps operating normally.  Because of the current loop, this will not affect calibration.

4-20mA current loops are not used or suitable for high speed applications, and the applications where they are most commonly used don't need high speed.  The speed limitation is due to the fact that by definition, a current source has a very high impedance, so cable capacitance will limit the frequency response even at quite low values of capacitance.  However, if the pressure in a large (perhaps LPG) gas tank is monitored, it will never change quickly under normal conditions.  If it does change fast the reason is likely to be visible! Even so, the current loop is certainly fast enough to show there's a problem.

Figure 2 shows a simple sender, which is actually a dedicated tester and is based on a design I did for a client who needed a 4-20mA checker/calibrator unit.  In this case, the sensor is simply a 200 ohm pot, and the range is from 3.977mA to 20.01mA.  Both are well within 1% of the design values.  Needless to say, the odd value resistors are either obtained using parallel resistors or trimpots for calibration.

Please note that there is a circuit all over the Net that claims to be a 4-20mA tester.  It is no such thing.  You will find that there are several discrepancies between the text and the drawing, and the text refers to a 4-20mA signal that ramps up and down and also indicates that a PICAXE micro-controller chip is used.  The circuit shows a 7555 timer and some other stuff that is in no way, shape or form suited to testing 4-20mA interfaces. I have no idea who was the first moron who screwed up the text and (stolen) circuit, but countless others have done likewise and followed like a flock of dumber-than-average sheep.  The Net is now completely polluted with a circuit that is utterly useless for the claimed task.  No-one seems to have noticed that they have simply stolen a schematic and text that don't match.

If the supply voltage or series resistance is changed across the range limits (12-24V and 125 ohms to 500 ohms in this case, but specifically excluding the zener), the current changes by less than 0.1% using this very basic circuit.  The most critical part is the temperature coefficient of the resistors and zener voltage regulator, as these will have more effect than external electrical variations.  The circuit as shown relies on the zener regulator having an external supply, because the zener draws much more than 4mA.  This doesn't matter in this case, because it's a tester and is self contained and mains powered.

Figure 2 - 4-20mA Test Sender

This is the circuit of the tester, and it is not intended as any kind of real 4-20mA interface.  However, it can be used to test receiver units and their associated analogue to digital converters, software, etc.  That is exactly what it was designed for, and it works very well indeed.  There is a dearth of any published circuits for 4-20mA testers, and that's the reason this unit ended up in the Application Notes section.

To understand how it works, there are two very important parts of the circuit.  VR1 and R6 are the parts that determine the current.  VR1 set a voltage at the non-inverting input of U1, and due to the opamp insisting on making both its inputs the same voltage, exactly the same voltage will appear across R6.  VR1 can be varied from 160mV to 0.8V and the same voltage will appear across R6 because of the feedback around U1.  160mV across 40 ohms is 4mA, and 0.8V across 40 ohms is 20mA ... there is the 4-20mA current required.

Everything else in the circuit is simply there as support.  The zener diode provides a stable reference voltage from the already regulated 12V supply, the resistors around VR1 set the upper and lower voltage limits, and Q1 supplies the current.  The meter is simply there because this is a tester.  D1 gives the opamp a negative supply - it's only 0.65V, but enough to allow the inputs to work properly at very low voltages.  D3 and C3 help protect the MOSFET from external nasties, and C3 also prevents the MOSFET from oscillating with long leads attached.

These days, you'd be hard pressed to find a modern 4-20mA interface that uses anything other than a dedicated IC.  They are made by several manufacturers, and have a variety of special characteristics.  Simply select the IC that suits your sensor, and the IC does the rest.

One of the biggest problems with any 4-20mA interface is the minimum current of 4mA.  The minimum current is all that's available to the sender, including the sensor.  This is difficult to achieve in some cases, so remote power supplies for senders are not uncommon.  However, there are still plenty of applications where no remote power is needed, and this is the way the interface was originally designed to operate.  A 3-wire variant exists where the sensor needs additional processing.  The third wire is power - typically 12V or 24V DC.

While there is quite a lot of info on the Net about the new ICs that are used, there seems to be remarkably little that discusses the older discrete senders.  Figure 2 is an example, but a critical part of any measurement system is the voltage reference.  Specialised devices exist today, but many years ago a zener diode was as close as you'd get.  Choosing the correct voltage is important - only a very limited range of zener voltages have an acceptable temperature coefficient.  A 5.6V zener is generally accepted as having as close to zero tempco as you can normally expect, but this may not apply at the low current needed for a 4-20mA current loop.

These days, it much simpler to use a precision voltage reference, such as the LM4040 shunt regulator, available in several different voltages.  You may also use one of the many band-gap voltage references available - these are typically 1.25V and have excellent performance, but can be comparatively expensive.

The AD693 is pretty much a complete system on a chip.  The only thing you need to add is your sensor, and the data sheet has many examples and other info to help you to create a working sensor and sender unit.  Unfortunately, the application details are not intended for those who have no prior experience with 4-20mA interfaces.

Normally, you would simply read the voltage across the sensing resistor, but there is always the 4mA offset, so this has to be removed.  If the sense resistor is 250 ohms, 4mA leaves you with 1V across the resistor.  This is easily subtracted by using a circuit such as that shown below.

Figure 3 - 4-20mA Receiver

In the circuit shown, the input is buffered by an opamp.  This prevents the input circuit around U1B from placing a load across the resistor and changing the calibration.  It makes far more difference than you might imagine, but the same result can be achieved by increasing the value of R1 very slightly so that the total resistance (R1 in parallel with R3 + R5) remains at 250 ohms.  Without the buffer or any correction, the error is over 1%, yet current loop interfaces can be better than 0.1% accuracy and linearity.  A 1% error is therefore significant.

The offset voltage is also critical, and needs to be stable with temperature.  The arrangement shown is a very simplified version, but like the transmitter, a precision reference is needed for high accuracy applications.  U1B simply subtracts the reference voltage from the voltage across R1, so with no signal (4mA or 1V) the output voltage will be zero.

Should this voltage ever become negative, a fault is indicated (loop current missing or minimum is too low).  With no loop current at all, the output will be 0V.  At the maximum current of 20mA, 5V is developed across R1, 1V is subtracted, and the output is 4V.  This voltage may be amplified further if needed, and used to drive an analogue instrument (such as a meter) directly, or can be digitised and used by a data logger, computer or micro-controller based circuit, or read by a digital meter.  Resolution and accuracy are determined by the stability of the voltage references in the transmitter and receiver, as well as the sensor itself.

Any 4-20mA system can be set up with additional detectors.  For example, when the loop is active, there will be a minimum of 1V developed across R1, because 4mA will be flowing.  Should the loop be broken by a damaged cable or faulty connector, the voltage across R1 may fall to zero, or may show AC (typically 50 or 60Hz) due to leakage.  These abnormal conditions are easily detected and can be used to trigger an alarm.  It may also be necessary to provide input protection, using MOVs (metal oxide varistors) or TVS (transient voltage suppressor) diodes.  Many industrial systems can be particularly hostile.

In summary, while the 4-20mA current loop protocol is seemingly well past it's 'best before' date, it is still used in countless industrial processes.  It is a robust and well proven technology that refuses to go away because it is robust and reliable, and works where other more recent protocols may give nothing but trouble.  While it lacks the fancy attributes of many digital bus systems, it will work over almost anything that conducts electricity and is easily extended, shortened, tested or repaired in the most hostile of environments.  Don't expect it to disappear any time soon.

1. AD693 - Analog Devices, Loop Powered 4-20mA Sensor Transmitter
2. LM4040 - National Semiconductor, Precision Micropower Shunt Voltage Regulator
3. DMS-AN-20 - Murata, 4-20mA Current Loop Primer, Application Note
4. XTR115 - Burr-Brown, 4-20mA Current Loop Transmitters, Datasheet

Main Index

App. Notes Index