Getting belt weigher data into your PLC or SCADA

What output options does a belt weigher integrator offer?

Every integrator offers some subset of these. Which ones you have depends on the model, and often on which option card is fitted, so start from the manual for your exact unit and firmware.

Analogue 4-20mA: the classic rate output

The oldest and most robust option is a single 4-20mA current loop carrying the rate. It is scaled so 4mA means zero t/h and 20mA means a maximum rate you choose (for example the belt's design capacity), and everything in between is proportional. Because it uses current rather than voltage, it is not bothered by long cable runs or a bit of voltage drop, which is why it has outlasted almost everything else. The starting value of 4mA rather than 0mA is deliberate: it is a live zero, so a genuine zero rate still draws 4mA and a reading of 0mA tells the PLC the loop itself is broken.

Two things it will not do. It carries rate only, never the total, because a total is a running count and a single analogue level cannot represent one. And it is only as good as its scaling: the same 12mA means one rate on a loop scaled to 1,000 t/h and a very different rate on a loop scaled to 2,000 t/h, so the integrator's full-scale setting and the PLC's scaling have to agree exactly. Get the loop wiring and isolation right (know whether the integrator sources the loop or expects the PLC to power it) and a 4-20mA rate output will run for years.

Pulse and relay outputs: counting the total

To get the total across, most integrators offer a pulse output: a contact closes briefly for each unit of totalised weight, say one pulse per tonne, and a remote totaliser or a PLC counter input adds them up. It is the analogue trick in reverse, the total handled as a stream of counts instead of a level. It is simple and hard to break, but the PLC has to keep pace with the pulse rate and never miss one, so the weight-per-pulse is chosen to keep the pulses comfortably slow at full rate.

The same relay hardware usually provides alarm outputs: a contact that changes state on high or low rate, on a fault, or on a zero or calibration condition, wired straight into the PLC as a digital input. Simple, reliable, and enough for many older installations that only ever needed a total and a couple of alarms.

Serial comms: RS-232, RS-485 and Modbus RTU

RS-232 and RS-485 are wiring standards, not protocols. RS-232 is point to point and short range, one device at each end. RS-485 is the plant workhorse: a differential pair that runs much further and is multi-drop, so several devices can share one twisted pair down the belt line.

Modbus RTU is the protocol, the agreed language, that is usually spoken over RS-485. It works on a master and slave arrangement: the PLC or SCADA is the master and does all the asking, each integrator is a slave with its own unit address, and the master polls them one at a time by address. Over a single RS-485 pair you can therefore read rate, total and status from a string of integrators, which is a large step up from one wire per value. The cost is that you now care about addresses, baud rate, data format and the register map, covered below.

Modbus TCP: the same idea over Ethernet

Modbus TCP is Modbus RTU's register model carried over ordinary Ethernet instead of a two-wire pair. The way you read a value is unchanged: the same registers hold rate and total. What changes is the physical layer. Instead of daisy-chaining a twisted pair and handing out unit addresses, you plug each integrator into a network switch and give it an IP address, and the PLC talks to it over the site network. It is faster, easier to fault-find with standard network tools, and it drops the practical limits on distance and device count that RS-485 imposes. If your integrator has an Ethernet port and your PLC is networked, this is usually the tidier choice.

PLC-native industrial Ethernet: EtherNet/IP, Profinet and Profibus

Some integrators speak the PLC’s own protocol, so the weigher appears in the PLC program like any other I/O, with no register map to hand-build. EtherNet/IP is native to the Allen-Bradley (Rockwell) world and Profinet to the Siemens world, and on older Siemens plants you’ll still meet Profibus DP. The site’s PLC decides which you want, so match the integrator’s option card to the control system already in place.

Which comms gotchas actually bite in the field?

The comms rarely fail outright. They deliver a number that is quietly wrong, which is worse, so this is where most of the real work goes.

Work on or near a conveyor, or inside an electrical enclosure, needs isolation under your site’s procedures and trained, authorised people. If that’s not you, stop and call us on 1300 101 666.

• Every register map is different. There is no universal layout for where rate and total live. Each brand, and sometimes each firmware version of the same brand, puts them in different places with different scaling. Always work from the correct manual for the exact model and firmware in front of you, not a similar unit down the line.
• Scaling and the decimal point. A register usually holds a raw integer, not a nicely formatted number, and the integrator's display applies an implied decimal point that the raw value does not carry. Read the raw count as the finished figure and you can be out by a factor of ten or a hundred. Confirm the raw-to-engineering-units rule before you trust anything.
• Two registers, wrong word order. A total is too big for one 16-bit register, so it spans two of them as a 32-bit value. Modbus is a big-endian protocol, but brands differ in which half they send first, so if the two 16-bit words are assembled in the order the other end does not expect you get an absurd total, wildly too large, tiny or negative. This word-order mismatch is the single most common cause of a nonsense total, and swapping the word order is the usual fix.
• Float versus integer. Some registers hold a plain integer, others hold a floating-point value packed across two registers. Decode a float as an integer, or the reverse, and the number is meaningless. The manual tells you which each register is.
• Watchdogs and polling rate. Many links use a comms watchdog: if the PLC stops polling, the value is flagged stale rather than left frozen at its last reading, which is the safe behaviour. Poll too fast and you can swamp a slow serial device; poll too slow and a watchdog trips. Set the polling rate to suit the link.
• RS-485 wiring near drives. A multi-drop RS-485 run needs termination at the ends and correct biasing to sit in a known state when idle, and it needs a proper earthing and shielding scheme. Belt lines are full of variable speed drives, and their electrical noise couples straight into a poorly terminated or poorly earthed serial pair, giving intermittent comms errors that come and go with the plant. Get the termination, biasing and earthing right and most of that disappears.
• The SCADA total that does not match the integrator. This is the classic family of faults, and it has many parents: a word-order mismatch, a scaling or decimal error, missed pulses on a counter input, a reset that only cleared one end, or two systems totalising over different windows. We never chase it from the SCADA screen alone. We run it down end to end, because there is one number that cannot lie: the integrator's own display.

For the mechanical and signal faults that sit upstream of any of this, see common belt weigher faults, and for the speed signal that feeds the rate in the first place, see the belt weigher speed sensors guide.

What do we do on site?

Commissioning an output is a defined job, not a guess. We set up the chosen output on the integrator, whether that is scaling a 4-20mA loop, setting the weight-per-pulse, or configuring the serial or Ethernet parameters and addresses. We then verify the value where it lands, reading rate and total in the PLC or SCADA and comparing them against the integrator's own face at the same moment, end to end, so we are proving the whole chain and not just one link. Finally we document the working map for the site: which output carries what, the scaling, the addresses and the register layout as actually wired. That last step is the one most often skipped and the one that saves the next person, and it means a future change or a fault does not start from a blank page. For where these numbers surface once they are trended and reported, see reading your calibration report.

Frequently asked questions