SIWAREX MS on S7-200 CPU 226 for Weight Feeder Limits

Problem Overview

A retrofit specification calls for six independent loss-in-weight / belt weigh feeders, each handling a different bulk material. The proposed hardware is an S7-200 CPU 226 with one analog output (EM 232) module and six SIWAREX MS (7MH4930-0AA01) weighing modules, fed by six load cells, six speed sensors, and six MICROMASTER 420 (MM420) inverters driving the feeder belts. No TP/HMI is in scope. The architectural question is whether this combination can close a weigh-feeder control loop at all, and if it cannot, what is the minimum-cost change that makes the system work.

The short answer, drawn from the SIWAREX MS application notes and Siemens S7-200 hardware manuals, is that the SIWAREX MS is a static weight module: it digitises a load cell signal and reports a net weight to the CPU, but it has no digital input for an encoder/speed sensor, no PID controller, and no totaliser. Closing a belt weigh-feeder loop (load × speed = flow) is therefore a job the CPU 226 must perform in user code, with hard limits on expansion current, program memory, and execution time.

Belt Weigh Feeder Functional Requirements

A belt weigh feeder is a closed-loop controlled conveyor. The control target is a setpoint flow Q_set in kg/s or t/h. The controller measures the instantaneous flow and trims the belt speed to track the setpoint. The four functional blocks are:

1. Load measurement. One or more load cells under the idler measure the mass per unit length q [kg/m] of material on the belt.
2. Speed measurement. A pulse wheel or inductive sensor on the tail or head pulley reports belt linear speed v [m/s].
3. Flow calculation. Instantaneous flow Q = q × v [kg/s]. Totalised flow T = ∫Q dt [kg or t].
4. Speed control. A PI or PID controller compares Q against Q_set and drives the VFD reference to adjust belt speed.

Items 1 and 4 are mandatory. Skipping speed measurement leaves you with a volumetric feeder (loss-in-weight hopper) that uses a different control law; skipping PID and driving the belt at fixed speed is just a belt scale, not a feeder.

The signal chain Q = q × v is linear, so an integrator-only controller (PI on the flow error) is sufficient for most solids; derivative action is rarely needed and adds noise. The PI loop is the part the SIWAREX MS does not supply.

What the SIWAREX MS Actually Provides

The SIWAREX MS is a single-channel weighing electronics module designed to plug into the S7-200 expansion bus, addressed as a 16-byte input and 16-byte output area in the EM slot. Its feature set, as documented in the SIWAREX MS Library manual and the 7MH4930 device manual, is:

• Connection of one load cell (or up to four 350 Ω cells in parallel, max 4 × 350 Ω or 1 × 85 Ω equivalent).
• 16-bit weight output, configurable resolution up to ±32 768 parts.
• Update rate 10 ms (100 Hz), with adjustable filtering (low-pass IIR, 0.05–5 Hz).
• Zero/tare/calibration via S7-200 data words.
• Theoretical calibration, digital filter, and limit values.
• No digital input for an incremental speed sensor.
• No PID or totaliser block.
• No direct analogue output to a VFD.

The SIWAREX MS is intended for static scales: tank weighing, silo weighing, platform scales, and loss-in-weight control of a screw or vibratory feeder where the feeder speed is fixed and the discharge is gated by a slide. It is not a belt scale integration block. Confirming this in the SIWAREX MS configuration software (SIWATOOL MS) makes the limitation visible: the only process variable exposed is net weight.

S7-200 CPU 226 Hardware Envelope

The CPU 226 (6ES7216-2AD23-0XB0 and successor 6ES7216-2BD23-0XB0) is the top-end S7-200 CPU. The relevant published limits for the proposed build are:

The 660 mA Expansion Bus Limit

The SIWAREX MS at 5 VDC draws approximately 120 mA (per the SIWAREX MS Library manual, table 4). Six modules therefore need 6 × 120 = 720 mA. The CPU 226 (and CPU 224) supplies a maximum of 660 mA to the expansion bus, as called out in the S7-200 System Manual and reiterated in the SIWAREX MS Library readme. This is a hard physical limit, not a configurability issue. The CPU 226 cannot be modified to source more 5 V current.

Two workarounds exist, both with caveats:

1. Split into two CPU 226 stations. Three SIWAREX MS on one CPU 226 (3 × 120 = 360 mA, well within 660 mA), three on a second CPU 226. The two CPUs share data via the second RS-485 port (PPI/Modbus master) or via an EM 277 Profibus slave. Adds cost, breaks the single-CPU architecture, and complicates commissioning.
2. Use a CPU 226 XM or CPU 226 DC/DC/DC variant. Same 660 mA budget; this does not solve the issue.

The only S7-200-family CPU that supplies more 5 V expansion current is the CPU 226 CN/AC variant with an external 24 V booster, and even that is undocumented for this use. Practically, the single-CPU 6-module build is electrically invalid.

Memory Budget for the Software Workaround

If the customer accepts the workaround of running flow calculation and PID entirely in STEP 7 Micro/WIN, the V-memory budget for one belt is roughly:

The data side fits. The code side is the binding constraint. The SIWAREX MS basic library blocks (FB41-style weight reading, taring, calibration) consume about 600 bytes of code per instance; six instances plus 6 PI loops plus 6 totalisers run close to 5 KB. The CPU 226 user program is rated 8 KB after edit/compile overhead; the build fits, but there is no room for alarm handling, recipe storage, or comms to a future HMI. Adding a TP177 or a Modbus master to a feeder line overruns the budget.

I/O Address Map for the Proposed Build

The intended slot layout, given the seven-slot limit, is:

This is the first architectural problem after the 660 mA limit: a single EM 232 with 4 (or 2) analogue outputs cannot drive six MM420 inverters independently. Either:

• Add a second EM 232 (slot 7) and split the feeders across two CPUs, or
• Use a multi-drop RS-485 USS bus to the MM420 inverters and drop the analogue reference entirely.

Architecture Diagram (Build-vs-Buy Decision)

Two things stand out from the diagram: the dashed pulse-sensor lines run to the CPU 226 on-board HSC inputs (no extra EM needed), but the speed reference path to the inverters must be USS, not analogue, because there is no EM 232 slot left once the 7th EM is reserved for a DI module.

Software Workaround: Computing Flow and Totalising

The flow law is straightforward. The PLC reads gross weight from the SIWAREX MS in engineering units (kg) and the belt speed from the HSC in m/s, and multiplies them.

A representative STEP 7 Micro/WIN STL snippet for one belt is shown below. Variables are local to OB1 and the belt index is unrolled six times in production code (or written as a loop with indirect addressing on VW).

// Inputs
NETWORK 1  // Read SIWAREX MS gross weight (kg) into VD200
// SIWAREX MS FB "MS_FB" returns gross weight in VD600 (REAL32).
LD     SM0.0
CALL   MS_FB, VB100, AIW16, AQW16, VD200, ...

NETWORK 2  // Belt speed from HSC0 period measurement (m/s)
// HSC0 is configured for period mode with 1 ms timebase.
// Pulse wheel: 100 ppr, roll diameter D = 0.200 m
// v = pi * D / (ppr * T)  where T = measured period in s
LD     SM0.0
MOVR   0.0628318, VD210        // pi * 0.200 m / 100 ppr = 0.00628
/R     HSC0_PV_S, VD210        // HSC period in s
MOVR   VD210, VD220            // belt speed in m/s

NETWORK 3  // Mass per unit length and flow
// Weigh idler length L = 1.000 m
LD     SM0.0
MOVR   VD200, VD230            // m in kg
/R    1.0, VD230               // q = m / L in kg/m
MOVR   VD230, VD240            // load factor q
*R     VD220, VD240            // Q = q * v in kg/s
*R     3.6, VD242              // Q in t/h (VD242 = VD240 * 3.6)

NETWORK 4  // PI controller (parallel form, anti-windup)
// Setpoint Qset in t/h (VD260), Kp (VD270), Ki (VD272, 1/s)
// Sample time Ts = 100 ms (SM0.5 toggles 1/2/0/5s; gate with a TON of 100 ms)
LD     SAMPLE_OK
MOVR   VD260, VD280            // Qset t/h
-R     VD242, VD280            // e = Qset - Q
*R     VD270, VD284            // P-term
MOVR   VD284, VD288            // uP

NETWORK 5  // I-term with clamp
LD     SAMPLE_OK
MOVR   VD272, VD300            // Ki
*R     VD280, VD300            // Ki * e
*R     0.1, VD300              // * Ts
+R     VD304, VD300            // I = I_prev + Ki*e*Ts

// Clamp to -100..+100 %
LDR>=  VD300, 100.0
MOVR   100.0, VD300
LDR<=  VD300, -100.0
MOVR   -100.0, VD300

NETWORK 6  // Sum and output
LD     SM0.0
MOVR   VD288, VD310            // P
+R     VD300, VD310            // + I
MOVR   VD310, USS_SPEEDREF_1   // 0..100 % scaled to 0..4000 hex for USS

NETWORK 7  // Totaliser, 32-bit kg
// Increment by Q [kg/s] * Ts [s] each cycle; store in VD320 (DINT, kg)
LD     SAMPLE_OK
MOVR   VD240, VD330            // Q in kg/s
*R     0.1, VD330              // * Ts
ROUND  VD330, VD332            // kg this cycle (REAL -> DINT)
MOVD   VD320, AC0              // previous total
+D     VD332, AC0              // new total
MOVD   AC0, VD320              // store back to V memory

PI Tuning Starting Points

For a belt weigh feeder with 1 s nominal belt transit time and 50 kg/m maximum loading, reasonable PI starting values (truncated from Ziegler–Nichols open-loop step) are:

Tune in this order: (1) set Ki = 0, raise Kp until the loop oscillates with constant amplitude, (2) cut Kp to 50 % of that, (3) raise Ki until the oscillation is critically damped, (4) verify with a 10 % setpoint step.

Recommended Alternative 1: SIWAREX FTC on S7-300

The SIWAREX FTC (7MH4900-2AA01 / -3AA01) is a multi-channel weighing + control module for the S7-300 / ET 200M. It is the Siemens-blessed path for belt weigh feeders, gravimetric feeders, and differential discharge weighers. Its relevant features versus the SIWAREX MS:

Pair the two FTCs with an S7-300 CPU 315-2 PN/DP (or a CPU 314 if no Profinet is needed) and you have headroom for an HMI, Profinet to a SCADA, and recipe handling. The cost is roughly 1.6–2× the S7-200 build, but the engineering risk collapses because the FTC implements the weigh-feeder law (load × speed) on the module itself, not in user code.

Recommended Alternative 2: Milltronics BW500

The Milltronics BW500 (now marketed as the Siemens SITRANS WW500 belt scale integrator) is a standalone belt scale controller with a single channel of load cell input, a single speed-sensor input, on-board PID, on-board totaliser, and a Modbus / Profinet / Analogue output option. It closes the loop in firmware with no external PLC. For a 6-feeder retrofit where the customer is cost-sensitive, the typical bill of materials is:

• 6 × BW500, one per belt
• 6 × load cells (already in the design)
• 6 × speed sensors (already in the design)
• 6 × MM420 inverters (already in the design)
• 0 × PLC. The BW500 is the controller.

Eliminating the S7-200 entirely removes the 660 mA limit, the memory budget problem, and the speed-sensor wiring problem. The trade-off is that the BW500 does not speak USS to the MM420 by default — it uses a 0–10 V or 4–20 mA analogue output, which is fine because the BW500 already has a 4–20 mA process output per unit. The MM420 accepts 0–10 V on terminal 3/4 and that closes the loop with no PLC in the path. For Modbus or Profinet aggregation of all six feeders into a SCADA, the BW500 supports Modbus RTU on its RS-485 port, and a Modbus master (which can be a TP, a SCADA, or a small S7-1200 if needed) reads the six units in a poll cycle.

Decision Matrix: Which Path to Take

Commissioning Steps for the Workaround (If You Must)

If the customer insists on the S7-200 + 6 × SIWAREX MS architecture and accepts two CPU 226 stations, proceed in this order.

1. Power budget check. Measure 5 V current at the expansion connector with all six SIWAREX MS inserted. The reading must be < 660 mA per CPU. If it is not, the design is invalid; revisit slot count.
2. Mechanical zero each SIWAREX MS. With the belt empty, run SIWATOOL MS, send CAL_CMD_ZERO, verify the net weight reads 0.000 ± 0.001 kg.
3. Calibration with test weights. Apply a known mass (typically chain weights equal to design load, e.g. 50 kg/m × 1 m = 50 kg) and run CAL_CMD_SPAN. Repeat for each of the six channels; record the mV/V reading for traceability.
4. Speed sensor scaling. Drive the belt at 50 % of rated speed using a handheld tachometer. Adjust the pulses-per-revolution in the HSC configuration until the PLC reports belt speed within ± 0.5 % of the tachometer reading.
5. PI tuning. Run each belt empty at a 10 % setpoint step. Raise Kp until sustained oscillation appears, halve Kp, then raise Ki until the step response is critically damped. Record Kp, Ki for the recipe.
6. Totaliser verification. Run a known mass (e.g. 100 kg calibrated weight, repeated 5 passes) through the belt. The totaliser reading at the end must match the calculated mass within ± 0.5 %.
7. Loss-of-weight test. Cut material feed; the controller must ramp the belt to maximum speed to maintain setpoint, then fault on empty belt within 5 s.
8. Non-volatile backup. Store Kp, Ki, setpoint, and totaliser to a recipe in the data log area protected by the CPU's retentive memory map. Document the recipe for each of the six materials.

Verification and Acceptance Tests

Before hand-over, the following checks must pass and be logged:

Troubleshooting Matrix

Field-Proven Caveats

• End-of-life of the S7-200 family. Siemens formally withdrew the S7-200 from new sales in 2017; spare-parts availability for the CPU 226 and EM 232 is constrained. A design based on the S7-200 in 2024+ is a service liability, not just an engineering compromise.
• SIWAREX MS calibration drift. In dusty environments the load cell zero drifts 0.1–0.3 % per shift. Auto-tare at every belt-stop or a slow zero-track (5 LSB/min) is mandatory, not optional.
• USS at 9 600 baud. Six inverters polled at 9 600 baud 8N1 take ~50 ms per cycle of USS traffic, which is significant on a 100 ms scan budget. Set P2010 to 19 200 baud only if cable length is < 50 m and EMC is good; otherwise stay at 9 600.
• Belt speed sensor mounting. A proximity sensor on a clean pulley face is far more reliable than a tail-pulse wheel that collects material. Specify the sensor location in the mechanical drawing, not at the wiring stage.
• Totaliser non-volatility. The CPU 226 retentive range is VW0–VW511, but only if the super-capacitor or battery cartridge is healthy. Document the retention time (typically 50 h on a fresh BC293 battery) and replace every 2 years.
• EMC for load cell cables. Run the 4-wire load cell cable in a dedicated conduit, min 200 mm from VFD power cables. Crossing at 90 ° is acceptable; parallel runs are not.

Summary Recommendation

The proposed S7-200 CPU 226 + 6 × SIWAREX MS architecture does not close a belt weigh-feeder loop without violating at least one hard limit:

1. The 5 V expansion budget of the CPU 226 is 660 mA; six SIWAREX MS draw ~720 mA.
2. The 7-slot EM limit leaves no slot for a digital input module to read the speed sensors (the on-board HSC inputs can read them, but no slot is left for the EM 232 required for analogue speed references to the MM420 inverters).
3. The user-program budget of ~8 KB cannot fit the SIWAREX MS library, six PI loops, and a totaliser without a future-proofing gap.
4. There is no on-board PID, no on-board totaliser, and no speed input on the SIWAREX MS — all of these have to be coded and validated in STEP 7 Micro/WIN.

The right Siemens answers are SIWAREX FTC on S7-300 for a new build (FTC has on-module PID, totaliser, and speed input; S7-300 has 5 V budget for two FTCs), or Milltronics BW500 per belt for a budget retrofit (BW500 is a complete belt scale controller; no PLC required for the loop, a small S7-1200 can aggregate Modbus if SCADA integration is needed). The S7-200 + SIWAREX MS combination is appropriate for static scale retrofits (silos, tanks, hoppers) and loss-in-weight feeders with a fixed discharge device, but it is not a belt weigh feeder controller.

FAQ