S7-300 MMC Data Retention: SFC84 WRIT_DBL for CPU 315-2DP V2.0

Problem Description: Weekend Data Loss on S7-300 CPU 315-2DP

A production conveyor application counts boxes and stores the running total in a Siemens S7-300 Data Block (DB). The line is powered down every weekend. When the CPU is switched back on Monday morning, the count value has been reset to 0 rather than continuing from the value present at the previous power-off.

The hardware in question is a SIMATIC S7-300 CPU 315-2DP with an MMC (Micro Memory Card) as the only load memory, programmed with STEP 7 (SIMATIC Manager) V5.3 + SP3. The CPU firmware reported by the user is V2.0. This combination of firmware version and load memory type is the key to the correct root cause and the correct remedy.

Root Cause: Why the DB is Reset to Zero on Power-On

The PLC does not, by default, mirror live runtime values from work memory back to the load memory (the MMC) at every scan. The model to keep in mind is:

• Load memory (MMC): the project, the initial DB contents, and the program.
• Work memory (RAM): where DBs are mirrored at RUN time and where the live counter value lives.
• Retentive memory: a small, battery-backed or non-volatile slice of work memory that survives power-off.

On the S7-300 firmware family, the retentive behaviour of a DB is governed by three things, and the S7-300 CPU 315-2DP with firmware V2.0 follows the rules in the table below.

For this specific system — CPU 315-2DP firmware V2.0 — the retentive behaviour of every DB is implicit and is, by definition, retentive. That means the count should survive a power-off as long as the work memory is preserved across the outage. If the count is in fact being lost, the most likely root causes are:

1. No backup battery fitted in the CPU. The S7-300 has a buffered RAM region that is kept alive by a lithium backup battery. If the battery is missing, dead, or has been removed, the work memory is wiped on every power-off, and the next start uses the initial values stored on the MMC.
2. Wrong MMC type or an MMC that has been re-initialised. Certain MMC operations (format, delete, re-download) force the next cold start to use the initial values from the project.
3. CPU is performing a cold restart (OB102) instead of a warm restart (OB101) or hot restart (OB100). A cold restart reloads the work memory from the load memory and uses the DB initial values.
4. DB was marked "Non-Retain" in a project edit while the firmware was reported incorrectly. The reported firmware is V2.0, so this property should not be active, but verify the firmware actually present at runtime (see Verification section).

Confirming the Real Firmware and the Buffer Status

Before changing code, validate the actual CPU firmware and the buffer configuration in STEP 7 > PLC > Module Information (online > accessible node > Module Information).

If the diagnostic buffer reports a Cold restart as the last startup, the DB initial values are used regardless of firmware, and you will need to either (a) change the startup mode to Warm restart in the CPU properties, or (b) save the value to the MMC with SFC84 before power-off (see below).

Solution Path A: Fix the Retentive Mechanism (No SFC84 Required)

For CPU 315-2DP firmware V2.0, the simplest, most reliable fix is to ensure the retentive path is intact. Carry out the checks below in order.

A1. Verify or Install the Backup Battery

• Open the CPU front cover and confirm a 3.6 V lithium backup battery is fitted (e.g. 6ES7 971-1AA00-0AA0, type "A").
• Measure the battery voltage under load; replace if below approximately 3.0 V.
• Set the battery switch on the CPU to the "BATT" or "ON" position; older CPU 315-2DP have a toggle on the front panel.

A2. Set Retentive Areas for the DB

Open the CPU hardware object in STEP 7 and select Retentive Memory. Define the area of work memory that should survive a power-off. The minimum needed is the DB number and a byte count large enough to cover the counter variable plus any neighbouring tags you want preserved.

With the battery present and the DB declared retentive, the work memory is preserved across the weekend power-off, and the counter will resume from the last value automatically. SFC84 is not required in this case.

Solution Path B: Use SFC84 "WRIT_DBL" to Save the DB to the MMC

If a backup battery cannot be used (some sites prohibit lithium cells on the shop floor, or the user wants a true non-volatile persistence), the alternative is to write the DB from work memory back to the MMC on demand using SFC84 "WRIT_DBL". This makes the MMC a non-volatile mirror of the live DB.

B1. What SFC84 Does

SFC84 copies the contents of a DB in work memory to the corresponding DB in load memory (the MMC). After the call, the next POWER-OFF / POWER-ON cycle or STOP / RUN transition will use the values that were present at the moment of the last SFC84 call.

B2. Sample Ladder Call

Network 1: Save DB10 to MMC on power-fail event
      |     P#+Save_DB10     Save_DB10   ( Edge flag )        
------| /|-----( )-----( S )------------+-----------------------/
      |                                |
      |  "PS-PowerFail"                |  WRIT_DBL
      |     Save_DB10        DB_NR     |   EN   ENO  RET_VAL  BUSY
      |   |   |  (S)         10        |---] [------(SFC84)------
      |   |   |                       |     DB10                
      |   |   |                       |   Save_DB10  (RE trigger)
      |   |   |                       |   /         \
      |   |   |                       |  +-----------+
      |   |   |                       |   RET_VAL -> MW 100
      |   |   |                       |   BUSY    -> M 110.0

Re-trigger the call on a rising edge of Save_DB10. If BUSY = 1, do not call SFC84 again; the call is single-instance per cycle.

B3. Structured Text Equivalent

// One-shot rising edge from power-fail detector
IF "PF_PowerFailEvent" AND NOT "Save_DB10_Trig" THEN
    "Save_DB10_Trig" := TRUE;
    // Triggers SFC84 below
    "Save_DB10_Req"  := TRUE;
END_IF;
IF NOT "PF_PowerFailEvent" THEN
    "Save_DB10_Trig" := FALSE;
    "Save_DB10_Req"  := FALSE;
END_IF;

// SFC84 call
IF "Save_DB10_Req" THEN
    "RetVal_SFC84" := WRIT_DBL(REQ := "Save_DB10_Req",
                                DB_NR := 10);
END_IF;

B4. MMC Write-Cycle Limit (Critical)

Recommended trigger sources for the conveyor count DB:

• Operator "Save" button on the HMI.
• End-of-shift time signal.
• Detection of an imminent power-off via a hardware signal on a digital input wired to a UPS / line monitor.
• OB81 (power fault) if a buffered detection mechanism is in place.

B5. SFC84 Return Value (RET_VAL) Error Codes

Solution Path C: Detecting Imminent Power-Off and Triggering SFC84 in OB81

The user asked specifically whether the PLC can detect the loss of mains and run a shutdown routine to call SFC84. The S7-300 supports this through OB81 (Power Supply Fault OB) and an external line monitor. The procedure is:

1. Wire a line-voltage monitor (e.g. a Phoenix Contact or Siemens SITOP UPS module) to a digital input on the S7-300. The signal must assert at least 1 to 2 seconds before the DC 24 V supply collapses to give the SFC84 call time to complete.
2. Configure the SITOP / UPS module for buffer time > 1.5 s, with a "power-fail imminent" relay contact wired to input I 0.0.
3. In OB81, evaluate the input. Because OB81 runs on power fault and on return of power, distinguish the two events with a flag set in OB1 each scan, and only act when the flag disappears (i.e. a power-down event was just seen).
4. Inside the power-fail branch of OB81, call SFC84 with a one-shot to write the count DB to the MMC. Do not call other heavy SFCs; you have only a few hundred milliseconds to a few seconds of buffer time.

Sample OB81 Skeleton

FUNCTION_BLOCK FB 1 "FB_PowerFailSave"
VAR
    SaveDone : BOOL;
END_VAR
BEGIN
    // Rising edge of mains-OK lost, but DC 24 V still present
    IF NOT "MainsOK" AND NOT SaveDone THEN
        "RetVal_SFC84" := WRIT_DBL(REQ := TRUE,
                                    DB_NR := 10);
        SaveDone := TRUE;            // one-shot, do not retry
    END_IF;

    // Reset latch once power is fully restored and CPU is in RUN
    IF "MainsOK" AND "OB1_Running" THEN
        SaveDone := FALSE;
    END_IF;
END_FUNCTION_BLOCK

OB 81 "Power Supply Fault OB"
BEGIN
    // FB_PowerFailSave.DB10 instance
    "InstPowerFailSave"(MainsOK := "I0.0_MainsOK");
END_ORGANIZATION_BLOCK

Comparison of the Three Solution Paths

Verification Procedure

After applying the fix, perform the following verification to confirm weekend-style behaviour does not return.

1. Online > Monitor > force the count DB to a recognisable value, e.g. DB10.DBD0 = 12345.
2. Power down the PLC completely (turn off the line isolator; do not just STOP the CPU).
3. Wait at least 30 seconds; longer than the typical SITOP buffer time, to mimic a true weekend.
4. Power the PLC back on.
5. Connect online and read DB10.DBD0. It must still be 12345. If it is 0, the retentive mechanism is not active: recheck battery and DB retentive settings, or verify that the SFC84 path actually executed in OB81 before power-down.
6. Inspect the diagnostic buffer for the last startup type. It should report "Warm restart" for a CPU 315-2DP without a configured cold-restart OB102.
7. If the MMC-based path is in use, run the test five times consecutively. If the count is consistent across all five cycles, the SFC84 is firing reliably and the wear concern is bounded by the trigger rate.

Troubleshooting Matrix

Field Notes and Best Practices

• For an S7-300 CPU 315-2DP with firmware V2.0, the Non-Retain tick box in DB properties is shown by the editor but is functionally ignored at runtime. The DB will always be retentive if the backup battery is in place. Do not rely on the tick box to disable retentivity on this firmware.
• If you upgrade the firmware above V2.1.0 in the future, re-evaluate the Non-Retain setting on every DB. The default is retentive, so leaving the tick box empty is the safe choice for production data.
• Where possible, do not use SFC84 on an S7-300 for routine save. The MMC endurance is a maintenance liability. The S7-400, S7-1200 and S7-1500 have different mechanisms (e.g. retain tags, recipe DBs) that are better suited to this use case.
• For the conveyor count application specifically, the cleanest approach is to keep the value inside a retentive DB and let the backup battery handle the weekend. Add an HMI save button that triggers SFC84 only as a belt-and-braces safeguard.
• Always store production-critical counts in a retentive DB or, better, mirror them to a higher-level SCADA so the PLC is not the only place the value lives.

Related STEP 7 / Siemens Documentation

For deeper reference on the SFC84 block, the S7-300 retentive memory model and the SFC error codes, see the official Siemens automation support manuals:

• SFC84 "WRIT_DBL" — description and parameters
• Retentive behaviour of S7-300 CPUs and DBs at POWER ON / POWER OFF
• S7-300 CPU 31x retentive memory — configuration and diagnostics
• S7-300 CPU 31xC and 31x Manual — Chapter 4: Retentive areas