Troubleshooting Siemens MicroMaster 440: F002 DC Link Overvoltage with r0026 / Terminal Voltage Mismatch

The Siemens MicroMaster 440 (MM440) is a robust general-purpose inverter, but field engineers frequently encounter a confusing condition in which the value reported in r0026 (smoothed DC link voltage) does not match the DC bus voltage measured directly at the DC+ and DC- power terminals with a digital multimeter (DMM). When the discrepancy is large — for example, r0026 displays 803 V while the DMM at the terminals reads 543 V on a 400 V-class drive — the drive will frequently latch F0002 (DC link overvoltage) and refuse to run, even though the actual measured bus voltage is well below the inverter's nominal over-voltage threshold.

This reference walks through the underlying causes of the mismatch, the correct way to measure the DC link on a PWM inverter, how to interpret r0026, and the parameter changes that prevent nuisance trips on loads that briefly regenerate.

1. Problem Description

A typical field report involves a 400 V three-phase MM440 driving a winder / unwinder used to draw copper wire onto a rim:

• Mains voltage measured at the line-side terminals: ~400 V AC L-L.
• DMM across DC+ and DC-: ~543 V DC.
• Parameter r0026 on the BOP/AOP: ~803 V DC.
• Drive trips F0002 (DC link overvoltage) during or shortly after a run command.
• When a second MM440 is connected to the same motor / bus, it does not report an F0002 — confirming the issue is internal to the suspect unit.
• Reducing the supply with a variac lowers r0026 in lockstep; the trip is suppressed when the reading falls below ~720 V.
• Once a load is mechanically attached to the motor shaft, the drive begins to report F0453 (motor stalled) and refuses to rotate.

The combination of symptoms — high r0026, missing under-voltage trip, and F0453 under mechanical load — points to an internal sensing problem and an application-side energy-regeneration problem. Both must be addressed.

2. DC Link Voltage Fundamentals

The MM440 uses an uncontrolled 6-pulse diode rectifier on the input, followed by a large electrolytic DC link capacitor bank, an IGBT inverter stage, and (on most frames) an internal brake chopper. The DC link voltage is not a flat DC value; it has two distinct components:

1. Average (DC) component, set by the rectifier:

V_DC,avg ≈ 1.35 × V_LL  (3-phase, no load)

V_DC,peak ≈ √2 × V_LL  (no load peak)

2. High-frequency switching ripple from the IGBTs and the 6 × line-frequency (300 Hz for 50 Hz mains) rectifier ripple.

For a 400 V ±10 % three-phase supply, the unloaded DC link sits between 540 V and 680 V DC. Under load it droops to roughly 540 V–560 V DC. The MM440 over-voltage threshold for a 400 V-class unit is approximately 760 V DC (it varies slightly with frame size and firmware); a reading of 803 V in r0026 would normally be above this trip level.

2.1 Expected r0026 by Voltage Class

Any value in r0026 more than 10 – 15 % above the "Expected V_DC (loaded)" column, with the motor at rest and the drive enabled, is suspicious and warrants further investigation.

3. Why the DMM Reading Can Disagree with r0026

There are three families of root cause for the discrepancy. Treat them as a checklist; in practice more than one is often present.

3.1 The Multimeter Reading Is Low (most common)

A digital multimeter set to DC volts is a true-DC instrument; it does not "read average vs. RMS". What it reads is the average of the voltage applied to its input over its measurement window. Three measurement problems are typical on a switching inverter DC link:

• Common-mode noise and high dV/dt. The DC– bus is bouncing with respect to protective earth at the IGBT switching frequency (typically 4 – 16 kHz on the MM440). A hand-held DMM referenced to earth ground (rather than to the DC– terminal) couples noise into the input and can read several tens of volts low.
• Slow ADC / long integration window. A DMM with a 1-second averaging window cannot follow the rectified 6-pulse ripple; the displayed value is the long-term mean, which is slightly below the peak.
• Input divider loading. The DMM presents ~10 MΩ. The MM440 DC link has a 1–10 MΩ bleeder, so the meter can pull the bus down a few volts when the drive is at idle. This is a small effect, not a 250 V effect.

3.2 The r0026 Reading Is High (internal sensor or scaling)

r0026 is derived from the MM440's internal DC-link voltage divider, scaled by an op-amp, and digitized by the control board's ADC. The signal chain is:

DC link + → R-divider → op-amp buffer → ADC → parameter scaling → BOP display (r0026)

Failure modes along this chain that can produce a falsely high reading include:

• Drift in the high-impedance divider resistors (a common failure on drives 8+ years old, especially in humid environments). The lower leg of the divider increases in value, raising the sensed voltage.
• Offset / gain drift in the buffer op-amp on the control board. A 50 mV op-amp offset at this divider ratio is ~20 – 50 V at the DC link — easily enough to push r0026 into the F0002 region.
• ADC reference drift, often correlated with control board temperature.
• Sticking relay on the pre-charge / bleeder circuit, leaving extra capacitance on the bus so the natural no-load bus is much higher than expected.
• Wrong firmware / parameter set for the installed hardware (e.g. a 230 V parameter set loaded into a 400 V unit). The displayed value is then in "230 V units" and reads 1.7× the true value.

When the variac test in the field report reduced r0026 in lockstep with the input voltage, the divider/ADC chain is responding to a real change at the DC link — which does not rule out a scaling error, but does rule out a fully open transducer. The most common finding on a working drive that nonetheless "reads high" is divider drift.

3.3 The Motor Is Regenerating

An overhauling load (a winder paying out, a descending hoist, a fan spinning down) pumps energy back into the DC bus. The MM440 absorbs it briefly through the bus capacitors; if the ramp-down is too short or no brake chopper / resistor is fitted, the bus rises and F0002 trips.

A second hallmark of regeneration, which appeared in the field report, is F0453 (motor stalled) appearing on the same drive. Under mechanical load, the controller can momentarily command more torque than the drive is configured to deliver, the IGBTs fold back, the motor stalls, the mechanical inertia re-enters the bus via the diodes, and r0026 rises until F0002 (or F0453) latches. A second MM440 on the same motor will likely show the same fault if the application is the cause; if it does not, the issue is split between the application and the suspect unit.

4. Required Tools and Prerequisites

• Basic Operator Panel (BOP) or Advanced Operator Panel (AOP) for parameter access.
• True-RMS DMM, CAT III 600 V or higher, with fresh batteries. A Fluke 87V or equivalent is recommended; low-cost meters often lack the bandwidth and isolation required.
• Optionally, an isolated oscilloscope or a high-voltage differential probe (e.g. Tektronix P5200) to view the actual bus waveform, not just the average.
• A copy of the drive's parameter set (upload to a MMC card or via DriveMonitor / STARTER) so it can be restored after the test.
• Access to the line-side disconnect, locked out.

5. Step-by-Step Diagnostic Procedure

1. Capture the current parameter set. Read r0026 with the drive disabled (no run command). Record r0027 (output frequency), r0030 (motor rated current), and r0035 (motor temperature). Save the full parameter set with DriveMonitor or write down P0003 = 3, P0010 = 0, scroll all parameters to a notebook.
2. Measure the supply at the line terminals (L1, L2, L3 to PE and L-L). On a 400 V nominal supply, expect 380 – 440 V L-L. Anything above 440 V L-L is an over-voltage condition and will raise the DC bus by the same ratio. If the supply is over-voltage, the issue is upstream (transformer tap, large loads on the same bus) and the inverter may be doing the right thing.
3. Measure the DC link correctly.
   • Lock out and wait the discharge time.
   • Set DMM to DC V, 1000 V range.
   • Place the COM lead on DC– and the V lead on DC+, both on the power terminal block — not on the cabinet chassis, not on PE.
   • Remove the lockout and power the drive, no run command.
   • Record the value once it stabilizes (typically within 5 seconds).
4. Compare DMM to r0026. For a 400 V class drive, the expected DMM value with the drive idle is 540 – 580 V (no load) and 530 – 555 V (idle with internal pre-charge). r0026 should match within ±5 %.
   • If DMM ≈ expected and r0026 is far higher (e.g. 803 V): suspect divider / ADC drift on the control board.
   • If DMM is far lower than expected (e.g. 543 V on a 400 V class with no F0002 normally) and r0026 agrees with the DMM, the drive is fine — the DMM measurement is the problem (see §3.1).
5. Test the sensor chain with the variac method. If the field supply is variable, sweep the line voltage from 360 V to 440 V L-L. Both the DMM and r0026 should track linearly with the line. The slope of r0026 per volt of line should be ~1.35. If the slope is steeper (e.g. 2.0), the scaling is wrong — a sign of divider drift or, more rarely, a wrong parameter set.
6. Check the parameter set for a 230 V / 400 V mismatch. Compare P0100 (Europe / North America / kW / hp) to the rating-plate voltage. If a 230 V parameter set was loaded into a 400 V drive (or vice versa), r0026 will read in the wrong units. Reset to factory defaults with P0010 = 30, P0970 = 1 and re-commission — but only after recording the existing set.
7. Inspect the control board and divider. With the drive de-energized and the bus verified dead, remove the control board cover. Look for:
   • Discoloured or cracked SMD resistors in the HV divider area (typical: two large SMDs in series, plus a low-side divider).
   • Electrolytic capacitor venting or bulging.
   • Solder joints that have gone dull / fractured, especially around the divider and op-amp.
   • Corrosion from humidity / oil mist.
   A failed divider resistor is the single most common field cause of an MM440 that "reads high" on r0026 but measures correctly at the bus with a scope.
8. Decide between repair and replacement. See §8.

6. Configuration Parameters to Mitigate the Application Side

If the diagnostic in §5 confirms that the drive's sensing is correct and the supply is within tolerance, the F0002 / F0453 combination is almost certainly an energy-regeneration problem on the winder. The following MM440 parameters are the ones to address, with their default values shown for a 400 V / 5.5 kW frame:

For a winder that is paying out a heavy rim, the recommended sequence is:

1. Set P1121 to at least 2× the deceleration time the application actually needs.
2. Confirm that an external brake resistor of the correct ohmic value and power rating is wired to DCP / R and DCN. The MM440 manual lists resistor values per frame (e.g. 39 Ω / 200 W for a 5.5 kW 400 V frame).
3. Enable the Vdc controller with P1240 = 1 so that brief transients extend the ramp automatically.
4. If the drive is also latching F0453, the motor data (P0304 – P0311) and the motor-stall detection (P2177 / P2178) must be checked. A motor that "stalls" during a transient regen event is a sign the current limit is too low for the load — increase P0640 (motor overload factor) and verify P1525 / P1530 (torque limits) are set to the application requirement, not the default 150 %.

7. Fault and Parameter Reference

The MM440 status word and the read-only r0026, r0027, r0030 variables are described in the Operating Instructions, "Parameter List" chapter, and in the parameter-list appendix that ships with each frame size. Always consult the parameter list for the exact firmware version in your drive; values and ranges change between firmware versions.

8. Repair vs. Replace Decision

The MM440 has been on the market since the late 1990s; many units in the field are 10 – 20 years old. Once the divider or control board is suspect, the decision between repair and replacement is largely economic.

• Repair makes sense when the unit is a current-production frame size, a control-board replacement (Siemens part number varies by frame; e.g. 6SE6400-1CB00-0AA0 for the smaller frames) is available from Siemens Spare Parts with a short lead time, and the customer can tolerate the downtime.
• Replacement with a current SINAMICS V20 / G120C is usually the right call for drives that are out of production support, that have already had one board replacement, or that sit in a harsh environment (humidity, oil mist, dust) that will repeat the failure mode. The mechanical footprint is similar but not identical; mounting adapters are available.
• Always confirm the F0002 root cause is on the suspect unit and not on the supply / load before condemning the drive. A drive replaced under a "bad sensing" hypothesis that turns out to have a regenerating load is a drive that fails again in six months.

9. Verification

After the corrective action, the following checks confirm a clean return to service:

1. With the drive idle and the line at nominal, r0026 must read within ±5 % of the value predicted by V_LL × 1.35.
2. A slow ramp-down of the loaded motor (full speed → 0 in 10 s or more) must not raise r0026 above the F0002 threshold.
3. A fast stop (OFF3, default 5 s) must not latch F0002 if the application can tolerate the deceleration; if it does, P1135 is too short or the brake resistor is undersized.
4. Loaded run at rated current for 30 minutes must not latch F0453, F0011, or F0004.
5. Trend r0026 over a full work cycle with DriveMonitor / STARTER. A clean signature is a flat line at the loaded bus voltage with brief excursions to the brake-chopper switch-on level during decel.

10. Field-Proven Caveats

• The drive will appear to "self-heal" at lower supply voltage. That is consistent with a divider that is slightly out of spec — at 380 V L-L the apparent overvoltage is hidden, at 415 V L-L it is exposed. The fault is still there.
• Replacing the BOP / AOP does not change r0026. If the panel is showing a different value than DriveMonitor, the panel itself is suspect, not the drive. Always cross-check with a second method.
• Standard DMMs can read low on a switching bus. Do not trust a hand-held meter alone; the second method is either a true differential probe on a scope, or the drive's own r0026 (which is already filtered and scaled).
• A winder that draws copper onto a rim is, by physics, both a motor and a brake at different points in the cycle. Even with a brake resistor fitted, a winder that is asked to maintain tension during a sudden stop will regenerate. The control loop, not the brake chopper, has to absorb most of that energy.
• F0002 and F0453 in the same fault log almost always means "stuck on a transient over-voltage that the controller interpreted as both an over-voltage and a stall". Fix the bus-side issue first; F0453 usually clears once the F0002 stops latching.

11. Quick-Reference Diagnostic Matrix

12. Related Standards and Reference Material

The DC link measurement and over-voltage behaviour of an industrial inverter are constrained by the following documents; consult the latest revision for binding values.

• IEC 61800-2 — Adjustable speed electrical power drive systems, Part 2: General requirements — rating specifications for low-voltage adjustable frequency a.c. power drive systems. Defines the test conditions for DC link over-voltage behaviour and the tolerance on the indicated DC link voltage.
• IEC 61800-5-1 — Adjustable speed electrical power drive systems, Part 5-1: Safety requirements — electrical, thermal and energy. Defines insulation, earthing, and the safe discharge time for the DC link.
• UL 508C — Power Conversion Equipment. North American standard for industrial drives; defines the over-voltage trip tests.

Manufacturer-specific reference material:

• MicroMaster 440 product page — overview, frame sizes, accessories.
• MicroMaster 440 Operating Instructions — full parameter list, F0002 / F0453 descriptions, brake-resistor sizing tables.
• Siemens Industry Online Support — search for "MM440 F0002" and "MM440 r0026" for application examples and firmware notes.