A hair dryer held near a capacitor for ten seconds is a diagnosis, not a coincidence. Warming an aluminum electrolytic drops its equivalent series resistance by a large factor, and a capacitor that has dried out over years only holds ripple down once it is warm. On a cold start it presents high ESR, the rail it filters sags and ripples, and the drive's current-sense or gate-drive circuitry reacts by tripping. The number that matters is not shop ambient at 50 °F — it is ripple voltage on the DC rail, measured AC-coupled, cold versus warm.
The Rail That Actually Trips
Two large electrolytics in parallel reading 27 VDC are not the main DC bus. A 2 hp drive fed from three-phase 230 V runs a bus near 1.35 × VLL, roughly 310 VDC; single-phase input gives about 1.41 × VRMS under light load. A 27 V rail is an auxiliary supply — control logic, gate drives, analog reference, or the armature/field supply for the fractional-horsepower DC feed motor riding on the same panel.
That distinction sets the whole investigation. If the failing capacitors sit on the gate-drive rail, undervoltage there means the output transistors never fully saturate; VCE(sat) climbs, switching losses climb, and the current the drive measures climbs with them. If they sit on the analog reference, the overload threshold itself moves. If they feed the DC feed motor, that motor draws more current to make the same torque against a sagging, ripple-laden supply. All three end at the same place: an overload trip that has nothing to do with the spindle.
Reading the 30-35 Hz Ceiling
The drive gets to 30-35 Hz and stops. That ceiling is a load ceiling, not a frequency limit. Under a constant-volts-per-hertz profile, output voltage, switching loss, and auxiliary rail loading all rise with commanded frequency, so a marginal supply survives at 10 Hz and collapses at 33 Hz. Fault codes stored in the drive at the moment of trip will say whether it is an instantaneous overcurrent, an I²t thermal overload, or a bus/supply undervoltage that the panel interprets as a fault.
Two symptom facts narrow it hard. First, the trip persists with the 2 hp spindle motor disconnected — that removes cold grease, stiff spindle bearings, thickened way lube, and any mechanical drag from the suspect list. Cold gearbox oil is the usual winter culprit on machine tools; it is ruled out here. Second, the fault clears with localized heat. A bimetallic overload relay in the panel behaves the opposite way: cold ambient makes it trip later, not sooner. So confirm which device latched — a drive fault on the display, or a mechanical OL relay you have to press to reset. If it is the relay, the drive is genuinely pushing real current through it and the answer lies upstream in the supply, not in the relay.
ESR Versus Temperature
This is chemistry and heat, not logic. An aluminum electrolytic's series resistance is dominated by the conductivity of the liquid electrolyte in the paper separator. Electrolyte conductivity falls steeply as temperature drops; manufacturers publish it as an impedance-versus-temperature multiplier on the datasheet, referenced to 20 °C and 100-120 Hz. A healthy capacitor has enough margin that the cold multiplier is harmless. One that has lost electrolyte through the seal over a decade starts warm-ESR already out of spec, and the cold multiplier pushes it past the point where it can shunt ripple current at all.
Ripple voltage follows directly: Vripple ≈ Iripple × ESR, plus the capacitive term. Double the ESR and you roughly double the ripple the downstream regulator or gate driver sees. Two caps in parallel share ripple current; if one goes effectively open, the survivor takes the full ripple current at twice the ESR and heats itself further. Electrolyte freezing is a red herring — that happens far below a 50 °F shop, and a frozen capacitor tends to vent rather than merely misbehave.
Diagnostic Procedure
Do not probe bus or rail terminals with the drive energized unless you are equipped for it; after removing power, wait for the bus to bleed and verify below 50 VDC at the capacitor terminals before touching anything.
| Measurement | What it should be | Where to take it |
|---|---|---|
| Trip source | Drive fault code, or OL relay contact | Drive display / relay reset button |
| Main DC bus level | ≈1.35 × VLL (3-ph) or 1.41 × VRMS (1-ph), light load | Bus + / bus − terminals |
| 27 V rail, cold vs warm | Stable from 0 Hz to full speed, within regulator tolerance | Across the paralleled caps |
| Rail ripple, AC-coupled | Small and load-proportional; a cold-to-warm change is the fault | Scope across cap terminals |
| Capacitor ESR | At or below the can's datasheet value, 100-120 Hz, 20 °C | ESR meter, caps out of circuit |
| Capacitance | Within printed tolerance; both caps matched | LCR meter |
| Output current per leg at trip | At or below Baldor nameplate FLA, legs balanced | Clamp meter on U / V / W |
- Reproduce on demand with freeze spray instead of a hair dryer. Chill one component at a time and watch for the trip. Cold isolates; heat does not, because a hair dryer warms the capacitor, its solder joints, and the terminal block simultaneously.
- Record the fault code and the frequency at trip before clearing.
- Scope the 27 V rail while ramping the drive from 0 Hz toward the trip point, cold. Watch ripple amplitude and the DC average together.
- Power down, discharge, and lift both capacitors. Measure ESR and capacitance individually — in parallel, a good cap masks a dead one.
- Retorque every DC link, capacitor can bolt, terminal block screw, and bus bar joint in the panel. Thermal cycling loosens aluminum-to-copper joints, and warm-restores-operation is equally the signature of a high-resistance connection expanding shut.
- Replace both capacitors as a matched pair: same or greater capacitance, same or higher voltage rating, 105 °C low-ESR grade, ripple current rating at or above the original. Observe polarity and the vent scoring.
- Inspect the input diode bridge and any snubber or bleeder resistors while the caps are out.
Verification
Verify cold, at the start of shift, before anyone has run the machine. Ramp to full commanded speed with the spindle motor disconnected first, then reconnected, and clamp each output leg at steady state and during acceleration. Currents should sit at or under the Baldor nameplate FLA and match leg-to-leg within a few percent. On the scope, rail ripple should stay flat as frequency rises rather than growing toward the old 30-35 Hz cliff.
Then run a real cut for twenty minutes and thermal-scan the panel: capacitor cans, terminal blocks, heatsink, and the overload relay. A capacitor running noticeably hotter than its neighbor is still failing, whatever the meter said at room temperature.
Recurring Failure Modes on Small Panel Drives
Oil-filled capacitors do not behave this way. Motor-run and power-factor caps use a stable dielectric film and show almost no ESR shift over shop temperatures; if the cold-sensitive part is oil-filled, look elsewhere — usually at a contactor coil, a start capacitor circuit, or a connection.
Three others repeat on this class of machine. A lost or weak input phase makes bus ripple explode and produces exactly the load-dependent trip described here, so measure all three input phases under load before condemning the drive. An overload parameter set from a previous motor's FLA will trip early forever regardless of temperature — read the drive's motor-current setting against the nameplate. And a drive that has sat unpowered for months develops leakage in its bus capacitors that only reforming or replacement will fix.
If ESR and capacitance both check good, ripple is clean cold, connections are tight, and the input phases are balanced, the fault has moved onto the control board — current-sense amplifier, gate-drive supply, or the trip comparator itself. At that point substitute a known-good drive of matching rating to confirm the drive is the failing element, then take the fault code and your measured rail data to the drive manufacturer or a board-level repair house. On an obsolete fractional-to-2 hp drive, a like-rated replacement sized from the Baldor nameplate is frequently cheaper and faster than component-level repair.
Frequently Asked Questions
Why does my VFD trip on overload only in cold weather?
Aluminum electrolytic capacitor ESR rises steeply as temperature falls, so a capacitor already degraded from electrolyte loss stops filtering ripple when cold. The resulting rail ripple and sag drive the gate-drive or current-sense circuitry into an overload trip that disappears once the part warms up.
Why does the drive trip at 30-35 Hz but run fine below that?
Output voltage, switching losses, and auxiliary rail loading all rise with commanded frequency under a volts-per-hertz profile. A marginal supply holds up at low speed and collapses once loading crosses the threshold, so the ceiling is a load limit rather than a frequency limit.
Why did a hair dryer fix it, and can I trust that as a diagnosis?
Heat lowers electrolyte ESR and also expands loose metal joints back into contact, so a hair dryer cannot distinguish a bad capacitor from a bad connection. Use freeze spray on one component at a time to isolate the fault, since cold applied locally identifies the exact part.
Why does the drive still trip with the spindle motor disconnected?
That removes all mechanical load, which rules out cold grease, stiff bearings, and thickened lubricant. The remaining current path is the drive's own losses plus the fractional-hp DC feed motor supply, pointing the investigation at the DC rail and control electronics.