My SFRA traces are not repeatable. What could be going wrong?
Poor SFRA repeatability is almost always caused by measurement setup issues rather than actual transformer problems.
|
Error Source |
Symptom |
Root Cause |
Fix |
|
Poor grounding |
Noise in the LF region (< 100 Hz) |
Ground loop or floating reference |
Connect the instrument to the tank ground pad; avoid shared paths |
|
Cable movement |
Trace changes between runs |
Cable flex changes capacitance |
Secure cables with tape; do not move during test |
|
Open test lead |
Flat trace (no peaks) |
Break in coaxial conductor |
Check continuity with the DMM before starting |
|
Incorrect lead connection |
Shifted peaks (all bands) |
Swapped injection/measurement ports |
Label cables; same color-coding every time |
|
Oil temperature variation |
Gradual magnitude drift |
Copper expansion changes geometry |
Record temp; compare only traces within ±5°C |
|
Remaining magnetization |
Distorted low-freq shape |
Core saturation from prior DC test |
Do SFRA before any DC test, or demagnetize the core |
|
Environment interference |
50/60 Hz spikes in the low band |
Nearby energized equipment |
Shut down adjacent equipment or use a notch filter |
Best practice workflow for repeatable SFRA:
1. Perform SFRA before any DC tests (IR, winding resistance) to avoid core magnetization.
2. Record oil temperature and ambient conditions.
3. Keep the same cable routing and connection points for baseline and follow-up tests.
4. Perform a "repeatability check" - take two identical measurements without moving cables. Traces should overlay perfectly.
5. Use phase-to-phase comparison (e.g., H1-H2 vs. H2-H3) as the primary interpretation method.