Short answer: SCR/thyristor switching can couple electrical interference into low-level thermocouple circuits through electric fields, magnetic fields, shared impedance and unintended ground paths. Fix the coupling route and topology first; filtering alone can hide noise while delaying real temperature changes.
Thermocouples produce small signals. A furnace panel may also contain high-current heater conductors, contactors, drives and switching edges. The resulting symptom can be periodic ripple, sharp spikes, an offset that changes with heater output, or several channels moving together when power switches.
Use the pattern as a diagnostic clue
- Does the disturbance follow heater firing percentage or a particular phase?
- Does it stop when the heater output is safely disabled under an approved test?
- Do all channels shift together, only one channel, or only probes on one furnace zone?
- Does the raw input show the event before application filtering?
- Does moving a cable or changing shield termination alter the pattern?
Preserve time-aligned heater-output or system-event information where available. Correlation is more useful than looking at a noisy temperature trace in isolation.
Map the coupling paths
| Path | Typical design check |
|---|---|
| Capacitive coupling | Separation from switched conductors, cable construction and shield strategy |
| Inductive coupling | Loop area, routing relative to high-current paths and twisted pair use |
| Shared impedance | Common supply/return paths, panel bonding and terminal layout |
| Ground loop | Junction-to-sheath connection, multiple earth points and non-isolated channels |
| Communication/system | Whether corruption occurs at the input, transmitter, network or software layer |
A controlled diagnostic sequence
- Define the symptom before changing anything. Save the rawest available temperature data, channel status, timestamps and the display/filtered version. Note whether the symptom is ripple, spikes, common offset, intermittency or data loss.
- Align it with furnace events. Compare the disturbance with heater output, contactor state, drive operation, zone demand and process stage where those signals are available. Correlation suggests where to look; it does not by itself prove the coupling mechanism.
- Map which channels are affected. One channel, one cable route, one input module, one zone and all channels point toward different portions of the measurement chain.
- Inspect the low-energy path. Verify thermocouple type, polarity, junction construction, extension cable, routing, shield termination, terminal condition, input configuration and isolation arrangement.
- Use only authorised comparisons. A suitable reference, simulator, alternate input or temporary routing may help separate sensor/cable, instrument and software layers when the plant procedure permits it.
- Change one controlled variable and verify. Record the before/after evidence. If several changes are made together, the graph may improve without revealing which condition mattered.
Worked plant scenario: ripple follows heater demand
Illustrative scenario: three thermocouple channels show a repeating ripple during part of a heating ramp. The ripple becomes more visible when the affected zone's heater demand increases, while channels routed in another tray remain comparatively steady. This pattern is consistent with a shared routing, shielding, bonding or input-topology issue, but it is not proof that the real furnace temperature is steady.
The team first retains unfiltered and displayed readings, heater-demand timing, affected channel identities and the cable/input map. It checks whether the pattern begins at the field input or appears only after communications or software processing. At an approved maintenance opportunity, qualified personnel inspect separation from switched conductors, cable loop area, shield termination, grounded-junction paths, common returns and input isolation. Protective earth is left intact.
If an authorised reference or temporary low-energy comparison remains stable while the installed path shows correlated disturbance, attention can move toward the field path and topology. If both the reference and installed channel move together, a real thermal event or a common influence remains possible. If the raw input is stable but the displayed value is not, investigate communications, scaling or software processing. The conclusion should state what the evidence supports and what remains unresolved—not simply “SCR noise fixed.”
Mitigate in a sensible order
- Verify sensor type, polarity, junction construction and cable compatibility.
- Correct instrument-cable routing and avoid unnecessary loop area.
- Review shield termination and plant earthing with a qualified electrical engineer.
- Use appropriate channel/input isolation where potentials or multiple grounded probes demand it.
- Then apply documented input filtering, integration and software smoothing that preserve required dynamics.
Do not disconnect protective earths or improvise unsafe isolation to make a graph look better. Furnace power and bonding changes belong to authorised plant personnel. Pyrolog scope is the measurement/data side; wider control or power-panel work may require coordination with the appropriate Multitek or plant team.
Common mistakes that hide the cause
- Applying heavy filtering first: smoothing may conceal spikes or delay genuine process movement without correcting the coupling path.
- Calling every irregular trace a ground loop: a damaged sensor, loose terminal, incorrect type, input configuration or communications fault can produce different but overlapping symptoms.
- Changing earth or shield connections casually: unsafe or undocumented changes can create hazards and make the final topology impossible to explain.
- Comparing unsynchronised data: heater output and temperature records with unrelated clocks can create a false impression of correlation or hide a real one.
- Using only the smoothed HMI trend: the diagnostic record should preserve raw values, quality/status flags and the filtering method where available.
- Making several changes at once: improved readings do not identify which route was corrected, and the problem may return during later maintenance.
Evidence worth retaining
Keep the raw and displayed traces, timestamps and time-source information; heater/zone demand or event signals where available; affected and unaffected channel list; sensor/junction and extension-cable identity; cable routes and shield/earth termination notes; input module, isolation and filter configuration; inspection photographs where permitted; each authorised test; the exact corrective change; and a verification run covering the operating condition that previously exposed the symptom.
Diagnostic boundary: these checks organize measurement evidence. The plant's electrical-safety rules, approved procedures and applicable requirements govern access, testing, modification and acceptance.