Bottom line first: when an SCR power controller (thyristor power regulator) is not given a command and not running, measuring 100-plus volts or nearly 200V with a multimeter at the output terminals is usually not the thyristor delivering power. It is a "phantom voltage" (often called induced voltage in the field) under no-load conditions — input-side voltage reaches the output through the RC snubber network across the thyristors, through device leakage current, or through stray capacitive coupling, creating a floating potential. Because a digital multimeter has a very high input impedance, a tiny current is enough to display a high reading. The way to verify it is to connect a matched test load and watch whether the reading drops quickly; but as long as any voltage is measured, the point must be treated as live.
This phenomenon is common at industrial electric heating sites: the equipment is powered up but no control signal has been given, or the load is not yet connected. An operator measures the output terminals and finds a substantial voltage, then starts suspecting that the thyristors have failed to turn off, that the trigger circuit is faulty, or that the equipment has an earth leakage fault. The sections below follow the order "where the voltage comes from → how to tell phantom voltage from real output → what to record on site → the safety boundary".
Why can the output terminals read 190V with no load connected?
Take a typical thyristor power controller. In the main circuit, an RC snubber network (also called a resistor-capacitor absorber or buffer circuit) is usually connected in parallel across each thyristor. It consists of a resistor in series with a capacitor, and its job is to absorb voltage spikes and limit the rate of voltage change during thyristor turn-on and turn-off, protecting the devices from overvoltage damage.
When the thyristor is not conducting, this RC branch is not fully disconnected. Taking single-phase AC input as an example, the line-frequency voltage on the input side drives a very small current through the RC branch. If the output is open-circuited, that current has nowhere to go, and the output terminals present a "floating potential".
A second source is capacitive coupling. Stray capacitance exists between wires inside the equipment, and between the thyristors, their heat sinks and the cabinet. The high input-side voltage couples to the output through these capacitances. In cabinets containing transformers, long cables or densely routed wiring, this coupling path is more pronounced.
The reason a multimeter can turn such a tiny current into a high reading is its own high input impedance. The voltage range of a typical digital multimeter has an input impedance on the order of 10 MΩ (check the instrument manual for the exact value). By Ohm's law, a current of only 19 μA through 10 MΩ will display as 190V. In other words, a high reading does not mean the output can supply any real power.
So with no load connected, the output terminals may read tens of volts, 100-plus volts, or nearly 200V — but the corresponding current is typically in the microamp to milliamp range, orders of magnitude below the ampere-level currents of a loaded controller. That is also why "the reading disappears once the load is connected".
How do you tell phantom voltage from a controller that is really outputting?
The direct and effective method is to connect a load matched to the actual operating conditions to the output terminals and watch how the reading changes. The voltage class, power rating and insulation of the test load must match the site conditions; do not improvise with a random lamp or a piece of wire.
If it is phantom voltage, the reading drops quickly once a suitable load is connected — often to near 0V. At that moment the controller still has no command and is not running, and no sustained load current should appear at the output. The high reading seen before was just a floating potential under a high-impedance measurement condition.
Conversely, if after connecting a load the controller — with no command and no run instruction — still shows a significant voltage and a sustained current at the output, it must not be treated as phantom voltage. Shut the equipment down and check, one by one, whether the thyristors have failed short-circuit, whether the trigger circuit is abnormal, whether the control signal is truly at zero output, whether the output wiring is correct, and whether some other circuit is feeding power back into the output.
| Observation | No-load phantom voltage | Case that must be treated as a fault |
|---|---|---|
| Conditions | Output open-circuited, or the load side fully disconnected | With a suitable load connected, the condition persists with no command and no run signal |
| Reading behaviour | Readings differ noticeably between instruments with different input impedances | Reading is stable and accompanied by a sustained load current |
| Loaded behaviour | Reading falls quickly to near 0V once a suitable load is connected | Significant voltage and current remain with the load connected |
| Action | Record the site conditions and treat it as normal operation | Shut down immediately; check thyristors, trigger circuit, wiring and control signal |
Can you touch it with bare hands?
No. A phantom voltage is not a safe voltage.
Seeing a 190V reading on site does not, by itself, prove that it is only the weak induced voltage caused by the RC snubber. Wiring errors, device faults, degraded insulation or back-feeding from other circuits can all present as "voltage at the output with no load". Until the cause is confirmed, every measurement point must be treated as live.
The site rule is therefore simple: as long as a voltage is measured, treat the point as live; never touch it with bare hands, never use tools without verified insulation, and never change any wiring without locking out, de-energizing and proving absence of voltage.
Does this no-load voltage need to be eliminated?
Under normal use, usually not. Once the controller is connected to its real load — heaters, resistance furnace elements, silicon carbide or molybdenum disilicide elements — the weak leakage current has a discharge path, and the no-load reading normally drops noticeably or disappears.
One caution: do not remove or modify the RC snubber network across the thyristors just to "make the reading go away". The RC values are determined by the device voltage rating, its dv/dt withstand capability and the site conditions. Removing or arbitrarily changing them can leave the thyristors without overvoltage protection during switching — trading one question for a far more expensive problem.
If you are commissioning or troubleshooting and need to confirm whether the output can actually supply power, use a test load matched to the voltage class and appropriate instruments — do not rely on the no-load reading alone. If a significant voltage and current persist with a load connected and no command given, the question is no longer "should this be eliminated" but "stop and troubleshoot the fault now".
What conditions should be recorded during a site check?
Judging a no-load voltage depends heavily on the measurement conditions. The same output terminal, measured with different instruments in different states, can lead to completely different conclusions — which is why the site record matters more than the conclusion itself. The table below can be used directly as a troubleshooting checklist.
| Item to confirm | Why it must be confirmed | How to record it |
|---|---|---|
| Measurement terminals | Whether the no-load voltage appears between output phases or between output and earth determines whether the reading is a line voltage or a floating potential to ground — two different conclusions | Record terminal numbers and phases |
| Input voltage and frequency | The output reading is strongly related to the input-side voltage; an abnormal input shows up directly at the output | Record the input-side reading with the same instrument |
| Multimeter model and range | Instruments with different input impedances give different results on the same circuit; range selection matters too | Record brand and model, AC/DC range, and scale |
| Actual load wiring state | "Not connected", "connected but open" and "connected but not energized" are three different situations | Photo plus a written note |
| Controller operating state | Whether it is powered, whether a command signal exists, whether a run instruction is active, and any panel alarms decide whether phantom voltage is a valid explanation | Record the panel state with a photo |
| Possible back-feed paths | Other power sources or bypass circuits in the same cabinet may back-feed the output, appearing as "no-load voltage" | Check the drawings and verify by de-energizing circuit by circuit |
Once these items are complete, nearly every "190V with no load" lands on a clear conclusion: either phantom voltage caused by the measurement conditions, or a genuine fault that requires shutdown. If any single item is missing, the conclusion can only be marked as unverified.
FAQ
Q: The output terminals of an SCR power controller read about 190V with no load connected. Does that mean the thyristors are not turning off?
A: This reading alone cannot tell whether the thyristors are off. The voltage range of a digital multimeter typically has an input impedance on the order of 10 MΩ, so a microamp-level current displays as a high voltage. The RC snubber network across the thyristors, device leakage current and stray capacitive coupling can all leave the output at a floating potential under no-load conditions. To assess the thyristors, connect a suitable load matched to the voltage class and measure again: if the reading drops quickly to near 0V, the earlier reading was no-load phantom voltage; if a significant voltage and sustained current remain, the device should be shut down and investigated.
Q: Will the voltage definitely disappear once the load is connected?
A: Not necessarily. In most cases, once the controller is connected to its real load — heaters, resistance furnace elements, silicon carbide or molybdenum disilicide elements — the weak leakage current has a discharge path and the no-load reading drops noticeably or approaches 0V. But if the load side has a wiring error, a failed device, degraded insulation, or another circuit back-feeding the output, a significant voltage and current may remain even with the load connected. So "it disappears with a load" is only a reference for identifying phantom voltage, not a guarantee that holds at every site.
Q: Why do different multimeters give different readings on the same output terminal?
A: Because the no-load output voltage is highly sensitive to the impedance of the measuring circuit. A typical digital multimeter voltage range has an input impedance around 10 MΩ, while meters with a low-impedance (LoZ) mode, analog multimeters and clamp meters behave quite differently. Instruments with different input impedances connected to the same floating potential will disagree. Record the meter model, range and scale together; when comparing readings, use the same meter, the same range and the same measurement point.
Q: Does this no-load voltage affect normal temperature control?
A: Under normal operating conditions, usually not. When the controller runs with its load, the output power is set by the command signal and the load current is at the ampere level; the microamp-to-milliamp coupling current under no-load conditions does not change the loaded voltage-current relationship, and temperature control is determined by the control method and load characteristics. However, if a significant voltage and current persist with a load connected and no command given, that is no longer a question of "does it affect control" — the equipment or wiring has a fault, and the equipment must be shut down first.
Q: Can I briefly short the output with a wire or a small lamp to discharge it?
A: Not recommended. First, a phantom voltage is not a safe voltage — until you know why the voltage is there and between which terminals, the point must be treated as live. Second, at 380V and above, and on high-voltage power control cabinets, casually shorting the output or attaching an unselected lamp can cause arcing, device damage or personal injury. The correct approach is to de-energize, verify absence of voltage, lock and tag out, and then have qualified personnel follow the site safety procedures using a test load matched to the voltage class and power, together with appropriate instruments.
About the timeliness of this article
This article summarizes general engineering practice and field feedback as of September 2026. It addresses the common situation in which the output terminals of a thyristor power controller show a measurable voltage with no load connected and no command given; it is not tied to any specific model. The field reading cited (about 190V) comes from user-reported feedback without instrument records or waveform evidence and is used only to illustrate the phenomenon; it does not constitute a root-cause determination for that site. Concrete judgement should be based on measured site data, device datasheets and the power controller manufacturer's technical agreement. To confirm whether a specific unit has a fault, qualified personnel should de-energize and inspect it according to site safety procedures.
Related Technical Articles
Thyristor Voltage Sharing: Static Grading vs RC Snubber
How to Choose an SCR Power Controller: Load, Parameters, and Control Methods
Application Case
KULUN 10 kV High-voltage Power Control Cabinet Passes Factory Tests
