Use a Panasonic TQ2-5V relay to compare the electrical power needed at two steady coil voltages. First find voltage brackets for pickup and dropout. Then approach one intermediate coil voltage from zero, and approach it again after first pulling in the relay. The contacts may settle in opposite states at essentially the same coil voltage. A series shunt turns that comparison into a measured coil-power exercise.

Pickup and holding are different conditions

A relay uses current in a coil to move an armature, the mechanical piece that transfers its contacts. Pickup is the transition to the energized position when coil voltage rises. Dropout is the release when coil voltage falls. The armature changes the magnetic circuit as it moves. An already moved armature can sometimes stay in position with less current than was needed to move it. Hysteresis is this dependence on the direction of approach or previous state.

The exact component is Panasonic Industry TQ2-5V, orderable number ATQ209, with standard through-hole printed circuit board (PCB) leads. It is a single-side-stable polarized relay. It returns to the de-energized state when the coil loses its drive. Its two changeover poles form a double-pole, double-throw (DPDT) contact arrangement. A normally open (NO) terminal is open relative to its common terminal (COM) with the coil de-energized. The normally closed (NC) terminal is the alternate contact.

At 20 °C the manufacturer specifies initial pickup at no more than 75% of rated voltage and dropout at no less than 10%, or 3.75 V and 0.50 V for this exact 5 V coil. These are the respective limits under the manufacturer's conditions. They do not identify either transition voltage or establish a holding voltage for your relay. Panasonic recommends normal operation within ±5% of rated voltage. This short reduced-voltage experiment characterizes a particular relay, supply and fixture. Use the manufacturer's rated drive guidance when designing an operating product.

The TQ2-5V relay coil is connected between a variable source and a ten ohm shunt, with a reverse flyback diode across the coil. Independently fed pull-ups sense both normally open contacts. An illustrative graph shows opposite contact states at the same intermediate coil voltage.
Both sense nodes fall near G0 when their normally open contacts close. The dashed parts of the illustrative graph mark brackets between voltage steps. At about 2.18 V the synthetic example has different states depending on the previous drive. The drawing omits the DHO804 power adapter and separate chassis earth lead; connect them as described before adding any input or output cables. Full-size PNG · Editable SVG

Wire a safe, low-voltage coil and contact fixture

Before connecting any input or output cables, connect the RIGOL DHO804's rear Ground Terminal to earth ground using the supplied ground cable, as required by the DHO800 User Guide. Its Type-C power input does not ground the instrument through the power cord. Use the supplied power adapter. Keep the SPD3303X's protective earth connected. The probe ground clips and any signal cable shields share G0 as described below; they do not replace this required chassis earth lead.

Use a Panasonic ATQ209, a Vishay 1N4007 axial rectifier in DO-41, one Vishay MRS25 10 Ω ±1% shunt and two MRS25 1 kΩ resistors. The shunt is a small series resistor that converts coil current into a voltage you can read. Use a SIGLENT SPD3303X in independent output mode and a RIGOL DHO804 with four PVP2350 probes. The sense branch has its own supply output. No mains or external loads connect to the relay contacts for this experiment.

With both supply outputs off, compare the relay's orientation marking with the bottom view on the Panasonic drawing before soldering to a breakout. Referring to a bottom-view pin drawing while holding the relay from above can swap pins. Connect bench supply CH1 positive to coil pin 1. Connect coil pin 10 to the top of the 10 Ω shunt. The bottom of the shunt goes to G0, the common test reference. Put the diode directly across pins 1 and 10. Its band marks the cathode, which goes at positive pin 1. The anode goes to pin 10 above the shunt. The diode is reverse-biased in steady operation and gives coil current a local path during interruption.

PinsFunctionConnection
1 and 10Coil + and −TOP and SHUNT, diode across them
3 and 8Contact commonsBoth G0
4 and 7NO poles A and BSeparate 1 kΩ pull-ups to 3.00 V
2 and 9NC polesLeave open
5 and 6UnusedLeave open

Join both supply negative terminals and all scope ground clips at G0. Keep protective earth connected. Sense pole A at scope CH1, pole B at CH2, TOP at CH3 and SHUNT at CH4. Use each scope probe tip at the named node, with its clip only at G0. Coil pin 10 is separated from G0 by the shunt, so grounding it with a clip would bypass the current measurement.

Set the adjustable coil supply current limit to 50 mA. Its highest setting in the experiment is 4.90 V. Check that the actual upper node stays at or below 5.00 V and that the supply remains in constant-voltage mode. Set the second supply to 3.00 V and 10 mA. Each direct-current (DC) sense branch draws about 3 mA when closed, for about 6 mA total. The low-voltage resistive sense drive is above Panasonic's 10 µA at 10 mV minimum switching reference. These levels qualify the reading as open or closed without measuring contact resistance.

Calculate only the coil’s steady power

Let VTOP and VS be the measured TOP and SHUNT voltages. The coil voltage is VC = VTOP − VS. With the diode reverse-biased at a steady plateau, use IC ≈ VS/RS and PC ≈ VCIC. The coil-power value excludes shunt loss and the 3 V contact-sense circuit. It also excludes transients between voltage steps.

The TQ2-5V coil is listed as 178 Ω ±10%, with rated initial values of 28.1 mA ±10% at 20 °C and 140 mW nominal. With 4.90 V at TOP and nominal 178 Ω plus 10 Ω in series, expect 26.1 mA, about 4.64 V across the coil and about 121 mW. Using the listed 20 °C resistance range and a ±1% shunt, the simple fixed-resistance estimate gives about 4.61–4.66 V across the coil and up to about 28.8 mA. At 2.18 V across a nominal 178 Ω coil, the model gives about 26.7 mW. These are calculations. The actual current and coil temperature can differ. The same test fixture measures the applied terminal voltage and current in each stage.

Find two transition brackets, then revisit one voltage

  1. Verify the reading paths. With the coil at zero and the sense supply on, both NO voltages should be near 3.00 V. Confirm the probe clips and use DC coupling, matching 1× factors and high-impedance inputs. Check the scope's settled zero with each tip on G0 before moving it back to the test point. Set CH4 for its small scale, starting around 50 mV/div, and check the shunt zero near 0 V. A high-impedance DC voltmeter can check the settled supply and small shunt readings. For each reading stop the scope on a visibly flat part of the plateau after its dwell. Use the four settled mean readings from that capture.
  2. Make the two ladders. Record one row after approximately 2 s at each setting. Increase the coil supply from 0 V in 0.25 V increments through 4.75 V, then set 4.90 V. Record up-1. Immediately use the same settings in reverse for down-1. Use the same cadence and repeat with up-2 and down-2 after several minutes with the coil at zero. Keep a similar room temperature and note the ambient condition and timing. Use actual settled voltages from both probe halves of the coil circuit and both contacts, with one row per setting. Save the readings of the actual four nodes, not the front-panel voltage settings. A contact wandering, bouncing across a threshold or disagreeing with its other pole calls for a fresh capture.
  3. Choose a middle voltage from the evidence. A pickup bracket extends from the final open step to the first closed step. A dropout bracket extends from the first open step on the way down to the preceding closed step. Each step gives a bound, not the exact threshold. Use the --sweeps-only command with the 84 ladder rows. It finds a terminal-voltage band above the upper end of both dropout brackets and below the lower end of both pickup brackets, with 50 mV margin at each side and a 0.65 V measurement floor. If no guarded band remains, do not force a middle trial. Check voltage step readbacks and temperature, and repeat the ladders if needed.
  4. Make the two histories. From zero, raise the supply directly to the chosen range without overshooting. After about 2 s, record idle-mid with both contacts open. Return the coil to zero and verify release. Now set the coil supply to 4.90 V, wait about 2 s, and record boost with both contacts closed. Step straight down to the same measured coil voltage as the first trial. Capture hold-2s and hold-10s after about 2 and 10 s at that setting. Keep the contact sense supply and connections fixed. Return the coil to zero and record reset with both contacts open.

Decide from both contacts and the terminal voltage

The program accepts settled NO readings from −0.05 to +0.15 V as closed and 2.65–3.15 V as open. The two poles must agree. Each of the four staircases must show one transition in the right direction and correct first and last states. The three middle terminal voltages must lie inside the common guarded band and match within 35 mV. A broad resistance check screens for shunt mistakes or missing current, and the inferred middle winding resistances must agree within a ratio of 1.08. That check helps bound obvious run-to-run changes, without measuring the actual winding temperature.

Both held records must still show closure. The last one supplies the steady hold-power value, compared with the measured boost value. For example, the included constructed data give 120.32 mW at boost and 26.40 mW while held, about 78.1% lower during that plateau. Those values describe the sample data. They do not tell you how much total energy a production driver would save or how long a real relay would hold over temperature and age. The diode is present for every run, and no release-time comparison is made.

Run the complete offline analyzer

Use Python 3.9 or later and the standard library. Download the code, original figure, tests and two synthetic examples, or fill the blank step worksheet. Its columns record stage, zero-based step index, TOP, SHUNT, NO A and NO B in volts. Use index 0 for each of the five separate middle-trial rows. For the first command make a copy containing just the four ladders. For the second use all rows. The optional measured 10 Ω shunt value can be supplied with --shunt-ohms. Exit 0 indicates a qualifying record for the selected mode, 1 an unresolved physical comparison, and 2 a command or file error.

python3 relay_hold.py steps-only.csv --sweeps-only
python3 relay_hold.py trial.csv
python3 relay_hold.py synthetic/steps-only.csv --sweeps-only
python3 relay_hold.py synthetic/trial.csv
python3 -m unittest -v test_relay_hold
#!/usr/bin/env python3
"""Bracket relay thresholds and compare two histories at equal coil voltage."""
import argparse
import csv
from dataclasses import dataclass
import io
from math import isfinite
from pathlib import Path
import sys

SWEEPS = ("up-1", "down-1", "up-2", "down-2")
STEPS = tuple(0.25 * i for i in range(20)) + (4.90,)
TRIAL = ("idle-mid", "boost", "hold-2s", "hold-10s", "reset")
HEADER = ["stage", "step", "top_v", "shunt_v", "no_a_v", "no_b_v"]
EPS = 1e-10


class Unresolved(Exception):
    """Syntactically usable readings that do not qualify the experiment."""


def screen(condition, cause):
    if not condition:
        raise Unresolved(cause)


def within(value, lower, upper):
    return lower - EPS <= value <= upper + EPS


@dataclass
class Reading:
    stage: str
    step: int
    top: float
    shunt: float
    a: float
    b: float

    @property
    def coil(self):
        return self.top - self.shunt

    def closed(self):
        def contact(voltage):
            if within(voltage, -0.05, 0.15):
                return True
            if within(voltage, 2.65, 3.15):
                return False
            raise Unresolved(f"{self.stage} step {self.step}: ambiguous contact voltage")

        state = contact(self.a)
        screen(state == contact(self.b), f"{self.stage} step {self.step}: NO poles disagree")
        return state

    def current(self, shunt_ohms):
        return self.shunt / shunt_ohms

    def coil_resistance(self, shunt_ohms):
        return self.coil / self.current(shunt_ohms)

    def power(self, shunt_ohms):
        return self.coil * self.current(shunt_ohms)


def read_rows(path):
    with path.open("rb") as handle:
        data = handle.read(24_001)
    if len(data) > 24_000:
        raise ValueError("file exceeds 24,000 bytes")
    csv_rows = csv.reader(io.StringIO(data.decode("utf-8-sig"), newline=""), strict=True)
    if next(csv_rows, None) != HEADER:
        raise ValueError("expected " + ",".join(HEADER) + " header")
    result = []
    for line, row in enumerate(csv_rows, 2):
        if len(row) != 6:
            raise ValueError(f"line {line}: expected six columns")
        try:
            step = int(row[1])
            values = tuple(map(float, row[2:]))
        except ValueError as error:
            raise ValueError(f"line {line}: expected integer step and four voltages") from error
        if not all(map(isfinite, values)):
            raise ValueError(f"line {line}: nonfinite sample")
        result.append(Reading(row[0], step, *values))
    return result


def check_voltage(row, shunt_ohms):
    label = f"{row.stage} step {row.step}"
    screen(within(row.top, -0.05, 5.00) and within(row.coil, -0.05, 5.00),
           f"{label}: check source and actual coil voltage")
    screen(within(row.shunt, -0.004, 0.37), f"{label}: invalid shunt voltage")
    if row.top <= 0.05 + EPS:
        screen(abs(row.shunt) <= 0.004 + EPS and abs(row.coil) <= 0.05 + EPS,
               f"{label}: zero and shunt readings disagree")
    if row.coil >= 0.65 - EPS:
        screen(row.shunt > 0 and within(row.coil_resistance(shunt_ohms), 143, 218),
               f"{label}: inferred coil resistance outside the broad fixture screen")


def bracket(rows, going_up):
    name = rows[0].stage
    labels = STEPS if going_up else STEPS[::-1]
    screen(all(row.step == index for index, row in enumerate(rows)),
           f"{name}: incorrect step indexes or order")
    screen(all(abs(row.top - level) <= 0.055 + EPS for row, level in zip(rows, labels)),
           f"{name}: source does not match the planned ladder")
    for left, right in zip(rows, rows[1:]):
        delta = right.coil - left.coil if going_up else left.coil - right.coil
        screen(within(delta, 0.085, 0.350), f"{name}: actual coil steps are missing or nonmonotone")

    states = [row.closed() for row in rows]
    changes = [idx for idx in range(1, len(states)) if states[idx] != states[idx - 1]]
    screen(states[0] == (not going_up) and states[-1] == going_up and len(changes) == 1,
           f"{name}: expected exactly one contact transition")
    idx = changes[0]
    return tuple(sorted((rows[idx - 1].coil, rows[idx].coil)))


def compare(rows, shunt_ohms, sweeps_only):
    expected = [name for name in SWEEPS for _ in STEPS]
    if not sweeps_only:
        expected.extend(TRIAL)
    screen([row.stage for row in rows] == expected, "incorrect stage count or order")
    for row in rows:
        check_voltage(row, shunt_ohms)
    limits = []
    for index, name in enumerate(SWEEPS):
        start = index * len(STEPS)
        edge = bracket(rows[start:start + len(STEPS)], name.startswith("up"))
        limits.append(edge)
        print(f"{name}: {edge[0]:.3f} to {edge[1]:.3f} V terminal coil")
    low = max(limits[1][1], limits[3][1])
    high = min(limits[0][0], limits[2][0])
    lo, hi = max(low + 0.050, 0.65), high - 0.050
    screen(lo < hi - EPS, "no safe observed middle region across both sweeps")
    print(f"Candidate middle terminal voltage: {(lo + hi)/2:.3f} V"
          f" (inside observed guarded interval {lo:.3f} to {hi:.3f} V)")
    if sweeps_only:
        return

    idle, boost, held_early, held_late, reset = rows[-len(TRIAL):]
    screen(all(row.step == 0 for row in rows[-len(TRIAL):]), "middle trial requires step indexes of 0")
    screen([row.closed() for row in rows[-len(TRIAL):]] == [False, True, True, True, False],
           "contact history does not show expected idle/boost/hold/reset states")
    screen(abs(boost.top - 4.90) <= 0.055 + EPS and within(boost.coil, 4.40, 4.90)
           and abs(reset.coil) <= 0.05 + EPS,
           "need a full boost plateau and a near-zero reset")
    matched = [idle.coil, held_early.coil, held_late.coil]
    screen(all(lo < value < hi for value in matched), "middle voltage is outside the guarded region")
    screen(max(matched) - min(matched) <= 0.035 + EPS, "the two histories did not use the same terminal voltage")
    coil_r = [row.coil_resistance(shunt_ohms) for row in (idle, held_early, held_late)]
    screen(max(coil_r) / min(coil_r) <= 1.08 + EPS, "middle winding readings changed appreciably")
    boost_w = boost.power(shunt_ohms)
    hold_w = held_late.power(shunt_ohms)
    screen(0 < hold_w < boost_w, "steady terminal coil power did not decrease")
    print(f"Matched terminal voltage: idle {idle.coil:.3f} V, held {held_late.coil:.3f} V")
    print(f"Coil only: boost {1000*boost_w:.2f} mW, held {1000*hold_w:.2f} mW"
          f" ({100*(1 - hold_w / boost_w):.1f}% lower in this record)")
    print("QUALIFIED: two histories gave different contact states at the same coil voltage.")


def main():
    parser = argparse.ArgumentParser(description=__doc__)
    parser.add_argument("record", type=Path)
    parser.add_argument("--sweeps-only", action="store_true", help="analyze only the 84 ladder readings")
    parser.add_argument("--shunt-ohms", type=float, default=10.0)
    args = parser.parse_args()
    if not 9.85 <= args.shunt_ohms <= 10.15:
        parser.error("shunt resistance must be finite and within 9.85–10.15 ohms")
    try:
        compare(read_rows(args.record), args.shunt_ohms, args.sweeps_only)
    except UnicodeError:
        print("DATA ERROR: expected UTF-8 CSV", file=sys.stderr)
        return 2
    except (OSError, ValueError, csv.Error) as error:
        print(f"DATA ERROR: {error}", file=sys.stderr)
        return 2
    except Unresolved as error:
        print(f"UNRESOLVED: {error}", file=sys.stderr)
        return 1
    return 0


if __name__ == "__main__":
    raise SystemExit(main())

Check one candidate voltage against both directions

Suppose the lowest recorded pickup bracket starts at 3.00 V, and the highest dropout bracket ends at 1.25 V. With the 50 mV guard at both ends, does a 1.28 V actual coil voltage qualify for the two histories?

Show the answer

No. The guarded interval is 1.30–2.95 V, and 1.28 V is below it. A qualifying measured middle strictly inside that band would still have to keep the contacts closed for the post-boost samples and leave them open when approached from zero.

Sources and related experiments