Designing an Isolated EMG Solenoid Control Circuit

Mark Townsend8 min read
Application NoteOther ManufacturerSensor Integration
Licensed PE Working through this on a live machine? A Maine-licensed engineer can take it from here — included with IMD hardware, by the hour for everything else. Book an engineer

On the control panel, the EMG level may move while the valve command stays off, or the valve may chatter when the muscle signal crosses the threshold. Start by separating the design into a body-connected EMG input, an isolated decision stage, and a protected solenoid output. Never connect electrodes to a valve-driver circuit until a qualified medical-instrumentation engineer has reviewed the isolation, leakage-current, fault, and commercial-compliance design.

Read the symptom before changing hardware

An EMG electrode produces a small differential signal in a noisy environment. The solenoid coil is an inductive power load. Treating them as one circuit creates false triggering, ground-current paths, and output-stage failures.

Observed symptom Likely cause or first check
EMG display changes, but the valve never energizes Check threshold logic, output mapping, switch polarity, coil supply, and whether the low-side driver shares the intended logic reference.
Valve chatters near muscle activation The threshold lacks hysteresis, filtering, or minimum state qualification. Noise may be crossing the decision level repeatedly.
EMG reading jumps when the valve switches Coil flyback, supply disturbance, wiring coupling, or a shared ground is feeding switching noise into the amplifier.
Signal rails high or low Check electrode contact, amplifier input range, reference electrode, gain distribution, supply headroom, and input protection.
Driver becomes hot or fails Compare coil voltage and current with the switch ratings. Check the flyback path, duty cycle, gate or base drive, and load wiring.
Touching grounded equipment changes the reading Stop. A USB cable, oscilloscope, programmer, power supply, or communications lead may have bypassed the intended isolation boundary.

Check isolation and grounding first. Replacing amplifier parts or increasing gain wastes time when the real fault is a ground path or valve-generated interference.

Split the signal chain into controlled blocks

The correct architecture is not electrode-to-amplifier-to-solenoid. Use distinct functional blocks:

  • Electrode interface: Accept the differential surface-electrode signal while limiting body-connected current under normal and fault conditions.
  • Analog front end: Reject common-mode interference, apply only the needed gain, and filter unwanted content without saturating.
  • Signal extraction: Rectify or otherwise derive an activation measure suitable for control. Raw bipolar EMG is not a stable valve command.
  • Decision logic: Apply a threshold, hysteresis, state qualification, plausibility checks, and a defined response to invalid input.
  • Isolation boundary: Prevent hazardous current and switching noise from crossing between the body-connected section and the powered control section.
  • Valve output: Drive the coil through a rated switch and an inductive-energy suppression path.
  • Pneumatic control: Regulate pressure and flow so an electrical command cannot create uncontrolled actuator force or motion.

The controller does not program flow merely by energizing a coil. A simple on/off valve produces discrete pneumatic states. Proportional flow needs a valve and driver intended for proportional operation, plus an appropriate command method and feedback where the mechanical risk demands it.

Define the body-side safety boundary

Isolation is a system property. An isolated signal component does not make the assembly isolated when another cable bridges the barrier.

  • Draw the boundary on the schematic and identify every conductor that crosses it.
  • Account for power, signal, shield, programming, communications, enclosure, sensor, and test-equipment connections.
  • Keep the electrode-side power source and data path within the selected medical electrical architecture.
  • Define safe behavior for open electrodes, shorted inputs, saturated amplifiers, loss of power, processor reset, communications failure, and a stuck output.
  • Make de-energization the default state unless the risk analysis requires a different state.
  • Provide an independent way to remove pneumatic energy; software alone is not a protective measure.

Battery operation can remove one mains path, but it does not settle the safety question. Charging leads, USB links, grounded instruments, conductive enclosures, and valve-side wiring can reintroduce a hazardous path.

For a commercial body-connected product, treat IEC 60601-1 as a minimum cited requirement to evaluate. Have a medical-device regulatory specialist identify the product classification, applied-part requirements, leakage-current limits, insulation requirements, fault tests, documentation, and any additional standards that apply. Passing one isolation check does not demonstrate product compliance.

Select the valve output stage from coil data

The referenced concept uses an H-bridge, but a normal unidirectional solenoid coil does not need bidirectional current. Replace it with a low-side switch when the valve datasheet permits that topology. Retain a bridge only when the load explicitly requires current reversal or a special drive profile.

Read these values from the valve datasheet or coil label before selecting parts:

  • Nominal coil voltage and allowable voltage range
  • Steady-state current or coil resistance
  • Inrush behavior, if specified
  • Permitted duty cycle
  • Required pickup and release behavior
  • Manufacturer restrictions on suppression devices

Rate the switch for the actual supply, coil current, inductive transient, thermal environment, and fault condition. Include a flyback or other suppression network selected for the required release performance. A simple recirculation diode usually reduces the turn-off voltage but can slow coil-current decay; that can delay valve release. Verify the chosen suppression method against the valve and switch documentation.

Separate the dirty coil-current path from the EMG analog return. Route the supply-to-coil-to-switch loop as a compact power loop. Place suppression at the inductive load or driver as the physical design requires, and keep switching conductors away from electrode and amplifier input wiring.

Build the design in a controlled sequence

  1. Write the operating states. Define valve-off, valve-on, invalid-EMG, controller-reset, loss-of-power, emergency-stop, and pneumatic-energy-removal behavior.
  2. Characterize the valve separately. Use a current-limited source and a suitable test load arrangement. Confirm coil current, driver temperature, pickup, release, and suppression without connecting any electrode circuit.
  3. Develop the EMG front end with a simulator. Use an isolated signal source or approved patient simulator. Check input range, saturation recovery, common-mode response, noise, and electrode-disconnect behavior.
  4. Create the control variable. Convert the acquired signal into an activation measure. Add filtering appropriate to the control objective, then add separate on and off thresholds so noise cannot toggle the valve repeatedly.
  5. Add command qualification. Require a valid input state before permitting an output. Force the output off when the signal is out of range, the acquisition stage saturates, or the controller detects an internal fault.
  6. Insert and audit the isolation boundary. Review every power and signal crossing. Repeat the review with programmers, computers, chargers, oscilloscopes, and pneumatic hardware attached.
  7. Integrate with a non-body test source. Operate the complete electrical and pneumatic chain using simulated EMG. Test normal commands and every defined fault state.
  8. Obtain the required professional review. Complete the medical electrical, mechanical, pneumatic, software, and regulatory reviews before any human connection or commercial trial.

Verify the fix without using a person as the test load

Verification must show both correct control and safe failure. Record the test setup, supply conditions, simulated input, output state, coil current, pneumatic response, and pass criterion.

  • Sweep the simulated EMG level above and below both thresholds. Confirm one clean transition in each direction.
  • Inject representative interference and switch the valve repeatedly. The EMG channel must remain within its intended operating range.
  • Open and short each simulated electrode connection. Confirm the programmed safe state.
  • Interrupt and restore power to each subsystem. The valve must not pulse unexpectedly during startup, shutdown, or reset.
  • Disconnect communications and halt the controller. Confirm the independent pneumatic and electrical protective actions.
  • Check switch and coil temperature across the intended duty cycle and ambient conditions.
  • Measure isolation and body-connected safety parameters with the methods and equipment required by the applicable conformity plan.
  • Repeat noise tests with all normal external cables fitted. A successful battery-only bench test does not cover a grounded programming connection.

Do not accept “the valve moved” as verification. The pass condition includes predictable thresholds, bounded pneumatic action, suppression of false commands, and a safe response to each single fault identified by the design review.

Avoid the recurring design traps

  • Do not copy a schematic into a parts list. Part ratings depend on the electrode interface, supplies, valve coil, duty cycle, isolation architecture, enclosure, and regulatory classification.
  • Do not solve saturation with more gain. Find the DC offset, common-mode interference, electrode-contact problem, or missing reference path first.
  • Do not drive the coil from a logic output. Use a rated power switch, suppression, and a defined default state.
  • Do not treat an H-bridge as mandatory. Use a low-side switch for a unidirectional coil when the valve documentation allows it.
  • Do not use software as the only motion limit. Limit pneumatic energy and provide an independent means to remove it.
  • Do not test isolation with grounded bench equipment casually attached. Test leads can create the very current path the barrier was meant to prevent.
  • Do not begin human testing to tune thresholds. Establish the complete operating envelope with simulation and controlled mechanical loads first.

FAQ

Why does the solenoid chatter when the EMG signal rises?

The activation value is crossing one threshold repeatedly because of noise or normal EMG variation. Use separate on and off thresholds, appropriate filtering, and input-validity logic, then verify the result with a simulated signal.

Why does switching the valve corrupt the EMG reading?

The coil transient or supply-current loop is coupling into the high-gain analog input. Correct the suppression, grounding, physical routing, supply decoupling, and isolation layout before changing amplifier gain.

Why does an EMG solenoid circuit need isolation?

Surface electrodes create a conductive connection to the body, while valve supplies, computers, chargers, and test equipment can introduce fault-current paths. The complete system boundary—not one isolator component—must satisfy the applicable medical electrical requirements.

When should I stop troubleshooting and contact official support?

Stop before human connection if the isolation path, leakage-current limits, valve ratings, pneumatic force limits, or fault response remain unresolved. Escalate to the relevant manufacturers’ official technical support, a qualified medical-instrumentation engineer, and the applicable certification laboratory; a novice-built parts list is not an acceptable basis for a commercial body-connected product.

Back to blog