DC Motor and Actuator Control: H-Bridge, High-Side Driver, or a Purpose-Built Module

Updated August 2026

A programme rarely starts with the words “we need an H-bridge”. It starts with a list: a fresh-air flap, two louvre actuators, a tailgate latch, a washer pump, and somewhere near the bottom, a note that three of them have to move both ways. That last note is the whole specification, because it decides which half of the catalogue you are shopping in.

This guide is for OEM engineering buyers and body-electrics owners scoping DC motor and actuator control on a commercial vehicle or a machine. Where the control function sits relative to the body controller and the power management unit is covered in the BCM, VCU and PMU comparison, and the module stack as a whole in the Smart Control Modules technical guide. What follows is narrower: why direction changes the part you buy, how to size channels against stall current rather than running current, and how to write the requirement so a supplier quotes a drive stage instead of guessing at one.

1. Direction is the dividing line

Most smart drive channels on a vehicle are high-side drivers. One switch sits between the positive rail and the load; the load's other end goes to chassis ground. Close the switch and current flows, open it and it stops. Add PWM and you vary how much of the time the switch is closed, which dims a lamp or softens the pull-in on a solenoid. What you cannot do, with any duty cycle, is make the current flow the other way. There is only one path.

An H-bridge is four switches drawn as the two uprights of an H with the motor as its crossbar. Close the top-left and bottom-right switches and current crosses the motor one way; close the other diagonal and it crosses the other way. Open the two low-side switches together and the motor coasts; close them together and it brakes into its own windings. One load, four useful states, and the reason the part costs more than a high-side channel: four switches instead of one, plus the interlock that stops both switches on one upright being closed at the same time, which would put a short across the battery.

Between those two sits a third answer that is easy to overlook because it is not electronic at all. On heavy body functions the motion is hydraulic, and direction is chosen by a solenoid valve. The electrical side is back to on/off — energise this coil to extend, that coil to retract — so the controller needs relay or high-side outputs, not bridges, even though the thing moving on the vehicle clearly moves both ways.

Sorting the load list into those three columns before anyone opens a catalogue removes most of the later argument. Lamps, horns, heaters, beacons, single-acting solenoids and any pump that only ever runs one way are high-side loads. Flaps, dampers, louvres, valve actuators, window and mirror motors are bridge loads, along with a pump that has to reverse. Tipper bodies, tailgates and PTO functions are usually hydraulic, which puts them back on relay and high-side outputs with the direction decided by the valve block.

Once a bridge is needed, the next question is who sells it and how much of the behaviour comes with it.

2. Which layer you buy it in

Once a load has been settled as electric and bidirectional, a bridge is driving it whether you buy one or not. A wiper controller has one inside it. An anti-pinch window module has one inside it. The question is rarely whether the bridge exists; it is which layer of the supply chain owns the behaviour wrapped around it.

The table below places the relevant Youlai parts by what they actually drive, which is not always what their names suggest.

The EBX‑2050 advertises twelve PWM channels, but those channels are high-side: they are sized for solenoid valve coils, not for reversing a motor, and a programme that reads “12 PWM” as “12 motor channels” discovers the difference during integration. The EBX‑2208 is the mirror image: nothing in its name says bridge, yet a wiper motor is a bidirectional load and the module drives it accordingly, with the park-position feedback loop and the reverse-to-park behaviour that a general-purpose bank would leave you to write.

The EBX‑951 is the general-purpose row. Eight bidirectional channels: seven rated 5 A with PWM speed control and per-channel current sensing, one rated 10 A with current sensing and no PWM, all sealed to IP67 for a chassis-side position where the protection grade has to survive road spray. It is what you buy when the actuators are yours, the behaviour is yours, and no catalogue part is shaped like your function.

3. Sizing the channels: running current, stall current, duty cycle

A buyer reads the actuator data sheet, sees a running current of 2 A, and allocates it to a 5 A channel with a comfortable margin. Then the actuator drives into its end stop, the motor reaches zero speed, the back-EMF that was holding the current down disappears, and what is left to limit it is the winding resistance and the supply. Stall current is a multiple of running current rather than a margin on it, and how large a multiple is a property of that particular motor — the actuator data sheet or a measurement on the mechanism is the only place the figure can come from.

Three numbers per motor decide the allocation, and only one of them is on the front page of most actuator data sheets:

• Running current under the load your mechanism actually presents, which is not necessarily the figure measured on a free shaft.
• Stall current, and just as importantly how long a stall lasts. An actuator with internal limit switches stalls for milliseconds during inrush. One that drives into a mechanical stop and waits for the operator to release a switch can stall for as long as a finger stays down.
• Duty cycle — what fraction of the time the channel is energised, and whether several channels are energised together. A flap that moves for three seconds an hour is a different thermal problem from a pump that runs continuously.

On the EBX-951, that arithmetic is what decides which load takes the single 10 A channel. Its asymmetric layout, seven 5 A plus one 10 A, is not an accident: real actuator sets are asymmetric, with one heavy consumer and a group of small ones, and paying for eight heavy channels to serve one load is a poor trade.

The current-feedback range on the seven 5 A channels is 6 A — deliberately above their 5 A drive rating, so an overload shows up as a reading rather than as a value pinned at the top of the scale. Only up to 6 A, though: a motor whose stall runs well past that is having its most interesting condition measured at the limit of the measurement. The drive rating decides which channel the load goes on; this decides how much you will be able to see once it is there.

Ambient temperature belongs in the same conversation. Every semiconductor drive stage carries the heat it dissipates out through its housing, so the same channel doing the same work has more margin on a cab-floor bracket than on a chassis rail beside an exhaust. The module's own range, −40 to +85 °C, is the envelope it survives, not a promise that every channel delivers its full rating at the top of that envelope. Tell the supplier where the box is mounted, not only what is plugged into it.

4. What PWM and current feedback actually buy you

PWM speed control

Pulsing the bridge at a duty cycle between 0 and 100% sets the average voltage across the motor, and therefore its speed. On an actuator that translates into approach rate, which is what makes the difference between a louvre that arrives at its stop and one that slams into it. Soft-start matters for the same reason at the other end: ramping the duty cycle up rather than applying full voltage instantly reduces the inrush the harness and the battery see, which on a vehicle with many actuators is a real load-budget item rather than a refinement.

The EBX-951 makes the PWM frequency adjustable across 10 Hz to 2 kHz on its seven 5 A channels, and that range is a trade-off rather than a menu. At the low end, current has time to decay between pulses, so torque is delivered in visible steps and the motor can be heard working; at the high end, the current is smoother and any acoustic artefact moves out of the range that irritates a driver, at the cost of more switching loss and a more demanding electromagnetic profile on the harness. Which end of that range a given motor wants depends on its inductance and on how close it sits to the cab, which is why the parameter is set during integration and confirmed on the vehicle rather than copied from another programme.

The single 10 A channel has no PWM. It is direction and full drive only, which is the right answer for the heavy actuator that just has to get there, and the wrong answer for one that needs a controlled approach. That distinction, not raw current, is often what decides which load takes the channel.

Per-channel current sensing

Each bridge channel reports what its motor is drawing. That single number is the raw material for most of the clever behaviour buyers ask about: detecting a jam, inferring that an actuator has reached its end of travel without fitting a limit switch, cutting a channel before a stalled motor cooks itself, or noticing over a season that a linkage is stiffening.

What matters commercially is where the boundary sits. The module provides the measurement; the interpretation — what current counts as a stall on your mechanism, how long it must persist before it means something, and what happens next — is defined per programme. The same 4 A reading is normal on a stiff damper in winter and a fault on a free-running flap. If you want the interpretation delivered as a validated feature rather than as integration work, that is the argument for the function-module layer: the EBX‑2206 ships anti-pinch torque detection and travel self-learning as specified behaviour, which is the same physics with the decisions already made and tested.

Two protections sit underneath all of this and are not programme-specific. Each 5 A channel carries a built-in freewheeling diode, so the inductive kick from switching a motor off is absorbed on the board instead of being pushed back onto the harness where it becomes everyone's EMC problem. And a three-stage supply-voltage watchdog guards the whole bank against what a commercial-vehicle 24 V rail actually does: an alarm with a slow-blinking red LED when the supply stays below 18 V, a driver cut-off below 9 V, and a shutdown above 32 V. Cranking sags and a mis-connected jump-start are ordinary events on a truck, and a bridge bank without that state machine turns them into warranty claims.

5. Where the drive stage sits, and on which vehicles

The EBX-951 takes its commands through 4 digital inputs and 8 analog inputs, and its specification lists no CAN or LIN interface. On a site where every other module talks over a bus this reads at first like a shortcoming; in a drive stage it is a design position.

A hardwired drive stage needs no seat on the vehicle message map, no source address, no agreement about what an actuator does when a frame goes missing, and no place in the network load budget. Switches, limit switches and position sensors wire straight to it, and the module does what the wiring says. Where the deciding logic lives is then a separate architectural question: it can be an upstream controller driving the digital inputs, or it can be the switch panel itself. The trade-off between hardwired and bus-attached command paths, and when each is the right answer, is worked through in the CAN, LIN and hardwired switches comparison; the whole-vehicle version, including where a drive stage belongs relative to the control layer, is in the E/E architecture guide.

The practical consequence is a scoping question, not a technical objection: if you want an operator to press a button on a CAN keypad and have a flap move, something upstream has to turn that message into a discrete signal. Either a body controller such as the heavy-truck BCM does it, or the requirement points at a bus-attached module instead. What the catalogue does not offer on the bus-attached side is a general-purpose bridge bank: the bus-attached parts that do contain bridges — the EBX‑2206 window module, the EBX‑2163 and EBX‑2315 door and window controllers — carry them inside a fixed function, so they answer only if your load happens to be that function. Naming this in the RFQ is what stops it surfacing at integration.

The same requirement arrives in a different shape on each platform:

• Heavy-truck cabs. Bidirectional loads are numerous but individually small, and most of them already have a dedicated part: windows and mirrors on the EBX‑2163 or EBX‑2315, wipers on the EBX‑2208, retractable steps and central locking on the EBX‑2207. The cab is the platform where the function-module layer usually wins, and the power window and door control guide covers that side of the vehicle in detail.
• Construction and agricultural machinery. The opposite picture, and the natural home for a general-purpose bank. Cooling flaps, radiator louvres, fuel and DEF valve actuators and attachment functions vary from machine to machine, so no catalogue function module fits, while the mounting position is exposed enough that IP67 stops being a preference. That is why the EBX-951 sits alongside the sealed distribution hardware in the construction machinery solutions set.
• Bus and coach. Passenger-facing actuators — saloon ventilation flaps, destination-sign mechanisms, luggage-bay latches — are bidirectional and often specified late in the build, after the main architecture has been frozen. A bank of general-purpose channels absorbs late additions better than a function module chosen for a function that then changes.
• Tipper and dump bodies. Frequently not an electric-motor problem at all. Top cover, tailgate and lift are hydraulic, so the controller drives valve coils rather than bridges — which is exactly what the EBX‑963 does, with the direction decided in the valve block and the electronics reduced to sequencing, interlocks and remote control.

Across all four, the requirement that outlives the part number is the load list with its currents and its duty cycles. Modules get superseded; the fact that a particular flap actuator stalls at 8 A does not.

6. How to write a motor-control requirement a supplier can quote

A drive stage is quoted against a load list, not against a part description. The six items below turn “we need a motor control module” into a channel allocation, a price and a lead time. The commercial half of the package — volumes, samples, timing, quality deliverables — is covered by the RFQ checklist; this is the technical half specific to actuator control.

1. The motor list. One row per actuator: what it moves, running current under load, stall current, how long a stall can last, whether it needs speed control, and whether it has internal limit switches or drives into a stop. This single table decides the channel count, the channel ratings and which load takes the high-current channel, and it is the item most often missing from a first enquiry.
2. Direction and simultaneity. Which loads are genuinely bidirectional, and which of them can be energised at the same time. Simultaneity is the part programmes skip, and it is what turns a set of per-channel ratings into a thermal question.
3. Command source and feedback. What drives the inputs — cab switches, an upstream controller, or a bus message that something else has to translate — and what comes back: position potentiometers, limit switches, pressure or temperature sensors. Name the sensor type and its output range, because that is what sets up the analog input configuration.
4. Behaviour at the edges. What should happen at end of travel, on an over-current event, when the supply sags during cranking, and when the operator releases a switch mid-travel. These are the questions that get answered by default if nobody writes them down, and the default is rarely what the programme wanted.
5. Environment and mounting. Where the box lives, the ambient temperature there, the required IP grade and the wash-down regime. A cab-floor position and a chassis-rail position are different parts of the same catalogue, and sealing is decided by position rather than by preference.
6. Harness interface. Connector family, sealing requirement and whether an existing harness has to be matched. The EBX-951 lands on four sealed TE connectors — three 23-pin bodies colour-coded black, blue and natural, plus one 8-pin — with per-pin allocation issued against your harness drawing at quotation. If the connector set is fixed on your side, say so early, because it is a hardware constraint rather than a preference; the connector guide covers how that choice interacts with sealing and current.

The motor list and the edge behaviour are the two to insist on: write the list before choosing a part, and settle the edges before integration starts.

7. What to look for in a supplier

A drive stage is a part where the invoice is for hardware and the risk is in the details around it: thermal behaviour under your duty cycle, what the protection state machine does on a bad rail, and whether anyone will still take a call in year eight.

• Willingness to work from your load list. A supplier who answers “yes, 5 A per channel” has not answered the question. Ask instead how they allocate your specific motors across channels, what they assume about simultaneity, and how ambient temperature at your mounting position changes the answer. That reply tells you whether the sizing has been done or deferred.
• Protection described as behaviour, not as adjectives. “Fully protected” means nothing. Thresholds, timings and what the module actually does mean something — on the EBX-951, an under-voltage alarm below 18 V held for 1 s, a driver cut-off below 9 V held for 0.5 s, a shutdown above 32 V held for 0.5 s, each with a distinct red-LED pattern so a technician can read the state without a laptop.
• Validation you can inspect. A bank of switching drivers on a vehicle harness needs credible transient and EMC evidence, and the usual commercial-vehicle set covers ISO 16750 and ISO 7637 alongside the quiescent-current budget that matters for an always-on position. Youlai validates in an in-house environmental laboratory with EMC pre-compliance equipment, and certified testing is performed at third-party accredited laboratories when the programme requires it.
• Honesty about the function boundary. The most useful answer to “can it detect a jam?” is a description of what is measured and what has to be defined, not a yes. A supplier who separates hardware capability from programme-specific logic at quotation stage is a supplier who will not surprise you at validation.

Quality system and PPAP level are on the IATF 16949 and OEM sourcing pages.

A motor list, even a partial one, is enough to say whether the requirement lands on the EBX‑951 as a general-purpose bridge bank, on a function module, or on a hydraulic controller. The Smart Control Modules technical guide, the BCM, VCU and PMU comparison, and the module selector cover the wider stack.

For drawings, a channel-allocation review against your motor list, or a sample request, please use the contact page or message +86 134 6767 4786 on WhatsApp. Typical reply within one business day.