Power Sequencing and Ramp Control on a Board

A board with several supply rails must bring them up in an order that the devices can tolerate. Where the order is wrong, a device can be damaged before the firmware has a chance to run, and the damage may not appear until the product is in the field.

Why Sequencing Matters

Many devices have a specification for the maximum voltage difference between two of their pins. A core supply that rises before the input and output supply can forward bias an internal junction and cause current to flow through a path that was never designed for it.

The reverse order can produce a latch up, where a parasitic structure inside the device turns on and holds, drawing a large current until the device is destroyed.

The requirement is stated in the device data sheet as a sequencing order and a maximum difference. Where the design ignores it, the failure appears as an unexplained early life failure. Our component reliability notes describe how the requirements are collected.

Inrush and Ramp Rate

A large capacitor bank draws a large current when the supply is applied. The inrush current can exceed the supply’s rating, cause the supply to shut down, or blow a fuse.

The inrush is controlled by a series element that limits the initial current or by a soft start circuit that ramps the output. A simple thermistor works where the supply is applied once, and a resistor with a bypass switch is used where the input may be cycled.

The ramp rate also matters for a device with a power on reset circuit that requires a minimum rate. A supply that rises too slowly can leave the device in an undefined state. Our power integrity notes describe the supply distribution.

Methods of Sequencing

A sequencing controller drives the enable pins of the regulators in a defined order, with defined delays. It is the most flexible and the most explicit method.

A cascade arrangement uses the power good output of one regulator to enable the next. It requires no additional part and it depends on the power good threshold of each regulator, which may not coincide with the requirement.

A resistor divider or a simple RC delay on the enable pins is the cheapest method and the least precise, because the delay depends on the threshold and the tolerance of the parts. Our switching regulator notes describe the enable circuitry.

Power Down and Brown Out

The power down sequence matters as much as the power up sequence, and it is harder to control because the supplies discharge at rates set by their loads.

Where a supply falls faster than another, the same voltage difference that was avoided at power up appears at power down. The remedy is a discharge circuit on the rail that would otherwise stay high.

A brown out, where the input falls below the minimum but does not disappear, is the hardest case. The device may partially reset, and the firmware must detect the condition and protect any non volatile data. Our design release checklist notes where the behaviour is recorded.

Reset and Its Timing

The reset signal must be held until all of the supplies are within specification and stable. A reset that is released early allows the device to start in an undefined state.

A dedicated supervisory circuit monitors the supply and holds the reset for a defined period after the level is reached. It is more reliable than a simple RC network, whose timing varies with the tolerance of the parts.

The reset should also be asserted when the supply falls, so that the device is held in reset rather than running at a marginal voltage. The threshold and the hysteresis determine how the device behaves during a brown out.

Verification

The verification is a measurement of all of the rails during power up and power down, with the time between them and the intermediate voltages recorded.

The measurement should be made at the device pins and not at the regulator output, because the trace and the decoupling network change the waveform that the device sees.

The test should include the worst case conditions: the fastest and slowest ramp the supply can produce, the maximum load and the minimum load, and the temperature extremes. Our quality notes describe how the measurements are recorded.

Design Checklist

Collect the sequencing requirements of every device and reconcile them into one order. Where two devices conflict, choose the order that satisfies the more sensitive one and check the other against its absolute maximum ratings.

Provide an explicit sequencing method rather than relying on the coincidence of the regulator thresholds. Provide a discharge path where a rail would otherwise remain high at power down.

Provide a supervisory reset that holds the device until the supplies are stable and that asserts during a brown out. Record the measured waveforms for the production configuration.

Process Control and Verification

On a design of this kind, brown out is the item that decides how the rest of the board is arranged. Running a first article through the same checks as the production panel confirms that the two agree, and that comparison is the cheapest form of process control available at prototype stage. Keeping a sample from the panel turns a dispute into a measurement, because the same coupon can be re-examined by both parties without rebuilding the batch.

Reviewing the design before the data is released is cheaper than correcting it after the panel is in the tank, because every step downstream inherits the decision made at the front end. Documenting the assumption is part of the design work, and a short note on the drawing prevents a question that would otherwise arrive a day later and cost a day of schedule.

Process Control and Verification

On a design of this kind, brown out is the item that decides how the rest of the board is arranged. Running a first article through the same checks as the production panel confirms that the two agree, and that comparison is the cheapest form of process control available at prototype stage. Keeping a sample from the panel turns a dispute into a measurement, because the same coupon can be re-examined by both parties without rebuilding the batch.

Reviewing the design before the data is released is cheaper than correcting it after the panel is in the tank, because every step downstream inherits the decision made at the front end.

Process Control and Verification

On a design of this kind, brown out is the item that decides how the rest of the board is arranged. Running a first article through the same checks as the production panel confirms that the two agree, and that comparison is the cheapest form of process control available at prototype stage. Keeping a sample from the panel turns a dispute into a measurement, because the same coupon can be re-examined by both parties without rebuilding the batch.

FAQ

Is sequencing necessary on a board with a single supply? There is no order to control, and the reset timing and the ramp rate still matter.

Can the sequencing be done in firmware? Only after the device is running, which is after the damage may already have occurred. The hardware must handle the power up order.

What does gopcb provide for power sequencing designs? We provide a consolidated sequencing requirement from the device data sheets, a sequencing method with measured delays, inrush limiting, discharge paths for power down, supervisory reset circuitry, and power up and power down waveforms measured at the device pins at the load and temperature extremes.