Programmable DC Power Supply

What is a Programmable DC Power Supply

A DC power supply takes an energy source, usually single-phase or three-phase AC mains, and produces a stable direct-current output whose voltage and current are held within tolerance despite changes in the load or the input line. The signal chain is consistent across nearly every design: the incoming AC is first rectified into raw DC, that raw DC is filtered to smooth the ripple, and a regulation stage then trims the output to the commanded value. The word "programmable" means the set points are not fixed by a knob alone but can be commanded electronically and read back over a communication interface, so the instrument can run inside an automated test sequence without a human turning dials.

Functionally, the instrument is built from four blocks: (1) the input and conversion stage, which rectifies and conditions the source; (2) the regulation stage, a series pass element or a switching converter that does the actual voltage and current control; (3) the metering and feedback stage, which senses output voltage and current and closes the control loops; and (4) the control and interface stage, a microcontroller front panel with a remote bus that accepts set-point commands and returns measured readings. When the metering stage reports values back to the operator, that path is called readback, and its accuracy is specified independently of the set-point (programming) accuracy.

The instrument differs fundamentally from a fixed wall adapter or an unregulated transformer pack in three ways. It actively regulates against line and load disturbance rather than letting the output sag. It enforces a current limit, protecting both the device under test and the supply itself, rather than delivering unlimited fault current. And it exposes both a manual front panel and a remote programming interface, so the same set point can be reached by hand on the bench or by a script in production. Those three properties are why the programmable DC supply, not the adapter, is the reference instrument of electronics test.

Historically the building blocks arrived in stages. Practical DC regulation began with the vacuum-tube and then transistor series regulator in the mid twentieth century, which gave the low-noise linear supply still prized for analog work. High-frequency switching regulation matured in the 1970s and 1980s, trading some output cleanliness for a large gain in efficiency and a large reduction in size and weight. Remote control was standardized when the IEEE-488 (GPIB) bus and later the SCPI command language let one controller drive a whole rack of instruments with portable commands. Modern supplies now add LAN, USB, and LXI networking, on-board sequencing and data logging that can offload to a dedicated data logger, and modular mainframes that mix supply, source-measure, and electronic-load channels in one chassis.

In application terms the category spans an enormous range. A small triple-output bench unit may deliver tens of watts at up to 30 V and a few amps for circuit-board bring-up, while a modular system mainframe can deliver well over a kilowatt across multiple precision channels for production ATE, and high-power rack supplies reach hundreds of amps for battery, electrolysis, and burn-in duty. No single instrument serves all of this. The work of selection, covered in Chapter 6, is matching the load profile, accuracy need, and control environment to a specific topology and channel count.

Power Supply Types and Form Factors

Programmable DC supplies divide along two axes: the regulation topology (linear versus switching, treated in Chapter 3) and the form factor and channel architecture, treated here. Form factor maps closely to use case, so identifying the right class first narrows the search before any specific spec is compared. The table below summarizes the main classes a buyer will encounter.

Single-output bench supplies are the simplest class: one isolated output, a front-panel knob or keypad, and often a basic USB interface. They suit single-rail prototyping, field repair, and education. Multi-output bench supplies add two to four independent outputs in one box, typically a pair of higher-voltage rails plus a low-voltage logic rail. The Rigol DP832, for instance, provides three channels rated 0 to 30 V / 0 to 3 A on channels one and two and 0 to 5 V / 0 to 3 A on channel three, a layout that mirrors the common need to power an analog stage, its complement, and a digital rail at once.

Modular system supplies are the workhorses of ATE. A mainframe accepts plug-in output modules, so one chassis can mix high-current, high-voltage, low-noise, and source-measure modules to match the device under test, all controlled over one LAN or GPIB address. The Keysight N6700 modular platform is a representative example, offering basic, precision autoranging, and source-measure module types in a single mainframe rated several hundred watts to over a kilowatt. This modularity is what lets one instrument replace a shelf of dedicated supplies.

High-power rack supplies serve battery formation, electrolysis, magnet, and EV-component testing, where outputs reach hundreds of amps or over a thousand volts and several units are paralleled or stacked for still higher power. Many of these support autoranging, where the output traces a constant-power curve so a single unit covers a wide voltage-current area, and the most advanced are regenerative, returning sink energy to the grid instead of dissipating it as heat during battery discharge or load test.

Source-measure units (SMUs) blur the line between supply and meter. An SMU sources voltage or current and simultaneously measures the other variable with high resolution, down to the microampere and nanoampere region, and operates in two or four quadrants so it can both source and sink. SMUs are the tool for semiconductor I-V curves, leakage measurement, and precise component characterization, where a plain supply lacks the measurement fidelity. They are a distinct category but increasingly appear as modules inside the same system mainframes as ordinary supply channels.

Linear, Switching, and Hybrid Topology

The single most consequential design choice is the regulation topology, because it sets the trade-off between output cleanliness, efficiency, size, and cost. Three approaches dominate: the linear regulator, the switching regulator, and the hybrid that combines them. The table below contrasts their engineering character.

Linear topology steps the AC down through a power transformer, rectifies and filters it, then passes the result through a series pass transistor controlled by an error amplifier that compares the output to a reference. The pass element drops the surplus voltage as heat, which is why efficiency is modest, typically 30 to 60 percent, and why a high-power linear supply is large and heavy. The reward is an exceptionally clean output: because there is no high-frequency chopping, ripple and noise can fall below one millivolt peak-to-peak and into the hundreds of microvolts RMS. For analog front-ends, sensor bias, RF, audio, and precision metrology that cleanliness is decisive.

Switching topology first rectifies the AC straight to high-voltage DC, then a transistor chops that DC at a high frequency, from tens of kilohertz up into the megahertz range, drives a small high-frequency transformer or inductor, and rectifies and filters the result, the same conversion principle embodied in a standalone DC-DC converter module. Delivering energy in switched bursts rather than dissipating the surplus pushes efficiency to 80 to 95 percent and shrinks the magnetics dramatically, so a switching power supply of the same power is far smaller and lighter. The penalty is switching residue: ripple at the switching frequency and broadband noise spikes, typically several to tens of millivolts peak-to-peak, which can couple into a sensitive device under test.

Hybrid (pre-regulated) topology is the common compromise in modern bench instruments. A switching pre-regulator coarsely tracks the required output a few volts above target, handling the bulk power conversion efficiently, and a linear post-regulator then performs the final, clean voltage drop. This keeps the pass transistor dissipation small (since it drops only a few volts) while preserving linear-grade output noise. The result is a supply that is reasonably efficient, runs cooler than a pure linear unit, and still meets low-ripple specifications. Many laboratory supplies advertised as low-noise use exactly this arrangement internally.

Two refinements appear at the high-power end. Autoranging lets the output follow a constant-power hyperbola rather than a fixed rectangular voltage-current box, so a single instrument can deliver high voltage at low current or high current at low voltage within the same power rating, covering the work of several fixed-range supplies. Regenerative (four-quadrant) designs add the ability to sink current and return that energy to the mains instead of burning it in a resistor bank, which is essential for efficient battery cycling and electronic-load duty. Both refinements are built on switching converters, since only switching topology can move power bidirectionally with acceptable efficiency.

CV, CC, and the Operating Envelope

Every programmable DC supply runs two control loops at the same time: a constant-voltage loop that tries to hold the set voltage, and a constant-current loop that tries to hold the set current. Only one can win at any instant, and which one wins is decided by the load, not by a switch the operator flips. Understanding this crossover is the key to using the instrument safely, because the current setting is also the device-under-test protection limit.

In constant-voltage (CV) mode the supply holds the output at the programmed voltage and lets the load draw whatever current it needs, up to the current limit. As long as the load demands less than the current limit, the voltage loop is in control and the output voltage is flat. This is the normal mode for powering a working circuit at a fixed rail. In constant-current (CC) mode the supply instead holds the output current at the programmed value and lets the voltage fall to whatever the load requires to pass exactly that current. CC mode is used to charge batteries at a controlled rate, drive LEDs, and protect a board during the first power-up.

The transition between them is called automatic crossover. Start in CV: if the load resistance falls or a fault increases the demanded current past the limit setting, the current loop takes over, the supply crosses into CC, and the voltage drops to keep current at the set ceiling. When the load eases and the demanded current falls back below the limit, the supply reverts to CV automatically. Most instruments show a CV or CC annunciator so the operator can see which loop is active; seeing CC unexpectedly during bring-up is the classic early warning of a short on the board.

Plotting the set points on a voltage-current plane gives the operating envelope, the area of output combinations the supply can deliver. A conventional supply has a rectangular envelope: any point up to the maximum voltage and maximum current, but not both at once if that would exceed the power rating. The table below maps the four user-selectable regulation behaviors offered by advanced supplies, which extend the simple CV/CC pair.

Two further behaviors extend the envelope on capable instruments. Constant power regulation holds output power, tracing the hyperbolic boundary that autoranging supplies exploit to cover a wide voltage-current area within one power rating. Four-quadrant operation, found on SMUs and regenerative supplies, lets the instrument both source and sink current in either polarity, which is required to charge and discharge a battery or to emulate a bidirectional bus. A two-quadrant battery module, for example, can source charge current and sink discharge current on the same channel. When sinking, a regenerative unit returns the energy to the mains rather than dissipating it as heat.

Key Specification Parameters

A datasheet may list thirty parameters, but a manageable set drives the buying decision. They group into three families, following the standard manufacturer convention: accuracy and resolution, stability (line and load regulation), and AC characteristics (ripple, noise, and transient response). Each is explained below, with representative figures from published bench-supply datasheets to anchor the magnitudes.

Programming (setting) accuracy and readback accuracy are both written as plus-or-minus (percent of setting + fixed offset). Programming accuracy states how close the actual output is to the commanded value; readback accuracy states how close the supply's own internal measurement is to the true output. As an example, the Rigol DP832 specifies voltage programming accuracy of 0.05% + 20 mV and voltage readback accuracy of 0.05% + 10 mV on its 30 V channels. The Keysight E36300 series quotes programming and readback accuracy under 0.04%. Always evaluate the full formula at your operating point: the percent term dominates near full scale while the fixed offset dominates near zero, so a headline percentage alone is misleading.

Setting (programming) resolution is the smallest step you can command, and readback resolution is the smallest change the meter can display. Resolution is not accuracy: a supply can resolve to 1 mV yet only be accurate to plus-or-minus 20 mV. Resolution matters when you need to nudge a rail in fine steps, for instance margining a logic supply, while accuracy matters when the absolute value must be correct. Bench supplies commonly resolve voltage to the millivolt and current to the milliampere; precision and source-measure modules push resolution far finer.

Line regulation and load regulation are the stability family. Line regulation is the output change when the AC input voltage and frequency vary over the allowed range; load regulation is the output change when the load current changes, conventionally from no load to full load. Both are written as plus-or-minus (percent of setting + offset) and tighter is better. The Keysight E36300 series, for example, specifies line and load regulation at 0.01%, and the Rigol DP832 specifies load and line regulation under 0.01% + 2 mV for voltage. Poor load regulation shows up as a rail that sags every time the device under test draws a current pulse.

Ripple and noise, also called PARD (periodic and random deviation), is the residual AC on the DC output, specified separately as RMS and peak-to-peak voltage over a stated bandwidth, conventionally 20 Hz to 20 MHz. It is the parameter that most distinguishes a low-noise linear or hybrid supply from a basic switching unit. The Keysight E36300 series specifies normal-mode noise under 350 uVrms; the Rigol DP832 specifies under 350 uVrms and 2 mVpp over 20 Hz to 20 MHz. For RF, audio, sensor, and analog front-end work, low ripple is essential; for digital and motor loads it is far less critical.

Transient response time is how fast the output recovers to within a stated band after a sudden load step. It matters whenever the load current changes abruptly, as a wireless device does when it switches from idle to transmit. The Keysight E36300 series quotes a fast transient response time under 50 us, and purpose-built fast-transient supplies (such as the Keithley Series 2300 family designed for battery-powered devices) hold the droop within roughly 100 mV under very large load steps. A slow supply lets the rail collapse during the current spike, which can reset or mis-test the device.

Several protection and feature parameters round out a selection. Overvoltage (OVP), overcurrent (OCP), and overtemperature protection set the trip thresholds that safeguard the load and the instrument. Remote sense (4-wire) measures voltage at the load to cancel lead drop, important at high current or long leads. Series and parallel capability extends voltage or current beyond a single channel. Output isolation rating states how far the output may float above chassis ground. And the control interface, covered in the next chapter, determines how the supply integrates into an automated system.

Selection Decision Factors

To turn the preceding chapters into a specific model, work down the ordered list below. Most selection errors come not from a single wrong number but from deciding a later detail before an earlier one is fixed, for example choosing a brand before defining the load envelope. These steps double as an RFQ template.

1. Output envelope: Define maximum voltage, maximum current, and total power, and confirm whether you need them simultaneously. A rectangular envelope is cheapest; an autoranging (constant-power) envelope covers a wider voltage-current area in one unit. Size the envelope so the normal operating point sits comfortably inside it with margin for transients.
2. Channels and isolation: Decide single versus multi-output, and whether outputs must be independent and isolated so they can be floated, stacked in series, or paralleled. Confirm the output-to-ground isolation rating against the highest rail you intend to float.
3. Topology and noise budget: Choose linear or hybrid for low-noise analog, RF, sensor, and metrology work; choose switching where efficiency, power density, and weight dominate and the load tolerates ripple. Match the ripple and noise spec (RMS and peak-to-peak, over 20 Hz to 20 MHz) to your device's sensitivity.
4. Accuracy and resolution class: Distinguish basic bench work (programming accuracy near 0.05% is fine) from precision and source-measure duty (sub-0.04% and fine resolution). Read the full plus-or-minus (percent + offset) formula at your operating point, not just the headline percentage.
5. Regulation and transient response: Specify line and load regulation (0.01% class for quality bench supplies) and transient response time if the load is pulsed. Fast-transient or wireless-device loads need recovery in the tens of microseconds.
6. Control interface and software: Confirm the remote bus (USB, LAN/LXI, GPIB, or RS-232) and that the supply speaks SCPI and ships an IVI driver, so it drops into existing test software. LAN/LXI suits rack automation; USB suits a single bench. Verify on-board sequencing, data logging, and trigger features if you need them.
7. Protection and safety compliance: Require OVP, OCP, and overtemperature protection with settable thresholds, and confirm the unit is certified to the relevant safety standard (IEC/EN 61010-1 for measurement and lab equipment, or IEC/EN 62368-1 for ICT and AV equipment) for your market.
8. Remote sense and connection: Specify 4-wire remote sense for high-current or long-lead loads, and confirm the terminal type, maximum sense-lead drop, and wire gauge needed for your current. Low-voltage high-current rails (battery emulation, 1.2 V logic) almost always need remote sense.
9. Total cost of ownership: Weigh purchase price against efficiency (heat and electricity at high power), calibration interval and cost, channel density (one modular mainframe versus several boxes), and the value of regenerative energy recovery for battery and load-test duty.

One dimension is easy to overlook at purchase but governs the instrument's life: serviceability and ecosystem support. Check calibration availability and interval, firmware updatability, driver and SCPI command stability across firmware revisions, spare-module availability for modular systems, and local repair and calibration service. Keysight, Rohde & Schwarz, Tektronix and its Keithley line, B&K Precision, Chroma, TDK-Lambda, and Rigol all maintain documented SCPI command sets, IVI drivers, and regional service, which matters far more over a five to ten year service life than a small difference in headline specification.