Linear and Switching DC Power Supplies for Low Noise Lab Work
Introduction: Linear and switching DC power supplies differ in ripple, efficiency, and heat, so the right choice depends on the measurement task rather than on a universal rule.
On a bench that is chasing microamps or millivolts, the power supply often decides how small a signal can still be trusted. A regulator with switching edges at a few hundred kilohertz puts that energy on every trace, through the cables and the ground return, and a leakage current or a bridge offset can disappear into it. Linear supplies are usually the first choice for that kind of measurement, yet they are not automatically the right answer, because they burn power and run warm. The three trade-off lines below — ripple, efficiency, and heat — are what separate a good choice from a convenient one.
Why Linear Regulation Produces Lower Ripple in Low Noise Laboratory Work
A linear supply regulates by putting a pass transistor in series with the load and driving it as a controlled resistance. The feedback loop compares a divided sample of the output against a stable reference and adjusts the pass device until the difference is small. The voltage dropped across that transistor, multiplied by the load current, turns into heat, which is the price of the architecture. There is no switching node in the power path, so the output spectrum carries no discrete tone tied to a conversion frequency. What remains is reference noise, error-amplifier noise, a small amount of mains-related ripple that the loop cannot fully reject, and slow thermal drift. Those contributions are low in frequency and small in amplitude next to the switching residue of a typical converter, which is why a linear bench supply is the reference point in low-noise laboratory work.
1. Linear Pass Elements Avoid High Frequency Switching Ripple
A switching converter first turns its input into a high-frequency square wave, then filters that waveform back into DC through an inductor and capacitors. Whatever survives the filter reaches the load as ripple at the switching frequency, at its harmonics, and sometimes as jitter that spreads energy across a band, along with common-mode currents that travel on cables and grounds. A linear pass element never creates those edges. It regulates continuously, so there is no sharp dV/dt step to couple into the load, no reverse-recovery spike from a rectifier, and no switching magnetic field sitting next to a sensitive probe. The residual is smaller and mostly slow, which is exactly what a low-level DC measurement can tolerate.
2. Continuous Regulation Keeps Output Noise Closer to the Set Point
Continuous regulation also shapes how tightly the output tracks what was dialled in. A linear loop with useful bandwidth corrects on a microsecond timescale, so load changes that are slow relative to that loop are absorbed before they appear as an error at the terminals. Line and load regulation in the tenths of a percent, plus a few millivolts of offset, describe that behaviour. Benchtop linear instruments carry the same idea into their interface: the MATRIX MPS-1000 series, a single-channel linear family covering 0 to 150 V, 0 to 10 A, and 36 W to 360 W, specifies 0.1 mV voltage resolution and 1 µA current resolution, offers SENSE remote compensation to cancel lead drop at the load, and shows voltage and current waveforms on a 4.3-inch display. That resolution defines how finely the set point can be placed when the circuit under test draws microamps.
How Switching Supplies Trade Noise for Efficiency and Size
The switching approach gives up that quietness in exchange for efficiency and size. A switching supply chops its input into a high-frequency waveform, passes it through a transformer or inductor, then rectifies and filters the result. Because the active device spends most of its time fully on or fully off, very little energy is burned as heat, and efficiency commonly lands between 80% and 95%. A linear supply cannot beat the ratio of output voltage to raw input voltage; producing 5 V from a 24 V rail means most of the power becomes heat in the pass transistor. That difference is what lets a compact switcher deliver high current, accept a wide input range, and run from a modest cooling path. The trade is that ripple and common-mode noise become design problems rather than non-issues. Their size depends on switching frequency, filter design, layout, grounding, and shielding, and a well-engineered switching supply can be quiet enough for a great deal of laboratory work. Extra LC filtering, low-ESR capacitors, post-regulation, and spread-spectrum modulation all reduce what reaches the load. Heat deserves equal attention on the other side: a linear supply that dumps tens of watts inside a closed rack raises the ambient temperature around the very circuit being measured, and the drift that follows can be larger than the ripple it was meant to avoid.
Matching the Topology to the Measurement Task in a Laboratory
The decisive question is not which topology is better but what the bench actually needs to resolve. When the smallest signal of interest lives in the DC to few-kilohertz band — leakage current in an off transistor, offset drift in a sensor bridge, a microvolt thermocouple output — supply noise sits directly in the measurement band, and a linear supply is usually the safer choice. When the device under test has its own post-regulator, or the measurement band sits far above the switching frequency, switching ripple can be filtered, averaged, or simply ignored. Transient response is the second axis. When a load steps from light to heavy, a linear loop typically corrects within microseconds and settles smoothly, because the control element is continuous and the output capacitor is not fighting a large series inductor. Switching supplies usually carry an output LC filter, which introduces a resonant pole; the compensation network keeps the loop stable, and settling can be slower or show ringing when the step is large. Recovery time and the size of the deviation are the numbers that matter, and they need to be small compared with the event being measured. A well-compensated switching supply can meet demanding dynamic specifications, so the comparison is about the budget for the task, not about the label on the instrument. In practice, the decision often comes down to the thermal budget and the signal budget at the same time. A rack full of linear supplies needs real airflow, and a bench that cannot provide it will drift no matter how clean the DC output is. A switching supply that frees up that heat and space can be the better engineering decision even for a precision project, as long as the residual noise is checked at the load terminals instead of assumed. Many laboratories end up with both: a linear supply for the sensitive first measurements, and a switching supply for the loads that would make the linear unit sweat.
Conclusion
Ripple, efficiency, and heat rarely pull in the same direction, and that is the whole story of linear versus switching supplies on a laboratory bench. Linear regulation earns its place wherever the measurement lives close to the noise floor, because continuous control keeps switching-frequency content out of the output and lets the set point be defined in tenths of a millivolt and in microamps. Switching conversion earns its place wherever current, input range, size, and thermal headroom matter more than a few millivolts of ripple that can be filtered or averaged away. Reading the smallest signal, the measurement bandwidth, and the airflow in the rack tells you which side of that line a given task falls on. Engineers who want a concrete reference for the linear side can review the MPS-1000 series documentation, which describes the resolution, remote sensing, and waveform display mentioned above.
FAQ
Q：Why do linear DC power supplies produce lower ripple than switching supplies?
A：Because the pass element regulates continuously instead of chopping, there is no switching frequency and no harmonic series in the output. The remaining noise comes from the reference, the error amplifier, and low-frequency mains residue, and it is both smaller in amplitude and lower in frequency than typical switching ripple. That matters most when the measurement sits in the same low-frequency band as the supply noise.
Q：When is a switching DC power supply still useful in a laboratory?
A：It is useful whenever current demand, input voltage range, rack space, or cooling limit what a linear unit can do. A compact switcher can power high-current loads all day without dumping heat into the rack, and it works well when the device under test has its own regulation or the measurement band is far from the switching frequency. With filtering and sensible layout, the residual noise is often acceptable.
Q：How does transient response differ between linear and switching supplies?
A：A linear loop generally reacts within microseconds and settles smoothly after a load step, since the control element is continuous and the output capacitor is not working against a large series inductor. A switching supply carries an output LC filter that adds a resonant pole, so settling can be slower and may ring when the step is large. Compare recovery time and deviation against the event you are measuring.
Sources / References
Understanding Linear and Switching Power Supplies - Texas Instruments
Texas Instruments Measuring Power Supply Transient Response
Uncertainty Analysis Basics - Keysight
Comments
Post a Comment