DC-DC Converter Module (Isolated & Non-Isolated, OEM)

Isolated vs. Non-Isolated: When Topology Choice Is a Safety Requirement

The decision between an isolated and non-isolated DC-DC converter is not a cost decision — it is a circuit architecture decision driven by safety, noise, and system grounding requirements. Getting this wrong at the design stage is expensive to correct during production.

Non-isolated converters (buck, boost, buck-boost) share a common ground between input and output. They are appropriate when:

• Input and output ground rails are intentionally connected (single-supply systems, on-board regulation)
• Output voltage is lower than or close to input voltage (buck), or input is lower than output (boost)
• No risk of high-voltage transients appearing on the input rail — for example, a 24V industrial PSU feeding an on-board 3.3V rail for a microcontroller

Synchronous buck converters from Chinese manufacturers operating at 300–600kHz switching frequency achieve 90–95% efficiency at moderate loads. The efficiency advantage comes from replacing the Schottky rectifier diode with a second synchronous MOSFET, eliminating diode forward voltage drop (~0.4–0.6V at full load). At a 5A output, this recovery of 2–3W as heat is significant in a compact module.

Isolated converters introduce a galvanic barrier between input and output. Isolation is required when:

• The input rail is mains-derived or carries a hazardous voltage (>60V DC or >42.4V AC peak per IEC 62368-1 definitions)
• Output ground must float relative to chassis or input ground — typical in industrial sensors measuring signals referenced to earth while the power system has a separate earth
• Common-mode noise injection from the input rail would corrupt a sensitive analog measurement circuit (strain gauges, precision ADCs, pH sensors)
• Creepage and clearance distances mandated by IEC 62368-1 or IEC 61010-1 require physical separation

Topology mapping:

• Flyback — Most common isolated topology for <100W output power. Single transformer winding (primary + secondary). Cost-effective but higher output ripple than forward converters. Standard in 1W–30W industrial sensor power supplies.
• Forward converter — Single-ended forward topology; lower ripple than flyback at equivalent power levels. More common in 20–100W modules. Requires a transformer reset mechanism (RCD clamp or two-switch design).
• Push-pull / full-bridge — Used above 100W where flyback magnetizing current becomes impractical. Less common in modular catalog products; typically appears in custom-designed power supplies.

Chinese manufacturers commonly label flyback modules as “isolated” without specifying the isolation test voltage or the creepage/clearance distances. Request the isolation voltage test specification (typically 1,000V DC hipot for 1 second, or 1,500V AC for 1 minute) and the working voltage rating. A module rated for 1,000V DC isolation used in a 230V AC mains-derived application provides inadequate margin — IEC 62368-1 requires reinforced insulation of 2× peak working voltage plus 1,000V minimum.

For industrial IoT applications where isolated power feeding RS-485 or analog sensors is a system architecture requirement, we evaluate modules against the target IEC standard during our inspection process before bulk order release.

Efficiency at Real Load Points: What “92% Efficiency” Actually Means

Module datasheets report peak efficiency — typically measured at 50–75% of rated output current at a specific input voltage. This number tells you the best-case conversion loss but is irrelevant for sizing thermal management in most real applications.

Load profile matters more than peak efficiency. Consider a 5A/5V synchronous buck module with a published 93% peak efficiency:

Load current	Typical efficiency	Power dissipated
0.25A (5% load)	72–80%	0.31–0.43W
1.25A (25% load)	88–91%	0.21–0.34W
2.5A (50% load)	91–93%	0.27–0.33W
5A (100% load)	89–91%	0.69–0.83W

Light-load efficiency collapses in non-synchronous designs because gate drive and quiescent current losses dominate at low output power. A module consuming 15mA quiescent at 24V input dissipates 360mW regardless of output load — at 100mA output load on a 5V rail (500mW output), this quiescent loss alone represents 42% overhead.

For battery-powered IoT modules operating in sleep cycles, always request the light-load efficiency curve (typically 1% and 10% of rated load) from the manufacturer. Reputable Chinese suppliers (MORNSUN, CINCON) publish full efficiency-vs-load curves in their datasheets. If a supplier provides only a single efficiency figure with no load specification, treat that datasheet as incomplete.

Where to apply efficiency data:

Thermal dissipation calculation: P_loss = P_output × (1 − η) / η. A 20W output module at 90% efficiency dissipates 2.2W as heat. At 85% efficiency the same output dissipates 3.5W — a 60% increase in thermal load that can push a borderline natural-convection design into forced-air territory.

For OEM applications where the module will be certified as part of a larger system, efficiency data feeds directly into the end-product energy label calculations (DOE Level VI for external power supplies, ErP Lot 6 for EU). Our sourcing team can provide test reports showing full load-efficiency curves from the factory’s production test, not just the datasheet claim.

Thermal Derating and PCB Layout: The Margin That Disappears in Enclosures

DC-DC converter modules are rated at 25°C ambient with natural convection. In a sealed enclosure with other heat-generating components, ambient temperature at the module may be 50–70°C — halving or eliminating the available thermal headroom before derating kicks in.

Thermal derating curves specify how output current must be reduced as ambient temperature rises. A 3A-rated buck module may derate linearly from 3A at 40°C to 1.5A at 85°C. If the module operates at 2.8A continuous inside an enclosure where internal ambient reaches 65°C, it is operating near or beyond its derated rating — which causes premature electrolytic capacitor aging and eventual inductor saturation under transient conditions.

Request the derating curve, not just the temperature range. “Operating temperature: -40°C to +85°C” means the module will function — not that it can deliver rated current across that range. The distinction matters for industrial deployments.

PCB layout requirements for non-isolated modules:

The controller IC and power inductor are the primary heat sources in a synchronous buck converter. Module manufacturers specify a minimum copper pour area on the host PCB — typically 10–25 cm² of 1oz copper connected to the exposed thermal pad or mounting pad. Insufficient copper area reduces heat spreading and creates a hotspot that degrades the inductor’s DC resistance over time.

Critical layout rules for integrating DC-DC modules onto a host PCB:

• Place input and output bulk capacitors as close as possible to the module pins; long traces between capacitor and module create parasitic inductance that increases voltage overshoot during load transients
• Separate the high-current switching loop (MOSFET, inductor, output cap) from sensitive analog measurement traces — 5mm minimum separation, with a ground plane pour between them where possible
• Do not route signal traces through or under the switching node area; the switching node on a buck converter swings rail-to-rail at the switching frequency and couples capacitively to adjacent traces

Forced airflow vs. natural convection. A 10W module (approx. 1.1W dissipation at 90% efficiency) in natural convection requires approximately 25 cm² of external heatsink surface or PCB copper pour to maintain <20°C temperature rise above ambient. Adding 1 m/s airflow across the module surface reduces the required heatsink area by approximately 60%. For dense assemblies with multiple high-power modules, budget airflow in the mechanical design — retrofitting forced airflow to a sealed enclosure is expensive.

For power electronics products destined for industrial or outdoor environments, our inspection service includes thermal imaging of modules under rated load inside a representative enclosure — not just at bench level with open airflow.

Chinese Supplier Landscape: Verified vs. Re-labeled

The DC-DC module market in China splits clearly into two tiers: manufacturers with established certification programs and engineering support, and trading companies re-labeling unverified modules with unfamiliar brand names.

Tier 1 domestic manufacturers:

MORNSUN (金升阳) is the largest dedicated DC-DC module manufacturer in China by volume, headquartered in Guangzhou. Their catalog spans 0.25W to 300W isolated modules with genuine UL/CE certification on each model number. Efficiency specs are conservative and reproducible. Their OCPT (overcurrent protection threshold) is well-characterized in the datasheet. Lead time from factory is 15–25 days for standard catalog; custom label printing available at 2,000-unit MOQ. Pricing is 15–30% above unbranded alternatives.

CINCON Electronics (台湾, China manufacturing) — Established OEM supplier for DC-DC modules in the 1W–75W range. Better documentation than most mainland manufacturers; CE/UL certifications are third-party tested, not self-declared. Commonly specified by European industrial equipment integrators. Lead time 20–35 days.

Tier 2 alternatives (Murata / RECOM functional equivalents): Several Guangdong manufacturers produce pin-compatible alternatives to RECOM’s R-78 series (non-isolated, 1A SIP) and to Murata’s MEJ and MGJ series (isolated, 1–2W). These alternatives are 40–60% cheaper than the original. The functional parity varies: switching frequency, input capacitance, and thermal performance can differ significantly from the part being replaced.

Verification checklist before ordering from an unknown supplier:

1. Request the CB test report (IEC 62368-1 or IEC 60950-1) — not the CE Declaration of Conformity alone. The DoC is self-declared; the CB report requires a Notified Body. Verify the test report issuer is an IECEE-recognized CB Testing Laboratory.
2. Request the UL file number and verify it at ul.com/database. UL certification on a Chinese module that does not appear in the UL database is fabricated documentation.
3. Ask for the isolation voltage production test record — confirming that each unit is hipot-tested, not just the type-approval sample.
4. Request five samples before committing to bulk. Measure output voltage accuracy, switching frequency (with an oscilloscope at the switching node), and output ripple at 50% and 100% load. Compare against datasheet values.

Our audit service covers factory visits to MORNSUN, CINCON, and shortlisted domestic alternatives — we verify production test coverage, PCB traceability, and component BOM compliance before you commit to tooling or bulk purchase.