We will be using an ElmorLabs PMD2 that monitors and measures voltage, current, and power flowing through the CPU, GPU and motherboard power cables.



Our testing methodology will now be using OCBase’s OCCT Enterprise Edition software for power loading and monitoring. Tweaking the OCCT Stability Test, we can set different power loads for the PSU easier rather than playing around with various software to achieve set power loads. All testing is conducted using the 230V residential power standard.



Methodology
The ElmorLabs PMD2 features several monitoring points which includes two 8-pin EPS connectors (EPS1 and EPS2), three +12V 8-pin PCIE (PCIE1, PCIE2, and PCI3), one 12V-2x6 connector (HPWR) and readings on the 24-pin which include +12V, +5, and +3.3. For the +12V output on the peripheral/molex cable, we will be using a multimeter.

Power load will be at 10% increments using the OCCT Stability Test running for 5 minutes – as much as our current testing setup allows which is currently 750W~800W with the following testing configuration below:



Results



In our test configuration, the RTX 3070 is powered via the 12V-2x6 (HPWR) cable, the RTX 4060Ti is powered by the PCIE3 cable, the PCIE1 and PCIE2 are unpopulated but is remained plugged in to the PMD2 device and will continue reading voltages. The EPS1 and EPS2 connectors are shared in a single EPS cable with pigtail connectors, a similar cable setup with the SAMA B850 power supply unit.

With our test configuration, system load peaked at 791W based on the PMD2 readings – whereas the SAMA B850 peaked at 800W with the same test and configuration. The MONTECH BETA 2 has slightly looser voltage regulation compared to the SAMA unit.



OCCT Stability Test
Overall +12V regulation remains well within ATX tolerance across all load levels. The PCIe and native 12V-2x6 connectors show tight control, generally staying within roughly ±1.1% at maximum load, with minimal fluctuation under transient conditions. This indicates stable GPU power delivery and consistent regulation on the primary +12V rail. Graphs below are readings during the 70% workload test and maximum possible workload test with our given configuration which came around ~93% load.



The EPS1 and EPS2 connectors show slightly greater droop under higher loads, peaking at just over -1.5%. This is expected since both connectors share a single pigtail cable, resulting in more noticeable voltage sag compared to the independently routed PCIe lines. As reflected in the graphs, the PCIe rails remain flatter while the EPS rails show deeper dips, though all readings remain well within ATX specifications.



The +3.3V rail consistently operates slightly above its nominal value and shows comparatively looser regulation across the load range. In contrast, the +5V rail remains tightly controlled, maintaining deviation within 1% throughout testing. Comparing the BETA 2 with the recently reviewed SAMA B850, the MONTECH BETA 2 850W boasts better quality 105°C Japanese-brand capacitors while the SAMA B850, with an all-Taiwanese capacitor construction, appear to be superior in terms of voltage regulation.

Acoustics
The MONTECH BETA 2 850W’s cooling fan spins at all times. At low loads, our sound meter picks up noise levels around 47~48 dBA near the cooling fan vent. (average room noise level with PSU off is ~45 dBA).



During maximum load testing, where total system draw reached approximately 791W, we recorded noise levels of around 62 dBA at the front of the open-frame test system near the GPUs and CPU cooler (see image from top of the page). Readings taken at the rear of the PSU, close to the power switch, were similar. As we are unable to fully isolate individual component noise, measurements were taken at key points around the system — near the GPUs, CPU cooler, and PSU.

Based on our observations, the PSU is not the dominant noise source under load. The loudest contributor remains to be the two triple-fan graphics cards (RTX 3070 and RTX 4060Ti) operating at maximum fan speeds. No coil whine was detected from the BETA 2 850W throughout testing.