Introduction

The Gigabyte UD uses the same platform as the P-GM series, but with improvements to iron out all issues. MEIC, the OEM behind these units, didn't put on a good start in the desktop PSU market, but Gigabyte seems to have faith in the OEM. The UD line consists of three models with 750 W, 850 W, and 1000 W maximum power output. I reviewed the updated flagship model featuring a 600 W PCIe 5.0 connector, so it is time to look at the lower capacity units that are not ATX v3.0 compatible and lack the new PCIe connector, which means I won't have to run the extra-demanding transient response tests with 120%, 160%, 180%, and 200% of the PSU's maximum capacity.

Speaking of the new transient response requirements, or power excursions as the ATX specification calls them, I remember having a conversation with Intel about it, and I asked why they didn't push GPU manufacturers to implement stricter DC-DC circuits on their designs to restrict power spikes instead of having PSU manufacturers factor in transient loads of up 200% on their designs! With the release of the NVIDIA RTX 3090 Ti, power spikes are nowhere close to 200% of the maximum-rated capacity for very short periods of time. Why am I mentioning all of this? Because as I see it, the new GPUs use better power circuits to filter out most power spikes, so these crazy PSU transient response requirements won't be necessary—the ATX v3.0 requirements will be overkill, leading to higher PSU prices for no reason. Nobody can tell for sure, and we will have to wait until the new GPUs are available to conduct the simple test of taking several PSUs that fail to meet the ATX v3.0 transient response tests and installing them in real systems to figure out if they can support the GPUs and there are no issues.

To measure GPU power spikes more efficiently, I am in the design phase of a special Power Measurements Device (PMD) for high polling rates without using Hall effect sensors but shunts to remove unwanted EMI noise from the equation. If everything goes as planned, which is highly unlikely given the current worldwide situation, the PMD will be ready this summer, and I will put together an article to present it.

Specifications

Photos

The box is small, and at the face is a photo of the compact PSU with the modular panel exposed. More information is on the backside.


Protection inside the box is good, with packing foam surrounding the product.


The bundle only includes the necessary modular cables and AC power cord, set of fixing bolts, and user manual.


Nothing out of the ordinary about the exterior design. The fan has a punched-out grille with honeycomb-style perforations. I would use a less restrictive fan grille to increase airflow.


The power specifications label covers the bottom almost entirely.


The modular board has ten sockets.


At only 140 mm long, this PSU is compact.


Here are some more photos of the PSU from various angles.

Cables and Connectors

A longer ATX than EPS cable is an odd choice. Usually, it is the other way around. I want both EPS cables to be 650 mm long. The number of available connectors is satisfactory, but the PSU lacks 12+4 pin PCIe, which will be essential soon.


The number of peripheral connectors is adequate, but there is no need for a floppy connector. Moreover, the distance between peripheral connectors is only 115 mm.


The PSU came with a US power cord.

Component Analysis

Before reading this page, we strongly suggest looking at this article, which will help you better understand the insides of a PSU.



Compared to the original platform in the PG-M units, this one received a few upgrades for tighter protection features and larger heatsinks to keep the FETs cooler. Despite the PCB's small dimensions, the design is nice and clean, allowing for plenty of airflow. Capacitors nowadays are like diamonds, hard to find and very expensive, so MEIC had to use Lelon caps on the secondary side, which have good specifications. If these specifications are not too far removed from real-life, the caps won't be the problem. However, the low-quality fan claiming to have a hydraulic bearing only has a rifle bearing and is the same as in the P-GM 750 W and 850 W units.


The transient filter is complete. There is also an MOV for protection against power surges.


I found a discharge IC in the transient filter; it provides a small efficiency boost.


An NTC thermistor lowers inrush currents. A bypass relay supports it.


The two bridge rectifiers can handle up to 30 A combined.


The APFC converter uses two NCE Power FETs and a single JF boost diode. The bulk cap is by Chemi-Con, so it is of high quality. I would like it to have a 420 V rating, though.


The APFC controller is a Champion CM6500UNX.


The two NCE Power primary switching FETs are installed in a half-bridge topology.


The resonant controller is a Champion CM6901T6X.


The main transformer is next to the parts for the LLC resonant controller.


The FETs regulating the +12 V rail are hidden by the two black heatsinks attached to the silver heatsinks through screws. To remove them, I would have to apply excessive heat to the PCB, close to the +12V FETs, which would probably damage them, so I chose not to. I might need this PSU for more measurements in the future.


The electrolytic caps on the secondary side are by Lelon. There are also eight polymer caps from the same brand.


Two DC-DC converters generate the minor rails.


The standby PWM controller is a PR8109T IC, and the 5VSB secondary rectifier is an SP10U45L SBR.


The supervisor controller is a Weltrend WT7502R.


Four polymer and four electrolytic caps on the modular PCB further reduce ripple.


Soldering quality is great!


This is the same fan as in the lower-capacity PG-M units. It is supposed to have a high-quality hydraulic bearing, but when I broke it down in the P750GM review, I found a plain rifle bearing instead. Still, a rifle bearing is far better than a sleeve one.

Test Setup