SK Hynix SSD quick review

Lit — Tue, 30 May 2023 03:01:31 +0000

SK Hynix PE6110 1.92TB

Recently I bought an m.2 SSD made by SK Hynix, the PE6110 in their enterprise entry-level lineup. The PE6110 is offered in 3 capacities, 960GB, 1920GB, and 3840GB. Here I have the 1920GB version. On the PCB front, we can see clearly the controller, printed with the SK Hynix logo. The NAND flash chips are soldered both on the back and on the front under the product label. On the back side, there’s also the single RAM chip, 2GB LPDDR4 3733. Under the product label, there’s also a good number of tantalum capacitors which will offer excellent power loss protection.

As the SSD is in 22110 form factor, most motherboards including mine are incompatible as 2280 is the most widely supported and popular m.2 SSD form factor. As a result, I also bought an m.2 to PCIe socket adapter that supports 22110. I also put on an aluminum heatsink on the front to help it dissipate heat. However, the capacitors for power loss protection are taller than both the controller and NAND chips, and are obstructing the proper installation of the heatsink, so it’s making poor contact with the heat-generating components.

Front: Controller and NAND flash chips under the label

Back: RAM chip and more NAND flash chips

Heatsink and PCIe adapter

First Boot - SMART Info

First time booting with the PE6110 installed, there was no trouble at all, and the initialization process of the SSD was quick and simple. After performing some benchmarks, I checked its SMART data. Sure enough, the drive was new and healthy. The temperature is higher than my other drives, but still acceptable. It idles at 10 degrees Celsius higher than my other NVMe SSD. With a power consumption of up to 7.0W during write operation, it shouldn’t be as big of a problem as for example Samsung’s PM983 which draws up to 8.7W during write. Hopefully, the janky heatsink won’t backfire on me in the future.

Benchmarks: Crystal Disk Mark

The official advertised sequential read/write speed is 3200/1900 MB/s, and random read/write at 185k/74k IOPS, as displayed on their official website: Enterprise SSD | SK hynix.

The benchmark revealed that SK Hynix’s claim on their SSD performance is accurate, with actual sequential read/write speed of 3400/1900 MB/s, it compares well with mainstream PCIe 3.0 SSD. Though the write speed might seem to be lower than consumer SSDs, it should not be seen as a disadvantage. As enterprise SSDs focus more on reliability, endurance, and long-term performance, they’re usually not set to use SLC cache, which boosts performance at the expense of depleting drive life faster.

The provided drive writes per day (DWPD) is 1, which means 1.92TB of data can be written to the SSD every day for the whole warranty period. That translate into 2.1 petabytes written (PBW) for 3-year warranty or 3.5 PBW for 5-year warranty. That is more than what most consumer SSDs offer. For instance, the Samsung 970 EVO Plus 2TB is advertised with 1.2 PBW over a warranty period of 5 years. The endurance of the PE6110 is definitely more than any consumer’s needs.

LED lamp quick fix

Lit — Sat, 20 May 2023 13:46:56 +0000

Broken LED lamp - Exploded Capacitor

Two weeks ago, the LED lamp in my living room went black shortly after giving a loud “pop” sound. Out of frustration, I tried turning the light switch on and off repeatedly, and occasionally it flashed for a brief second, before going dark again. I knew the LED arrays were intact and it was certain that the control or power supply circuitry had failed.

The LED lamp was disassembled and upon opening the plastic box that contained the PCB, I immediately spotted there was a blown capacitor. The black plastic base of the capacitor is visible and has separated from the metal housing.

To fix it, I simply desoldered the blown capacitor and replaced it with a new one. The original one was 16V 100uF, I replaced it with a 25V 100uF because I thought maybe the voltage spiked too high for whatever reason and resulted in an exploding capacitor. Though in hindsight, overvoltage was probably not the cause.

Hooking it up to mains and watching it light up again, I was pleasantly surprised by how easy the fix was. To make sure the fix indeed worked and that there weren’t more failures, I kept the lamp on for a few hours before installing it back to the ceiling. During the test, I found that the circuit produces more heat than I would have thought. The plastic box containing the circuitry was warm to borderline hot. It was measured with my multimeter that the plastic casing reached more than 40 degrees C upon running for a few hours continuously. Remember, it reached this temperature under an open environment. When it’s up against the ceiling, it’s essentially in a closed environment with no airflow so the temperature will be even higher.

Though the exact cause for the failure of the capacitor can’t be determined, I think excess heat should be the main reason that it went kaput. It’s reasonable to think that the thermal performance is unideal and it’s probably a design flaw that is hard to overcome. If this exact failure happens again, I’ll replace the capacitors with polymer capacitors that have higher temperature ratings or simply drill some holes for ventilation.

P.S. This happened more than two weeks ago, so why am I writing about it just now? I had final exams so that’s why. It also means I’m free from exams now and I think I can spend some time in the coming weeks to start completing some of the webpages.