In order to measure the voltage drop of a PSU under load, all components and peripherals need to be connected.
If everything is connected, how can you measure the voltage drop using a multimeter? (e.g, for the last HDD in a daisy-chain connection.)
Do you simply leave the last HDD in the SATA chain unplugged, and take a measurement at the end of the connector itself? Would this be considered an accurate reading, even though there is one less component (an HDD) drawing from the PSU?
You need to install a “breakout plug” (my name) at the last device that allows connection of the voltmeter to the power conductor to the device, such that the device is a drain on the system while the voltage applied to it is being measured (other side of the voltmeter at ground potential).
If there’s a free hard drive plug, of course, it’s easy enough to measure there. Failing that, leaving one drive disconnected would be “close enough” IMO.
Other options would include:
If all you’re interested in is +12 and +5, the latter might be the easiest, as those pads are usually pretty easy to access on a drive.
No splitter for SATA power on hand, but I did do this:
Which read +12.1 VDC on the 12V contact of the SATA power cable at the end of the daisy-chain (where the last HDD would have been installed).
I confirmed this again with a different multimeter, just in case one was giving an inaccurate reading.
All drives were undergoing long sefltests, and I stressed all CPU cores with stress -c 8, so as to put as much strain on the PSU as possible.
Yet the last HDD on the SATA chain (when installed) would consistently disconnect, spin back up again, disconnect again, and repeat, over and over. It didn’t matter which drive it was. The consistent culprit was “last drive on the SATA chain”.
EDIT: It was two-fold. If all other HDDs were removed, then the drive on the “bad SATA plug” at the end of the chain would run without any problems. If all drives were installed, then the last plug on the chain would always cause this problem on whichever drive happened to be connected to it. So both conditions had to be met: all six HDDs installed + last plug on the SATA would cause whichever drive to keep failing.
I eventually bit the bullet and just ordered a new PSU, which immediately resolved the issue.
I’m not sure if it was the daisy-chain cable from the former PSU, or the PSU itself which was the problem.
For what it’s worth, here is the problematic PSU. (Yes, I know. It’s a lesser known brand, and I’m not sure why I originally purchased it five years ago.)
The lesson of the day: Don’t try to save a penny with your PSU purchase. Stick to reputable brands and models, even if they cost slightly more. I went with a modular Corsair. It wasn’t worth trying to further diagnose the issue, considering that a failing PSU could put your (new) data at risk.
Fond memories of my travel toolkit before hotels had RJ45’s “everywhere” into which we could plug our modem cables … I held back on mentioning such measures for fear of a bad outcome somewhere…
Back to @Winnie’s experience - didn’t actually test with 6 drives hooked up apparently - there’s also the issue of speed of response and resolution of the meter to be considered.
Anyway, Winnie, glad you have it fixed and shared your situation for the benefit of the readers here.
Honestly, I’m of the mind that it’s better to go with the simpler option (replacing the PSU), than spending time trying to figure out what is exactly wrong with it, and possibly try to “fix” it.
I believe jgreco (from the old forums) emphasized that a good PSU is just as important for your NAS server as are the drives.
I’ll stick to highly-rated models from now on.
How could anyone question a brand whose slogan is “Stable Supply, Less Spend”. 
It just speaks “professional” and “reputable”.
Reminds me of DFI LanParty motherboards: “UV on, Tweak fun!!”
You may very well have had an issue unrelated to the actual PSU capacity and totally related to how they brought that power to the drive or the individual voltage busses / rails.
Power to the drive - how thick are the wires, how beefy the connectors. Voltage drop is not your friend and tends to mess with anything plugged in at the far end first. The skinnier the wires and the higher the amp draw, the worse the problem. I doubt the 5VDC bus is a major culprit as it only powers the electronics on HDDs but the 5VDC bus will be more sensitive to voltage drop in absolute terms than the 12VDC one. A dirty connector may interfere also.
I dealt with a lot of voltage drop issues w/12VDC systems, the most expensive and time-consuming was replacing about 80 ft of #2 with 0000 wire. That took a long time and cost $$$ but finally met the 5% rule for ABYC. I generally over-dimension all my wiring now, it’s cheap insurance.
The other issue is the busses themselves. The PSU may not have a monster single 12VDC rail / buss to drive the motors on each HDD - it may have multiple segmented busses designed to separate / isolate issues with the GPU card from the HDDs, for example.
Many OEMs will ship cable dongles designed to stay in safe limits (i.e. 3 SATA HDDs per SATA plug in modular models) but users are free to attach extra outlets and then the fun begins.
