WD Easystore and 3.3v power issue

A

alaricljs

Limp Gawd
Joined
Apr 6, 2011
Messages
347
So here's an interesting story, maybe I'll get some feedback, but mostly I expect this to be an FYI for anyone else looking at these drives. I have a Norco RPC-4224 case w/ hotswap drive cages. The backplane power is fed by Molex connections having 5v and 12v lines so I assumed there would be no issues. In early November I bought a pair of 10TB Easystores that I shucked and stuffed into my server. They work just fine and I had no reason to expect issues on future drive purchases.

I just bought another pair of Easystore 10TB drives and shucked them. Interestingly the manufacture date is 2 months before my other drives. They refuse to spin up unless they have the 3v pin isolated. All the pertinent #s between the drives are identical firmware, model #, P/N #, everything.

I was expecting that this 3.3v issue was handled by the firmware. Am I wrong or are they abusing the firmware #s by having functionally different FW? Any chance someone knows a method to pull firmware from one drive and put it on another?


Thanks!
 
So the issue is that there is some voltage signal on the 3.3V line in the SATA connector? But its not 3.3V though right? Since I don't think any PSU's output 3.3V on the SATA.

I have heard this issue with the EasyStore (just picked up a 10TB unit), but I wanted to make sure of what the problem is. There should NOT be any voltage on the 3.3V? And the tape prevents that?
 
The SATA spec was changed so that one of the 3.3v pins was reassigned to be a power disable function (aka PWDIS) to be compatible with the SAS specification. Older PSUs and backplanes that supply 3.3v to SATA connector have a problem with this. Newer ones that account for the change to the spec don't.
 
alaricljs said:
So here's an interesting story, maybe I'll get some feedback, but mostly I expect this to be an FYI for anyone else looking at these drives. I have a Norco RPC-4224 case w/ hotswap drive cages. The backplane power is fed by Molex connections having 5v and 12v lines so I assumed there would be no issues. In early November I bought a pair of 10TB Easystores that I shucked and stuffed into my server. They work just fine and I had no reason to expect issues on future drive purchases.

I just bought another pair of Easystore 10TB drives and shucked them. Interestingly the manufacture date is 2 months before my other drives. They refuse to spin up unless they have the 3v pin isolated. All the pertinent #s between the drives are identical firmware, model #, P/N #, everything.

I was expecting that this 3.3v issue was handled by the firmware. Am I wrong or are they abusing the firmware #s by having functionally different FW? Any chance someone knows a method to pull firmware from one drive and put it on another?


Thanks!
Click to expand...

Check the actual part numbers on the drives, I’d be surprised if they are the same. You need to tape the pin or get a array.
 
ryan_975 said:
The SATA spec was changed so that one of the 3.3v pins was reassigned to be a power disable function (aka PWDIS) to be compatible with the SAS specification. Older PSUs and backplanes that supply 3.3v to SATA connector have a problem with this. Newer ones that account for the change to the spec don't.
Click to expand...

So if you use a 4pin to SATA adapter, it will not have the 3.3V at the SATA terminal? Since 4pin only carries 12V and 5V.
 
Ocellaris said:
He’s plugging the drives into a backplane though...
Click to expand...
Which is odd to begin with since the backplane is powered by molex connectors and he has other 10TB shucked drives that function fine.
Additionally reddit wd compendium shows the 4224 should work fine (which is confirmed by some of your already working drives).
Are the drives on the same backplane row as the other drives that are working? Have you tried other slots further down?
Have you verified the drives are spinning up and/or molex power snugly plugged into the plane they are connected to?
 
It seems his backplane is creating 3.3v from the +5v or +12v since his new drives do work when 3.3v is isolated.

As far as firmware abuse goes, the PWDIS feature is not controlled by firmware. It's a feature built into the power circuitry of the drive, If pin 3 is pulled high (i.e it has power applied to it), the power circuit will cut power to the rest of the drive's electronics. It's meant to be able to power cycle a hung drive without having to either power down the entire system containing it or having to physically pull the drive and reinsert it.

It's luck of the draw whether you get a SATA 3.3 (i.e. PWDIS pin 3) drive or a SATA 3.2 (i.e. +3.3v pin 3) drive since it seems WD just pulls the EasyStore units from whichever supply chain is active, gives it a white label with a new model and serial number, and then slaps in an enclosure.
 
ryan_975 said:
It seems his backplane is creating 3.3v from the +5v or +12v since his new drives do work when 3.3v is isolated.
Click to expand...

I'd be interested to know if the tape is removed if the drives still functioned now(reseating may have caused it to work rather than the 3.3v issue?)
Generally all of those white label drives are the same especially since he noted that: "All the pertinent #s between the drives are identical firmware, model #, P/N #, everything."
So it makes me wonder if it was a fluke or missing power rather than the 3.3v issue (since most backpanes, especially molex powered ones, aren't generally affected)
 
As ryan said, some WD externals have the power disable feature and some don't. There's no marking to distinguish them. That's why you can't trust reports of "X PSU is compatible", because people just assume that their drive has power disable.
 
Finally got the time to cover up the pin on both drives and get them into the backplane and as I expected they work fine that way. So apparently my RPC-4224 puts something on that line. Whether it's 3.3v or not I don't know, I don't have fine enough leads for my DVM to figure out what the voltage is.

Spartacus09 while indeed 2 rows of my server were not getting 12v due to a bodged custom cable I tore the whole thing down and fixed it before running an exhaustive test across all backplane ports for basic functionality and then checking the 2 new WD drives. All 4 drives are labelled as though they were identical units aside from SN & WWN so whatever WD did with these it's not possible to tell by looking that there's a difference. For anyone wondering, my November `18 purchase has Aug 2018 drives and my April `19 purchase has June 2018 drives.
 
alaricljs said:
Finally got the time to cover up the pin on both drives and get them into the backplane and as I expected they work fine that way. So apparently my RPC-4224 puts something on that line. Whether it's 3.3v or not I don't know, I don't have fine enough leads for my DVM to figure out what the voltage is.

Spartacus09 while indeed 2 rows of my server were not getting 12v due to a bodged custom cable I tore the whole thing down and fixed it before running an exhaustive test across all backplane ports for basic functionality and then checking the 2 new WD drives. All 4 drives are labelled as though they were identical units aside from SN & WWN so whatever WD did with these it's not possible to tell by looking that there's a difference. For anyone wondering, my November `18 purchase has Aug 2018 drives and my April `19 purchase has June 2018 drives.
Click to expand...

Thats definitely odd thanks for testing for us.
 
Put a super tiny strip of tape on every single spinning rust drive worth buying (shucked drives are way way cheaper than anything else, with a bigger array even having a local onsite self-warranty with spares is still cheaper).

OR

Remove a wire from your PSU sata cable that has never been used and never will be used. If your PSU is not trash or has modular cables this is extremely easy to do.

For some reason "common internet wisdom" always picks the former.
 
Last edited:
The kaphlon tape is cheap and probably less daunting than damaging cables or unpinning them for some folks.
 
You must log in or register to reply here.
Back
Top