Cutting power to SATA HDD to save heat and energy

[soldar]

I have a Seagate Barracuda, 750GB, ST3750640NS, 3.5", SATA HDD which I was using just for backups but since my main drive failed I have called this one into active service again.

I want to use it as a secondary storage drive, not as the OS boot drive.

BUT, it gets hot. Too hot for my taste. I assume it is OK but I am thinking that, as I will only be accessing it very sparsely, I could turn it off most of the time and only turn it on when I need it.

I assume just cutting the 12V and 5V supplies would be fine and would not damage anything. I do not know how the OS, Win XP, might deal with it but I am hoping nothing catastrophic. In the worst of cases I would need to restart the computer.

I assume it would be safest to cut both voltages but maybe cutting only one is safe? This would make installing a switch easier as I would not need double pole. I suppose I could always install a relay.

I would welcome any advice, tips or knowledge on the subject.

[DimitriP]

I assume just cutting the 12V and 5V supplies would be fine and would not damage anything. I do not know how the OS, Win XP, might deal with it but I am hoping nothing catastrophic.

Assuming and hoping don't sound very "engineery".

Go to your Power settings and tell windows to turn the drive off when it's not being used.
Or replace it since it's getting "too hot".
But whatever you do, don't assume and hope.

[dobsonr741]

Friendly suggestions: get the basics right first: be able to provide proper airflow to cool the drive. Regardless of cutting power.  Get rid of Windows XP. Upgrade. It’s obsolete.

Then look for the spindown option on the OS you replaced XP with. Also, the drive needs to support spindown, which might not always the case.

[madires]

AFAIK, support for hot-plugging is part of the SATA standard. You could cut the power or put the HDD into sleep mode (via a command).

[magic]

Absolutely spin it down by software, every disk made in this century supports this.

Cutting power shuts down the SATA link and makes the software consider the disk gone for good. For all the software knows, the disk may now be plugged into another machine and being modified by it. Any data buffered by the software and not flushed yet will be discarded and lost.

[soldar]

AFAIK, support for hot-plugging is part of the SATA standard. You could cut the power or put the HDD into sleep mode (via a command).

Thanks. I know hot-plugging is part of the SATA standard and that is why I assume it is OK to power down the drive but I just wanted to see if anyone has any actual experience because XP has its own quirks. 

I do not believe XP has any user way to put a HDD into sleep. I have seen some utilities but I really do not want to spend too much effort on this and I figure they might not work the way I want or at all.

After all, this drive might not last that long anyway and it is not worth putting too much effort into it.

I had a caddy for an extractable HDD and that would have worked but I have no idea where it is. I may still have it or I may not.

Cutting the two voltage cables with a switch is really easy so I think I will just try that. I have power cable extensions so I do not even have to mess with the computer itself. Just put the switch on an extension. 

[RoGeorge]

Some motherboards used to have this feature, for example this one, where it explicitly states the Hot Plug feature
https://www.asrock.com/mb/Intel/Fatal1ty%20Z77%20Professional/
and it came with an onboard power off switch too (which was a great feature to have to protect yourself against ransomware and such), however I remember there were some Windows "drivers" to power down the HDDs from the OS.  It was also posible to turn the power on/off from BIOS, without any drivers.

WinXP used to have some modes with immediate write (without any cache), but I don't know if that deals with the cache located inside the HDD's too, or if it is for the OS cache only.

In Linux it can be all managed with hdparm and some scripts to specify the exact behavior you wish.  Don't know how to do the same in WinXP, but it should be possible.

[magic]

Virtually all SATA hardware is hotpluggable and will survive, it's your data which may not.

Same deal as yanking out a pendrive, except more risky because even a consumer OS may reasonably assume that at least internal disks won't disappear at any random moment. I dunno how long NTFS may keep dirty data around, why would anyone want to find out the hard way if there are better options?

I remember that with Windows 98 it was possible to set the disks to spin down after some period of inactivity. Wouldn't be surprised if XP had it too.

[DiTBho]

Virtually all SATA hardware is hotpluggable and will survive, it's your data which may not.

Only if the hdd slid has rigid rails, and the connector on the backplane is really hot-plug.
It's the difference between a cheap HDD enclosure, and a professional one, which costs 3x times, not only for rails and connectors, but also for the air-flow.

[DiTBho]

it gets hot. Too hot for my taste

it sounds bad air-flow.

[magic]

I have hot plugged (and unplugged) SATA HDDs to my consumer grade desktop computers countless times. Never any issue as long as common sense is observed (unmount and remove logically before disconnecting). The order of connecting and disconnecting power/data cables doesn't matter. The precharge pins on the SATA power connector limit inrush current as long as the HDD is implemented correctly. Although they won't be effective anymore if you control the power with a switch or relay.

Recently I added a passive panel with two dual-voltage powered eSATA connectors, internally connected with the PSU and the motherboard SATA ports, so I can attach bare 3.5" SATA disks with a single external cable. Again, fully hotpluggable and no issues whatsoever.

[amyk]

How hot is hot?

As others have said, spin it down via software; XP definitely has the setting for this.

That said, it being a Seagate, you should keep an eye on the SMART attributes since they are known to not be a particularly reliable brand.

[soldar]

The drive is on its lasts legs but will have to do for a while and for a purpose. Before anyone is tempted to say "just buy a new drive" please think how much you are willing to contribute.

SMART info:
Temperature: 53C
Reallocated sectors count 138 (stable for quite some time now)
Power on count 558
Power on hours: 50765 (5.8 years)

I don't remember where I got it from. Probably pulled from some old computer.

I put a switch to disconnect the power and things went more or less as I expected. No HW problems but Win XP does not update. If it is started with the drive then the drive always appears and if I try to access it Windows just waits forever. If I start Windows with the drive off and power it up later Windows is not aware and it does not show up. More or less what I expected. So updating the status requires a reboot.

I will probably have it powered off most of the time until I can replace it.

I purchased the ASUS P5Q SE mobo in 2010 and I guess it is also showing its age. When I pulled on the sata cable connector the plastic part from the mobo broke and pulled out leaving the 7 metal pins on the mobo. It looks to me like with great care I could probably put the plastic back in place and maybe hold it with hot glue stick or epoxy. For now I do not need it as I have other sata connectors available. But all my gear is showing its age. Just like the owner.

[DimitriP]

It’s obsolete.

Wrong thread

As others have said, spin it down via software; XP definitely has the setting for this.

[magic]

I put a switch to disconnect the power and things went more or less as I expected. No HW problems but Win XP does not update. If it is started with the drive then the drive always appears and if I try to access it Windows just waits forever.

I am morbidly curious what happens if the disk is disconnected and then connected back.

Bonus: what happens if it's disconnected right after some files are modified, then the files are modified on another system, then it's connected back...
 

[golden_labels]

Blindly powering down the drive will eventually lead to data or file system corruption. There are at least two layers of caches, which must be flushed before the drive can be powered down. You will not be able to avoid that with just flipping the switch on power lines. This is in particular true for primitive file systems like FATs. NTFS, ext or btrfs are at least reliably repairable (mostly), but FATs may go completely bonkers.

If you don’t care about speed, consider an USB attachable external HDD. They are imperfect, and the cheaper ones were reported to fail with large amounts of data being written at once. Many also bail out on the first I/O error and just shut down. But Windows will see them as USB sticks, which can be properly and safely unmounted. There are also enclosures of this kind, where you put your own HDD. In that case make sure it handles a drive of the size you have. And, unless somebody confirmed this particular enclosure model causes no trouble, rather don’t attach a HDD with data on it. Some were not using the entire drive space or even assuming they can store their own data on the drive (the same way Gigabyte’s BIOS backup was once nuking user data).

Some general notes:

Any SATA drive should be hotpluggable and should survive that with no issues. Whether manufacturers of consumer devices follow the requirement beyond “it doesn’t die on first attempt” is to my knowledge unknown. I never seen any mention of anybody testing this. If a controller on the motherboard can handle this is a separate thing: it’s possible it will not be able to see any drive attached after booting. I have seen mixed reports on that for older mobos. Same for Windows XP: the system must actively rescan the bus to recognize all new drives.

Any modern HDD, back to even later PATA drives, can emergency-park heads. I didn’t see any research on the topic, but after that became a standard feature I didn’t hear a single report of a power failure causing heads crash into the platter. Which was a quite common occurance in earlier times. That fall in reports is something I can take as a sign of this not being faked and working reliable enough for rare enough situations.

Frequent spin-down/spin-up and parking cycles were shown to considerably drive failure rate up. Search for opinions on how that worked in the WD Caviar Green series for details. But that was for cycling happening up to a hundred times a day. This isn’t what you asked for, but it may be the case, if you use the option DimitriP mentioned above.

I’ll skip extended comments on using Windows XP in 2024 on anything that is not a separated environment.

“I did this foryears and it worked” isn’t a good evaluation method for something that is already known to be a low-probability event. It just indicates it’s not leading to a complete disaster, but we already know that.

If you turn off a drive, with any method, consider letting it lie for a moment. The platters will not stop immediately. If you pick up the drive instantly, the gyroscopic effect will not only cause unneeded stress on the bearings, but your brain may get confused with signals from the hand and you’ll drop the drive. HDDs don’t like being dropped.^{(citation needed)}

[Berni]

Just install a SATA drive cage, then you can just pull the drive sled out slightly to disconnect it. (And just set that SATA port as being hotplug in bios)

Other alternative is just buying a cheap external USB HDD that you can disconnect

[Nominal Animal]

I could turn it off most of the time and only turn it on when I need it.

No, just put it to SATA sleep or standby (doesn't matter which, they're the same on this drive).

Here are the drive technical specs.  In sleep or standby, it consumes 0.8W (106mA @ 5V + 23mA @ 12V).

Having the drive report 53°C indicates you have insufficient airflow.

Even at 30°C ambient, on my quiet machine with all Noctua fans running at 800-900 RPM with three 3.5" SATA HDDs they reported at most 46°C; at 20°C ambient, 36°C typical, so pretty stable +16°C.  (Even back then, my machines all ran Linux, and the hdparm command could be used to put the drive to sleep/standby immediately.  One of the drives was a "backup" or archival one, the other two were in RAID0/1 and in very frequent use.)

Then again, I do not trust Seagate HDDs as anything except weights: I had too many problematic units back in the day.

[DiTBho]

Any SATA drive should be hotpluggable and should survive that with no issues.

*ONLY if * see my post above. I won't repeat.

[DiTBho]

Having the drive report 53°C indicates you have insufficient airflow.

Same thing I thought.
Then well, I'm one of those who is willing to spend 200 euros on a 4 bay HDD box with certified connectors, anti-vibration cuscions and right ventilation for disks that must work 24 hours a day.

[nctnico]

I could turn it off most of the time and only turn it on when I need it.

No, just put it to SATA sleep or standby (doesn't matter which, they're the same on this drive).

Here are the drive technical specs.  In sleep or standby, it consumes 0.8W (106mA @ 5V + 23mA @ 12V).

Having the drive report 53°C indicates you have insufficient airflow.

I agree. Lack of airflow is the problem which must be solved first.

[coppice]

Having the drive report 53°C indicates you have insufficient airflow.

A lot of computer cases are really bad for this. They put so much emphasis on getting air to flow across the CPU and GPU, that they completely ignore the rest of the case. If there are alternative places to mount the disks, try them. They can vary enormously in their cooling flows.

[soldar]

Regarding the temperature thing, yes, obviously if I put a fan on it the temperature goes down but this drive seems to get much hotter than others I have had or have and it makes me think it might be faulty.

Still, my experience with electronics and heat is that very often they are poorly designed for heat dissipation. I have had many experiences with devices which were insufficiently cooled. The main chip on my home security NVR gets way too hot and I have the cover of the case off and a small fan running right on top of the chip. The TDT TV tuner same thing. One of my computers has a ATI Mobility Radeon HD 5430 which has no fan, only passive heat sink, and it overheats so I have jury-rigged a fan over the heat sink. I have encountered many situations where cooling is just insufficient. In some cases the chip was going bad and heat production increased until the thing just failed.

I have a crate full of computer fans so it is not complicated to just rig one over the drive but, at this point, it looks like the drive is on its last legs and not worth spending too much effort on it.

And the less I mess with things the better. Trying to fix or improve one thing often leads to breaking another. I have damaged a SATA connector on the mobo and it might be fixable but I would probably have to take the mobo out because I need to align the pins with the holes in the plastic part and I cannot do that looking from above. Bummer.

As for putting the drive to sleep via software I just do not trust it because the OS can access it beyond my control.

I might play with putting a fan over the drive. That seems the easiest thing to do.

What is clear is that putting the drive in a regular 3.5" bay with others stacked above and below will not do.

[Nominal Animal]

this drive seems to get much hotter than others

Have you verified that with a surface thermometer, or relying on the internal thermocouple in the drive itself?  It could be just a faulty thermocouple (or rather, the circuitry measuring it).

In the worst case, its motor grease is aging making it sticky (increasing its sticktivity) and increasing power draw and thus increasing heat generated.

Still, my experience with electronics and heat is that very often they are poorly designed for heat dissipation.

For sure, that's why I started building my own.

I've seen many that try to have laminar flow for the processor heatsink.  That can easily cause the regulators near the processor on the board to get too hot.
I typically use one additional fan at low RPM inside the enclosure to ensure non-laminar flows and sufficient mixing, plus a few baffles on the input fans.
It is not difficult to do –– although I so wish I had had a thermal camera to check! –– so it is kind of surprising big manufacturers get this so wrong so often.

As for putting the drive to sleep via software I just do not trust it because the OS can access it beyond my control.

In Linux, the kernel correctly resets the drive from all sleepy modes automatically, so it was never a problem.  That is, I've never needed to "wake up" a SATA drive.

Some drives, like WD Greens, are pretty aggressive sleepers by default, so adjusting the settings (also doable via hdparm in Linux) is definitely useful.

What is clear is that putting the drive in a regular 3.5" bay with others stacked above and below will not do.

That's how I had the three drives in my quiet setup, but with an input fan (Gentle Typhoon was even better than Noctuas, due to the semi-obstructed flow) in front of the cage.  Spreading them out more would have been even better, but there was no room in that cage.  Airflow on the sides of a HDD drive does not help much, what matters is airflow on both large flat surfaces.

[soldar]

Have you verified that with a surface thermometer, or relying on the internal thermocouple in the drive itself?  It could be just a faulty thermocouple (or rather, the circuitry measuring it).

I rely on my fingertips.

In Linux, the kernel correctly resets the drive from all sleepy modes automatically, so it was never a problem.  That is, I've never needed to "wake up" a SATA drive.

I'm not sure I understand but precisely what I do not want is the OS waking the drive automatically because it is doing some internal housekeeping.

For now I might opt to just jury rig the drive and a fan on it. This seems easier than shutting off the power. Then I can experiment further with the power saving options.

All my computer boxes are a shoddy mess inside with things hanging by the wires and not secured as they should be.  The PSU is 14 years old and it works ok except that at some point it seemed like the output caps were not quite filtering as they should so I connected some caps at one of the Molex output connectors and that seemed to correct the problem and so they have been hanging there, wrapped in cardboard and tape, for a decade. Molex connectors are a PITA because often the pins will not line up with the holes and it is impossible to connect them.