He10 disk fun

I recently had one of my older 48kHour He10’s start to throw SMART errors that included an inability to finish a short test, etc. Since I have gotten my money’s worth, I thought I’d extend my usual work life to home and have a look inside this disk. Specifically, I’m interested in how they build them differently from air-filled mechanisms.

First, I peeled off the end protectors, which are basically thin strips of metal designed to protect the ends of the drives where insertion damage is most likely. Specifically, they protect the Al layer you see peeled below from nicks and like damage that would allow the Helium to escape.


The top layer itself seems to consist of a thick Aluminum tape-like layer that is tightly adhered to the underlying lid, covering all screws and sealing to the edges of the case itself. This is presumably the main seal. The actual metal lid fits inside the die-cast disk case, allowing the top edge of the case to be machined smooth and tightly-bonded with the Al-Tape around all the edges.

Using thick Al tape makes sense, He presumably diffuses very very slowly through the stuff, esp. at low pressure. Al is also used as a diffusion barrier in candy-wrappers - it’s what keeps your snickers bar fresh even after weeks in a vending machine - though in the case of the candy-wrapper the layer is only a few atoms thick and the molecule of interest is Oxygen, not the much-harder-to-contain Helium. Sputtering is your friend.

Inside, screw holes and like penetrations through the steel lid were covered with small sealing stickers as with most hard drives, save one. Presumably, that’s how the Helium was filled into the mechanism once they had evacuated the shell, followed by lidding with the thick Al Tape. I presume this is a batch process…


Once you unscrew the lid, the extra sealing around the edges becomes apparent, along with what I presume is some sort of filter / getter assembly glued to the underside of the lid. I presume all the Helium being inserted into the case is blown through this assembly. Or perhaps it’s simply there to absorb any moisture.


I have yet to take it off and take a closer look at it but the bottom reminds me of the Gore-Tex air filters that used to be pretty common on air-filled drives for pressure-equalization w/o allowing water to get in. It’s one of the better uses for Gore-Tex.


The platters / read/write heads are fairly similar to what I’ve seen in the past, save for the huge cutout in the bottom for the filter I mentioned above. Now that they have been extensively sprinkled in dust, I doubt they’ll be very useful.


The heads look pretty conventional and I was amazed just how many were in there.

5 Likes

I decided to research a bit further and stumbled across this post from HDD Surgery, where they took one of these units apart on a YouTube video. According to those folk, the edge of the lid (i.e. what I called Al-Tape) is actually laser-welded to the case. Makes sense, it would make for a even better seal than just relying on the adhesive in the tape.

In this screenshot you can see how all the holes on top of the case cover are covered in small stickers, save for the one that goes through what I thought was a filter / getter.

2 Likes

In your opinion, did it appear the platters were stacked closer together that a typical hard drive? How many heads? 16?

Crazy how small stuff gets these days. Even IC’s have atom wide paths, remarkable.

Now for the big question, did you power the drive on and see if it would partially work after opening it up?

Key here would be to spin it up in something that speaks SATA natively so we can see whether reported helium levels match expectations. Too bad I cannot do that in a uncovered state since all my external drive enclosures feature USB interfaces.

An open case would be totally egregious re: the load on the spindle motor since the air flow would not start rotating as readily inside the case as it would with the lid on (surface effect). Never mind the spindle bearing assy being totally unhappy about the case lid screw no longer supporting it from one side (i.e. operating like a front - load clothes washer)

The head crashes would be glorious though unlikely to be as loud as they were in the removable DEC Hard drives from the VAX 750 / PDP11 at school. Those platters were huge, heavy, and required thick gloves under frisbee conditions.

I reckon it’s 7 platters in there (count the orange rest spots). Obviously, more capacious drives either have more platters or greater areal density. Some drives are over 20TB now, i.e. twice the capacity of what you see here. Presumably, areal density is the deciding factor. :grin:

These days, with modern-ish bridge chips, smartmontools works pretty well across USB/SATA bridges. Worth a shot.

Because you allowed white mold to grow and bloom on the platter itself. :roll_eyes:

Maybe next time don’t put your NAS server in a humid environment?

#RookieMistake

white-mold-bloom-infected-hdd

2 Likes

Must be an HGST / WD helium-filled drive, since the Seagate Exos (at least the ones I have) lack the “SMART 22 attribute”, which reports the current helium level, even though they are helium-filled. :frowning_face:

Last time I checked, the drive was reporting 100% Helium. I would love to be able to see the drive spin. I may try and dig out an older SATA drive enclosure that cannot handle <2TB drives but which should be able to read SATA codes. Then spin it up in a safe environment and see what happens.

Hopefully, not a reprise of some of those famous washing-machine self-destruct vidoes that shouldn’t be repeated due to these yabos not understanding the risks they were taking.

I repaired and operated a ruggedized version for submarines. I would tear them apart, rebuild them, all while on the sub. It would take a good person at least 12 hours to do the job and a terrible person less time because they just slapped it together and made it work good enough. This was not good because every 105 days we would rebuild one of two units. One crew rebuilt one, the other crew rebuilt the second. The bad part is when the other crew did a poor job, the hard drive would fail while we were on alert patrol. It would be my sole responsibility to repair it. Grasping both the mechanical and electrical concepts were easy for me. I would be able to push electrons through what felt like a gazillion flip-flops to make the circuits work.

But yes, 20 heads approx 1" in diameter slapping up against the iron oxide 14" platters, LOUD! and very distinct. Telling the Weapons Officer and ultimately the Commanding Officer that we are in reduced capability was never a fun thing. But depending on the problem, no more than 12 hours later I would have the thing up and running again. Head alignments were easy, things are much different these days.

1 Like

I can’t imagine trying to recreate a clean room environment on a sub but I have no doubt it can be done by folk as talented as you are.

In the VAX , a head crash would be detected quickly and the heads would park almost instantaneously. But it would never be quick enough - so we could always recreate which platter had the head crash by looking for the shiny, spiraling groove that the head would dig into the platter as it was dragged to the head-parking-area.

DEC was a funny company, probably the first to implement differential pricing by putting artificial software brakes on a CPU to throttle it’s output based on the price point the customer was willing to spring for. Ditto technicians being able to double the HDD capacity by reconfiguring jumpers to enable the heads on the other side of a HDD to start working also… etc.

The equipment was built like a tank, likely sold at cost for the most part, but hoh-my the maintenance contracts were expensive. We estimated the school had spent 3x the cost of the mainframe on maintenance over the 9 years the school had that minicomputer.

Years before the last century, I had to swap out a removable 5MegaByte disk cartridge on a DEC PDP 11/34A. The disk spun down fine, and unlocked the cover, which is supposed to mean the heads were parked.

But, low and behold, as I removed it, a little extra resistance was noticed. That was when I saw the heads had not retracted. Worse, they suddenly jumped free as the disk cartridge finished removal, then the heads wacked together :-(.

After the service guy fixed the drive, I asked if I could keep the heads. However, he needed them for “proof” of failure.

The good thing about this, is that it appears the heads did not crash. Nor did they scratch my disk platter. After repair, all data was safe.


In regards to computers at sea, I helped design and write software for a U.S. Navy sonar system. We used a Sun SS20 work alike, with removable disk drive. 8mm tape and much older re-writable optical media, (if I remember correctly).

For that system, they initially were using a T1 line between the sonar head unit and the SS20. I said why? We were developing both sides, why not Ethernet? (Though not TCP as that was too much over head… for more simple point to point data comms.) So that is what we did, 10Base2, (aka thinnet using BNC & Coax).

1 Like

I was laughing my ass off… No such thing as a clean room on a submarine. Your Vax drive didn’t have one, the heads and platters were basically out in the open and you may have a filter that may have worked somewhat. Smoke was the enemy on the submarine, it was 5X larger than the laminar gap. And back in the day, there was a lot of smoking going on.

1 Like