[SOLVED] H3 died in 4 months

[fenixil]

My H3 does not show signs of life - when power is connected system LEDs are off, no HDMI signal.

I found this issue with identical symptoms. I use hardkerne's 19V PSU, voltage drop on the 15 amp fuse is huge - 14V. Fuse is ~58kOhm if board is not connected to the PSU for more than a day, after I connect power it becomes ~28kOhm and grows slowly.
I hear a faint coil-wining-like noise from the back side of the board, one of the inductors has 0.28V on it, and others are 0 (see images).

For the context - I used this board as NAS - I have 2 HDD 3.5"" attached via USB3.0, 12V is taken from the board's connectors. 2 months ago, the board stopped booting as one of the drives was constantly restarting. I found that if I disconnect drives and then connect them one after another everything is running fine even in the long run. It was super inconvenient though as every time the board rebooted I had to disconnect the drives manually, let it boot up, and connect them back again. A couple of weeks ago, this workaround didn't help. I left only 1 drive connected via SATA, however, the board periodically just powers off. Logs were clean and I have not found the pattern when that happens - it just turned off itself. As suggested here I ran memtest86+ and it was running for 24 hours without any errors and then the board turned off itself again. After I restarted it, memtest86+ lasted only 2 hours and again - a shutdown. As suggested here I decided to make a "cold boot" - I disconnected PSU, removed the battery, plugged in PSU and memtest86+ start running - something was off as I expected BIOS to get to its defaults. I didn't wait for 5 minutes before I plugged in the DC, that was the issue. After I did that my board stopped showing signs of life.

I use an ethernet expansion board, it completely covers the fuse, so my current bet is that it's degraded over time, however, I'd like to get advice on the forum before I do something stupid and completely burn my board. Any suggestions, please ?

[fenixil]

It's nearly impossible to remotely find out what the state of your hardware is.
However, if inductor noise occurs, it is not a problem with the fuse itself, but rather one of several DCDC converters connected to the fuse is damaged and the system does not operate.
BTW, which HDDs have you used? Can you share the model name and brand?


Anyway, try sending us the board for a RMA since our warranty period is about four months (16 weeks).
We can extend it a couple of more weeks for you convenience.

[fenixil]

Hm, I checked the invoice, and looks like I purchased it much earlier - December 2022 :facepalm:, 4 months ago I started observing issues, 2 months ago it stopped booting so I had to workaround the issue, couple of days ago it stopped showing signs of life at all. Sorry for the confusion.
I use 2 14TB IronWolf PRO (ST14000NE0008).

Conceptually fuse should have a low resistance when it's healthy (mOhms) and huge resistance when it's blown up (MOhms). I don't know how to interpret my 28-58 kOhms - it's something in-between, so I treat it neither healthy nor blown-up (degraded?), does that make sense? What's the fuse's resistance on the healthy boards?

[mctom]

The schematic doesn't say what kind of fuse this is, other than 125V 15A.
But it does look like a SMD blow fuse, so I wouldn't suspect it's a polyfuse.
If your measurement is correct then there is something wrong with this fuse, I guess.

If you're in doubt, you can replace it, given your board is outside warranty anyway.

[fenixil]

You are right. Actually measured value is about 0.2 Ohms.
Therefore, it is very worth to try replacing the fuse.
A compatible fuse part number is Bel Fuse 0680H9150-05. You can find it in Mouser or Digikey.

[fenixil]

I have great news! After I replaced the fuse, the board came alive! Scrubbing the data now, but I think the issue is solved. Thanks, everyone for your help!

[rooted]

Congratulations on the resurrection of your board.

[mctom]

Great work!

Now the question is why this fuse failed in such a weird way, but I wouldn't worry about that too much unless it fails again.

[fenixil]

why this fuse failed in such a weird way

@mctom I disassembled the old fuse but didn't see anything suspicious. I googled similar cases and found the thread where someone said that fuses can degrade over time due to oxidation or evaporation. During active use of HDDs power consumption is up to 50W, but it's ~2.5A with 19V PSU which is way lower than the fuse's nominal. Another hypothesis is that it was overheated due to the ethernet expansion board. Next week I'll get my FLIR camera back, so I'll investigate in that direction and post my findings here.

[mctom]

That's a cool idea, to check how's it working with thermal imaging. Maybe it is constantly stressed thermally. After all, fuses are blown directly with temperature.

But I'm suspecting a defective fuse that failed prematurely. As you mentioned the computer randomly shut down before it failed completely. Then you measured it in kOhm range when it should have been either sub-Ohm or in MOhm range.