Synology DS1817+ power supply death...happily now back!!

Looks like my trusty Synology DS1817+ which has been going strong for many years, has suffered a severe power supply fault and died. There was a mains fault (trip and reset a minute later) that probably was the straw that broke its back - and it was running OFF the UPS at the time, as that needs new batteries...

I'll do some surgery later and see if there isn't an obvious fuse inside anywhere which has blown, but looks like I'm power supply shopping.

Just hope the rest of it is OK 😭

Edited by Pheasant (Wed 13-Nov-24 21:19:30)

Well there's not a trace of black soot in there at least, but not much else obvious either (to the untrained eye). I'm not messing about with a mains powers supply.

Have ordered a new PS and hopefully it will restore it to life,

Did you test the power supply with a multimeter?

There’s a thread here (slightly different model) that might be relevant: https://community.synology.com/enu/forum/1/post/1318...

Thanksfor the link. As far as I can see they're referring to a DS1815+ there, which is a generation older than mine, hence the chat about the Atom bug and soldering in resistors etc.

Here's a video of one getting pulled apart. Unfortunately the Mobo position on the later DS1817+ is right at the bottom of the unit, underneath all the drive bays. A real pita (not bread) to get into.

I can just sneak a peak at the CMOS battery here, but its too tight too get both probes from a DMM on there from outside. So it will need to be further surgery...

Surgery (and googling) continues...🤣

Mobo is out. Wasn't actually all that bad after all. Simply required the bottom cover fully removing (6 screws) and then take out 7 securing screws. After that disconnect carefully release one micro ribbon cable connector and gently lever off the Mobo from 2 remaining IDC connectors from the drive backplane. Or something like that!!

Checked the existing 2032 CMOS battery. Reading was 3.2032 V. I think it's OK but while I was in there have replaced it with a 'fresh' (well it was in a packet in my battery box!) Energizer 2032 button cell which is reading 3.314 V according to Messrs Fluke.

Now I'm in actually some doubt as to whether this board may be suffering from the dreaded C2000 Atom series bug.....more googling before putting it all back together! 🤣

Power supply check now. Out of the chassis test: it's powering up / fan is running when green + black are jumpered. All the +12V and +5V outputs I've checked look OK too...

Leads me to suspect it's the power switch sub-board or Mobo. I've read some stuff about a faulty transistor on the DS1815+ units....

Apparently you need to test 2032 cells with a small load (say 100 ohm) to get the true (actual depleted) voltage. Testing with them just open circuit apparently doesn't tell you much, as a fully depleted cell can still measure 3V (unloaded with just a DMM) but when a small load is applied the voltage drops like a stone.

That's something I've learnt today. Also how to switch on an ATX-type power supply through the header pins!

Edited by Pheasant (Wed 13-Nov-24 20:14:58)

you can get a little plug thing or you can make your own. some high end psus have a connector to power up two psus simultaneously

I made a DIY jumper with a bit of wire I had lying around. That was good enough for this evening's little job.

Anyway am pleased to say the new CMOS battery did the trick. We're back in action and fully operational - on the kitchen floor and topless 😂. But fully running nonetheless!!

Wooohooo.

Edited by Pheasant (Wed 13-Nov-24 20:57:47)

Good news.

Somewhere in the back of my mind I seem to remember a similar problem caused by the CMOS battery - but I can’t recall whether it was one of my Synology NASs or something else.

Indeed. Thank you for your help. I certainly didn’t expect it to be the CMOS battery, but your post led me on the path and saved me from needlessly getting a replacement power supply. So thank you very much 🙏

feel bad that i didn't pick up on the cmos weirdness. I've seen enough failing batteries in motherboards to notice it. But the weirdness is such that you don't initially pick it up as that and you believe it to be something else.

Not related (but only in fault diagnosis), I saw an aio watercooling pump fail this year, it took me about 3 days to work out that the pump had failed, the computer was turning off at random times which could have been psu related and i went with that. I only realised that it was the pump when i accidentality put my hand on it. It was my first pump failure tbh.

Don’t feel bad at all. Just one of these weird things. It’s easy to sometimes just target what you instinctively believe the problem to be rather than apply a logical, step by step diagnosis

Guilty your honour 🙈

Again a good read with a positive outcome. Glad you are up and running again.

Cheers. Also £93 back on my credit card now too, for a falsely accused power supply - and just minus one 2032 button cell for a couple of quid: that'll cover the few (low alcohol) lagers I downed whilst tearing it apart last night 😀🤘

I assume you brought a multi pack of them for a couple of quid

Yeah whatever Amazon charge for 6/12 of them on Prime. There's always some kicking about as 2032 is a popular size for keyfobs on cars and gate remotes etc.

I can beat that ...

I was asked about a shower pump failure - it kept starting up every 30 minutes. A "pump specialist" told him that the pump was failing and he needed a new one at about £600 plus installation labour of £200. I gave him a diagnosis that took 10 minutes to fix with a 10p tap washer. A pity I was not on a commission of 25% of savings!

You should have at least got 25% of the cost of the washer.

In reply to a post by Taras:

Not related (but only in fault diagnosis), I saw an aio watercooling pump fail this year, it took me about 3 days to work out that the pump had failed, the computer was turning off at random times which could have been psu related and i went with that. I only realised that it was the pump when i accidentality put my hand on it. It was my first pump failure tbh.

Well, for odd hardware failures, I had a 486 computer I was loading with OS/2 It worked on my other computers, but it almost consistently failed on this computer, while unpacking one specific file. If I got past that one file, it was fine and the system would run. But for some reason, I needed to redo it several times with the same result, again almost consistently. The mystery was resolved when I opened the case and found the cpu fan had taken itself off the processor and was hanging in free air. Once I had refixed it, there was no more problem. really odd that it hung consistently on unpacking just one specific file.

In reply to a post by DFScale:

really odd that it hung consistently on unpacking just one specific file.

Might have been one chunky subroutine, or specific bit of maths in unpacking a certain section of that file, where the processor had a sustained peak/burst - just enough to trigger a micro-thermal event (without a fan to drop temps) and trigger a reset...

In reply to a post by DFScale:

Well, for odd hardware failures, I had a 486 computer I was loading with OS/2 It worked on my other computers, but it almost consistently failed on this computer, while unpacking one specific file. If I got past that one file, it was fine and the system would run. But for some reason, I needed to redo it several times with the same result, again almost consistently. The mystery was resolved when I opened the case and found the cpu fan had taken itself off the processor and was hanging in free air. Once I had refixed it, there was no more problem. really odd that it hung consistently on unpacking just one specific file.

Nice... reminds me of a 386sx that every file you copied from the 3.5" 1.44 FDD would be randomly corrupt. So from the factory the pre-installed DOS and Win3.0 would randomly crash. The only reliable way to discover the faulty floppy drive or controller was to unzip a file from the floppy to the HDD using DOS based pkunzip as it reliably failed...every time. Same floppy on another PC worked every time.

Took a nightmare with support to get that motherboard replaced; expensive machine back in 1990

My first 'home' PC was an Acer 386SX with 4MB RAM running at a blistering 20 MHz....with a socket upgrade to a 486DX. Oh heady days!!

In reply to a post by Pheasant:

My first 'home' PC was an Acer 386SX with 4MB RAM running at a blistering 20 MHz....with a socket upgrade to a 486DX. Oh heady days!!

Apricot 8086 DOS based machines first, then eventually had use of an Amstrad PC1512 for a while, until parents bought an amstrad portable...great keyboard, useless tiny LCD screen, and an NEC multisync grey scale screen. Dual floppy with 2400bps modem, was amazing. Then upgraded with a 2.5" 20mb HDD from a small outfit in Reading made the machine faster than most companies 286 and 386 machines...!

I built my own PC from parts (when PCWorld was a small shop in south london, not yet owned by Dixons/DSG) and that was a 386sx. Ran that for years. Moved to a Pentium 75 as they were what all the cool kids had at uni, ran cooler and quieter than the 60 or 66...

Ancient history now.

Edited by jchamier (Thu 14-Nov-24 20:53:13)

I started on an Intel SDK80 with hex keypad and then an SDK86 which used 8080 and 8086 respectively ...

I had an Amiga 4000/040 and i loved it to bits, i had the emplant card too and managed to get mac emulation working ..

fun times

Downer was floppies at 1.7mbytes - they ran at half speed with double the data than the old 880Kbytes drive which meant it took 4 times as long to fill a floppy up.

oh and the 68040/25 was a toasty cpu

I was also working with COSMAC - using the RCA CDP1802 the first CMOS CPU. Had a TWIN 8" FDD and that was enormous. The drives had to be calibrated weekly.

Perhaps that file was larger than the others. Decompressing is processor intensive.

In reply to a post by Pheasant:

Don’t feel bad at all. Just one of these weird things. It’s easy to sometimes just target what you instinctively believe the problem to be rather than apply a logical, step by step diagnosis

Guilty your honour 🙈

Completely dead 2032 button cell battery confirmed by battery tester.

Passed using OC test on digital multimeter. Don't trust O.C voltage in a DMM (without a load).

Sometimes bad news is good news.