Hi check out what I found in the dumpster. It's a Synology Ds4 1/5 plus 4 by Nass drive-up This is absolutely fantastic I've never found well I've found a nice in the dumpster before. was like some old crusty think but this one still has the protective film on the front and it looks like it hasn't been used at all. There's absolutely no dust inside the fans in there.

It looks like it's virtually never been what not used for any extended period of time at all. Unfortunately, it didn't come with the power adapter and it does need a hundred watt 4 pin job' down here and they actually run about at least 80 bucks something. But anyway I've already got a Ds4 1/8 The power supply is the same so I can actually power this thing up so it is an older model I believe Think it's discontinued now, but it's worth like seven. You can still buy it for like seven.

Eight hundred Australian bucks. Absolutely brilliant. What's the catch? Well, I posted and tweeted a photo of this. The thing you got to follow me on Twitter because that's where I post all my dumpster photos and find some that a lot of them don't even make it to a video like this one.

And ordinarily I wouldn't have done a video on this, but it was. It's almost too good to be true because as somebody in Twitter Thank You very much pointed out that there's a famous No. An issue with this is not just our Synology drawers, but actually the Intel Atom c2000 processor used in this thing. There's a famous bug in this and we'll take a look at the details of that that actually caused this thing to fail.

And ironically based on my last video, it was found in a shopping trolley. So yeah, if you haven't watched that video, definitely check it out. Oh and by the way, no, it doesn't come with any drives, but it has all the holders in it and actually seriously, it looks like these things that never been used. Anyway, that's this power this sucker up with the My Other Power Boy paradigm and unfortunately it does have this noan problem with this Ds4 1/5 Plus which runs the Atom C 2 a variant of the Atom C mm processor and the the blue power light here is not supposed to be blinking like that, is supposed to be solid and all the drive lights are not supposed to be solid orange like this.

So it looks like it has this fault in this bug that causes degradation apparently on the atom CPU in this thing. But rather than just toss this thing out, that would have a go because there is actually a fix for these things. By the way, Yes, I have put a 1 hard drive in there and it makes are no different and that is likely why they toss this thing out. Maybe they had it run in for a little while and it failed because as I said like, it's very clean.

So this is an indication that it's just not booting up at all. It's not going through the boot processor and eventually if I just leave it for a while, it will actually just switch off. By the way, to get this thing open is a real bugger. These two plastic clips in here.

this metal frame actually goes under that and you've got to get your fingers up under there and like pull it out. and but once you do, it does actually come apart like that. but you get it's not an obvious and will inside. not a huge amount on this side.

I Like the extender board for all the SATA connectors SATA and power of course and that just goes into a focus you bastard that just plugs into a right angle connector on the mainboard and Bob's your uncle so that's really neat and tidy. Then you've got to just an I/o board at the back for your Ethernet and your USB Z's but the main board. it's in here. That should do it for those playing along at home.

There's your backplane board and I Love using now PCI connectors for border board interconnects. I've used them for you know, test interfaces and all sorts of things over the years. They're just cheap and got tons of contacts and they're really handy and reliable that whole boards going to lift out now except for some shielding tape on there, but puffing that today we're in like Flynn There you go, There's our main Synology processor board that's going to be our culprit, the Atom processor. And once again, this is not arson Ologies fault.

it's a systemic fault in the die of the in the design inherently in the design of the Atom c2000 processor and all variants of it. So hundreds of manufacturers would have been bitten by this until C 2000 bug. unfortunately. anyway, because this is otherwise a very nice raid rival a lot of people get a Synology crap should have left in the dumpster.

Whatever. this looks like it's going to be fairly reliable. And these electrodes here? these are all polymer ones probably put the kettle on. You can tell they're not a wet electrolyte type because they don't have your typical cross arrangement pressure vent in the top.

all they had to have at an alum. ID two more there okay these days, but you know they've got a reputation model number for those playing along at home. and copyright 2014 Anyway, interesting that they got the firmware ramaa flash up on ROM old school up on the little daughter board there. They do that for reasons of production efficiency.

perhaps because these things aren't particularly cheap. so they're not shaving off every dollar of these things they aren't sold into. You know, they're really high volume consumer market. They're more business oriented.

You know, like 700 bucks retail for this thing or whatever? 800 bucks? Aussie Yeah, Yeah, you can afford to do that little sort of thing at Mei because you can have a mass programmer for example, may be faster to actually program these off board in a big custom-designed gang programmer than it is to you know, have it. do it via the USB port or whatever you know, Ethernet port or whatever. So yeah, yeah, 2014 date code: 5a plus years old. But anyway, it's like a bought one.

So according to the interwebs, the fix is to put a 100 Ohm resistor between pin 1 and pin 6 there. and that's it. apparently. Um, so yeah.

I'm curious to know where what that's actually doing and where that goes. But yeah, I don't like the odds of being able to trace this unless we have a schematic because you know this is like a 8 layer board, 10 layer board. something like that. Pin 6 here.

this actually buggers off over to here and there's a zero Ohm jumper. our three five six. they're not sure where that goes Also goes to this unpopulated resistor and it buggers off. under here, drops down a via and I think it goes off somewhere over under the to the processor.

But yeah, I'm not going to follow the money too much so it's not a huge amount we can actually glean from looking at that. Unfortunately, unless we really spend a long time at it, couldn't be bothered. and unfortunately Googling doesn't find any info on the pinout for J-3 what that is or whatever. And of course no schematics.

but if you do happen to have that information, please leave it in the comments down below. Well, that's not pretty is it? But Oh I've actually run out of 100 Ohm axial resistors. Do you believe it? There was nothing left in my draw. So I had to put 251 Ohms in series.

So we've got a hundred and two Ohms or gosh darn it. Well, it's not gonna work now. Anyway, let's put this back together and a pair it up and blink 'ti blink. but ah ha.

look at this. the orange lights aren't on so I've installed one hard drive just to make it do something. but it certainly changed. so I'm gonna leave that for a bit.

Aha, look at this.look status blinking orange light disk one green, solid blue so it just took like a minute to boot there. and winner winner chicken dinner that has fixed it. Likely to work as a lot of people have done this modification and they said yep, but got it back up and running after it died. But there is something about apparently degradation of the chip perhaps but that Synology have said they actually extended the warranty product and others that use the C Mm chipset and in full knowledge that hey, these things they reckon that's their engineering analysis anyway is that these you know these chips don't die well at least not for the warranty are term that they extended it for anyways so we know we're no chicken dinner.

There it is. This station a Founders D is for 1/5 plus so we can connect into that end-user License Agreement year blah blah and we should be in like Flynn and that is local address one my two, One Six eight, twenty thirty. your mileage may vary set up. There you go.

So yeah, it knows that we're booting this for the first time installed the latest DSM for security fixes dis station manager. Yet we can install that now so all data and hard it will be removed to Linda's station manager. Ten minutes come back. but anyway, it's it's working, it's it's fine.

There's nothing wrong with it. So I expect you. you put in all for this and you set them up in whatever configuration you want. Bob's your uncle and there you have it.

We are good to go system. Health good. Everything's hunky-dory Uptime Three minutes. it's lasted three minutes.

It's gonna last three years right? Hmm I don't know. No workers and we've got the one use drive. So now if you want to use it, you'd put in your however configure you drives however you want. It's fine.

So practically a brand-new Ds4 one Five Four drive NASA in the dumpster. Well, it was faulty because it did have this C 2000 Intel Atom bug. But it's fixed and it's raring to go. Should I Trust it.

Okay, so what I'm gonna do is just probe the signal here on pin 6. pin 2 is actually ground. I'm using my low inductance probe here because we've actually got as you'll see a 25 Meg signal. Look at that and that's a bit.

How are you doing? That's not my probe doing all that funny business in there. that is the actual signal. So that's that's the clock. They have it.

It's pretty poor stuff and yeah, my probe is compensated. Everything's hunky-dory so it looks like they are potentially like shorting out two clocks here. and wow, that's got to be over the low threshold at that point. So that's like third, like your CMOS levels I Don't really like that.

sort of like porch so so to speak in there. that's like a classic when you have like two signals shorter together, that's not good at all. This is after the mod of course. Apparently the output of the clock fails or deteriorates on the chip and that's the actual are silicon fault, it's a failure mode apparently.

the output driver sort of maybe deteriorates over time. juda whatever process you know they're they're using on the die there and so that's a really awful clock. I Don't like that at all, but it's working and pin number one that looks like it is just the rail I Couldn't measure that before but it looks like yep, one, two, three point, three volts. so it's pretty solid.

but you can see some noise on that and that's just the noise superimposed on that so you know that's perfectly fine. Nothing wrong with that, Sorry, they're just. basically it's a low impedance pull up a hundred ohm pull-up resistor to the to the rail on that clock. So it's a bit how you're doing, but it does the job in it's better than like a baby.

Better solution than like having to run mod wires and cut tracks and all that sort of jazz. Let's just check that. I have pulled the resistor. So this is the original failed configuration and oh there it is.

Yeah, look at that. Let me capture that. So that's actually the same frequency. so this is really fascinating.

It's like it's about ground there. You can see that the ground is like in the center of that waveform so it's almost as if like this is like an input that's picking up noise like it. or it's a high impedance. you know, tri-state output or something like that.

Or it's an output driver that has failed. which I Think some people have speculated for this thing o'clock output because it is coming from that hundred ohm resistor. So but why Hour 25? Why it comes good when we simply pull that up to 3.3 volts. I Don't know if that clock is coming directly from the chip and simply pulling it up hard, pulling it up with a hundred ohms solves the problem.

then. ah man, that's dodgy as ah, but like I said, it works. But this, if anyone knows the exact details of, you know the failure mechanism of here. And if we can, hopefully get the UH schematic.

it's rather puzzling. Anyway, when you pull it up it, the clock comes good. When you have no pull up, it's bad. It's not like it was originally floating, it's failed.

I Mean you know this was a working design and it just fails because there's no one problem on the chip. So the output driver is failing. But you pull it high, hard, pull it high and it works. Okay, so yeah, apparently this was a big deal in February 2017 in Intel Atom C 2000 bug is killing products from multiple manufacturers.

As I said, it's nothing to do with Synology Everyone was I impacted and the problem is the SOC which is the system-on-chip the actual C mm processor, the LPC clock out zero and or LPC clock out one pins lobe. That's part of the low pin count a bus. You'd have to go into the topology of the C mm processor to know what that is. I'm sure you know Intel designers out there now I'm talking about Anyway, it may stop.

function in. it just may stop functioning. Um, if the LPC clocks stopped functioning, the system will no longer be able to boot. It depends on how the system is are designed, of course some I'm sure some systems weren't actually affected anyway.

Um, that was a dead Indian didn't tell yous MVAs to squash the reporting of this anyway. I I Have no idea. Anyway, a week later, a Synology announced product status update and the Four One Five Plus was one of the. as of today, all of our products used This component are performing in line with Synology X' quality standards.

Bla bla bla Testament to our confidence in the reliability. We're extending the warranty on the products utilizing this specific component by an additional year, so that's gonna last at least a year according to Synology now. and Anandtech have a great write-up on this. I'll link all all these down below for those playing along at home.

I Ryan Smith Great write-up here. This was February 8th 2017. It's not a new thing, but anyway, just discover things in the dumpster don't I and it's A. We won't go into huge details, but we'll actually search the data sheet for the LPC clock out pin because it generates a well in this case 24 measured 25 megahertz signal that then can power other stuff inside your product.

Some you know boot ROM and all sorts. You know boot functionality and all sorts of things like that. Here you go, it tells you. along with legacy I/o devices, second most common device to hang off, the LPC is the boot ROM bias owing to the fact that is a simple device that needs a little bandwidth and this is where the C mm floor truly rears its.

I'm gonna add ugly head is that Yes, in this particular case it looks like that's how the design of the Sumela genus is happening. They're using the this low pin count bus to interface with the external boot ROM and well by doing that and of tons of other manufacturers doing that. Um, it's come a gutter and they've failed unfortunately due to some sort of degradation of the silicon. It's weird, but it does happen here.

It is early circuit degradation and this is actually from the Intel errata from because they have to release a router and this is quite common. Go look at any complex microcontroller or CPU from any manufacturer and you'll find often pages of things that simply just don't work. Um and they they just advertise this feature and they go oh sorry no serial port - no that doesn't work sorry in these versions of silicon. sorry now bad and here it is.

Here's the specification. Update here and we can have a look: February 12 It's had lots of changes over the years but the one we're interested in is system may experience inability to boot or may cease operation. You think the system may experience inability to boot la blah blah VIPs that will stop functioning and no longer be able to boot and they don't offer things why? And then you can go down to the summary table to see which are steppings of the silicon are affected. Number Nine down here.

look B 0 and C 0 steppings of the diet. No fix where we just sorry don't have a fix for all these things. All these problems sorry we just don't have a fix and this does not just an Intel thing. It happens to practically every manufacturer of these modern.

The processors and micros are so complex and they copy them from other designs and copy elements from it. And yeah, they can simulate until the cows come home and they can test their dies until the cows come home. But until it gets out there and designers use it, do we find and report bugs like oh I think I've mentioned this before I Used to pick 24 F-series chip once and I couldn't figure out why my serial port wasn't walking and working and sure enough I checked the latest errata for the chip and buried right down the bottom like this it you know it said something along the lines of us sorry the serial port to just doesn't work and on serial port one we swapped the pins we got they we goofed that up so you know it was like no fix and I here's a fix and then I had to put my chopped up my circuit put mod wise in to make the damn serial port work again. Anyway, yeah, these things happen.

And of course if you go over to the Eevblog forum it's got everything Evie plug. If you're not on the Eevblog forum. seriously it's the world's best electronics forum. Absolutely like everything's on there.

And yeah sure enough Intel Atom see 2005 is akin now 2017 based on posted this and and all sorts of issues and somebody talked about. Sure enough the Synology D is four and five box died and needed to do the work around. So yeah, alright so if go over to the darkness over here and we search for LPC clock out Zero. There it is and frequency are twenty five megahertz.

It looks like it is a fixed twenty five megahertz. I provided two devices requiring LPC clock. So it's just a internal clock generated from the main processor clock. Whether or not it's it's phase synchrony, it is internal, but whether or not that's important so you could potentially get around this fault.

even if this pull-up didn't work by putting in like a having a little mod board with your own twenty five megahertz oscillator and then pairing and then using that to power whatever boot circuitry, roms, or whatever legacy bus devices you've got powered on there. But as I said, the synchronization could be a problem so. but you could also have clock sync as well. so it's possible even if this pull-up might a resistor fix, didn't do it.

so, this could still potentially be fixed in products. If you haven't had enough incentive to do so, you could design a little mod board I've done videos on designing mod boards and they can. You can potentially have that with a little oscillator and some clock sync circuitry I would imagine. and then LPC host signals say no.

33 megahertz huh? Okay, so it might be the LPC serves as a PCI to is a bridge. oops sorry, my screen capture was way off there. Oh no dumbass. Dave Okay, yeah.

professional. YouTube I'm sure and it can support up to two loads here. It's a 3.3 signal so you know it's obviously just like a standing totem pole driver output: Y This particular one is affected I Don't know. You'd have to talk to one of the you know silicon designers and to know exactly what's going on there? There you go.

that's interesting, fix and down. What should I do with this should I actually go out and spend 80 bucks on a power adapter and then also I gotta fill it up with hard drives. Geez yeah! I think to fill it up with to match my other one I'd have to fill this up with like eight nine hundred dollars worth of hard drives I've got four six key Nass hard drives I think there are Western Digital Reds or something you know to fill that up and get a duplicate mirror drive I could have them in different locations and back it up. It could potentially be a cheaper sub backup solution than what I use at the moment.

I actually backup my entire raid drive to Backblaze which is an online cloud service provider and so all up there so you know and that's actually quite expensive to backup that amount of data. So anyway, should I trust this thing or not, please leave it in the comments down below. If you've got any experience or seen any other stories about people apply this fix and well they've come and guts. are you know, six or twelve months later? So I don't know.

it's a bit dodgy. But anyway, and if you know if Synology actually do this fix or did this fix because they I think they still sold this unit after they knew about this problem. So and apparently they would. Yes, so you were a good version of it, a fixed version.

So anyway, let us know what you think. If you enjoyed that video, please give it a big thumbs up and as always, discuss it down below and over on the forum and subscribe to my library channel. I'm by the time this video goes up, I'm probably going to hit 10,000 subscribers. It's insane.

Catch you next time.