Intel NUC DC53427HYE 5 sec blink issue

[TimelessNL]

Hi,

I recently acquired an Intel NUC DC53427HYE. I know this device is almost 9 years old  but for some headless projects it can still be of value.
there is however one big issue with the thing. It does not (unless...) boot, and the only signs of life is a blinking blue LED once every 5 seconds.
This does seem a somewhat common issue as seen here, here, here, here and here. The latter is even from Intel themselves. I did write a response on such a topic to find out if the repair lasted.

To summarize what I did, I heated the board for a couple of seconds directly on the CPU/Chipset dies and when the board was still warm (not hot) I tried to turn it on to see if that made the CPU work which it didn't. I did the same thing on the opposite side around the CPU area and to my surprise the board booted just fine. Normally I think reflow is not very worthwhile since it is most likely only temporally. But what made me curious is that the board works fine to this day (2 days) and survived multiple (soft)reboots. Only when I turn it off completely (with/without removing power adapter) it goes back to its 5sec blink pattern unless I give it a hot-start.
That and since Intel mentions "The power-on circuit may have failed" I hope that the device can be repaired.

Also this all strikes me, and reminds me of this issue EEVblog also once covered.
I attached some pictures of the board from another forum, so they aren't mine. I forgot to take pictures when I had it open last time (currently doing a runtime test)
I could however not find any visual sign of wear/tear on the PCB.

Kind regards (and new member)
TimelessNL

[TimelessNL]

I was also looking around on Ebay just for fun and noticed some are selling these units with the exact same problem description like here

[rob77]

i would try to troubleshoot as any other laptop..NUC is closer to a laptop design than a desktop mobo. check all rails, check the system management controller (most probably that nuvoton micro) .. there must be a chip creating standby 3V3 and 5V - usually a buck controller for 3V a nd 5V rails have a LDO for the stanby voltages... e.g. RT8243 in HP laptops with 4th and 5th gen intel cpus...etc..

[abdulbadii]

see a thermistor the NTC one, might be culprit?

[amyk]

i would try to troubleshoot as any other laptop..NUC is closer to a laptop design than a desktop mobo. check all rails, check the system management controller (most probably that nuvoton micro) .. there must be a chip creating standby 3V3 and 5V - usually a buck controller for 3V a nd 5V rails have a LDO for the stanby voltages... e.g. RT8243 in HP laptops with 4th and 5th gen intel cpus...etc..

This one is closer to a regular desktop mobo. The Nuvoton in the lower right corner isn't an EC, it's an LPC SuperI/O. I can't see the part numbers on the other ICs but nothing stands out as looking like an EC.

[TimelessNL]

Thanks for the replies

check the system management controller (most probably that nuvoton micro)

This one is closer to a regular desktop mobo. The Nuvoton in the lower right corner isn't an EC, it's an LPC SuperI/O. I can't see the part numbers on the other ICs but nothing stands out as looking like an EC.

When the system is booted this chip can be read from i2c, but to me its reading seem fine with the system idling?

Code: [Select]

root@proxmox# sensors
nct6776-isa-0a30
Adapter: ISA adapter
Vcore:         856.00 mV (min =  +0.00 V, max =  +1.74 V)
in1:           904.00 mV (min =  +0.00 V, max =  +0.00 V)  ALARM
AVCC:            3.33 V  (min =  +2.98 V, max =  +3.63 V)
+3.3V:           3.33 V  (min =  +2.98 V, max =  +3.63 V)
in4:             1.06 V  (min =  +0.00 V, max =  +0.00 V)  ALARM
in5:             0.00 V  (min =  +0.00 V, max =  +0.00 V)
in6:           256.00 mV (min =  +0.00 V, max =  +0.00 V)  ALARM
3VSB:            3.30 V  (min =  +2.98 V, max =  +3.63 V)
Vbat:            3.23 V  (min =  +2.70 V, max =  +3.63 V

see a thermistor the NTC one, might be culprit?

Can that be a cause why the system can only be 'warm' booted?
I can be mistaken but shouldn't a NTC affect the inrush current do to it being warmed up, and when pre-heated it should be limiting it even more?

I can't see the part numbers on the other ICs but nothing stands out as looking like an EC.

If it is of any help, I found some other pictures on the internet and also one of my own (that I forgot about)

As a side note: The system is still up and running (6 days)

[TimelessNL]

I got a small update. The NUC has been running fine for the last few weeks so I think I can probably say that the SoC itself is fine.
Today I decided to reboot the NUC which went fine, but after removing the power it was back to its old behavior.

Then I decided to heat the board with a more precise heatgun tip and discovered that the problem lies in the area attached to this post.
I suspect it might be any of the 4 'large' brown capacitors, since if let it cool for a while and only apply heat in specific area it boots.
But maybe you guys have something better in mind? or a way to troubleshoot this any further?

EDIT:
I used some paper to shield some components from the heat and pinpointed it to a cluster of diode's or transistors?

[MK14]

I suspect it might be any of the 4 'large' brown capacitors

Is it just me, or does it look like there use to be 5 of those capacitors, and the remnants of one, very nearby remain (solder pads plus some unidentifiable stuff), which maybe fell off while you were messing/repairing it ?
Or it could be supposed to be empty pads there ?

[TimelessNL]

Can you point me to the missing or remnants? because If I compare the pictures in the OP (which are not my board) they look the same I think.
I also used some paper as a shield to focus the heat and it seems that the 3 transistors/diodes are at fault (see updated attachment of my last post)

As a side note I'm heating the board with 100°c on the lowest possible fan setting.

[MK14]

Can you point me to the missing or remnants? because If I compare the pictures in the OP (which are not my board) they look the same I think.
I also used some paper as a shield to focus the heat and it seems that the 3 transistors/diodes are at fault (see updated attachment of my last post)

I think you were in the middle of editing your post, while I was making mine. You're right, the board which is NOT yours, looks the same. It was worth a try though.

If it was me, I'd be tempted to try and narrow it down (via heating or cooling with ice spray), down to the component level, if practicable.

Pity it doesn't say which component(s) fail, somewhere on the internet, and/or if it is on there, make it easier to find.

[TimelessNL]

Yeah I was a little bit to eager post my new findings and immediately started to troubleshoot further and discovered some more
I found those K2N and 2305 on AliExpress and they are quite cheap, maybe I just replace them and see what happens 

[TimelessNL]

Mmh... I cannot find any component online that specifically has 2305 printed on it.
So I'm unsure what the component is and where to get a replacement.
Any hints?

I also noticed these 2305's can also be found around the power button.

EDIT: I heated one of the legs of the 2305 with a soldering iron and that made it boot, so I'm almost certain its that mosfet/transistor that is at fault
I also measured the voltage on the device:
When boot fails one of the legs remains 3.3v but when attempting to boot briefly goes to 0v (see boot_fails.jpg)
When I heat up the continuous 3.3v leg and the device successfully boots one of the legs remains 3.3v and the other 0v(see running.jpg)

So to me a mosfet capable of 3v3 should be enough?

[rob77]

probably a P-chan mosfet ?

https://www.vishay.com/docs/70833/70833.pdf

probably it's switching 3V3 to some other circuitry... so either the mosfet is bad or the circuitry behind it has a intermittent short.

you can replace the mosfet with any P-chan with with similar current rating, Rdson and gate threshold as the 2305 has.

[TimelessNL]

you can replace the mosfet with any P-chan with with similar current rating, Rdson and gate threshold as the 2305 has.

But that is kind of the problem, I have no idea what specs the 2305 part has.
The Si2305DS you and I linked in this thread does not have the 2305 digit code printed on the part but A5 (or A5SHB) instead.

[amyk]

It might not even be a MOSFET. I would refrain from replacing it until you find a positive match. Reset/supervisory circuits often hav 3 pins too.

Unfortunately Google is absolute shit when it comes to finding part markings now. Whether this is a deliberate attempt at sabotaging repair, you decide...

[Monkeh]

https://www.alldatasheet.net/view_marking.jsp?Searchword=2305

Seems reasonable to me.

[amyk]

https://www.alldatasheet.net/view_marking.jsp?Searchword=2305

Seems reasonable to me.

None of those companies seem like something Intel would use in their mobos... and the only one I could find a picture of, has an extra dot on the mark:

https://www.4project.co.il/product/393967

But the pinout and behaviour does match a PMOS, so putting one in shouldn't cause anything to blow up even if it wasn't.

[rob77]

or you can just desolder it and see if it boots.. and if it boots then you'll see what peripheral is missing

[TimelessNL]

I finally received the new MOSFETs, unfortunately replacing the suspected bad MOSFET had no positive effect. However pre-heating the board made it boot with it installed.
So I soldered the old one back and looked at other components, then I started randomly heating some capacitors with the tip of my soldering iron until I found one that made it boot.

So I removed the capacitor and replaced it with some random SMD capacitors (that I extracted from an old SBC I had laying around) but neither of them worked.
I then tried to boot the board with no capacitor installed (see attachment), and you know what...? it works every time, no issues whatsoever and to my knowledge all peripherals are detected.
So for now I keep the capacitor removed but I have no possible explanation why this even works, reduced capacitance can lead to random issues but that it works without any capacitance my mind.

[petterb]

I have the exact same issue with my DC53427HYE and I also used the same solution(s). I could boot when using a heat gun to heat the capacitor but had problems to put the PCB back in the case quickly enough so I ended up removing the capacitor. Once I removed the capacitor the unit is booting just fine every time, however I see two issues; (1) the fan is running at max speed continuously and (2) the built in Ethernet port does not seem to work anymore. Im suspecting that maybe the BIOS settings related to Cooling could play a part here as I had just changed to the "Quiet" when I started to experience the problems. If I manage to put the capacitor back I could possibly explore this further.

[amyk]

The capacitor may have been partially shorted and pulling down a standby rail voltage.

[SergeiF]

Sorry for resurrection of this topic, but since there was no solution I thought I might add something.
I also have this problem, and applying heat to the area also "fixes" the problem.

I went little bit further and simply putting the soldering iron against K2N(A6) transistor (see the photo attached) allows the machine to be turned on for rest of the session.
Unfortunately, I tried applying heat directly to the bigger transistor next to it and it also has same effect. EDIT: this was one off /EDIT. The heat on the third transistor doesn't affect issue.

I suspect heat is a red herring (at least regarding the transistors) and simply "overrides" the BIOS/CPU by temporarily breaking down the junction and turning the circuit on.

I will continue investigating but without circuit diagram it is going to be very difficult.

[SergeiF]

it looks like that K2N transistor is this: https://www.diodes.com/assets/Datasheets/MMST3904.pdf

An NPN transistor. Measuring it in the circuit confirms that is it.

When in fault state:

E: 0V
B: 0.6V short pulse once in couple of seconds (otherwise 0V)
C: 0V short pulse once in couple of seconds (otherwise 3.2V)

When in on state:

E: 0V
B: 0.6V constant
C: 0V constant.

Bridging C and B with 1kOhm resistor turns on the machine.

[SergeiF]

I have removed the cap next to that transistor and it also fixed the issue.
I measured the cap outside of the circuit and it measured ~6.5uF (6.8uF?).
The capacitor was across the base and emmiter of that K2N transistor.

I have applied 5V across the capacitor and it didn't break down. It also held the charge (measured after disconnecting the supply).

Now, the machine reliably starts without the capacitor and I am tempted to give up like the OP did. I highly suspect there is an issue with a capacitor elsewhere in the circuit, ie: a bootstrap or reset logic is not triggered with correct timing.

[gamalot]

you can replace the mosfet with any P-chan with with similar current rating, Rdson and gate threshold as the 2305 has.

But that is kind of the problem, I have no idea what specs the 2305 part has.
The Si2305DS you and I linked in this thread does not have the 2305 digit code printed on the part but A5 (or A5SHB) instead.

NCE2305

https://www.lcsc.com/product-detail/MOSFET_Wuxi-NCE-Power-Semiconductor-NCE2305_C113493.html