Hello everyone, I am new to the world of Mikrotik devices, but I hope to get some guidance or help from the community in my case.
Let me start by saying that I got second-hand a little-used Mikrotik HEX S (RB760iGS). The problem I have with it is that after powering it up, the buzzer gently “whirs”, the chip (UT4421G) marked with a red square heats up a lot. This is not constantly, but happens 3-4 times in a row, after which the device is able to start up normally a few times and then there are again the problems I described.
I have noticed that if the Mikrotik is in a problematic state, just touch the area marked with a yellow rectangle with your finger and the device starts normally, just.
Unfortunately, I don’t have access to a PoE power supply, but in my opinion, after tracing the paths, it wouldn’t do anything, because the problem is deeper in the power section. I have tried many power supplies (from 12V to 36V with different current capacities), unfortunately with no results. I also soldered out the capacitors (100uF 63V and 680uF 6.3V), but measurements show that they are in good shape. I am not a trained electronics technician, but I can handle a soldering iron and have access to a soldering station, so replacing components would not be a problem.
I wonder if the problem is at all on the hardware side of the device? Or is the problem somewhere in the software/bootloader?
For more information, I will be happy to share.
Of course if this is the wrong section then please move it, as long as not to the trash
Hi! I got 2 HEXs boxes I use one as a switch and another as a Dude host. I had no stability issues with them for ages.
So if I were you, an I would have lot of time to spend, I would check AliExpress for that power MosFET IC (UT4421G) and also for the capacitors. Order them and replace them. They come cheap and if you are not unlucky, they are mostly decent quality.
Then you should do a fresh netinstall.
Regards
Edit: take your time and roughly measure all the resistors around that IC as well, replace if measured values don´t align with marked values. Like they show a short or a break if they shouldnt. (Remember: there are fuses and 0ohm resistors as well, so check markings…)
If when you touch the area (i.e. you exert some pressure on some components) the thingy works it could also mean a (invisible or almost invisible) cold solder point.
A failed component is of course possible, but it shouldn’t change behaviour if you touch it.
You could try to see how the board behaves when freezed (zip bag in refrigerator for one hour or so) and when heated (hair drier level of heat), the hex in theory can work -40°C to 60°C.
Re-tinning the solderings on those components in the yellow rectangle is however right on the border of what I personally consider too tiny to be possible without some good equipment and quite a bit of experience, the smaller resistors or fuses are smaller than the size of the capacitors pads.
The main issue if you need to replace some components is to identify them and source them in small quantities, the UT4421G you can probably find in lots of 10, the other ones might easily be available in lots of 50 or more.
Even if they seem fine, now that you have desoldered them, do replace the capacitors, it is always a good thing to do.
If it is “for fun” it is OK, of course, otherwise it is simply not worth the time or money, three or four lots of components at (say) 5-7$/€ each makes easily 20 $/€, sure you can then fix some other 10 hexs with the same defect, but it is improbable that you will ever need them again.
Thank you for your help and guidance. Tomorrow I will start working on my board, I will also start by looking for components to replace. This is a learning opportunity for me, so I am eager to take on the challenge, so keep your fingers crossed.
I will follow your instructions and I will let you know about the progress
Sorry for my long reply, but I was need some time to order components and resolder them. And over 3 weeks of testing it’s looking good for my Hex S, I was never experienced any symptoms I descripted in #1 post.
Components that I replaced:
100uF; 63VDC and 680uF; 10VDC THT capacitors
BZX84B10 Diode Zener
AO4421 unipolar MOSFET
Estimated cost is about $15 for the components and shipping
Great, some unnecessary CO2 emissions avoided!
Btw. if you plan on further repairs, get yourself a cheap Chinese rework station (for hot air soldering), then resoldering items gets much easier.
Did you identify which of the components you replaced was actually faulty?
I’ve got a Hap AC3 that’s not powering on, the adapter is OK (24V). When I checked the board with a thermal camera and powered it on - the only component that was heating up (38-44C) is the UT4421G. The rest of the board was “cold”.
I’ve ordered a couple but it’ll take a week or so..
That UT4421G is a power MosFET, which can dissipate 2W, so it can get hot under normal conditions. So that´s not necessarily the component you want to replace. Check the components around for shorts/breaks & burn marks. Check also the electrolytic capacitors for leaks, burns or rupture, but they are your primary concerns. Replace all suspicious passive components, before you replace the active components. When switching on first after replacements, keep the thermal camera on, to find out if a component get´s quickly hot and destroys some other component, cutting power to itself.
I don´t know, as we have no component level docs and I also don´t want to start figuring out by disassembling mine, but that MosFET might be a part of a power supply. So if some part after it has failed, the rest of the circuit might just not get supply.
Idle or under load? Marginal power adapter might output close to 24V when idle but drop voltage under load. And failing capacitors also mean very uneven output voltage which isn’t shown by normal voltmeters, oscilloscope does OTOH. The uneven supply voltage can disrupt device …
I think UT4421G is definitely the culprit… took your suggestion about testing with a different adapter - 12V 2A.
UT4421G lit up like a Christmas tree on the thermal cam - 178 C. No other components nearby showed any indication of heating.
