Community discussions

MikroTik App
 
bderleta
just joined
Topic Author
Posts: 8
Joined: Thu May 04, 2023 3:01 pm

RB951Ui-2HnD PoE-out issue (was: Powering mAP lite from RB951Ui-2HnD)

Thu May 04, 2023 3:08 pm

Hello,

This is my first post here, so please move to relevant category if this one is unsuitable.

I'm trying to power up mAP lite connected to ether5 port of RB951Ui-2HnD powered via 24v passive injector on ether1, also tried powering RB951 via DC jack, no luck so far - the mAP lite does not power up. Also tried crossover cable from RB's ether5 to mAP lite. The mAP is correctly working when powered either by microUSB port, or directly from the same 24v injector (ubiquiti one) that also successfully powers the RB951. I tried both PoE auto on / forced on in RB951 ether5 settings. I tried various patchcords otherwise working well and passing conductivity tests, so it's rather not a cable issue.

Both devices have installed latest ROS 7.9 with firmware 7.9. Any idea?
Regards, Bartosz
Last edited by bderleta on Sat May 06, 2023 1:25 pm, edited 1 time in total.
 
bderleta
just joined
Topic Author
Posts: 8
Joined: Thu May 04, 2023 3:01 pm

Re: Powering mAP lite from RB951Ui-2HnD

Sat May 06, 2023 2:52 am

Ok, there is something wrong with PoE-Out on RB951Ui. The DC+ pair of ether5 is floating at ~6V (decreases under slight load) and not driven anywhere by router regardless of ether5 poe setting (off, force on).

First, I followed the overvoltage damage manual and checked:
- terminating diodes - all 75 ohm
- TR102 pins voltage drop - all around 0,32 ohm
- D2 - open loop; D4 - around 3V drop measured in circuit, open loop when soldered out; D1000 - open loop; all those have proper forward drop of ~0,31V; when device is powered, the port side of D4 does not any leak voltage though

Then I powered the device and took following measurements from both sides simultaneously:
- Q1001 - if that's a UT4421 p-mos, then Vgs = -0,385V regardless of PoE setting; bringing its drain to ground enables the port so I assume it's ok
- Q1000 - if that's a n-mos, seems ok, not driven
- Q1004 - if that's a n-mos, seems ok, not driven, can't find what's driving it - only culprit is Q1003, but I can do it forever this way blindly...
- Q1 - Vgs -9,88V, seems ok, powers source of Q1001.

Scanning around Q1004, I probed voltage on U1000's pin 1 and... the poe turned itself on despite I disabled the output in RoS. Further probing caused the voltage output to toggle between on and off, red diode coming back and forth accordingly. Something driven Q1004, but only by around 0,65V. When poe is working, voltage on U1000's pin 1 in reference to ground is around 0,16V, when off - 3,3V.

Long story short, even touching around U1000 makes PoE out (and its switching in ROS) working until next power cycle. IDK it's a glitch or a sign of, for example, damaged totem pole output.

Is my device really damaged? If so, is it beyond repair?
Last edited by bderleta on Wed May 10, 2023 2:02 am, edited 1 time in total.
 
sc2vdp
just joined
Posts: 2
Joined: Tue May 09, 2023 12:49 am

Re: RB951Ui-2HnD PoE-out issue (was: Powering mAP lite from RB951Ui-2HnD)

Tue May 09, 2023 1:09 am

Just registered after 10+ years lurking here and owning several Mikrotik home routers/APs since I suddenly have this same problem with PoE out but with a hEX S instead.

This hEX S (ROS 7.9 and boot loader firmware 7.9) is unable to power either one of hEX, an RB750GS or an wAP AC when it's powered from the DC jack with the stock power supply (24V, 1200mA).
PoE out has been working perfectly fine with the wAP AC for a couple years straight with the same stock PSU. I stopped using when I put away the wAP AC but now I wanted to set it up again I am stumped with this. Tried netinstall-ing it to 7.9 but same behaviour.

I didn't mess and probe the board's electronic components like the user above but I did measure the DC+ pin voltage of a plugged in cable (cable works, it powers either of the above devices with an injector) and it would seem to be stuck on ~10V regardless of the PoE setting (forced on or off). The red LED on port 5 never turns on but the port works otherwise fine.

It can be of course board damage but my gut feeling (wishful thinking?) points more to a bug in some ROS update/firmware within the last months.

Anyone else finding the same? Worth contacting official support or through here? Device is well out of warranty (~2-3 years old)...
 
sc2vdp
just joined
Posts: 2
Joined: Tue May 09, 2023 12:49 am

Re: RB951Ui-2HnD PoE-out issue (was: Powering mAP lite from RB951Ui-2HnD)

Tue May 09, 2023 12:47 pm

Turns out this was fully correct and "solved" my problem with the hEX S. Leaving the device unpowered for 10mins brought the PoE out back to life...
Had same problem...
Spend a week on it (it was a bit more annoying)
If a power cycle does not help.

Solution:
  • make all the good settings (forced on etc)
  • connect your poe client
  • shutdown the router (maybe a reboot before )
  • unplug the router
  • drink a coffee/tea etc
  • plug the router
  • enjoy
Still Mikrotik peoples, improper shutdown should not do something like these.... Because even if possible at UPS shutdown signal to be notified the router to shutdown, when power on network should be the first up.
viewtopic.php?p=796256#p802988
 
bderleta
just joined
Topic Author
Posts: 8
Joined: Thu May 04, 2023 3:01 pm

Re: RB951Ui-2HnD PoE-out issue (was: Powering mAP lite from RB951Ui-2HnD)

Tue May 09, 2023 5:46 pm

Replaced U1000 with TI's LM339PWR, the device behavior is the same - once powered, PoE out don't work, unless touched U1000's pin 1 with anything, or some resistors (in the R/C set of 8 on the right from U1000), touching another one from the set breaks it back. I suspect some of the transistors, though I have almost no experience, can I ask for a bit of help on it?
 
rolling
just joined
Posts: 16
Joined: Wed Mar 11, 2020 5:15 pm

Re: RB951Ui-2HnD PoE-out issue (was: Powering mAP lite from RB951Ui-2HnD)

Tue May 09, 2023 8:59 pm

Hi,

try with another power supply more powerfull.

regards.
 
bderleta
just joined
Topic Author
Posts: 8
Joined: Thu May 04, 2023 3:01 pm

Re: RB951Ui-2HnD PoE-out issue (was: Powering mAP lite from RB951Ui-2HnD)

Wed May 10, 2023 2:01 am

Thanks for response.
Earlier I tried powering it from Ubiquiti 1 amp 24v passive PoE injector and 800mA DC power adaptor bundled with the device.
I just tested with lab power supply set to 5A limit - device powered up properly consuming about 75 mA, but the PoE behavior is the same, broken unless touched around the comparator.
A fragment of service manual for the device covering that portion (how the PoE out is driven), or a bit of insight from Mikrotik team would make things a lot clearer, potentially save this device from bin and make me very grateful.
 
bderleta
just joined
Topic Author
Posts: 8
Joined: Thu May 04, 2023 3:01 pm

Re: RB951Ui-2HnD PoE-out issue (was: Powering mAP lite from RB951Ui-2HnD)

Mon May 15, 2023 3:43 am

Turns out this was fully correct and "solved" my problem with the hEX S. Leaving the device unpowered for 10mins brought the PoE out back to life...
Glad it worked for you. Unfortunately, in my case, did not help at all - I strictly followed the procedure and PoE out is still dead. Even downgraded to 6.48 long-term, nothing changed.

I'm giving up, if I'd need a working PoE out I would just solder an external switch grounding the final pmos gate forcefully.
Well... I like their toys, and ROS, but regarding PoE stuff, even TP-Link is more reliable and compatible.
 
bderleta
just joined
Topic Author
Posts: 8
Joined: Thu May 04, 2023 3:01 pm

Re: RB951Ui-2HnD PoE-out issue (was: Powering mAP lite from RB951Ui-2HnD)

Tue Nov 28, 2023 4:28 am

I gave up with repairs for a while - with no documentation it is usually unfruitful. But today I needed it for some quick tests in the field and I gave it another chance. Of course, the device did not repaired itself. I measured again all the transistors involved in PoE out switching and the interesting behavior repeated - when touched either 1OUT or 1IN+ of U1000 (LM339), PoE out started to be controllable until power down. I replaced carefully the chip again, paying attention not to overheat it nor touch with bare hands, applied a lot of flux for desoldering so several elements moved a bit and maybe cold joints were reflown by the way. And... it works now every time I power on the device. Wow.

So, the conclusion is - either LM339 is so fragile, or elements around (mostly the divider block at the right from U1000 was heated) had some cold joint(s).

Image

...and the mAP lite is happily powered :))

Image

Who is online

Users browsing this forum: No registered users and 28 guests