i have lot of rb450, rb450g, and rb433,
on all mikrotiks there are connected batterys,
some mikrotiks are working on 12v 45ah, others 24v 14ah.

on every restart they are stopping working, then i need to go on tower unplug poe and then plug it back after that its working.


also after some days they are stopping working, also i need to go on tower unplug jack and plug it back, its terrible becouse i have lot of points and it seems i need to walk all day to restart them.


i have firmware 4.16 (with v5 rc7 it was same)
i have pppoe server on all mikrotiks. and i am using radius for authentication, also i am using ip->firewall mangle and queue with pcq.

please maybe some can help me?
i wrote to mikrotik support but no answer

How long is the cable from PoE to RouterBoard?
I recall I had an issue like this a couple of years ago,
and changing the power supply from 12V to 24V resolved the problem.
I see you are using 24V power supplies, but may be they could not
be enough for the length of the cable or the quality of it.
Just a guess.

I don't know about the rest of the RBs, but the 433 input is 10-28v. If you changed the batteries to 24v, the bulk charge voltage from a standard charger is about 29.5v. If the charger is just float charging them (about 26.5v), they will be fine. It could be the voltage on the battery dropped and the charger went into bulk charge, and the voltage drop on your supply line was not enough to drop it into limits. In this case, the power leads could be too short.

hello
you think this is battery problem?


cable length is less then 15m

I year ago, my college bought 10m of FTP cable to connect a PC with the switch.
It turned out it was faulty after some testing why network was not working.
I think this is such a rare example, that it could not happen again, but who knows.

Mine is just a guess. I understand that you can not even access the log to see what could be
possibly going on, but worth trying.

Not 100% sure, but im pretty sure i have read somewhere that running RB on batteries, doing so using POE is not such a good idea. May be you could try using the power jack on the board rather than the POE?

i have lot of rb450, rb450g, and rb433,
on all mikrotiks there are connected batterys,
some mikrotiks are working on 12v 45ah, others 24v 14ah.

on every restart they are stopping working, then i need to go on tower unplug poe and then plug it back after that its working.


also after some days they are stopping working, also i need to go on tower unplug jack and plug it back, its terrible becouse i have lot of points and it seems i need to walk all day to restart them.


i have firmware 4.16 (with v5 rc7 it was same)
i have pppoe server on all mikrotiks. and i am using radius for authentication, also i am using ip->firewall mangle and queue with pcq.

please maybe some can help me?
i wrote to mikrotik support but no answer

downgrade to 4.11 and problem will solve. there is a bug in the version 4.12 and later . http://forum.mikrotik.com/viewtopic.php ... 71#p247771

Not 100% sure, but im pretty sure i have read somewhere that running RB on batteries, doing so using POE is not such a good idea. May be you could try using the power jack on the board rather than the POE?

I have Routerboards running on batteries using PoE no problems, installations dating back to 2004. So its not POE/Power injector that myth is debunked. Using a passive PoE w/ a DC input voltage is no different than running it on DC power from the power jack or running it off a standard AC -> DC PoE injector.

@shrek777:
To troubleshoot this problem effectively you should probably eliminate the power concerns by plugging these devices into a reliable power source and rebooting to see if you can reproduce the symptoms. Or using a multimeter to ensure the batteries are producing a constant operating voltage with the load connected.

You could try downgrading to the 4.11 release as omidkosari suggested. I've yet to confirm the bug he's reporting and have not heard any confirmation from Mikrotik, but it appears others have had success with this release.

I'd be interested to hear what your findings are once you fix this problem.

This bug appeared in RouterOS v3.x and disappeared in 4.x . but unfortunately in again back in 4.12 an later . i have a lots of support tickets and forum posts which can write here .
unfortunately all of my findings were in production and mikrotik staffs just ignored the problem . now i afraid that this problem comes back in 5.x

I have a similar problem. I have a tower that is solar powered. I have 4 12 volt batteries. the voltage is 29.5 voltes and never discharge. I am using a RB433ah with 4.11 os. It has been working since September fine. today it died. I have 28+ volts coming from the batteries. I tried hooking to ac source with 24 volt poe injector and it works fine.

could it be that 28-29 volts is too much? I am not sure how to cut the battery output down to 24 volts. even if i could it would drain the batteries I would think.

Also I am trying to figure out how to have the RB433 monitor the voltage and show me in winbox or the dude or email me but so far I have not figured out how to program it. the wiki says it can be done but the examples are hard to understand.

Any help would be welcome.

I think anything over 28v is too much. I use 12v solar/battery units without a problem. I don't know what type voltage regulators they use in the RouterBoard units, but some regulators I used in the past will shutdown if over or under voltage.

The voltage monitor script in the wiki is a little difficult to understand, but you should be able to "copy-and-paste" that code into your router. It is a two-script set. Change the email addresses, schedule the main script to run every hour, and you should be set. The most difficult part is getting the router set to send email.
http://wiki.mikrotik.com/wiki/Monitor_i ... B333/433AH
And for help with the email part:
http://wiki.mikrotik.com/wiki/Email_set ... leshooting

Any questions about it, ask.

I don't use Winbox or the Dude, so I do not know the capabilities there.

thanks surfertim I have setup the script and other things like you said. i have test email from RB411 and got it. I will see if voltmonitor works.

i went back out to the tower and hooked up rb433 to 12 battery and it is working for now. I sure wanted the 24 volt to work. It did for 5 months with no problems. just today if failed. NOTE we have sub freezing weather this week but batteries are fully charged.

voltages are 29 on batteries, 1.5 amps and higher charging and .2 amps load (rb433ah). now the load is hooked directly to one battery and not the solar charging circuit.

david

could it be that 28-29 volts is too much? I am not sure how to cut the battery output down to 24 volts. even if i could it would drain the batteries I would think.

If you need to reduce voltage, just install silicon diodes in series between the power source and the load. Each silicon diode has a voltage drop of 0.6 Volts. Choose diodes with a higher current rating than the load requires. This method does not drain batteries or generate any significant heat and is very stable and predictable.

If you are in the northern hemisphere, this is the time of year that overvoltage shutdowns will happen. My battery chargers are all temperature compensated, so the bulk charge voltage is about .2v higher now (cold) than in the summer (hot). But you will start seeing that for yourself when you get the first few seasons of emails from the router. I have hundreds of reports over the last couple years to compare.

ADD: I have used the solution offered by bburley with great success. I used 3 diodes in series to insure no overvoltage. As I recall, I used 1N4001. This option is not exactly free. It involves about a 7% power loss in the diodes. I prefer 11-15v supply with no additional loss.

ADD: I have used the solution offered by bburley with great success. I used 3 diodes in series to insure no overvoltage. As I recall, I used 1N4001. This option is not exactly free. It involves about a 7% power loss in the diodes. I prefer 11-15v supply with no additional loss.

IN4007 is probably easier to find and it is even rated for higher voltage.

I'm not so sure about which way to look at power loss. The load, in order to maintain it's rated load in watts, will draw more current at lower voltages. This still does not neccessarily mean more "power loss". If you measure the voltage and current where the load (and diodes) connect to the battery, and multiply the values together, you will get power used in Watts. Do this with and without the diodes and compare results.

Power loss is a physics thing. It is not just a good idea, it's the law!

It boils down to a voltage drop at a specific current. The formula that applies here.
P = I x E
It is the amount of power dissipated by the device at a specific voltage drop at a specific current.
If E = 0.6 (voltage drop) and I = 1 (current), then P = .6W per diode.

You are correct SurferTim

I have enough experience that I should have known that It has only been 3-1/2 years since I traded electronics for networking and I am already showing signs of looking at things funny

thanks I have added several diodes in series and so far all is well. I am NOT getting any email from RB433ah so I have something set up wrong.

However I have another tower with RB433ah runing 411 OS. it does not show voltage in system health. all my other tower show voltage usually around 23 volts but this one shows 0.0

any ideas?

thanks again

david

I would upgrade to V4.16. Earlier V4.x versions had problems with the voltage reading. I had one version showing 28v all the time. ??

You should have two scripts in "/system script". voltmonitor and voltreport.
You should have voltmonitor scheduled to run once an hour in "/system schedule". You don't schedule voltreport. It is called once a day by the voltmonitor script.

Look in "/system schedule" and insure there is a run count for your script.

ADD: Insure you have NTP client set correctly, or it is always 1970.

the voltmonitor script shows 22 times and the voltreport shows 1
i just set them up yesterday. I have not gotten any email yet but i did force one and received it yesterday when i tested it.

I have read there was a problem is OS above 411 so I have not upgraded yet.

Thanks

david

If the run-counts are correct, then check the email addresses on both scripts. There are two email addresses. I send the low voltage alerts (email address in voltmonitor) to my Blackberry account and the reports (email address in voltreport) to my main business account.

Did you enable the email logging? You should check the log for more details on the email send.

The voltmonitor script only sends the alert emails if the voltage drops below the lowvoltalarm voltage. That variable is in tenths of a volt, so 125 is 12.5 volts.

ADD: Also remember the indicated voltage is going to be (.6v times the number of diodes) less than the battery voltage.