I cannot understand why Mikrotik keep on discontinuing products without suitable replacements. I specifically refer to routerboards that can be used as internal AP's with PoE (802.3af) switches. I also cannot believe that we are the only people that are begging Mikrotik for 48v 802.3af compliant routerboards?

More recently the CRD was discontinued and we were advised that the RB600 is the replacement (at double the price at 3 x the physcial dimensions). Now the RB600 gets replaced by the RB800 which is once again even much more expensive than the RB800 (http://forum.mikrotik.com/viewtopic.php?f=3&t=39984). An 802.3af compliant internal AP is now going to cost us more than 3 times the price than 3 months ago. The fact that it is more powerful is irrelevant for the puposes af an internal AP since a more powerful CPU is not a requirement for this type of device.

This situation makes it very difficult for us to continue to use our favourtive hardware for internal AP's. If you have 5+ AP's connected to a switch, power injectors are simply not practical and makes for a very messy installation and for this reason we connect the AP's to PoE switches. Unfortunately Mikrotik no longer has any Routerboard other than for the RB800 that can be used in this configuration.

Can other forum users that have the same requirements/problems please advise accordingly?

Thanks,

Nico

give these a try: http://ubnt.com/8023af

Thank you I can confim that these covert work well with the 411 and 433 boards we've tested thus far.

Regards,
Nico

Hi All:

I am testing the POE port of the RB800 with a 802.3af switch and I am not able to power it up. The switch is a HP V1910-24G-PoE (365 W) Switch (JE007A) (ex 3COM) and according to the documentation it supports 802.3af mode A only.

The same RB800 can be powered through the LAN1 POE port using with a passive POE splitter with a 48V power supply, but this way we are using the spare pairs (pins 4,5,7,8). The RB800 user manual states that this routerboard supports power over data lines, but in my test seems not to work.

Does anyone knows whether the RB800 works in 802.3af mode A?

Thanks,

Julio

Hi All:

I am testing the POE port of the RB800 with a 802.3af switch and I am not able to power it up. The switch is a HP V1910-24G-PoE (365 W) Switch (JE007A) (ex 3COM) and according to the documentation it supports 802.3af mode A only.

The same RB800 can be powered through the LAN1 POE port using with a passive POE splitter with a 48V power supply, but this way we are using the spare pairs (pins 4,5,7,8). The RB800 user manual states that this routerboard supports power over data lines, but in my test seems not to work.

Does anyone knows whether the RB800 works in 802.3af mode A?

Thanks,

Julio

their datasheet says:

IEEE 802.3af Power over Ethernet (PoE) ready: PWR models can provide up to 15.4 W per port to power standards-compliant IP phones,
wireless LAN access points, Web cameras, and more

15.4W is not enough if you use even one wireless card. I guess if you unplug them all, the RB800 will turn on, but anyway, this switch doesn't give enough power.

these are the power requirements for RB800:

~9W without extension cards,
maximum – 35+ W (25+ W output to extension cards)

Thank you very much for your answer.

I am testing the RB800 without any extension card at all, so 15.4W should be enough. I have also tested the switch with other POE devices like IP phones without any problem.

Could you confirm that the LAN1 port of the RB800 works fine with 802.3af mode A?

Thanks in advance,

Julio

please send your serial number to support. we will investigate.

Dear all:

Could you provide brand/model of switches that work fine with RB800 powered by 802.3af?

Is 802.3at POE+ (up to 30W per port) supported on this routerboard?

Thanks in advance,

Julio

just to let you know, we received your support ticket, and our engineers will do some testing with boards from the same batch as yours.

Hi folks:

I tested five different RB800 with our POE switch and two of them work fine with 802.3af mode A, and the other three don't. The serial numbers of the two good ones end in Fxxxx, while the other three end in 1xxxx or 2xxxx. Is there any specific batch of RB800 not 802.3af compatible?

Thanks,

Julio

We also tested some RB800 devices with 4 different af and at devices, all of them worked fine. Email support with exact test information so we can see what is the problem.

+1

We have some sites with standby backup links with rb800, one radio card only.
CPUs had to be lowered to 400MHz because they where rebooting (due to 'overload' events on the switch).
When those links become active, they keep alive only for some minutes, after that, they reboot (they take longer time than usual to do it).
This behavior has been observed also on two other rb800 acting as AP. CPU has also been reduced.
When traffic rises over 2/4 Mbits on that APs, they reboot.

Thank you. I have already sent the serial numbers to support.

May be our switch is not 100% 802.3af compatible... could you provide brand/models of the switches that were tested OK with RB800?

Thanks in advance,

Julio

give these a try: http://ubnt.com/8023af

Sorry, do NOT try these. We installed a bunch of them with UBNT NSM2s and almost all of them failed after a couple of months. There are loads of complaints on UBNT's forum about this product failing due to heat, but ours were installed underground with a constant 20deg temp. Many, including us, are not using them anymore.

We sent them back to UBNT and they simply sent us another batch of the same product, no hardware changes at all, so the problem is probably still there.

15.4W is not enough if you use even one wireless card. I guess if you unplug them all, the RB800 will turn on, but anyway, this switch doesn't give enough power.

The 802.3af standard sets a limit of 15.4W per port. If the RB800 needs more than this, you shouldn't say it's 802.3af compliant! That is what 802.3at is there for.

And keep in mind that 802.3af/at and "48V passive PoE on spare-pairs" are not the same. We had loads of problems with this on the RB/CRD. 802.3af compliancy means that you can use a mid-span (Mode B, PoE on spare pairs) or a PoE switch (Mode A, PoE on data pairs). The 802.3af standard specifies that PDs (ie. APs, phones, etc) must support both Mode A and Mode B. A device which simply accepts 48V passive PoE on spare pairs might work with a midspan, but that doesn't make it 802.3af compliant.

Try these. I have used them with great success

http://www.streakwave.com/ItemDesc.asp?ic=IPA4524-I01

give these a try: http://ubnt.com/8023af

Sorry, do NOT try these. We installed a bunch of them with UBNT NSM2s and almost all of them failed after a couple of months. There are loads of complaints on UBNT's forum about this product failing due to heat, but ours were installed underground with a constant 20deg temp. Many, including us, are not using them anymore.

We sent them back to UBNT and they simply sent us another batch of the same product, no hardware changes at all, so the problem is probably still there.

I found this thread while searching for info Mikrotik POE support and thought I'd comment on the UBNT passive adapters. I prefer standards base POE especially for management (easy to power cycle) and I hate having a pile of injector bricks at dense installs.

I have a dozen of the outdoor version running for a full year now on tower sites in weather without any failures. I did notice some changes in the batches I have in service, most notably the boot at the top was changed from shrink wrap with adhesive (which did not adhere to the pigtail cable at all) to an injection molded boot. On the old shrink tube version I used rubber tape to seal it up after noticing the output cable would rotate slightly. They also say they made some internal changes with regards to overheating, but without being able to actually open them (impossible to do for the outdoor version without destroying it) there's no way to verify that claim.

One cause of overheat death in my experience is putting a too much cable length on the output side. It seems cable losses can be too much for them and they'll cook, so I always make sure to use no more than a 12" jumper between the output of the indoor version and the device it's powering. I've seen people connect them right at the switch and have 200 foot of cable on the output to the powered device. Then they say they're running too hot and dying, while ones placed alongside the powered device (while warm to the touch) don't seem to have a problem.

So far I've had great results with these powering both MT and UBNT products from 802.3af switches, but I too have read both horror stores and success stories for them. Count mine as a success with the above caveats noted. And hopefully this doesn't curse me.

We were using the indoor ones. They were attached right below a lighting pole, in an underground corridor. Ample cooling, no direct sunlight. Cables were very short: 10-20m switch <> adapter, and 2m adapter <> unit. All of them failed, twice.

When I asked if the new units were any different from the ones we were using, their response was along the lines of "this is what we have received and we don't know if they're different".

Unless UBNT is ready to prove to me that they have identified the problem and corrected it, I'm not deploying these any more, there are other adapters available which do the same thing and have proven to be more reliable. If anyone's interested I'll dig up the info.