I wasn't paying attention on my work bench and ended plugging a mikrotik gige poe into a poe+ af/at port. To my surprise the
cube60 that I was setting up powered on. I had not plugged in the DC psu at that point. I was not expecting af/at to work with
the injector inline. Now I am worried what would happen if I did plug in the DC power since the poe+ port is already powering
the device. Our standard small site setups start with dc injectors then upgrade to a hex poe or crs112 8p (if they didnt start with
one). Often when doing the upgrade from non poe switch to pow switch we just move the pigtails. I am wondering if this hasnt
been the source of a few failed poe ports. I tested with a 10/100 cable (only has 2 pair not 4) and the cube60 still powered up. I
love that both A and B modes are supported. What is the expected results of this non standard configuration? We are making
notes that this should be allowed to happen in the future.

Most Mikrotik devices support some sort of PoE in on (usually) one of RJ45 ports. One really should consult product page (or brochure) to see what kind of PoE is supported (passive, observe allowed voltage range! ... or standard 802.3 af/at) and on which ports.

For example, cube 60g ac supports 802.3 af/at (as well as passive with range of 12V-57V) while cube 60 lite only supports passive PoE with voltage range 12V-28V.

All MT devices which support multiple power inputs (e.g. PoE in and DC jack) can be connected to all of them at the same time (can be used as powering redundancy), the source with highest voltage is then used.

Thanks for the reply, however you did not understand what I said.

I am fully aware of mikrotiks powering options. As I originally said /this was not an intentional configuration and not addressed anyplace/
I did fail to list the device correctly but its able to be figured out from the context. It is a cube60g.

I have 8 ports wired to my work bench that are af/at poe. 8 others that are passive, 2x12v, 4x24v, 2x48v and 8 more that have no power.
Also have a couple gige poe injectors sitting on the bench.

While setting up sets of new cube60g's (vendor did not have any prepaired ones so i need to do it) I plugged a injector into a cube and then to one of my panel ports. That port happened to be one of the af/at ports (which I am fully aware cube60g use). This was not my intention as i thought I had used all the af/at ports.

Now the issue becomes you have an af/at port with a power injector in it, with a cable going to a device. Its linked at gige so all 4 pairs are connected and working. A typo 2 PD would need connected to pairs 3/4 that are used for passive poe. So what is the /expected/ event order when Type 2 is /active/ and someone plugs DC into the injector.

And an update.

Plug in ethernet to injector, injector into switch. Switch powers device as it should.

Plug the DC in first, switch does not bring power up (its not a managed switch so no logs)

Plug DC in after switch has powered device, /both are active/ and feeding power to the device. Switch is /not/ shutting off power. I use a tik on it and voltage went from 51v to 55v. This is notwhat I would expect but it does explain some malfunctioning ports on devices that were upgraded over time from passive 12v to devices with working poe+.

Im going to see if i have any other devices I am willing to risk long term testing of this as well as better voltage checking.

Anyone familiar with the with the circuit diagram for Mikrotik PD devices?

If you plug the PoE injectors male end into an af/at switch, a cable coming from the female end of the injector is connected to the Cube and it powers up, it means that you have a switch providing the power on mode A pairs. The injector is made as mode B, which is why the mode A pairs just goes through the injector untouched.
If you plug the DC cable before you connect the injector to the switch, your external PSU powers on the device and when you connect the injector to the switch, the switch does nothing as the device is already up and it is not possible to perform a "handshake".

Yes I am aware. My point here is that this is not documented anyplace and the full effects are unknown. I need to find my poe breakout kits to do some more testing. I suspect that depending on the end device there may be other interactions. Also I have not tested the non gige injectors and those might do yet something different. Best to just run 2 separated poe kits instead of doing in place upgrades.

If the end device would not support PoE on mode A pairs (some older RouterBoards did not support both mode A and B), nothing would happen as the af/at PSE would not detect anything, If the end device would support mode A but only passive up to 30V (hAP ac2, for example), nothing would happen as the af/at PSE would not detect af/at capable end device. Is there anything else you would like to know?
Yes, this is not documented anywhere, but RBGPoEs intended use is with the male end connected to non-PoE switches, routers, even computers. Also, all of our PSE devices uses mode B and you do not have this kind of behavior with our devices.