I have bought several units of RBFTC11, and they will be powered by gigabit 802.3at POE. According to documentation, one should use an “unshielded cross cable”.
My question is about the wiring diagram of the crossover cable, i.e. either Half-crossed (two pairs crossed: 1,2, 3 and 6, two pairs uncrossed: 4, 5, 7 and 8, for 10BASE-T or 100BASE-TX crossover) or a Fully-crossed (all pairs crossed, the only crossover for 1G but also working for 10M and 100M ethernet).
Maybe anyone has experience in powering this device with gigabit POE 802.3at, and can share which wiring diagram is working.
According to the technical support, it might depend on the switch powering the RBFTC11, and even a straight cable might work. But I would prefer not to test too much with powering, and go to the assurance of experience. Given that it will work at gigabit speeds, I would assume that I should use a Fully-crossed diagram, but it is not clear and even technical support seems to suggest Half-crossed (usually for 100Mbps).
I have successfully powered the RBFTC11 with a straight cable (not even a crossover cable needed) with a 802.3at POE+ mode A (power in pins 1, 2, 3 and 6) switch.
But what did you understand?
We mean the ethernet cables that have the pairs twisted on themselves,
not the “cross cables” to connect two PCs to each other, which by the way now with the auto-detection in the chips there is no longer any need… ..
Product page says PoE in range is 12-57V … and seeing range of input voltage already means that device supports passive PoE in (802.3 af/at is always 48V give or take). Product brochure (available in Support&Downloads tab) agrees with my assessment.
hAP ax2 comes with 24V power adapter (which is fine with RBFTC11) and can output up to 0.6A via PoE out port (at 24V that translates to 14.4W which is much more than needed by RBFTC11).
I purchased two RBFTC11s and neither would power from any of my 802.3af/at compliant POE switches (TP-Link TL-SG1005P and TP-Link TL-SG1210P). These switches work fine with other POE devices. The RBFTC11s powered up just fine using the supplied 24V passive injectors and wall plug.
I feel like these switches are not truly 802.3af/at compliant – they don’t seem to be doing the handshake to request POE power. I found multiple threads of others with the same issue:
Following up on this: Yes, it is poorly written. I reached out to Mikrotik and they explicitly suggested that I should use a crossover cable.
Nobody thinks UTP means cross, or that cross cable means STP/UTP ever. Twist is twist, cross cable is either a cross connect cable (which doesn’t really apply here) or a crossover cable (which appears to apply here) as per Support’s response:
Hello,
Please try to use a crossover cable to power the device.
Best regards,
The thing that is odd to me is that crossover cables have been out of use for quite a while since Gigabit handles that negotiation automatically. I won’t be able to test it for a few weeks, but I have made a pair of cables to be ready (since I certainly don’t have any now). So it must somehow be nonstandard if it requires a crossover, and that’s kind of unfortunate as it means you have to be very specific in your cabling and tracking your ports (it shouldn’t matter later, but who knows).
The FTC21 appears to be a 48v native device (and offers some more functions if you need it).
I guess that everybody is puzzled as to how the “crossover” cable could possibly work in this scenario. The thing is the following: in normal cable, 100Mbps data is using pins 1+2 and 3+6 … connected straight one-to-one. In crossed cable, 100Mbps data is using the same pins but pins are connected crossed: 1+2 on one side with 3+6 on other side (and vice versa). This crossing is handled by “auto MDI/MDIX” function. The rest of pins (4+5 and 7+8) are used for 1Gbps and are never crossed in UTP cable.
Now for 802.3 af Alternative A, pins 1+2 and 3+6 are used for power. And for Alternative B, pins 4+5 and 7+8 are used for power. So when connecting PSE and PD, which insist on different Alternatives, crossed cable should cross 1+2 to (e.g.) 4+5 and 3+6 to 7+8. But his kind of crossing doesn’t work for data. And “normal” crossed cable (crossing 1+2 to 3+6 and 3+6 to 1+2) doesn’t allow mixing both PoE alternatives, it only changes polarity when using Alternative A.
So it’s a mystery on how crossover cable is supposed to work in PoE … or if it’s actually “normal crossed UTP cable” at all … if it’s not, then MT should document it much better.
It seems as 48V native as FTC11 … both support passive PoE as well (FTC11 goes lower with minimum voltage). It’s only that FTC21 seems to properly support 802.3 af/at (possibly both Alternatives as a decent PD is supposed to).
An 802.3af/at/bt compliant PD must accept both Mode-A and Mode-B with either polarity, i.e. 1&2(+) with 3&6(-), 1&2(-) with 3&6(+), 4&5(+) with 7&8(-), 4&5(-) with 7&8(+). Additionally the power supplied over the network cable must be galvanically isolated from any other connections on the device, so devices which have no other copper connections, e.g. wireless access points or fibre media converters, this could be dispensed with.
Assuming a device already accepts the most common passive PoE, Mode-B 4&5(+) with 7&8(-), and can tolerate upto 57V, then some minimal additional circuitry which presents a suitable signature (19–26.5 k) and does not draw any power until the supply voltage is >20.5V would be recognised as 802.3af Class 0.
I would suspect that the Mikrotik PD implementation does not comply with the 802.3 standards, and their design only accepts Mode-A of a specific polarity. Using an ethernet crossover cable which swaps 1&2 with 3&6 allows the device to work with Mode-A of the other polarity, there was some documentation of this on the old wiki https://wiki.mikrotik.com/Manual:TOC/MikroTik_POE_in_compatibility_table.
Mode-A is the most common provided by PSE, likely for historic reasons so you only required two pairs for power and data to fast ethernet devices. Mode-B 802.3 PSE are unlikely to be able to negotiate correctly with a device which also accepts passive PoE voltages below 20.5V as this would confuse the signature detection.
@tdw
Yes, that sounds like the only possible explanation
As mentioned in the other thread the RBFTC11 comes with the RBGPOE: https://mikrotik.com/product/RBGPOE
which is declared as mode B (4,5+ and 7,8-).
A crossover cable would not cross 4,5 and 7,8, only 1,2 and 3,6 so it can convert only from “usual” mode A (1,2+/3,6- ) to mode A “reverse” polarity (1,2-/3,6+ ) or viceversa.
The only explanation could be that the RBFTC11 can be powered BOTH by mode B and ONLY one of the modes A, but which one?, the article from the wiki you just posted seems to imply that the “normal” one is correct (1,2+/3,6- ) as the crossover cable is required for those devices that have the “reverse” one (1,2-)(3,6+), but some devices such as the map lite and cap lite seem like reversed.
It is entirely possible that - for whatever reasons the RBFTC11 behaves like the map lite and cap lite.
Surely the good Mikrotik guys could have documented this properly.