I’m currently running my RB3011 from 24V, because I’ve always thought it needed 14V minimum from the user manual.
However, the web page at MikroTik for it says 10V minimum. It would be handy to be able to run from 12V as I already have a common 12V supply in the rack, but I don’t want to redo all the wiring only to find it isn’t reliable because it really does need >14V.
Anyone know which document is correct?
Most mikrotik devices will run reliably at the datasheet voltage, when fed via barrel/jack P4 connector
Some devices require slightly higher voltage when powered via “Poe-In”, generally ether-1, due to protection diodes (af/at poe requires a diode bridge, which adds a minimum of 1.4v drop)
10V is more than enough to run the RB3011 itself, 24v is recommended when powering other devices via ether10 - “poe-out”
Thank you - that’s what I was hoping to hear. I couldn’t imagine the regulators would really need that much headroom on the PSU input. I don’t use the passive PoE anyway; got a Cisco switch for PoE.
I’m having 900Mbps internet installed in a few weeks, along with changing to VOIP (with a local PBX I’m configuring now) from a copper landline, plus a new UPS, so it’s a good time to rewire the rack in the loft. Running from my existing 12V PSU which powers a few NASs eliminates another wall wart.
Hopefully the RB3011 will cope; probably be OK since I’m not a heavy user anyway and my rules are minimal. It’s only a home network (but one with now four VLANs and needing a 28 port PoE switch in addition to WiFi, 
).
It’s likely that RB3011 will be a bottle neck here. Offcial test results with a grain of common experience of forum users say, that RB3011 is able to comfortably route at speeds of around 500Mbps (“routing → 25 Filter rules → 512 byte [packet size]”). It can do more but with very optimized firewall filter rule set. Number of users doesn’t matter much, it’s actually beneficial to have multiple concurrent users because processing of separate TCP connections is spread over multiple CPU cores (not that RB3011 provides much of a target to spread processing, it’s got “only” 2 CPU cores).
BTW, both specs and product brochure specify input voltage range of 10V - 30V … the same for both barrel jack and PoE-in. So I don’t see where the idea about minimum 14V came from.
Hmm… I was wondering about that over the last hour, reading the test results. Might upgrade to an RB5009 when funds allow again.
As to the 14V… here:
RB5009 has 3 power inputs (and supports -48v), but has a minimum voltage requirement of 24v
Yes, depending which RB5009, I would either power the non-PoE out version from my PoE switch, or the PoE version from 48V to take advantage of the 802.3at PoE it has. That extra PoE output would be useful, but the PoE version costs a lot compared to a used RB5009UG+S+IN, and I’m running out of budget.
Nuts to the budget, I’ve just ordered the RB5009UPr+S+IN.
My wife can live on Ramen for next few months
If you are not aware that’s you that have just selfpromoted to Ramen eater in your Family … expect 5009 complains on missings visits to a hairdresser and manicure
I know, but I can pretend for a few days until the parcel arrives on Monday and she asks “What have you bought now?”. It’ll be worth it. I’m almost certain…