Hi all,

I'm testing the PowerBox for deployment on rooftops with a 30m Cat6 cable run from the 24V PoE injector into ETH1, then two UniFi Mesh APs + one LightBeam AC for backhaul, connected on ETH3 to ETH5.

Measuring while directly powered via the barrel connector over a 50cm cable, setup works fine, with a maximum load of 15W with three devices running concurrent bandwidth tests while connected to the UniFi Mesh.

If I then power over the 30m test cable, it all falls apart. The LightBeam gets powered, but the UniFis don't get powered. Enabling the "long PoE cable" option only causes the UniFis to power up, then shut down, with lots of audible PSU hissing going on. These are the measured values (if anyone needs explanation of columns etc., shout):
I've come up with the following ideas to test:

- Use a passive "PoE splitter" to "convert" the voltage at the PowerBox side to a barrel connector. May be useless if barrel + ETH1 PoE in are hardwired together.
- Use an active PoE converter from 48V at/af to 24V. I have found these which claim to supply up to 24W at 24V, but not tested: https://www.balticnetworks.com/ignitene ... erter.html
- Run AC to the rooftop, stick the power supply inside a waterproof box, and power the PowerBox via barrel connector (ugly, but hey...)

If anyone has any brilliant ideas or experience on similar setups, it'd be awesome to hear!

Try these:

https://mikrotik.com/product/rbgpoe_con_hp

Or a power supply with 28/30v

Check the specs of your wireless gear. If they can run on voltage higher than 24V (e.g. up to 48V), then go for 30V power supply (that's maximum that powerbox handles).

It seems suspicious to me that even without any load (0.04A is no load) you see 1.8V drop on 30m ethernet cable. And more than 5V drop at one quarter of an Amp, which would mean 19.6 Ohm resistance which is way more than Cat5 specifications require (less than 0.188 Ohm per metre or 5.64 Ohm per 30m). With maximum allowed resistance, voltage drop at full consumption (3 devices, speed test) should not be greater than 4.5V (15.12W at 19.5V equals 0.796A which, at 5.64 Ohm resistance, equals 4.488 Volt drop).
Unfortunately it seems that Cat6 ethernet cables have higher specified DC resistance (around double of that of Cat5e).

I'd say you should check your ethernet cable (possibly change it to Cat5e) and connectors, they should not drop voltage that much.

Thanks for the comments!

Check the specs of your wireless gear. If they can run on voltage higher than 24V (e.g. up to 48V), then go for 30V power supply (that's maximum that powerbox handles).

The Ubiquiti gear won't take higher - well it might, but it may wear out the caps faster etc. This is also installed at high altitude, high heat, so not ideal to push specs.

It seems suspicious to me that even without any load (0.04A is no load) you see 1.8V drop on 30m ethernet cable. And more than 5V drop at one quarter of an Amp, which would mean 19.6 Ohm resistance which is way more than Cat5 specifications require (less than 0.188 Ohm per metre or 5.64 Ohm per 30m). With maximum allowed resistance, voltage drop at full consumption (3 devices, speed test) should not be greater than 4.5V (15.12W at 19.5V equals 0.796A which, at 5.64 Ohm resistance, equals 4.488 Volt drop).
Unfortunately it seems that Cat6 ethernet cables have higher specified DC resistance (around double of that of Cat5e).

I'd say you should check your ethernet cable (possibly change it to Cat5e) and connectors, they should not drop voltage that much.

I've measured the test cable I have used (factory made with assembled connectors), and it has between 18 and 21 ohms per cable, which is what you have worked out. In some websites I have seen specs for Cat6 at 23 ohm / 33m, and for Cat5e at 10 ohm / 33m.

I'm ordering other cables to test with, but I may be limited by what's available in the install country.

Try these:

https://mikrotik.com/product/rbgpoe_con_hp

Or a power supply with 28/30v

Thanks for the suggestion, but this would add too much extra cost (converter + 48V injector + waterproof box on the roof).

Another idea: put a mUPS on the tower before powerbox. According to brochure it boosts PoE out to 24V if input is lower than that ... and you don't have to install batteries if power outages are not a concern. Price tag is not outrageous either.

Another idea: put a mUPS on the tower before powerbox. According to brochure it boosts PoE out to 24V if input is lower than that ... and you don't have to install batteries if power outages are not a concern. Price tag is not outrageous either.

Mups drops Poe dto. 12v in battery mode, with this bad cable it will result in shotdown because of low voltage

Another idea: put a mUPS on the tower before powerbox. According to brochure it boosts PoE out to 24V if input is lower than that ... and you don't have to install batteries if power outages are not a concern. Price tag is not outrageous either.

Mups drops Poe dto. 12v in battery mode, with this bad cable it will result in shotdown because of low voltage

According to brochure mUPS drops to 20V when in battery mode. If batteries were not attached, this wouldn't happen at all (actually voltage would drop to 0).
But if, as I suggested, mUPS is mounted on the mast, having this bad cable on its input, output is connected to powerbox with very short cable (so voltage drop on that cable would be really low). This way powerbox would receive almost 24V on its PoE in (almost) regardless the power draw. mUPS would be used as voltage regulator.

If brochure statements are correct.

So, new results are in, with a super-duper-wondercable ordered from Amazon (20€). Cat6, 30m, supposedly "high quality". Measured resistance 5.8 ohms vs. 21 ohms from the grey cable. This time, the PowerBox PSU + injector keeps the three devices powered, even during the same speed tests on three devices at once (also tested with one device on one UniFi and two on the other).

With the Mikrotik PSU, voltage at the PowerBox doesn't drop below 21.7V. I have determined the critical voltage for the UniFi Mesh devices to be 20V, anything below that and they start randomly rebooting. The LightBeam seems to hold up better.

The results below are from a bench PSU so that I could set supply voltage and measure current at the same time.

The powerbox pro or omnitik AC is much better than the regular powerbox

The PowerBox Pro has a self-consumption of 3.1W vs. less than 1W for the PowerBox. With the LightBeam and two UniFis connected, it's drawing 15W with no traffic, vs 11.5W of the PowerBox. It's also a lot more expensive