I'm still from the old school.

I don't run POE up towers unless there's no other way. It's one thing to run POE up a pole on the roof of a client's home or business. It's a different kettle of fish when there are multiple radios on a comms tower.

In this specific case, the high site provides us with clean 220V AC power. The site owner has a 3KVA UPS for his WISP clients and backs that up with a generator.

I went out and bought a Meanwell 150W 15V power supply, wired it up, ran 6mm armoured wire up the mast, terminated the wire onto a circuit breaker, ran additional wiring to connect 2 x UBNT rockets down the line (hence the 15V) and connected up a RB433 via the pext jack. wlan1 got a R52Hn to connect to our backbone. wlan2 got a CM9 that we use to connect to the router from our laptops and tablets when we are on site and wlan3 has a dBii F.50pro that's connected to a 19dBi UBNT sector for client access. There's also a RB750 up there and when I connect the UBNT kit, everything will meet up on the RB750.

It's a recipe that I know and love.

Problem is, this time it just doesn't work.

I had problems from day one. The backbone link was messing me around with signal quality of between 35% and 45%. I tried various combinations of Nv2, 5GHzA and 5GHz N and the link was messy. Then, the F.50Pro decided it would just die randomly for no reason. Everything was working. The radio showed the two test clients were connected, but there was no traffic. I couldn't even ping or MAC telnet into the client radios. As soon as I disabled and re-enabled the F.50Pro eveything would work properly again for a while until the radio froze again.

Today I went to the site, pulled the 15V PSU and replaced it with a 24V 100W PSU.

Touch wood, so far so good.

Signal quality on the backbone link has jumped to 84%. The F.50Pro no longer freezes.

What I don't understand is how can a 4A 24V PSU be performing better than a 10A 15V PSU when the router is supposed to work on anything between 10 and 28V ?

I use those same F.50Pro cards direct from 12V batteries at other sites and they don't give problems.

Am I missing something blatantly obvious here?

Am I missing something blatantly obvious here?

Voltage drop on power wire - you can't beat Ohm's law.

Edit: May be IEEE PoE or passive PoE(used by MT) is not as bad as you think - designed to work over 100m of 4 wires of UTP.

One thing you could, or perhaps should have done was check your actual line voltage and or amp draw on the power side using a multimeter and see if there

was a problem with low amps/volts.

Perhaps you have or had a faulty transformer that was dropping amps under load? I'm not so sure if the voltage drop would necessarily would cause it, since

you are telling us that you have AC at the top of the tower? if so, short cables voltage drop shouldn't be too much. You're Using POE out of a Rocket GPS to

power a RB433? Or the other Rocket?

I don't understand your power layout very clearly.

UBNT Rocket Specs their poe out at 12W -- not enough for a RB433

Your total draw for your setup with everything will be ~45 Watts or so Maximum.

That means ~1.4A @ 24V or ~2.2A @ 15V.

I would suspect low voltage, a faulty power supply, or regulators not able to supply what you need.

Draw Breakdown

RB433 25W Max
RB750 3W Max
Rocket 8W Max

I would look at the 15V power supply closely, test it under load. Then also think hard if you made any configuration changes that could have come in to play

at the same time you swapped out your power supply.

-Brad

Am I missing something blatantly obvious here?

Voltage drop on power wire - you can't beat Ohm's law.

Edit: May be IEEE PoE or passive PoE(used by MT) is not as bad as you think - designed to work over 100m of 4 wires of UTP.

As you say, but Ampere's Law also comes into play, so even if you are running that excellent new UBNT double shielded cable, there is still a magnetic field around the cable that causes complete and utter chaos with nearby two way radio antennas. I have seen this more than once on sites.

That is why we run thick, shielded wires (6mm or 10mm for example) up towers for DC and we also run them up a leg where there are no two way radio antenna cables.

lazerusrm:

There's no AC power up at the top of the mast, but the cables going up are thick enough for voltage drop to not be a factor.

We run DC up there. We then run that DC into a circuit breaker and then run the DC into a DIN rail terminal block.

From the terminal block, there is one lead going to the RB433 and another lead going to the RB750.

There are then three more leads going to POE injectors. Three ethernet ports on the RB750 are connected to the LAN side of the POE injectors. One ethernet port on the RB750 is connected to ether1 on the RB433. We don't use ether2 or ether3 on the RB433 as we prefer moving the processing load onto the RB750 while the RB433 does the wireless connectivity. We would then run a 1 - 2m STP from each POE injector to each rocket.

It's a simple setup that I've done more than once. I'll be making up another one on Monday, but this time with a RB435G and will post some pictures.

In other news, I had the 15V PSU on the bench through the night, powering 3 rockets, each with a 19dBi sector, and everything's still fine this morning.

This is all a mystery to me.

As you say, but Ampere's Law also comes into play, so even if you are running that excellent new UBNT double shielded cable, there is still a magnetic field around the cable that causes complete and utter chaos with nearby two way radio antennas. I have seen this more than once on sites.

i referred Ohm's law. As you using shielded wires, i think inducted energy is minimal

That is why we run thick, shielded wires (6mm or 10mm for example) up towers for DC and we also run them up a leg where there are no two way radio antenna cables.

Fine, how do you explain that move to from 15V PSU to 24V solved problem? Can you disclose voltage on RB433 power connector - of course measured under load?

That is why we run thick, shielded wires (6mm or 10mm for example)...

WOW! I'm not a metric wire user, but the conversion on those sizes is not logical. That is some BIG stuff.
6mm = 2 AWG (1/4" diameter conductor)
10mm = 3/0 AWG (almost 1/2" in diameter conductor)

Are you sure that is the wire you are using?

i referred Ohm's law. As you using shielded wires, i think inducted energy is minimal

I used to think so too. Unfortunately the taxi company with the digital repeater in the rack next to us in the comms hut, LMR 400 cable running 30m up the towert and a pair of VHF dipoles just above me know differently. When I insert a POE injector, their radios die. They pay the landlord more rent than I do, so when the landlord says; "No PoE". It's "No PoE".

Fine, how do you explain that move to from 15V PSU to 24V solved problem? Can you disclose voltage on RB433 power connector - of course measured under load?

If I could explain this, I wouldn't have started this thread.

Measured under load, the voltage was 15V. Now, measured under load, the voltage is 24V. I have a DIN rail terminal block up there, so it's easy to pop a multimeter in and take a reading.

The point of this thread is to try and work out why a 24V 4A PSU works, while a 15V 10A PSU doesn't work on site. Yet the 15V 10A PSU works just fine on the bench.

When you have wires this thick going up a mast, you don't worry about voltage drop.

Are you sure that is the wire you are using?

See picture above.

...and I thought I was using big wire at 12ga AWG (2mm).

That is why we run thick, shielded wires (6mm or 10mm for example)...

Tim,

I have a site that I own where I take 30A 240V from a substation ( brings 11,000V down to 240V) 485m up the side of a mountain using 25mm shielded cable. I lose exactly one volt between the substation and the battery charger.

...and I thought I was using big wire at 12ga AWG (2mm).

No man! I'm not wiring a high level brake light onto a Land Rover

2mm is about 40 amps max rating with about a 9 amp power transmission rating
6mm is about 180 amp max rating with about a 90 amp power transmission rating.
10mm is about 320 amp max rating with about a 240 amp power transmission rating.

Those would be some bad-ass brake lights!

Those would be some bad-ass brake lights!

Africa's not for sissies

I KNOW when you refer to "sissies", you are talking about someone else.

How far is that power line run from the power supply to the routers?

ADD: On a personal side, I have always wanted to surf South Africa. You know it is featured as one of the best surf spots in the world as far back as the movie "Endless Summer" in the '60s.

You have great surf and "great whites". No surf spot is perfect.

EDIT: After that new pic below, I have no more to add. I did not want to step on your new post. Appears to be a very nice setup!

I envy you for the surf, but not for the creatures there. An 8 foot bullshark was bad enough. Sixteen feet of great white, and I also would have been a "blood spot".

I KNOW when you refer to "sissies", you are talking about someone else.

I have yet to meet a sissie surfer.

How far is that power line run from the power supply to the routers?

About 60km. It's two different sites. I own the one where I ran the power cable. I rent space where I have the power problem.

ADD: On a personal side, I have always wanted to surf South Africa. You know it is featured as one of the best surf spots in the world as far back as the movie "Endless Summer" in the '60s.

You have great surf and "great whites". No surf spot is perfect.

We have some great surfing right outside my front door. Nibbles are few and far between. Last year there was one person who was eaten. All they found was his arm and a big red blob in the water. He was body surfing. The year before that there was one teenager who was tasted just down the road from where I live. As long as you're not stupid, surf sober, don't surf at dawn or dusk and don't surf in dirty water, they'll leave you alone.

Believe me when I say they are BIG. Most people here surf with the knowledge that if one of them takes you, your body should go into immediate shock and you should black out, so it should be quick. 18 foot great white vs 6 foot man is no contest.

On an aside, here's a panel I'm making up for a new site that I'm developing:

It's currently hanging on a picture hook against a wall of my office as I'm having big problems with the construction company who are failing dismally in their attempts to waterproof the hut. I want to move in, but can't.
The power cable, obviously goes to the wall socket.

Power goes to the main circuit breaker and then on to two more circuit breakers.

The first circuit breaker controls the blue and white Meanwell power supply.
That power supply provides power to the little Planet ATA.
The Planet ATAs are the cheapest things I've been able to find in this country that have IP addresses.
It's in the "wild" so if there's a power failure, it'll die.

The second circuit breaker controls the timer.
The timer switches power on and off to the big Meanwell 155W "security" power supply.
I have a bank of 3 x 105AH 12V batteries that will connect to the Meanwell.
From the Meanwell, power goes to the 5 circuit breakers.
One is for the cable going up the tower.
One is for the RB250GS
One is for the RB750
Two are spare.

If power goes off, everything except for the ATA will continue to be powered. Dude will pick up that the ATA is down and we can start the process of getting power reinstated to the site, or take a generator up if it's more serious.

The 3 x 105AH batteries should give us about 60 hours of run time for 2 x RB433, 2 x RB750, RB250GS and 6 x RocketM5. It's a big site.

OK, maybe I do have more. Is that 60km of DC power line? Or the AC run to the power supply? That is 60,000 meters?

At 0.000203 ohms/meter for 10mm wire, that would be 12 ohms. Sounds like you are losing most of your power in the wire.

EDIT: OH! My bad! That is only one leg. The ground wire has as much. So 12 * 2 = 24 ohms total.

That is why we run thick, shielded wires (6mm or 10mm for example)...

WOW! I'm not a metric wire user, but the conversion on those sizes is not logical. That is some BIG stuff.
6mm = 2 AWG (1/4" diameter conductor)
10mm = 3/0 AWG (almost 1/2" in diameter conductor)

Are you sure that is the wire you are using?

it's 6mm sq. so it is something like 9 to 10 AWG and 7 AWG for 10mm sq., i am right eben?

@petrn: My bad there too. At 60,000 meters, even 3/0 AWG (85mm your way) is way too small for low voltage DC.

Guys! The two sites have nothing to do with each other. I merely mentioned the second site to Tim as we were talking about wire thickness and the fact that I used 485m of 25mm cable to take AC power up a mountain from an elecrtrical substation and only experience 1V loss over that distance.

They are 60km apart as the crow flies. It's three hours drive from the one to the other. I own the one where the 485m cable is. I rent space on the one where I am having problems with the 15V PSU, now replaced with a 24V PSU.


Size Diameter

AWG Metric Circular inch mm
mm2 mils

20 0.518 1,022 0.0369 0.938
18 0.823 1,624 0.0465 1.182
16 1.309 2,583 0.0587 1.491
14 2.081 4,107 0.0740 1.880
12 3.309 6,530 0.0933 2.371
10 5.261 10,383 0.1177 2.989
8 8.366 16,510 0.1484 3.770
6 13.302 26,251 0.1871 4.753
4 21.151 41,741 0.2360 5.994
2 33.631 66,371 0.2976 7.558
1 42.408 83,693 0.3341 8.487
0 53.475 105,535 0.3752 9.530

Edit: OK, so that's a mess. Best you go http://www.stealth316.com/2-wire-resistance.htm

I misunderstood. I thought it was 60km from the very nice power supply above (your pic) to the routers. That was the distance I was asking about. If 485 meters, that is about 1km round trip. That is only 2/3 ohm with 25mm wire. But that will still limit the amount of current you can use at the routers. At 15 volts on the power supply end, you could approach the lower limit of the MT boxes at about 5 amps (10v or 50w) at the router end.

The 485m AC cable is an aside.

Let's get back to the basics.

The problem install has 21m of 10mm cable. DC 15V 10A PSU don't want to drive RB433 so nicely, yet DC 24V 4A PSU drives RB433 very nicely.

To make things more complicated, DC 15C 10A PSU works well on the bench.

The 485m AC cable is an aside.

Let's get back to the basics.

The problem install has 21m of 10mm cable. DC 15V 10A PSU don't want to drive RB433 so nicely, yet DC 24V 4A PSU drives RB433 very nicely.

To make things more complicated, DC 15C 10A PSU works well on the bench.

On the bench with 21m of cable? I think that PSU doesn't work with capacitive load of that cable.

Okay so, under load your power measurement was giving you 15V at the TOWER. So your power supply is probably giving 16.5 V or so directly from its

terminals, right?

Short cabling is used from this power supply to all associated units so voltage drop from this point should be somewhat negligible.

If this case is truly an isolated incident, and you ONLY swapped the power supplies. Perhaps it is possible

that you have a point of resistance somewhere in the circuit that is dropping voltage to the equipment. Did you ever take a voltage reading from MT ROS on

the RB433 while under load with the problematic power supply?

This would be something to check on the bench.

It might be possible that if you were getting low voltage directly at the RB board due to a bad connector somewhere in the chain, that an internal voltage

regulators on the RB433 itself, or the voltage regulators on the Wlan cards themselves were dropping out.

If you want to dive into it a bit more, you can start probing the Voltage Regulators on the RB433 and measure input / output voltages and check specs and

datasheets to see what their operating input voltage ranges are. It could be that you were low enough under load to cause the board itself to have enough

power, but that the WLAN cards themselves were not getting the current they need for normal operation. one way would have been to test if this was a

possibility would have been simultaneous bandwidth tests where you put all the cards on the RB433 on receive mode, and then monitor performance. Then

put all three cards in a bandwidth test, but transmitting simultaneously and see if there was a greater problem there.

Just trying to share some ideas to consider. Interesting problem

-Brad