Solar Rocket?

[air2net]

I'm in the process of setting up a greenfield community wISP. I'm going to use sectored Rocket M5's, feeding to Nano M5 CPE, using exclusively 11n and, hopefully, Airmax.

I would like to power the Rockets (and sometimes the Nanos) with solar. I have a couple of questions concerning power consumption and the power supplies provided with the units.

First, am I correct that both units can handle 10 to 24 volts on the POE? If so, why does the Rocket come with a 24V supply and the Nano with 15V? Or is this another misprint in the specs? (I know from another post they both use max 8w, with 4-6w typical).

Second, am I correct in assuming, for both units, that the injector and power brick are combined in the same unit normally used for Ubiquiti products?

If these assumptions are correct, I would like to use some of those little triangular DC injectors, available from various suppliers, powered directly from a 12V deep cycle battery, charged by solar. A single 40w panel seems like enough for three sectored Rockets on the same pole, and a 15w panel for each Nano. That should provide enough excess power to keep the batteries charged, with enough excess capacity for night-time and darker days (very few of those here in Arizona).

Having lived on solar for many years, I know that some solar charge controllers create a voltage spike, as the Sun rises, and the panel is switched back into charge mode. I have seen the voltage go rapidly from the 12V overnight value to 14V or more (rising more slowly to around 16V as the Sun gets higher). Is this spike going to cause any problems for these units?

To be on the safe side, I'm planning, at first, to use nothing but a simple blocking diode, just to prevent reverse current back to the panel at night. This should smooth out the voltage changes, but there remains the (much smaller) problem of sudden shading or unshading of the panel by moving clouds.

My question is whether these units can handle the changes in voltage? I would hate to have to stick an inverter between the battery and the standard power supplies -- that would waste more power than the units actually use.

Anyone have any better ideas?

[sunwize]

A DC-DC converter (regulator) should keep the 12vdc nice and smooth if the radio is affected by the solar charge controller output. Let me know if your interested in a turnkey design for the project.

[kijoma]

hi,

relatively slow changes in voltage will be fine, the product has a switching converter inside it that will iron that out..

the only feature lacking in all these products (and i don't just mean ubiquitis) is a low voltage cut off..

if the voltage drops below the safe working voltage the units will invariably crash into an unknown state.. when the volts rise they then need a power cycle to wake them up again..

this is particularly annoying in rural areas when the grid power often dips for a short duration, leaving the radio hung... and resulting customers on the phone!

A simple circuit would use a comparator to detect the voltage from the output of the routers switcher and if it dips at all, turn off the switcher for ~5 seconds and re-soft start it again... i suspect this would require no more than 3-4 small parts if designed into the unit.

for an external version you would have to set the levels to switch off and on at (with sensible hysteresis).. when i get 5 minutes :icon_rolleyes: i intend on making such a circuit to fit inside a standard PoE splitter block so power droop/dip issues no longer leave radios hung..

cheers

[air2net]

Thanx, Bill. I had been mulling over the low voltage cutout problem as well, while waiting for the products to arrive. Finally got some nanos and rockets, today, but no sector antennae, unfortunately. Nevertheless, it's time to play, and see what we can get.

I had pretty much guessed the internal DC-DC converter could handle slow voltage changes, but I was looking for confirmation before I tried the smoke test.

Fortunately, the power requirements for a typical pole with three or four sector transmitters are not very high, and I think I can do away with the need for a solar charge regulator altogether. I am going to start by connecting the panel (and the POE injectors) directly to the battery, with nothing but a blocking diode in between, to prevent reverse current flowing back to the panel at night. (And fusible links to protect the equipment, of course).

In this scenario, the panel is cut out of the circuit, at night, when it is not producing any output. At dawn, as the Sun rises, the panel's voltage and current will rise slowly, overcoming the blocking diode and reconnecting into the circuit. This should always occur at about the same voltage as the battery and the rest of the system, with no rapid change. At maximum output, the voltage will always be under 18 volts and the excess current will simply charge the battery. As the Sun goes down, voltage and current will gradually decrease, until the blocking diode kicks in and cuts the panel out of the circuit again. During the night, I expect the battery potential to drop slowly from about 14V to about the nominal 12V.

I don't think it has to be very complicated. Later, I may have to add low and high voltage cutouts, for excessively dark and bright periods, but I may also be able to handle both problems simply by using a larger battery bank as a buffer for both.

I'll try it out and let you all know what happens (even if it smokes). I'll also take practical usage measurements, so we can come up with a generalized solution for everyone to use.

[twinkletoes]

I'd love to see designs for solar sites, including recommened panels, batteries, charging controllers, and low-voltage cut-off. Some of the SunSomething (can't remember now) controllers i've used in the past had a low-voltage cut-off...