Re: [Cerowrt-devel] solar wifi ap designs?

[Richard Smith <smithbone@gmail.com>]

On 06/09/2017 10:02 AM, Dave Taht wrote:

> My use case used to be covering hundreds of km in the Nicaraguan
> jungle. The prototype for that covers a mere 110 acres in the los
> gatos hills, trying to get stuff deep into ravines and so on.
> 
> I have at least a dozen areas where a wifi repeater would be useful,
> but I tend to meshy route things, so rather than repeaters, I have
> routers.

> the ubnt radios I presently use tolerate 6-24v, and eat less than 5w -
> so some of the things I've been watching rathole here don't apply.
> What I'd hoped for was an "all-in-one unit" much like the wifi camera
> that kicked off the thread.

I haven't done much of a search in the last 5 years but I've yet to see 
such a product.  I wonder how well the wifi camera stuff actually works 
measured against availability during minimal-sun periods.

Accounting for wost case is always what amplifies the requirements of an 
off-grid system.

NASA Surface meteorology and Solar Energy claims that for Los Gatos 
December is the lowest output.  Given a split-the-middle tilt alignment 
of 37 degrees it will receive average full-sun net of 3.5 hours.

I'll Assume your setup averages at half duty (2.5W) which is 60Wh/day.

Using my .7 number for PV system input-output that's minimum of a 25W 
panel with zero tolerance for low-sun days.

NASA has minimum irradiance across n-day periods.  For for 1, 3, and 7 
day periods in December the minimums are 12, 23, and 42 % of average. I 
did some spreadsheet numbers for those periods using 30, 40 and 50 W 
panels.  I summarized the results into a table of PV wattage, Min 
battery, and recovery time.  PV wattage is your input wattage, Min 
battery is the amount of energy storage you need to stay operational 
during the period, and recovery time is how many days of average output 
you need to recover the battery back to 100%.   Note that this assumes 
your storage device is 100% efficient. Real batteries would need to be 
over sized by some % to net out the required energy.

1-day

PV  Batt  Recovery
--  ----  --------
30   50    2
40   47    0.9
50   43    0.55

3-day

PV  Batt  Recovery
--  ----  --------
30  122    5
40  103    2
50   83    1

7-day*

PV  Batt  Recovery
--  ----  --------
30  173     7.5
40   90     2
50   8.5    0.1

* The 7-day numbers are a bit misleading.  Any 7-day period could 
contain a 3-day period. So even though the period math claims that for 
50W panel you only have 8.5 Wh of deficit you still need energy storage 
numbers from the 3-day period to deal with the worst distribution of the 
worst case.

So looks like a 50W panel and a 90ish Wh battery should do pretty good. 
Unfortunately, this is all average based so it still won't guarantee you 
stay up 100% but it should be pretty close.

Now contrast that with the best month June having 6.8 hours of full-sun 
and min % of 16, 51 and 59.

In the 3-day and 7-day periods the system doesn't go into deficit.

1-day

PV  Batt  Recovery
--  ----  --------
25   39    0.5
30   34    0.33
40   26    0.16
50   17    0.08

Event though the 3 and 7 day periods don't actually go into deficit you 
still need storage for ~2 days @ the 1-day minimum.  The 3 day period 
could be comprised of 2 really weak days and then 1 great one.  So sadly 
you would still need 70ish Wh of storage.

I included 25W to have a item that matched the best recovery in December.

If you aren't sizing for worst case and are ok if things stop when it 
rains for more than 1-day then you can get by with a lot less of a 
system, but for 95%+ uptimes  you need a lot of excess capacity.

> So I'm back to building a custom enclosure for edison batteries, a
> solar panel hopefully lacking in charge controller, and a nanostation
> m5 or rocket m5 as the gear.

If you omit the controller you will need to factor in the additional 
loss of forcing your PV array to operate at a fixed voltage.

You can think of PV as a voltage controlled current source.  At 
different voltages it will produce a different amounts of current 
producing an IV curve.  There exists a point on that curve where the 
voltage * current produces a maximal power output usually denoted in the 
specs as Vmp.  For most "12V" Poly-Si panels Vmp at operating temp is 
15-16V.  The exact value varies by panel makeup and by operating 
temperature.

If you directly connect the panel up to a battery then you will force it 
to operate at the battery voltage and thus its power output will be 
bound to that voltage point on the curve.

It certainly makes for a simpler system but depending on the situation 
the output power reduction of the PV can be significant.

The 1.6V charing voltage of a NiFe cell seems to align with a 12V
PV panel in that a a 9 cell NiFe would need 14.4V for charge.  But the 
devil is in the details.

If say for example your battery can draw more load than your PV can 
output then there won't be much voltage rise on the battery.  A low NiFe 
battery is 1.2V/cell so say the battery voltage is only 11 or 12V when 
it starts to charge.  Thats going to reduce your PV power output.

An MPPT controller can recovery that lost power by keeping the PV 
voltage at a maximum but that has to be contrasted vs the efficiency of 
the controller.  Depending on the system it may be a wash.

MPPT helps in smaller PV setups where the load is much larger than the 
PV input.  In these cases the load will overdraw the panel and pull the 
voltage way down which decreases the PV output even more.  The PV ends 
up operating way down its power curve.  You can have full-sun on the 
panel and still be no where near the expected output.  MPPT can prevent 
that from happening and keep the panel operating at maximum power.

Are you not wanting a controller due to cost and complexity or because 
you are concerned about conversion efficiency?

-- 
Richard A. Smith