The side facets are available for solar panels except for the one facet that has the magnetometer boom. Power during spin will vary as the facets change angle with respect to the sun. The worst case expected tilt of the spin axis with respect to the ecliptic is 30 deg.
The magnetometer boom can shade 2 of the facets during some of the spin. A shaded panel will drop its output voltage and be unable to contribute any power to the total even when it is only partially shaded. The net effect is quite small because the facet is only shaded over a short angular range where it wouldn't contribute much anyway. The difference in spin averaged power is only about 2%.
These plots show total output power summed over all facets relative to the full output of a single facet. Worst case tilt combined with a single panel failure gives a 1.1 spin average. This means a full sun output of 1.2W per panel should be able to supply the minimum power even with a single panel failed. A 20% efficiency solar cell would require a minimum of 44 square cm per facet.
There is considerably more area available than 44cm-sq per facet. A 80cm-sq panel is a reasonable choice. That gives 2.16W per panel. Nominal spin ave total of 3.39W. Worst case 2.38W
The solar power varies by about a factor of 3 during a spin. To effectively use all the power some way to store and return the energy is needed. A capacitor filter would be massive for this much power at such a slow rate. The load is cycled so a battery is required to average out the power demand as well.
The Specrolab dual junction cells have a max power output at 2.06V. Multiple cells can be wired in series for higher voltages at lower currents for the same total area.
Lithium-ion cells have a very high energy density and happen to be a good voltage match. The max charge voltage for Li-ion is 4.1V per cell, nearly an exact match to 2 solar cells.
80 cm^2
solar maxpower Li-ion charge
cells V mA cells voltage
2 4.12 1120 1 4.1
4 8.24 560 2 8.2
6 12.4 373 3 12.3
8 16.5 280 4 16.4
10 20.6 224 5 20.5
12 24.7 204 6 24.6
14 28.8 160 7 28.7
Lower currents (at higher voltages) mean less losses in the charge regulator etc. and less chance of magnetic interference. More smaller cells means more interconnects. A 4 solar cell panel and a 2 cell battery provides a good match with reasonable currents.
The battery must be large enough to provide power during the worst case eclipse portion of the orbit. This is about 8 hours. The nanosat will probably be setup to reduce power consumption by about a factor of 4 or more during this time. 8 hours at 250mW would require 2 Watt-hours. The pack should have more capacity than this so that it never becomes completely discharged. For at most 60% discharge a 3.3W-hr pack is needed to survive the longest eclipse. For a 2 cell battery this is a 450mA-hr cell.
Pack capacity remains above 80% down to -10deg C but drops to 40% at -20deg C. If the battery pack cools significantly during eclipse it needs to be sized larger.
To be effective at averaging the input power the battery must be able to accept a charge current as large as the peak solar output during spin. For 40cm-sq per facet with 4 cell panels, max current is 280mA. Recommended max charge currents for Li-ion cells vary from 0.5 to 1 times the nominal A-hr capacity. A 560mA-hr cell could take peak current without going over the 0.5C limit.
An 80cm-sq panel increases the max to 560mA. For a 0.5C limit a 1120mA-cell would be needed. At 80cm-sq there is plenty of extra power so wasting some at the peak isn't a problem. For the 30deg tilt case the peak drops below 500mA so a 1000mA-hr cell would maintain full efficiency when needed.
A 1000mA-hr Li-ion polymer cell is 23 grams (only 46g for a 2 cell pack).
The charge control circuit limits the charge current to 500mA when the pack voltage is below 8.0V. Between 8.0 and 8.2V the current limit is decreased to 0. The controller will allow the panels to supply current directly into the load while keeping the charge current at 0.
When current is needed the panel voltage will be slightly more than the battery voltage. The best power transfer will be at a battery voltage of about 8.0V which is a nearly charged pack. It is expected that the battery will be kept near full charge over most of the orbit. Once the battery is full, the switch will remain open much of the time and the panel voltage will increase to its open circuit value.
With 80cm-sq panels the output is at least 2.7W so there is a 2.7-1.3 = 1.4W surplus. The battery will recharge from a 2W-hr eclipse in less than 2 hours.
A low on resistance FET switch is used to minimize power loss. Schottky diodes prevent back current into a dark or shorted cell. The diode forward drop would be about .2V for 1A diodes used at lower currents.
Regulated output power is supplied via DC/DC converters. It's hard to find standard converter modules with this input range so it may mean making custom circuits. Nothing exotic is needed but some of the newer micro-power controller chips would be nice. Being limited to rad hard parts may sacrifice some efficiency.
The transmitter power is only on for a short percentage of the time so its converter can be disabled by the CPU until needed. The output voltage of this converter can be whatever is needed for the transmitter. When the transmitter is on it will need more power than is available from the panels. The battery will have to supply the difference. At 5W for the expected 10min transmit time that is only 0.83W-hr. This is well within 7.4W-hr (1000mA-hr).
Worst case current from the battery would be 800mA with transmitter on during eclipse with a (nearly drained) battery voltage of 6.5V. This is within the capability of a 1000mA-hr pack.
Battery cycle life shouldn't be a problem but the available specs don't help much in predicting exactly how well the pack will do for this situation. Li-ion cells spec about 20% loss in total capacity after 500 full discharge/charge cycles (about the same as NICd or NiMH). The curves don't show any rapid degradation within the 500 cycles or imply that 500 is a limit. For this application the battery will see mostly very small discharge/charge cycles. The only relatively deep discharge is the worst case eclipse. That orbit period is 3 days for only 122 cycles in a year. Very small discharge/charge cycles (less than 0.2%) happen at a much higher rate.