The current-voltage characteristics of the solar panel are similar to a current limited voltage supply. The blue curve shows a typical characteristic for a 4 cell panel based on the specifications for dual junction cells. The red curve is a simulation used for testing.
The actual panels will have a variation in output current during spin. Some checks were done with simulated spin modulation. As long as the peak current isn't limited, the battery pack averages the short term variations. A dc supply set to the average gives a similar result.
The following table gives the estimated current associated with a given solar panel area per facet. Best case is for the spin axis perpendicular to the sun direction. Worst case is for a 15 deg spin axis tilt and one failed facet. Peak is the maximum instantaneous current during a spin.
| Facet Area | Best Case (spin avg) | Worst Case (spin avg) | Peak |
| 20 sq cm | 110 mA | 77 mA | 140 mA |
| 40 sq cm | 220 mA | 154 mA | 280 mA |
| 60 sq cm | 330 mA | 231 mA | 420 mA |
| 80 sq cm | 440 mA | 308 mA | 560 mA |
The active load current increases somewhat as the battery voltage drops. The DC-DC converters supply constant power to the load so their input current increases when the buss voltage drops. Most of the idle current is drawn directly from the battery buss and it remains nearly constant.
The recharge with a 320mA panel took about 12 hours. The calculated total energy discharge of 20 W-hrs was not quite matched by the calculated energy input on recharge of 16 W-hr. The discrepancy is likely due to accumulated error caused by small inaccuracies in the current values. The rated capacity of the pack is 22W-hr so it seems likely that the discharge calculation is closer to the correct value than the recharge. In any case the pack is adequate to survive an 8 W-Hr eclipse.
At the start, current is limited by the panel and the panel voltage drops to about 8.5V. The panel voltage runs about 0.2V higher than the battery due to the drop in the isolation diode. When charging demand decreases the controller limits the input current and the panel voltage increases up to the panel open circuit voltage. The average power supplied by the panel drops to about 750mW when the battery is fully charged.
In this case the available panel current is more than the maximum load. The charge current eventually goes to zero and the load is supplied directly from the panel. The only time power would be drawn from the battery would be during eclipse or when the transmitter is on.
Test were run simulating a 850mW panel to check operation with minimal input power. This is the calculated spin averaged output given 20 sq cm per facet on 5 facets.
For optimal power transfer the panel must run near its maximum power point (blue dot). To better simulate the curve in this region a series resistor was added (56 ohm) and the supply voltage was increased (14V). This simulator curve (red) diverges from the panel characteristic above 8.6V but the maximum battery charge is 8.3V so this portion of the curve is not used during recharge. When the pack reaches 8.3V the portion of the curve below 8.3V is not used. For data at full charge the simulator was changed to use 22 ohms and 10.7V (green) to better simulate the portion of the IV characteristic from 8.5 to 9V.
The power output from the panel stays nearly constant over the duty cycle because all available current is being used either by the load during active time, or for charge during idle time. Available current decreases slightly as the panel voltage increases.
When the pack finally reaches full charge, the average charge current goes to zero but idle charge current does not. It still has to replace the energy used during active time. A small charge current is needed at least at the start of the idle time. The average panel output (green) drops to about 750mW.
The previous runs have been based on typical cell I-V specifications. If for any reason the voltage output is significantly lower, the panel will not be able to charge the pack to full capacity. The pack is sized to have excess capacity but a lower panel voltage will also affect recharge time.
To simulate a 10% drop from nominal voltage, a supply of 8.5V 230mA with a 4 ohm series resistor was used (red trace). The green curve shows the 750mW power level. The panel average operating point will eventually move to the intersection between the green curve and the panel's I-V curve. For the nominal (blue) curve this intersection is above the pack voltage limit. For the low voltage (red) curve the final charge voltage will be determined by this intersection point. For this case there is very little excess power available for charging as the battery voltage approaches this point.
The final charge voltage reached was 7.95 Volts. Based on data from the 8.3V discharge curve, this represents roughly a 4 W-hr (20%) loss from a full capacity of 20W-hr at 8.30 Volts.
Starting with the pack at 7.9V, a discharge to 6.5V under nominal load took 16.6 hours. The calculated total energy output from the battery was 12W-hr. This is somewhat lower than the 16W-hr capacity estimated from the 8.3V discharge curve.
Recharge with a low voltage simulated panel took about 4 days however 90% of the energy was returned in the first 2 days. After 4 days the calculated total energy input to the battery was about 12W-hr which matches the energy output during discharge.
Increasing the panel voltage to nominal charges the pack another 4 W-hr giving a total input of 16W-hr. This is again less than the expected 20W-hr full capacity but is within the typical energy calculation errors.
These results show that a 10% degradation in panel voltage would have a significant effect on the worst case eclipse survival. For a short duration orbit with a large energy demand during eclipse, the lowered capacity combined with long recharge times could be a problem. If the voltage degradation is expected to be more than 10%, changing to a 5 cell panel should be considered.