I am testing a DC2038A-Q on using a 30Watt, 22 Vmax solar panel aimed at the south. It uses a 12V sealed lead acid battery (Yuasa, 2*7Ah).
The load is just under 100 mA continuous, delivered through a DC/DC converter to a 5V load. I remotely monitor the voltage from the panel and the battery voltage.
The charge voltage and the maintenance voltages have not been changed yet from the factory settings . I left these because they are suitable settings for evaluating the low light perfomance.
I get the following surprising results:
- battery charge takes about an hour in full sunlight, but the end voltage is always 12.41 Volts.
- when no direct light hits the panel (heavy cloud covering, 3 hours before dusk, 3 hours after down) the voltage of the panel is always about 0.2 Volt above the battery voltage. This is not optimal, MPPT does not seem to be working. When I use a different solar charger (EPEVER LS1042E ), the dawn and dusk periods do add to the battery charge. Reading the data sheet, resistance might be an issue, but the battery leads are not very long (0.4 metres). Besides, the DC2013A does have a 10uF capacitor over the output.
- overnight, the solar panel voltage drops to 0, but the charger does seem to remain active, coparing the voltage drop to the to the EPEVER charger. I calculate this to a current draw of about 30mA.
- With a fully discharged battery, the Power Path keeps the battery connected to the output. The output voltage is always identical to the battery voltage (in low light). Reading the data sheet, this should not happen. Power path is a way to feed the load with a discharged battery, when solar energy is available.
- there seems to be no way to protect the battery from a deep discharge (such as cutting off load under 9.5 Volts battery voltage).
Some of these problems can be solved by setting the charging voltage higher. What can I do to solve the other problems ?
Sibe Jan Koster
An update, for anyone interested:
1. I have solved some problems by connecting the load directly to the battery. This is also suggested in the data sheet. This will keep MPPT working in low light. The…
Thanks for posting your solution. I would have advised the same. Which issues still remain?
You can increase vcharge_setting to charge up to 13.77V, so you should be able to hit your target here…
I have done some more experiments, to document:
1. Low light performance
In low light, with no load on the system, the MPPT loads the solar panel with 200 mA at 15.75 Volt. That is 3.15 Watts Battery is charged with 250 mA at 13.04 Volts = 3.35 Watts (according to the sensors on the board).
When applying a load of 350 mA, the MPPT seems to stop. The current from the panel is 212 mA, at 12.92 Volts, the battery current is -146 mA at 12.88 Volts. Thus, the power from the panel has dropped from 3.15 Watts to 2.7 Watts, i.e. a 16% lower value. In these conditions, it would be better to continue charging the battery with 250 mA, using MPPT.
2. Battery charge not picked up again after a passing cloud
At full solar input, the charger switches to absorb charge (constant voltage, 14.45 Volt, 0.8A). When a cloud blocks the sun, the charger leaves the absorb charge state. After the return of the sun, the battery charger remains in 'charger off' state (the battery voltage was 13.02 Volts). The solar panel voltage collapsed and the battery had to supply part of the power to the 4.6 Watt load, even while the panel was at the time capable of delivering 14.45*0.8 = 11 Watts.
It was necessary to write a "false" to "suspend charger" to solve this.
Ideas anyone ?
1. I have solved some problems by connecting the load directly to the battery. This is also suggested in the data sheet. This will keep MPPT working in low light. The power path is then no longer functional.
2. The 'charger off' state is restored when the system scans the MPPT settings every 15 minutes. That should be done more frequently.
3. While v_charge_setting can be set by the user, no option is available for sealed lead batteries (i.e. 14.7 Volts charge, 13.7 Volts maintenance). As a result, only part of the battery capacity is available (in our case, about 60%).
You can increase vcharge_setting to charge up to 13.77V, so you should be able to hit your target here. For absorb and equalize stages, see vabsorb_delta and v_equalize_delta bitfields.