Only the panel itself.

 

Maybe the Panels used on the little yard lights dont have the internal diode D1 and it is the reverse leakage of the Panel itself that supplies base current to Q1. The larger panels (especially the ones sold for automotive trickle charging) do have D1. D3 is there just to reverse bias the Vbe junction of Q1, and could be a Schottky to minimize loss when the sun is shining...

It isn't too easy to open the customary black-blob chip in these garden lights to see how its done - but I have my own theory.

In the discrete component variety, its usual to find an SB diode (e.g. 1N5817 etc) to prevent the Ni-Mh cell discharging back into the solar cell when its dark.

Now suppose you use the B/E junction of an NPN transistor as the diode, and connect the transistor as common base to the battery negative - the negative lead of the solar cell is connected to the emitter. Pulling the emitter more negative has the same effect as pulling the base more positive - it makes the collector conduct.

The collector is used to shunt the base of the oscillator/inverter transistor that drives a boost choke to develop 3.4V from 1.2V in so it can drive a white LED.

 

Every little light I've seen has a blocking diode. I've been expecting them to build it into the little controller chip but so far they all still have a separate diode.

But I need to look again at one without the Cds cell. I think I've seen some but can't recall how they shut themselves off.

The last batch in my back yard (circa 2012) use a QX5252 as the charge/LED controller. No black blob but a discrete 4 leaded device. The data sheet indicated that diode is internal to the device. It needs to be there to keep the panel from discharging the battery at night.

These are available on EBay, only 3 external components to drive LEDs: Solar cell, nicad battery, and an inductor.

Data Sheet (warning: not engrish)

 

The last batch in my back yard (circa 2012) use a QX5252 as the charge/LED controller. No black blob but a discrete 4 leaded device. The data sheet indicated that diode is internal to the device. It needs to be there to keep the panel from discharging the battery at night.

The last one I took apart - the red leads from both battery & cell were soldered to the same pad, the black leads each had their own track that disappeared under the black blob. But I have also seen the black leads common once or twice.

 

Several years ago I built two systems to turn on LED's at dark and then turn them off at light. I used 5W solar panels. In the daytime the panels would charge a 12 SLA battery to run the LED's night. I used a 1n5817 between the output of the solar panel and a PB137 IC to charge the batteries. I used a 1N757A zener diode as a small load on the solar panel to detect darkness. I used a NE556 as voltage sensors to detect the night voltage to turn on the LED's and to turn them off in daylight. I used another Ne555 to PWM the LED's to control intensity and conserve power. There was another 555 timer on the board to flash the LED's on and OFF remotely if necessary. The systems ran continually for three years without problems until I deinstalled them because of a move.
I will be glad to post the circuit if members of the forum would like.

 

k7elp60
YourPost sound interesting can you post schematic-diagram.

 

k7elp60
YourPost sound interesting can you post schematic-diagram.

Here is the schematic. It was designed a built about three years ago. I'm figuring out all the technical specifications. When I get them figured out for sure
I will post it. Thanks for your interest. I also have PCB lay out and I have two working models with solar panels, batteries and LED lights, that I would consider trading for something else.