2001-08-09 | Example Xfig Diagrams | PV System on the Cheap
I built my solar system with a cheap lawnmower battery from Boxbrick, a 50 Watt PV panel from BP Solar (Solarex), a cheap 300 watt inverter, an MBR1045 Schottky Barrier Rectifier, an 8048 microcontroller, a ULN2803 darlington driver, an ADC0804, and a couple of relays. The Solar panel I used had a cool mounting system (multimount) that let me cover the bolts I put into my roof with the solar panel. You just use some bolts with 5/16" heads and slide the panel on. Here is a link to a pdf of the sx50. I bought this for $269. I can't find a better price for 50 watts anywhere. The whole system, including shipping, cost me less than $400 US in July 2001.
Here is the schematic:
The 2 1k resistors divide 5 volts, roughly, into 2.5 volts. The 2.2k and 4.7k resistors bring the voltage down so that 15 volts at the battery is just under 5 volts at the ADC0804 pin 6. After I built the circuit, I measured for three voltages and noticed the hex code at the ADC0804 (shown on my 8048 dev system display). F2 is about 14.5 volts, and the program turns off the charge relay at this point. The program leaves the charge relay off until the voltage gets to be about 13.5 volts (E3). If the voltage gets to about 12 volts (CF), the circuit turns off the inverter and waits until the voltage is back up to around 13 volts before it turns the inverter back on. This may seem odd, but without a load from the inverter, the battery voltage will pop right back up, so this keeps the cycling of the inverter to a minimum. I have the A/D converter in a continuous update mode by tying the INTR line to the WR line. The program starts the continuous update by requesting the first update by bringing the two pins low. I considered using power MOSFETs, but I didn't see enough of an advantage for this circuit. The simplicity of using relays, and the lack of risk of any meltdown going through the gate of the power MOSFET made me buck the general trend of the typical designs for solar regulators. Of course, my decision to use a microcontroller is a little unique also. With a microcontroller, though, I have much more flexibility in changing the operation of the circuit. In fact, some of the stuff I do with the inverter, like turn it off at a low voltage and wait until a higher voltage before I turn it back on, would be hard to do with analog without some digital circuitry (f/f). I chose the 8048 because you can get it in junk appliances and circuits. I believe that many IBM PC keyboards used this chip. It was sold for over 20 years. Certainly the choice of microcontroller is up to you. Any microcontroller with 2 8 bit ports should work with this circuit.
I drew this using XFig. Here is the fig file.
Here are some pictures of the battery:
See. It really is a cheap battery. Not recommended for solar systems, but I couldn't afford the correct one, so I wanted to test with this one and see just how bad it is. (If you are a vendor and want to sponsor the site by providing a couple Trojan T-105s, your name and link goes here. ;-) )
Here is a closeup of our MBR1045 diode mounted on a heatsink:
The wire with the yellow connector goes right to the negative lead on the solar panel. The right lead of the 1045 goes to the charge relay, and then to the battery. The left lead is connected directly to the case, which is connected to the heat sink. The reason why the 1045 is needed is that the voltage drop is only .45 volts. Other diodes have a larger drop. I found one at Mouser Electronics part# 511-STPS1045D for $1.38.The reason why you need a diode in there at all is that when it is dark, the solar panel will discharge the battery. I threw the circuit above into a plastic bag and used electrical tape to seal it in a Ponderosa cup (Hoss truly doesn't mind):
You can faintly see Adam on the right.
Here is the complete PV system on the cheap:
The inverter is just an inexpensive 300 watt one that I got at Costco.
Here is the program running on my 8048 dev system. The voltage on port 1 is d8, which is a little under 13 volts. Here is an assembly listing of the program. See 8048 Development System for more info. Here is a binary file (right-click, save, don't open) of the program that you could load into the dev system. I suppose you could burn the file into an EPROM directly if you wish, but it has my set points, which may be off since the resistors are not precision.
Here is a picture of the breadboarded 8048 dev system with the above circuit:
Here is a closeup of the 8048.
Ahhhh... '77 was when it was conceived, but the 29th week of '79 was when it was born in Malaysia.
Here is a closeup of the ADC0804.
Here are the relays, with the ULN2803 and LEDs.
So far, I've been able to generate most of the electricity I use for lighting my work area. Since most of the power is used when it is light out, I'm not really cycling the battery too much.
Completed and soldered PV Controller:
I'll put up some schematics eventually, but the only difference between this and the dev system article is that there is a 2816 EEPROM and a 74ls373 between the bus on the 8048 and the address lines on the 2816, w/ ALE on the 8048 connected to the LE pin on the 74ls373.
"It doesn't matter what you ride, it matters THAT you ride." --unknown bikerboxbrick
Articles tagged with boxbrick on Mud Hut Club:
2021-01-16: Little House of Information Technology