I took one of the new Ruby v1.0b boards and connected it up to a simple PCB I made for one-wire devices. Actually, I connected to a stack of four of them. It looks like this:

That is 20 DS18B20 temperature sensors with 2m of cabling connected up to 4 of my little one-wire bus PCBs.

The head unit looks like this:

As you can see, it is a simple board that allows you to build a linear network of 1-wire devices. At each end are simple 0.1″ pin holes to make spacers between the boards and to provide some rigidity when they are stacked as they are above. An interesting side note, I had a 4.7k resistor for the pull-up for the one-wire network. It worked fine for 10 sensors. When I stacked on the next 5, it didn’t work at all. As suggested here, I changed it to a 2k resistor and now all is good. Cleanly reads all 20 sensors, once per minute and sends the data to the net.

Incidentally, the DS18B20 chips are epoxied into a stainless steel hood to make them waterproof. They are sweet:

I have tweaked my code and am now receiving all the data. I already knew that there was some variability between tubes, but it can be > 10°F which was surprising. Here is an early graph of this afternoon.

Not too surprising, the first and last tube in the manifold get colder as they are near an opening in the manifold. When we get a nice sunny day, it will be interesting to track how the temperatures sync to the irradiance as well as to each other. Stay tuned…

It looks like this:

Apogee SP-110

and has 5m cable with three leads:

• Red: Positive from the Sensor
• Black: Negative from the Sensor
• Clear: GND

The pyranometer SP-110 has an output of 0.20mV per W/m². Full sunlight is 220mV (1100 W/m²).The full range is 0-350mV or 1.75 full sun. The SP-110 is self-powered. The ADC has an internal 2.048V V_REF so no need to mess with A_REF.

Since full sun output is 220mV if we multiplied it by 8x we would get a range of 0-1.76V which is below V_REF. The ADC has a configuration register where we can set the sample bit depth as well as the gain. Check the datasheet for all the config options.

To setup channel 1, one-shot conversion, 18-bits, x8 gain the binary would be: 10001111 or 0x8F. One-short conversion puts the ADC into standby mode. Or for the same with continuous mode it would be 0x1F.

In 18-bit sample mode, the ADC returns 4 bytes. The first 3 are data bytes, the last is the configuration. In 12,14 or 16-bit modes, the ADC returns 2 data bytes and 1 config byte.

One-shot conversion allows the ADC to go to low-power mode between conversion requests. For a 18-bit conversion, the ADC will draw ~ 36μA per conversion as < 1μA in standby (300nA). Doing 18-bit conversions continually draws about 138μA.

Using this little sketch I can get the desired output:

#include <Ports.h>
#include <RF12.h>

PortI2C p4I2C(4);
DeviceI2C adc(p4I2C, 0x68); // 0x68 address for the Analog plug with the pads unsoldered

byte Config;
int ack;

void setup() {
 ack = 0;
 // Serial
 Serial.begin(57600);
 Serial.println("\n[Bejouled MCP3424 Pyranometer]");
 ack = adc.send();
 if(ack)
     Serial.println("Got Ack");
 //adc.write(0x8F); // channel 1, 18 bit, 8x gain, one-shot conversion
 adc.stop();
}

void loop() {
 long raw;

 ack = adc.send();
 if(ack) {
   // Start new conversion
   adc.write(0x8F);

   adc.receive(); // Sets R/W bit high
   raw = (long) adc.read(0) << 16;
   raw |= (word) adc.read(0) << 8;
   raw |= adc.read(0);
   Config = adc.read(0);
   //Serial.print("Config: ");
   //Serial.println(Config,HEX);  // Configuration
   while(Config != 0x0F) {  // Loop waiting for the conversion to be finished
     Config = adc.read(0);
   }
   adc.read(1); // last poll

   adc.receive(); // Sets R/W bit high
   long data = adc.read(0);
   raw = data1 << 16;
   // check for negative values
   if(!data1) {
     raw |= (word) adc.read(0) << 8;
     raw |= adc.read(0);
   } else {
     raw = 0;
   }
   Serial.print("  Raw: ");
   Serial.print(raw);
   adc.read(1);
   adc.stop();

   long ir = (raw/8)*5; // Remove the 8x gain and multiply by 5W/m2 per mV
   Serial.print(" Irradiance: ");
   Serial.print(ir);
   Serial.println("W/m2");
 }
}