Category Archives: Weather

1-wire Power Injector

One way that I get reliable 1-wire communication is to use a power injector. This means that the along with the 1-wire signals, I inject a supply voltage into additional cable pairs.  Since  the most readily available commodity cable currently is 8 conductor Cat-5 cable, chances are that you also use this cable for 1-wire cabling and have spare wires going unused. Dallas called the 1-wire bus a MicroLAN, so I will continue to use their terminology here.

For short runs, where the total length of the MicroLAN is < 30 meters and the network only has maybe 6 to 10 slave devices on it, cable selection for use on the MicroLAN is reasonably simple, as even flat modular phone cable can work with a small numbers of 1-Wire slave devices. However, the longer the MicroLAN, the more important cable properties and therefore cable selection becomes more critical.  However, since I’m dealing with 30M or less, I’ll focus on that here and leave the research for  better cable techniques for when I run into a problem.

One of the remaining issues with using 1-wire driven from an Arduino pin directly is the fact that it is also providing power to all of the devices on the the 1-wire LAN when in parasitic mode.   While 1 or 2 devices are of little concern, I started to worry about building a reliable 24/7 full time  MicroLan configuration that would report weather data from 8 or so sensors, most of which were located at the garden bed outdoors.  It became clear that some of my devices were really not performing as desired  when “parasitically”  powered when I finally put all the pieces together.

Experimentation showed me that the signal from the Arduino pin was just not up to that task of powering and communicating with all the devices without some random miscommunication here and there.  I would get values like 185F and not the 70F I was expecting from a 1-wire temperature sensor, for example.

My solution was reasonably simple.  I would create an external “middle-man” device that would insert power into the CAT-5 cable after leaving the Arduino board.  Its really just a dual supply from typical commodity wal-wart power source.  The unmodified 12VDC supply will be fed into the cable on a spare pair and treated as “unregulated” and a 5V regulator will be added to another spare pair and considered to be “regulated”.  All code will now be written to comply with the “non-parasitic” mode of 1-wire.

The Design:
1wire-power-injector

The PCB:
PI

Current Weather Station

Arduino Mega

AAG One Wire Weather Instrument Series 2. (Wind Speed, Temperature, Wind Direction)  DS18B20 temperature, DS2423 (wind speed) DS2450(wind vector)

Hobby Boards: Humidity/Temperature Board – Built from my own parts

Rain Gauge (See my post about it)

BMP085 – Pressure Sensor

Internal Temperature via DS18B20

1-Wire Power Injector (see my “how to make a PCB guide):

In progress: COSM upload – nope.

FAIL:  Adding COSM/PACHUBE support ran me out of working RAM… so project for upload of data is scrapped at the moment.

Updated Rain Gauge

Newest Rain Counter

So I’m taking the time to re-do the rain gauge. The existing one clearly had some issues (hacks/fixes) and I wanted to also gain the use of counter B (if possible) for a different piece of weather gear.

Rain Counter version 7

I was also able to get a hold of some “liquid tin” from MG Chemicals to see if it compares well to Cool-AMP powder as a copper plating solution.

Results:

While the Liquid Tin is much easier to apply, I actually don’t like it as much as the Cool-Amp coating.

Weather Station – Hacked Rain Gauge

Converting a broken  LaCrosse WS-9004U 915MHz WIRELESS RAIN GAUGE into a “wired” 1-wire  0.02 inches per tip rain gauge.

The Lacrosse unit is an inexpensive rain gauge available from my local Fry’s store for about $24.  They are unusually finicky when being set up so I already knew I didn’t really have a lot of love for the device.  After about 6 months, my unit stopped sending/receiving updates and no amount of new batteries, resets or pairing attempts seemed to help.  I tossed it into my junkbox of parts.

Right around the time that MAXIM decided that making 1-wire devices was “tooo haaarrd”, (said in a whiner voice) and they killed off the DS2423 counter device, I decided to buy some before they were totally gone to see if I could make my junkbox tipping bucket hardware work in a “wired” mode.

It wasn’t really that hard.  I used a sample design from my WEATHERTOYS book, (AUTHOR: Tim Bitson), and  after ripping all the guts out of the device (wireless sender and battery holder, I had some room for some 1-wire related parts.

The original PCB  fits in a nice slot in the plastic case and used a reed switch aligned with a magnet attached to the center of the tipping bucket.  I used the dimensions to create a replacement board and everything fit except the CR2032 battery needed by the counter to not lose data when 1-wire was shorted or disconnected.

Original Board:

The CR2032 nicely fit into the area where the prior AAA batteries were housed.

The 1-wire connection is made by using a weatherproof gel-filled crimp splice available from the local hardware store.

The code (java in my case) is straight from the Weathertoys book with adjustments made relating to the side of the tip, 0.02 inches per tip.

Installation

The Guts

The main board and battery board

The backside

Note: Not my best work…  I attempted to re-orient the reed switch during testing, which only managed to a) break a reed switch and b) break a copper trace (fixed with blue wire).  In addition, I actually installed the wrong SMD part that I thought was a DS2423… which turned out not to be the case…

I suppose I could make a new board… but once this one started working… I just left it alone.

The gauge is currently off the station since I forgot to paint the plank, made from ash, used to mount the sensor.  Small ants were using the aging cellulose to create a fungus garden under the cover of the rain collector.  Silly me.  Now I must replace and paint the replacement wooden board.

Next Step.  Stop using a Linux server and JAVA and convert over to an Arduino  MEGA2560 as a weather station controller.

Weather Station Woes

I might have mentioned that my AAG WS603B Weather Station turned out to be the most oddball replacement for my original 1990’s Dallas 1-wire Weather instrument.  It worked reliably for 10 years.  I had version 1.0, which was rather difficult to program for as it used the DS2401 serial number for each wind direction.  This required a lot of extra  1-wire coding.

I’ll spare you some of  the gory details regarding the quality of the latest AAG unit though I should mention that it was incompatible with all 1-wire wiring standards to date.

The real problem with the new unit (other than reliability) was that the new case is clear (no so bad) full of colored LED’s (can you say SILLY?) and not water resistant. Even though the case seemed closed and the PC board was “sealed”, it still managed to die after 6 months of usage.   It took 6 months of trying to get the Windows software from them for the station… by then… the device was dead. I’ll not be buying anything else from AAG.

Anyway, one of the design ideas that AAG implemented was actually quite sensible. They used HALL sensors for Wind Direction instead of the fragile reed switches as used with other station designs. The Dallas Version 2 weather station (formerly sold by AAG) used a DS2450 and reed switches to track 16 positions of wind direction with just 1 magnet rotating. Only 8 sensors were needed as the intermediate positions were resolved when the magnet straddled two sensors at once.

Dallas Weatherstation version 2 Schematic:

I attempted to recreate this solution with Hall Effect Sensors. My first test circuit is below. Seems to work well.

So now my plan is to rebuild a replacement for the failed board in the WS603 housing.

Barometric Pressure

While looking at low cost pressure sensors in the Mouser Electronics catalog, I located the FREESCALE MPXAZ6115A as a possible sensor for my project. The sensor has the following statistics;

Device: MPX6115, MAX PSI 16.7, MAX kPa 115.

Since barometric pressure here hovers at around 100kPa or so, this sensor would do just fine. The analog output of the sensor is relative to the min/max pressure range of the sensor.

According to my initial tests, the sensor would output about 4.06 volts at 100kPa.

The built-in analog input on the Arduino would also keep the circuit simple and after a few tests I was able to determine the offset value I needed to get correct readings for the localized barometric pressure.
In my case… testing showed that the magic number is 0.13.

// Nominal Transfer Value:
// Vout = VS x (0.009 x P – 0.095)
// ± (Pressure Error x Temp. Factor x 0.009 x VS)
// VS = 5.1 ± 0.25 Vdc

float Vin;
float P;

void setup()
{
Serial.begin(9600);
}

void loop()
{
Vin = (5.0/1024.0) * analogRead(0);
Vin = Vin + 0.13;     // Offset Adjustment
 Serial.print(Vin);
 Serial.println(" Volts");

P=((Vin/5.0)+0.095)/0.009;
 Serial.print(P);
 Serial.println(" kPa");

Vin = (P * 0.2952999);
 Serial.print(Vin);
 Serial.println(" Inches of Mercury");
delay(2000);
}

I’m using a LADYADA Boarduino on a solder-less breadboard for testing. The sensor hookup is dead simple with only one exception that makes it tricky. The part I selected is designed to be surface mounted.

I decided to create a carrier board using the board layout software I prefer called SprintLayout from ABACOM in Germany. Other than 5V power and ground connections, the Vout from the carrier board goes directly to the Arduino Analog(0) pin.

To create the PCB board, I use the GOOTIE toner transfer method to apply the layout on the PCB for etching. (google search “gootie PCB” for more info)

Having developed a dislike for the chemical etchant that Radio Shack sells; Ferric Chloride, I have also adopted the etchant that
Gootie describes.

It is based on the swimming pool chemical Muratic Acid and Hydrogen Peroxide in a 1 to 2 ratio. It’s fast, non-opaque and does not require heating or excessive agitation.

Note: I also recently picked up a used GBC Creative Laminator at the local Goodwill for $14.00. It does an excellent job of applying the toner to the copper on the PCB to be etched. Using an hand iron was OK, but the results were not always predictable.

My method of making PCB’s at home.

 

Here is the Layout: