WiFi Weather Station

June 9, 2019 Andy Allen

I have a weather station problem. Well, a series of problems, really. I think I’ve had 3 of them over the past several years, but they all suffer from one or more problems.

They don’t tell me Dew Point
Relative humidity is inferior and I’ll fight anyone who says otherwise.
They lose connectivity to their sensors
I don’t need hyper-local data. I just want to know the temperature in my neighborhood, not literally in my yard. There’s no reason to deal with this.
They eat batteries
The outdoor temperature sensor in particular seems like it’s always crying for a new battery.
They’re ugly
Who wants THIS on their wall??

If I build my own, I can fix this. My aesthetic is mid-century modern IoT. I think it’s funny to make an internet-connected weather station and then put it in a retro-looking container with analog gauges.

The Architecture

Processor

I have a Particle Photon that’s not assigned to anything at the moment. It’s cheap, wi-fi enabled, powered directly by USB, and you can flash new software to it over the web. It’s actually really powerful and at $19, it’s not even the most expensive thing in the build.

Gauges

For the all-important dew point, I’m using a Juken X27.168 Stepper Motor. It has internal stops at each end of a 315° rotation, so it’s easy to find its zero position at start-up.

I want to display temperature and daily high temperature on one scale. To do that, I need a Juken X40 879B dual-spindle stepper motor.

They’re very low torque, but they can be driven at USB voltages, so it keeps the power situation simple.

Cabinet

I need a nice looking container to put the whole thing in. I like the look of this clock. It’s more than 8 inches across, so there should be plenty of room for everything inside.

I’m replacing the face, but I’ll reuse a couple of the hands.

Building It

Software

The Photon has plenty of memory for an HTTP Client, a JSON parser, and a library to drive stepper motors. Everything can be globally scoped and reused as long as the thing is powered on.

The first free weather API I found is OpenWeatherMap. It doesn’t have dew point explicitly, but as everyone knows:

        Cγ_(T,RH)
T_dp = ⎼⎼⎼⎼⎼⎼⎼⎼⎼⎼⎼⎼
       b - γ_(T,RH)

where

         ln(RH/100) + bT
γ_(T,RH) = ⎼⎼⎼⎼⎼⎼⎼⎼⎼⎼⎼⎼⎼⎼⎼⎼⎼
             C + T

and

b = 17.67, C = 243.5, 
T = temp (°C),
RH = relative humidity

So computing dew point and stepper motor positions is just a little algebra. I passed Algebra last century, and I still remember a lot of it.

The firmware is at https://github.com/powerfulmojo/WeatherMojo.

Particle variables and functions are documented here.

Electronics

The X27.168 and the X40.879B are driven by L293D dual H-bridges. They’re about 80¢ each when you buy 10. They have built-in flyback diodes to protect the Photon from any inductive-load-related nastiness that comes back from the motors.

Here’s what the design looks like:

It’s a lot of wires, but it’s not very complicated. Each H-bridge has 4 power pins, 4 ground pins, 4 processor connections, and 4 motor connections. I was able to squeeze everything onto a 24x18 perfboard.

Since the X40 has front contacts, I soldered that to its own board and used right-angle headers to connect jumper wires to it.

I had to fiddle with pin assignments to get the motors spinning in the right direction, but that’s all software. I did it after everything was soldered in place.

The Gauges

The face is a stainless steel disc I got from an eBay seller. I drew up some concept gauge layouts to see what I liked.

I printed the gauges on some water decal paper on my laser printer & applied it to the steel in one piece.

The hardest part of the project was actually attaching the gauge hands. The contacts on the temperature motors are on the front, so all of the connections have to be made between the motor body and the back side of the face. The spindles only stick out a few millimeters in the first place, so I spent a whole day driving back and forth to the hardware store trying out parts I might use to mount the hands to the motor.

If you try this, save the clock mechanism. It already has parts that fit the clock hands. That way you only have to adapt them to fit the motors.

The Finished Product

I love it! I probably ended up spending as much as I would on a decent weather station, but this one shows me data I actually care about.

The big hand shows current temperature while the skinny one shows the highest temperature seen today. Most of the time they’re overlapping but as the temperature cools down, it shows you how hot it got today. It resets every morning at 9:00am so you still have a little time to see yesterday’s high at breakfast.

Apple Keyboard Power Cable

February 23, 2014 Andy Allen

Power a keyboard with USB instead of batteries

I have an Apple Wireless Keyboard that I love, but it runs on batteries. This little project is how I put a USB wire on it to make it work without batteries. It's still a Bluetooth keyboard, it just runs off of USB power.

The Parts

To complete this, you'll need:

2 AAA to AA Battery Holders
AMS1117 3.3V 1A Voltage Regulator
USB Cable (any kind)
An Apple battery cover (optional)
Soldering iron, wire, hot glue, and miscellaneous tools

I got all of it for about $16 (not including the keyboard).

Assembly

1. Drill the metal bits

You'll have to drill holes in the battery cover and the "negative" end of one dummy battery. This was by far the trickiest part.

Make the holes just big enough for your USB cable to pass through with the insulation. I was not confident in my ability to nail this on the first try and I didn't want this to be a permanent conversion, so I bought another battery cover. If you're really committed, you could drill the one that came with the keyboard.

2. Run the wires

Here's what I'm calling the wires involved:

New red: A wire you supply that will connect the Vout pin of the AMS1117 to the positive contact inside the dummy battery.
New black: A wire you supply that hooks together the ground of the USB cable, AMS1117, and negative contact inside the dummy battery.
USB red: The red wire that's already inside the USB cable. It will supply +5V to the AMS1117
USB black: The black wire that's already inside the USB cable. It is the ground for all components.

You'll need to drill holes in one end of each battery so a wire can run between them. Run the new red wire all the way through one of the dummy batteries.

Cut the end off of your USB cable and find all of the wires.

Push the USB wire through the drilled battery cover and negative battery terminal. If you buy the same dummy batteries I did, the ends pop out. That's useful for soldering wires to the metal bits without ruining the plastic bits. You have to solder:

The new red wire to:
- the positive end of a dummy battery and
- pin 2 of the AMS1117
The new black wire to:
- the negative end of the other dummy battery and
- the usb black and pin 1 of the AMS1117
- The usb red to AMS1117 pin 3

Here's the pinout from the chip's data sheet:

When you're finished, it will look kind of like this:

3. Test it

Hook the USB cable up to a power source and make sure you get 3.3V between the positive and negative terminals of your dummy batteries.

4. Glue it

When I was done with soldering, I put a glob of hot glue on the usb wire and pulled it down into the dummy battery so it doesn't move. Then I assembled the batteries and glued the gap between them to keep them stuck together.

The battery cover is not glued to anything. It can spin on the cable. To hook it up, just take the batteries out of your keyboard and put this thing in instead.

Standing Desk Converter

February 17, 2014 Andy Allen

I've been thinking a lot about health lately, thanks to Clairvoyant's health challenge. And I've been reading about the health benefits of a standing desk (or maybe not). You can buy standing desk converters like these:

But they start at $75 & they don't look great, so I decided to build one. I prototyped it with stacks of books and came up with a 12" rise for my keyboard and a 20" rise for my laptop on its mStand. The mStand isn't required, but I use it when I'm sitting so it might as well stay. 18" wide is enough to hold a keyboard and trackpad comfortably.

I built it out of some 3/4" cherry ply I had leftover from the desk. I routed the edges so they're rounded and I put a brace at the top of the back to hold everything square.

If I were to build it again, I would change some things:

move the legs in about 3/4"
have shelves overhang the riser by about 3/4"
blend the roundover more carefully into the lower shelf
use a wider brace so it's more certainly square

But that stuff is minor. I'm pretty pleased with it.

There, I Solved It

September 14, 2009 Andy Allen

I was inspired by http://thereifixedit.com. I feel a kinship with the innovators whose work is featured there. After all, I just expanded my universal remote, and they built one from scratch. I decided to document my own attempts at problem solving. Hopefully the engineering is better than theirs, but I feel confident my problems had less reason to be solved in the first place.

The Problem

Sometimes I forget to close my garage door. I've left it open all day and all night. This is a perfect example of a problem that barely needed to be solved in the first place. Right in my wheelhouse.

Solving It

What I need is an unmissable indicator of my forgetfulness. I'm thinking of a light that comes on when the garage door is open. Ideally, one that I can see while watching TV or laying in bed.

Design

The concept is simple: There's an LED, a power source, a switch, and a couple resistors.

Parts

The switch is a magnetic reed switch. It's normally open, but if a magnet cozies up to it, it gets all closed. Perfect. I don't know the model number, because I just found it in my garage, left over from when they installed the security system. You could get one for about $3, and you could go wireless for $50.

The power source is a couple of AAA batteries. I happened to have a battery holder that size lying around, so that's what I went with. I think you could get a new one for $2.

The LED is a panel mount model I bought just for this. It is green, and snaps into a 1/4" hole.

According to my math, I need a minimum of 40 Ω of resistance to avoid burning up my LED @ 3V. All I had were a couple of 100 Ω resistors, so I went with that. In parallel, that gives me 50 Ω plus the resistance of the wire & switch, so I won't burn the LED up.

Then there's some miscellaneous wire, solder and heat-shrink tubing to keep everything neat. Oh, and a magnet. I happened to have one of those too.

Assembly

Most of the stuff sits inside the house. I picked out a nice location in a closet and drilled a hole for the LED. I soldered up the battery pack, resistors, LED, and the leads to the switch and fed them through into the garage.

At the other end, I just had to stick the switch in place and mount the magnet to the garage door. And by "mount," I mean "stick it to the garage door because it's a magnet."

Victory is Mine

After just a little tweaking to get the position of the magnet just right, I have a working light. It has already reminded me to close the garage door once, and I expect it will have a long life saving me from myself.

The "Do My Job" Button

June 22, 2009 Andy Allen

a.k.a. The USB Easy Button

It's A Fundamental Truth of Requirements Gathering that the longer you talk to someone about custom software, the more their answer boils down to two requirements:

Make me a button called "Do My Job."
Make it do my job.

Then they get thoughtful for a moment and add a third requirement:

Make sure only I can press the button.

Some of them will even joke "You know, like an Easy Button."

Those Easy Buttons from Staples are awesome, but they have one small flaw: they don't actually do anything useful. It's my aim to change that.

Let's Make a Button!

What we need is an Easy Button Hack, so I'm going to build a USB Easy Button. I found most everything I needed from jro's project on instructibles and a flickr photoset by tommybear. Plus, I've been dying to try one of these U-HIDs.

What you'll need:

An Easy Button (or Botón Fácil if it suits you better) $4.99
A U-HID Nano and USB Cable (with shipping) $42.00
A Modular Harness for the U-HID (optional) $9.00
Soldering iron & solder
Wire (if you didn't buy the wiring harness)
Dremel or a chisel
Hot glue or silicone
Small phillips screwdriver
PC running Windows XP

Step 1: Program your U-HID Nano

Plug the USB cable and wiring harness (if you're using it) into the U-HID Nano. If you're going to have the button send a single command like me, just leave the black (ground) wire and the gray (pin 10) wire long enough to reach the PCB. We won't be using the other 7. If you want to use the button to close more than one switch at a time (for example, to send "Ctrl + Alt + Del"), leave one wire per button, plus the ground wire.

I hooked the wires up to a temporary switch at this point for programming the thing. You could go ahead and hook it up to the Easy Button. Just scan down a little to see which contacts to use.

You program the U-HID Nano with U-Config, a software package available from the manufacturer. It's a pretty easy process, and the Technical Manual was easy to read.

I'm not going to go into the details here except to note that a driver install and firmware update were required to get it working on my machine. Both of those processes are clearly documented on their site. I set it up so that when pin 10 goes to ground, it will send the macro "L Alt, F8". It seems to send the scancodes fast enough that my machine counts it as a combination keypress.

Step 2: Take the Easy Button Apart

Turn the button over, and you'll see four black pads on the bottom. Pull them off to expose the screws. Be sure to save the pads so you can stick them back on. Go ahead and take the batteries out while you're there.

Remove all 4 screws to release the silver ring and red button from the assembly. You'll be left with just the guts of the machine. Inside, you should see a white button. That is the heart of the Easy Button and the only part of the original electronics we're actually going to make use of.

All of these things have to go, so desolder and discard them:

The black capacitor
The resistor closest to the button
The red speaker wires
The black & white power wires

If you don't know what those things are, don't sweat it. Take a look tommybear's photos on flickr. He's better at this than I am.

Remove the 2 screws, and take the PCB off of the assembly. Set the metal spring aside. We want to keep that because it gives us a satisfying CLICK when the button is pushed.

Remove the 4 screws that hold the little mezzanine level on. You may have to pry it up just a bit if the hot glue below is sticking it down. Take the metal slugs out and discard them. They're just stuck in there with a little hot glue and you can pick them out without ruining anything. I guess you could leave them if you have the room & like the heft. Remove the speaker too.

Use your dremel or a chisel to take out any little plastic bits that are in your way. Just remember that you have to leave the 4 posts that hold the mezzanine level up.

Step 3: Put the U-HID in the Easy Button

Find a way that the whole U-HID assembly will fit inside the button, then use some hot glue or silicone to stick it in place.

Replace the mezzanine, metal spring, and PCB. Now comes the soldering. The U-HID Nano is going to just sit there and wait for pin 10 to touch the ground pin. We're going to use the normally-open momentary switch inside the Easy Button to interrupt that connection, so the circuit is only closed while the Easy Button is depressed. Solder the gray and black wires as shown.

It would be a good idea to test it at this point to make sure it works before you put everything back together. Use a little hot glue to make sure the wires do not interfere with the holes in the PCB. Those are what keep the button aligned properly, and if they're blocked, you can't push the button.

Step 4: Let The Wire Out

Cut a slot in the silver ring that's just big enough for the wire to get out of the enclosure. I made mine a left-handed Easy Button (made to sit in the left side of the computer). You could hook yours up backward or cut a channel through the battery compartment and have it come straight out the "front."

Step 5: Replace the Ring and Button

It should only go back together one way: with the battery compartment farthest away from you, the "easy" label should be right-side up and the Staples logo on the ring will be directly toward you. Replace all 4 screws in the bottom of the case and use a little hot glue to stick the rubber pads back on. You're done with the hardware: you've built an Easy Button that actually does something.

Step 6: Write the Software that Actually Does Your Job

If you're able to buy software that does your job, you could use that too. Implementation of a security system so that only you can push your button is left as an exercise to the reader. For extra style points, make sure your software ends up playing a "That was easy" sound effect when the job is done.

Step 7: Assign a Hotkey to Your Software

If you are having the button send a key combination that is recognized by the system already (like Ctrl + Alt + Del), you can skip this step.

Create a folder in your Start Menu called "Easy Button" and create one shortcut inside the folder called "thatwaseasy.lnk". Right-click the shortcut and select Properties. Click the "Shortcut key" field and press the hotkey combination you told the Easy Button to send (in my case, Alt + F8). This shortcut has to be in your start menu or on your desktop. You can set the shortcut property of any shortcut, but it only works if the shortcut is in the right place.

TA DA!

For about $50 you could have one just like this. If you pay close attention to jro's project, you could use a recycled keyboard instead and do this for about $5. Now get out there and build your own.

Win32::AmbientOrb

June 15, 2009 Andy Allen

This page describes a Perl module I wrote to support a serial-port controlled Ambient Orb. See the Original Post for details. If you just want to get the package, you can download it directly or grab a PPD.

Manipulate an Ambient Orb through a serial port

NAME
SYNOPSIS
EXAMPLE
DESCRIPTION
EXPORTED HASHES
METHODS
AUTHOR

NAME

Win32::AmbientOrb - Manipulate an Ambient Orb through a serial port

SYNOPSIS

  use Win32::AmbientOrb;

EXAMPLE

  use Win32::AmbientOrb qw(:ALL);

  my $port = "COM1:";
  Win32::AmbientOrb::Port($port); #set port to COM1
  InitializePort();               #set serial port settings
  PagerIgnore(1);                 #don't listen to the pager network

  # go straight to red
  my @red = (176, 0, 0);
  DirectColor(\@red);

  # quick transition from blue to red
  my @blue = (0, 0, 176);
  TransitionColor(\@blue, \@red, 20, 5);

  # slow transition from red to blue
  TransitionColor(\@red, \@blue, 50, 100);

  # clean up
  CloseOrb();

DESCRIPTION

The Win32::AmbientOrb module lets you manipulate an Ambient Orb connected to a Win32 machine through a serial port.

It uses Win32::SerialPort for serial port communication.

EXPORTED HASHES

Two hashes are exported by default, %OrbColor and %OrbRGB.

%OrbColor

Maps from color names to Color IDs useful for ColorAnim( ). Available color names are:

 • Red          => 0
 • LightRed     => 1
 • DarkOrange   => 2
 • Orange       => 3
 • LightOrange  => 4
 • DarkYellow   => 5
 • Yellow       => 6
 • LimeGreen    => 7
 • PaleGreen    => 8
 • GreenMinus3  => 9
 • GreenMinus2  => 10
 • GreenMinus1  => 11
 • Green        => 12
 • GreenPlus1   => 13
 • GreenPlus2   => 14
 • PaleAqua     => 15
 • Aqua         => 16
 • DarkAqua     => 17
 • Cyan         => 18
 • DarkCyan     => 19
 • LightBlue    => 20
 • SkyBlue      => 21
 • BlueMinus2   => 22
 • BlueMinus1   => 23
 • Blue         => 24
 • DeepBlue     => 25
 • VeryDeepBlue => 26
 • Violet       => 27
 • Purple       => 28
 • LightPurple  => 29
 • Magenta      => 30
 • MagentaPlus1 => 31
 • MagentaPlus2 => 32
 • MagentaPlus3 => 33
 • MagentaPlus4 => 34
 • MagentaPlus5 => 35
 • White        => 36

%OrbRGB

Contains the mapping of color ids to RGB arrays useful for DirectColor( ).

METHODS

Port( [$port] )

Sets or returns the port that the Ambient Orb is connected to. Default is COM1:

InitializePort( )

Initializes the serial port for communication with the Ambient Orb. Settings are:

 • BAUD: 19200
 • PARITY: N
 • DATA: 8
 • STOP: 1

ClosePort( )

Cleans up the port object.

ColorAnim( [$color], [$animation] )

Changes the color of the Orb using the default color and animation settings. Colors are listed above in %OrbColor Animations are 0-9

        ANIMATION
 • 0 almost imperceptibly slow
        ...
 • 7 very fast
 • 8 crescendo effect
 • 9 heartbeat effect

DirectColor( \@rgb )

Sets the orb instantly to the color specified in the 3-element array @rgb. Red, green and blue values should be between 0 and 176.

TransitionColor( \@rgb0, \@rgb1, $steps, $wait )

Performs a slow transition from the color in @rgb0 to the one in @rgb1. Sets the orb instantly to the color specified in the 3-element array @rgb0, then sends $steps updates, one update every $wait milliseconds until it reaches the color specified in @rgb1. Red, green and blue values should be between 0 and 176.

PagerIgnore( $ignore )

Instructs the orb not to listen to the pager network. If you don't call this, the orb will keep changing its color to reflect whatever channel it was listening to before you plugged in the serial port.

AUTHOR

Andy Allen

Orb Walkthrough

June 8, 2009 Andy Allen

This page goes through how to set up your serial-controlled orb. The sections are: What You Will Need Connecting the Hardware Installing Perl and Required Packages Creating a Color Script Installing the Service Web-Connected Orbs Other Uses For the Orb

You Will Need:

HARDWARE

A Windows XP or 2003 Server host
An Ambient Orb (or several)
A Hardware Developer's Kit from Ambient (one per orb)
Long enough serial cables to put the orb where you want.

SOFTWARE

Perl (I used ActiveState). I've included ppm commands below.
Aldo Calpini's Win32::API package ppm install Win32-API
Dave Roth's Win32::Daemon package ppm install http://roth.net/perl/packages/Win32-Daemon.ppd
Bill Birthisel's Win32::SerialPort package I didn't use a ppd, I installed it the old-fashioned way (as described in the README, but you could try finding a ppd with a Google search
Win32::AmbientOrbppm install http://powerfulmojo.com/tools/ppm/Win32-AmbientOrb.ppd
The service files in AmbientOrbService.zip

Connecting the Hardware

Plug the serial exension cable (if you're using one) into your serial port.
Plug the HDK into the serial extension cable
Plug the Orb into the HDK.
Plug the Power adapter (supplied with the Orb) into the HDK.

The orb will go through its regular power-up routine, then start displaying the DJIA channel (or whatever it was most recently tracking).

Installing Perl and Packages

Install Perl and get all of the packages above installed in the order listed. If you use the ppm commands provided, it will link the modules' documentation right into your html documentation so you can see how to use everything. You will only be using Win32::AmbientOrb directly. It depends on Win32::SerialPort (which depends on Win32::API), and we will need Win32::Daemon to get it up and running as a service.

Creating a color script

The service will need a script to tell it what color to turn the orb. A color script can be written in just about any language, it just has to print an integer to the console representing the correct color. Possible colors are:

  	
      0
      Red      24      Blue
  
          3
      Orange      27      Violet
  
      6
      Yellow      30      Magenta
  
      12
      Green      35      Red-Magenta
  
      18
      Cyan      36      White

You can use any integer between 0 and 36. The ones I didn't list work out about like you'd expect (i.e. 9 is yellowish-green). My color script is in Perl and is available below. A blue orb means all builds are successful, a red orb means at least one build is failed, and a yellow orb means the CruiseControl server was unreachable.

Here is an example of a (very simple) batch file color script:

@echo 24

It's not very interesting, because it just returns the code for "blue." The default script looks like this:

 #!perl
 use strict;
 use LWP::Simple;
 4
 my $color = getColor();
 print $color;
 7
 sub getColor {
     my $color = 6;
    for (my $retry = 0; $retry < 3; $retry++) {
        if ($color == 6) {
            sleep 1 if ($retry > 0);
            $color = getColorRequest();
        }
    }
    return $color;
}
 18
sub getColorRequest  {
    my $color;
    my $dashboard = get('http://cchost/ccnet');
    if ($dashboard) {
        if ($dashboard =~ />Fail/) { $color = 0; }
        else { $color = 24; }
    }
    else {
        $color = 6;
    }
    return $color;
}

It is a little more interesting: it makes a request to a server named cchost and searches the response for the word "Fail." If it finds it, at least one build is failed and the orb should be red. If no builds are failed, the orb should be blue. In the event that the server is unavailable, it will retry the request twice more, but if it still can't reach cchost it will turn the orb yellow.

Installing the Service

Now that the orb is hooked up and you've got your color script written, here's how to install the service to make it go.

Put the files in place

Create a folder named C:\Program Files\AmbientOrb and copy everything in this .zip into it: AmbientOrbService.zip. Put your color script in the same folder.

Install the Service

You will need the script installOrbService.pl (or one like it) to install the AmbientBuild Windows Service on your host.

 #!perl
 use strict;
 use LWP::Simple;
 4
 my $color = getColor();
 print $color;
 7
 sub getColor {
     my $color = 6;
    for (my $retry = 0; $retry < 3; $retry++) {
        if ($color == 6) {
            sleep 1 if ($retry > 0);
            $color = getColorRequest();
        }
    }
    return $color;
}
 18
sub getColorRequest  {
    my $color;
    my $dashboard = get('http://cchost/ccnet');
    if ($dashboard) {
        if ($dashboard =~ />Fail/) { $color = 0; }
        else { $color = 24; }
    }
    else {
        $color = 6;
    }
    return $color;
}

You only need to run this script once to install the service. It takes one argument: -i for install or -u for uninstall, then prints a short status message indicating whether the action was successful.

Line 14 points to the script that will be started when the Windows Service starts. That's our orbService.pl script. If you write your own Windows Service, you would put its pathname here, using the short pathname (with no spaces). If you fill out a username and password in lines 12-13, the script will run as that account. If you leave it blank, it will run as System.

Note that line 14 should contain the orbService.pl script as written, not your color script.

Configure the Service

The service looks for a file named orbService.config, which can have any of the following values. If a value is not specified, it will use the default listed.

KEY

Default Value

Description

colorScript

C:\Program Files\AmbientOrb\ccnetWeb.pl

prints orb color code to STDOUT

webOrbs

[n/a]

serial #s of other orbs, comma-separated, no trailing comma

port

COM1

serial port to use

logFile

C:\Program Files\AmbientOrb\orb.log

verbosity

0 = silent, ..., 3 = garrulous

maxLogLines

10000

maximum lines to leave in log file

pollingInterval

30000

wait between updating the orb (ms)

sleepTime

2000

between processing service messages (ms)

waitForStop

20000

how long to expect service stop to take (ms)

errorColor

color to use on colorScript error (0-36)

errorAnim

animation on colorScript error (0-9)

While the service is running, changes to this configuration file will be automatically loaded the next time the script checks its messages (within 2 seconds by default). If you leave the service configured as shown, it will run the ccnetWeb.pl script every 30 seconds.

Web Orbs

In my environment, we have three orbs and only two HDKs. The service is installed on both hosts with HDKs attached, but that leaves one orb out of the loop. To update it, one of the servers is also configured to update the orb using its webOrbs configuration parameter. This orb is updated using Ambient's Web Developer API.

Other uses

This service was written to make it easy to monitor anything using an Ambient Orb. With a different color script and polling interval, you could make it monitor just about anything. During testing I had it monitor hockey scores, crude oil prices, and network traffic (not all at the same time). If you do something like that, I'd like to hear about it, so please let me know.

#!perl
###########################
# installOrbService.pl
###########################
use Win32::Daemon;
my $svcName = 'AmbientBuild';
%Hash = (
    name    =>  $svcName,
    display =>  $svcName,
    description => 'Ambient Build Monitor',
    path    =>  'c:\\perl\\bin\\perl.exe',
    user    =>  '',
    pwd     =>  '',
    parameters =>'c:\\Progra~1\\AmbientOrb\\orbService.pl',
);

my $action = $ARGV[0];
if (lc($action) eq '-i') {
    if( Win32::Daemon::CreateService( \%Hash ) ) {
        print "Successfully added.\n";
    }
    else {
        print "Failed to add service: ",
         Win32::FormatMessage(Win32::Daemon::GetLastError()),
         "\n";
    }
}
elsif (lc($action) eq '-u') {
    if( Win32::Daemon::DeleteService( "", $svcName ) ) {
        print "Successfully removed.\n";
    }
    else {
        print "Failed to remove service: ",
         Win32::FormatMessage(Win32::Daemon::GetLastError()),
         "\n";
    }
}
else {
    print "Unknown action ",
    "(use -i for install, -u for uninstall).\n";
}

Note that line 14 should contain the orbService.pl script as written, not your color script.

Configure the Service

The service looks for a file named orbService.config, which can have any of the following values. If a value is not specified, it will use the default listed.

KEY

Default Value

Description

colorScript

C:\Program Files\AmbientOrb\ccnetWeb.pl

prints orb color code to STDOUT

webOrbs

[n/a]

serial #s of other orbs, comma-separated, no trailing comma

port

COM1

serial port to use

logFile

C:\Program Files\AmbientOrb\orb.log

verbosity

0 = silent, ..., 3 = garrulous

maxLogLines

10000

maximum lines to leave in log file

pollingInterval

30000

wait between updating the orb (ms)

sleepTime

2000

between processing service messages (ms)

waitForStop

20000

how long to expect service stop to take (ms)

errorColor

color to use on colorScript error (0-36)

errorAnim

animation on colorScript error (0-9)

Web Orbs

Other uses

Ambient Orb Setup

June 1, 2009 Andy Allen

This page describes the process of hooking an Ambient Orb up to show a real-time information using the Ambient Hardware Developer's Kit.

Previous Work

The Orb provides a great, at-a-glance indication of build status for a continuously integrated software project. The subjects of Continuous Integration and its benefits are discussed in a number of other places, so I won't cover them here. Likewise, the use of an Orb using the Ambient Web API is covered admirably by Michael Swanson, so you can read that there.

Problems

The two problems I found were:

Update lag

The Ambient pager network does not change the color of an orb for 10 - 30 minutes. I can break a build in seconds. An ideal system would indicate the status of software builds in real time with no lag.
Integration of multiple software projects

I work in a shop that produces a lot of different projects at a time. We currently have eleven builds managed by CruiseControl.NET. Using NAnt events as described elsewhere did not provide an easy way to measure how many builds were successful moment-by-moment.

Solution

Lucky for me, the first problem can be solved with hardware. As a software guy, I love when that happens. Ambient produces a Hardware Developer's Kit. You put an Orb on one end and a serial port on the other, and you're in business: you can update the orb instantly to any color you like.

Unfortunately, the cable is only 3 feet long, making the orb's location a bit constrained. This was easily solved with a cheap 50-foot serial cable. As a bonus, the $45 (with shipping) HDK frees me from the $80/year subscription to Ambient's developer channel, so the hardware investment will pay for itself within the first year.

To have the Orb reflect the status of multiple software projects all at once, I polled the CruiseControl.NET server. For now, I'm using a screen scrape of the CruiseControl dashboard. With a little gumption, I could probably poll the same port that CCTray uses.

Keeping in mind that I might not always use CruiseControl as my CI solution, and that I might change the way I poll it for build status, I thought it would be nice to keep those changes insulated from the core orb updating service.

I'm a native speaker of Perl, so that's the language all of this is in. I used Win32::Daemon by Dave Roth to create a Windows service in Perl. The service talks on the serial port using Bill Birthisel's Win32::SerialPort package and updates the colors using my Win32::AmbientOrb package.

How to Do It Yourself

This Walkthrough describes how to set up the whole system.

Lazy Part 3

March 31, 2006 Andy Allen

Part 3: Talking to the Computer

For my next trick, I'll need an assistant: the USB-UIRT. This is that gadget that's going to receive infrared signals from the remote. I'll also need the Girder software to learn the codes my remote is sending and actually do something with them. For now, I'm just going to be satisfied with proving that my PC is hearing what the remote is saying. Making the software do anything about it is a matter for another day.

I'll only be needing a handful of buttons here. They're all standard play controls and appear on just about any remote.

Power (to open or bring focus to the application)
Play/Pause
Stop
Previous Track
Next Track
Shuffle
Repeat

Hardware Installation

Dude. That couldn't have gone much easier. Plug in the UIRT and the Plug and Play dialog finds it. You download the drivers from the USBUIRT site and tell the PnP wizard where to find them. Done.

Software Installation

This was also a cakewalk. Come to think of it, it was much easier than a cakewalk. I've participated in several of those and never walked away with a thing. Here there is no cake, but at least I have something to show for my effort.

Telling Girder About It

Go to File -> Settings, select "Plug-in settings," and check the box next to "USB-UIRT." Restart Girder and you're off & running.

Setting Up Your Remote

Go to Tools -> Add Remote Wizard. Click Next, name the remote, then start programming. It asks you to press any button to make sure it detects a remote at all, then it asks you to press each button that you want to use on the remote. After you finish this, Girder will be able to get signals from your remote.

All of this only took me an hour, and most of that was poking around trying to figure out what else Girder could be used for. I went ahead and picked buttons to use for song rating too. If everything goes smooth, I'll not only be able to skip sucky tracks, but ban them for good measure. Next, I'm going to make Girder listen for the signals from my remote.

Budget update: $143.44, 8 hours.

Lazy Part 2

March 27, 2006 Andy Allen

Part 2: Programming the Remote

I got the remote control in the mail today. It's a beautiful, silver OneForAll 8910.

Just to get things warmed up, I found the code for my TV and entered it. I got that up and running right away. Early win. Nice.

Uncharted Territory

I tried out the IR.exe and Remote Master software. It turns out there's one more handy piece of software, IrToWav. I'll explain that in a minute.

In brief, Remote Master creates the key mapping for the remote (which button should send what signal), IR.exe converts that map to a binary file the remote can understand, and IrToWav generates a .wav file that can load the upgrade to the remote.

How slick is that? The remote has a modem in it for one-way uploads of information. If you call OneForAll technical support, they can have you hold your remote up to the phone and load new devices and protocols. If you use IrToWav, you can do the same thing, but with a custom program you created. You don't even need the JP-1 cable to write to this remote.

There's a great tutorial on all of this on the hifi-remote site, and another good one here. The hifi-remote forums are also a gold mine of great information. If you just take two hours to read the sticky posts in the beginner forums and search for your particular remote, you'll be a beginner no longer.

To control the music player on my PC, I just selected a common universal code for a CD player (for some reason, code 0157 is used for 16 different CD players in my remote's catalog, but not for mine). I figure choosing a common code means there's a low probability that its carrier signal will be out of range for my IR receiver or some other nonsense. I plan to use these play controls to manipulate the Music Engine on my screen. I must say that this step of the process went smoother than I could have imagined. The remote is already controlling my whole home entertainment system (except PC), and I had a super-easy graphical interface to remap all of the buttons.

I didn't need the $14 JP-1 Cable to do this, but since I already ordered it, I'm adding it to the budget. I'll use it later when I'm in a nerdy mood.

Budget update: $39.84, 7 hours

A Different Kind of Lazy

March 26, 2006 Andy Allen

All is not as it should be with the world. When I'm on my sofa listening to Launchcast, I'm subjected to Nickelback songs. That's no way to live, and I'm willing to go to the very gates of hell and back to avoid walking 15 feet to hit the "Skip" button every time it happens.

The problem

Back when you had a TV that looked like this:

if you didn't like what was on, you just had to wait it out or get up. Wiser minds than mine realized early that this situation was untennable.

In June of 1956, the first consumer wireless remote control entered the home. It may be just coincidence that this came two months after As the World Turns was introduced on CBS.

Fastforward 50 years.

Now, an increasing amount of my entertainment content comes from my PC, not an audio component sitting with my television. When I'm listening to Yahoo Music Engine and their Launchcast station tries to pawn some Nickelback off on me, I have to get up and skip the track by hand.

This shall not stand.

My Goal

I'm going to use the same infrared remote control I use for my home entertainment system to interact with the Yahoo Music Engine on my PC. It won't be easy, and the fact is, I'll spend more time putting this system together than I would have spent walking over to the computer to skip Nickelback songs over the next year.

If you don't get why I would do this, stop reading here because it's not going to get any better.

Along the way I'll keep track of how much time and money I'm spending to get this done. At the end, I'll have a parts list and step-by-step plans to make this dream a reality in your own home.

The parts list so far

USB-UIRT with shipping, $53.61 from Promixis This is the hardware that is going to receive infrared signals from the remote and set events in motion.

Girder 4, $49.99 from Promixis This is the Windows automation software that's going to listen for the remote and take all the right actions. I have never worked with it before, so I don't know what techniques will be available for interacting with the target application. There seems to be some support in there for locating windows by title and such (maybe I can just wire it up to automatically press the "Skip" button if "Nickelback" appears in the title bar), and if mouse spoofing doesn't pan out, YME has a developer API that might be useful. We'll see.

One for All URC 8910 Universal Remote Control and 4 AAA batteries, $25.85 from Amazon (optional) This will command my whole array of audio/video equipment, as well as the PC. If you have any universal remote, it will work. I got this one because my old "universal" remote couldn't control my DVD player, but this one's got that sweet JP-1 port so I can program it with custom codes. Now there's no such thing as an infrared device I can't control.

JP-1 Cable 13.99 from DIY Gadget (optional) This is used to program the remote control. I could get by without this if I used the built-in (and super cool) modem updater, but... it's a gadget. Besides, if $14 is going to break the bank, maybe you should consider another hobby.
IR.exe free download from Hi-Fi Remote (optional) This will be used to write data to the remote over the JP-1 cable. You can skip this if you're not going to be buying the JP-1 cable or getting that picky about your key mapping.
Remote Master free download from Hi-Fi Remote (optional) This works in conjunction with IR.exe and provides a graphical interface to remap keys on the remote. Naturally, you can skip this part if you're going to skip the previous two items.
IrToWav free download from Hi-Fi Remote (optional) This can program an equipped OneForAll remote (like mine) without the JP-1 cable. I'm still going with the JP-1 cable because I want to be able to read programs from the remote as well as write them.

Cost

I expect to spend a little less than $150 (including shipping). I have spent 5 hours finding options and selecting all of these purchases. I expect I'll spend about 20 hours by the time this is over. I will add items to the total expenses as they arrive.

Budget update: $0.00, 5 hours

The Architecture

Building It

Software

Electronics

The Gauges

The Finished Product

The Parts

Assembly

2. Run the wires

The Problem

Solving It

Design

Parts

Assembly

Victory is Mine

a.k.a. The USB Easy Button

Let's Make a Button!

Manipulate an Ambient Orb through a serial port

Previous Work

Problems

Update lag

Integration of multiple software projects

Solution

How to Do It Yourself

Part 3: Talking to the Computer

Hardware Installation

Software Installation

Telling Girder About It

Setting Up Your Remote

Part 2: Programming the Remote

Uncharted Territory

Cost