If you haven't looked at and read the Project HeavyBlue section, I suggest you do so now or else none of this will make sense.

As a result of seeing what a peltier assisted cooling system can do if something goes wrong, I devised a VERY crude safety scheme which basically looked for any type of power failure, and upon seeing one, shut the whole computer down. This particular system was similar to a closed loop burglary system and used energized relays to maintain the loop. This had its own flaws and worked well enough for the time but I was always afraid of a burn up or freeze out. This worry prevented me from making any real upgrades to the rig and as a result it became very antiquated.

Eventually, upgrade was no longer a nice dream but a full on necessity, and I had to figure out a way to make sure that a failure of the cooling system or the computer wouldn't destroy expensive components. That is when the idea of a completely standalone monitoring and controls system would be necessary. At first it sounded like just a simple thing, use another computer with some sort of data acquisition and controls card to constantly just monitor the other computer. Obviously something like this had to be done since if the computer is monitoring itself and hangs, there's no longer a safety system. Then I got to thinking, there's only two ways I can monitor and control this properly: A linux box, or a Microcontroller with onboard IO. Since I didn't want to spend a lot of time and money getting drivers and hardware to work for Linux, I opted for the Microcontroller. Not that bad of an idea considering it was birthed at a pool table in a bar, in downtown Blacksburg VA huh?

I spent Spring and Summer of 2006 getting the parts I needed to make this safety computer and sure enough, after a couple of revisions to the code, it passed all of my I/O tests without error. It was now time to build the computer it was designed to protect, but first, here's how I made this safety system.

The brains of this system is the Remote Processing RPC-210 Embedded Controller, official information is available here. Providing 28 Digital, 8 Analog Input and 2 Analog Outputs with 512K of onboard memory, this board is a bit overkill, but future job expansion was in mind.

Unfortunately I/O is useless without sensor hardware. For the task of answering simple yes/no questions and/or on/off commands, relay contacts to/from digital lines, after the proper line level conversion of course is a breeze, the real fun begins when you start monitoring temperature and power usage for they be analog.

For the digital checks, the following are monitored (this list is not complete):

1. Whether or not the coolant pump is running, barring the possibility of a kink in the hoses, a differential pressure switch across the pump works quite well.
2. Whether or not the coolant loop has enough coolant to function, using a float style switch in the reservoir works perfectly.
3. Whether or not the power supply for the peltier elements is operating using a simple relay contact.

Of course, digital controls need to be used so somehow you have to convert TTL level outputs to something that can actually switch things.

And now for the analog inputs:

1. Temperature Monitoring: Thermistors in voltage divider networks, changes in temperature changes the voltages the analog ports see.
2. Current Monitoring: Current sense amplifiers used in tandem with current shunts (low resistance resistors).

The interface boards pictured above are necessary for signal conditioning to make sure no stray voltages can break any sensitive components.

All these ports and their uses are identified in my I/O List.

Of course I can't forget the Software that makes the whole thing work...

• Version 1.0.0
• Version 1.0.1
• Version 1.0.2
• Version 1.0.3
• Version 2.0.0
• Version 2.0.1

In addition to providing the necessary safety functions for Project HeavyBlue, Project BlackBox also broadcasts a heartbeat of sorts, every program cycle a formatted string containing all the vital data collected and calculated by Project BlackBox.

I realize that I have not given all the information necessary to fully explain exactly how Project BlackBox works, quite frankly I am thinking about selling the idea or custom making other BlackBoxes for interested people. To date, after looking at hall effect technology, I can shrink down the size of Project BlackBox to 2 or 3 5.25" computer bays, excluding the necessary power supplies of course. I am currently looking into using one of the digital output lines as the clock signal for a multiplexer, thus allowing the 8 analog inputs retrieve even more information.