Implementation Project

When I showed my idea to my friend Kevin he immediately thought of an application in his home for the MiniSBC – control his sprinkler system.  His current controller has the following characteristics:

1. Controls up to 12 zones
2. Has one minute resolution
3. Only one zone at a time may be activated
4. No communication method

Using an embedded device to control a sprinkler.

His home currently only uses eight of the 12 available zones so the MiniSBC is able to handle his requirements.  The programming is completely under our control so we can activate as many zones in parallel as we want – limited only by our power supply.  We can choose the RCM we want based on the communication technology we want to employ.  In this case he wanted to use Wi-Fi so we used the RCM5400W.

In order to allow for local control of the system an LCD is interfaced via one of the two RS232 ports.  The 4 x 20 character LCD we selected is from Matrix Orbital and is connected to the MiniSBC via one of its two RS232 ports.  In addition, a small printed circuit board was developed to allow for operator input.  This is a very simple board (see schematic 4) which has three LEDs (driven by LCD outputs), a five position joy stick switch and three push button switches.  A basic menu system was developed which enables the operator to enter and modify a watering schedule.  Some of the basic features of the Menu are:

• Enter the current date and time
• View the stored watering schedule
• Create and modify the schedule
  • Resolution is 1 minute
  • Any combination of zones may be activated
  • Once the schedule for a day is entered it can be copied to other days
• Store the schedule in flash

A Relay Board was initially developed to switch the control voltage to the sprinkler solenoids, see Schematic 5.  The board contains essentially only two circuits: 5V switching power supply and eight relays.  It is designed to use a “wall wart” (from a Rabbit product) which has two 24VAC secondaries at 800ma each.  The relays have 5VDC coils and are driven directly by the open drain outputs of the MiniSBC.

There are four connectors on the board:

• 6 pin socket for power
• 3 pin header to supply +5V to the MiniSBC
• 10 pin socket which mates to the output header of the MiniSBC
• 12 point screw terminal for connections to the sprinkler solenoids

The board is designed such that the output pin header of the MiniSBC can plug directly into the input socket of the relay board.  This allows for stacking the two boards without requiring a cable.  Since the RCM plugs directly onto the MiniSBC this yields a fairly small package with three stacked printed circuit boards.  The dimensions, in inches, are: 3.5 x 2.25 x 1.5.

The MiniSBC can be seen inside with a Wi-Fi RCM5400W inserted.

I decided to use a switching regulator on this board because I wanted higher efficiency than a series regulator would offer.  The RCM5400W draws about 600ma at 3.3V when transmitting.  That is almost two watts.  The LM2576 has about 85% efficiency which means that it will dissipate about 0.3 watts.  This means that no heatsink is required so I simply soldered the TO220 case to a copper pour.  If I had used a series regulator it would have had to dissipate over 14 watts (based on having a 24VAC input) between the regulator IC and a series resistor.  In this case the additional complexity is well worth the effort!

As it turned out, this board was not successful in switching the solenoids.  It took only a few on-off cycles to cause the relays contacts to fuse!  The relays are rated at 200VAC and 0.5 amps so I am a bit puzzled as to why they did not work.  I installed an R-C suppression network on each output to see if that would help but it did not.  All I can think of is that there must be quite a surge current when first applying the 24VAC to the solenoid through the relay contact.

I then decided to design a Triac based board, see schematic 6.  The Triacs I selected are the same ones as are in the original controller that Kevin has.  They are “Sensitive Gate” devices and work in all four quadrants.  They are rated at 600VAC and 4amps with a guaranteed maximum I_GT of 5ma.

You can see from the schematic that the power supply is identical to that of the relay board as are the input and output connectors.  The 74HCT540 is needed to invert the signal coming from the MiniSBC board which has normally off, open drain outputs.  As such, the output voltage is normally high which would cause the Triacs to turn on.

The outputs of the Triac circuits switch essentially opposite to those of the Relay Board.  The Relay Board is a “high side” switch in that the solenoids are connected between the relay outputs and ground.  The relays connect 24VAC to one side of the solenoids which would turn them on.  The Triacs are a “low side” switches in that the solenoids are connected between 24VAC and the Triac which completes the circuit to ground when turned on.

The size of the Triac board is same as the Relay Board and the connectors are in exactly the same places.  This allows the MiniSBC to use either one interchangeably.  Even though the Relay board did not work out in this application there are many applications where it would be a better solution.  One of which is that it can switch both AC and DC while the Triac Board can only switch AC.

The Third Revision with A/D:

The third, and final, board in this series (see schematic 3 and Fig. 3) adds the ability to use the A/D convertor which is present on several of the Rabbit 4xxx Core Modules.  This board needed some more room for the analog input circuit and connector so it is a little bit wider than the RCM41xx.  However, it can still be implemented with a two sided board.

Much more going on here.

The A/D on the RCM4xxx devices is the ADS7870 from TI.  It has 11/12 bits of resolution, eight inputs and a Programmable Gain Amplifier.  The inputs can be used in a single ended configuration or they can be paired to measure differential voltages.  All the voltages applied to the input of the ADS7870 must be positive so if you need to measure negative voltages you will need to implement some external circuitry.

The input circuit is a voltage divider that allows for an input range of 0 to +10 volts.  This input range can very easily be changed by modifying either or both of the input divider resistors as required.  The ADS7870 allows you the choice of any of these three voltage reference values: 1.15, 2.048 or 2.5V.  The schematic shows the divider with 100K and 33.3K resistors.  This provides a voltage division of four, so with a reference of 2.5V you will get a 0 to 10V range.  These same resistors will also allow a full scale input of 4.60V if the 1.15V reference is selected.  You can customize these resistor values on a per input basis if your application has different ranges to measure.

Notice that the sequence of the inputs on the connector is not channel 0, 1, 2…  I did this in order to make routing the traces easier.  If having the connector pins and the channels sequence together is important to you then you will need to do some rerouting.  I basically made an engineering decision that it was not important enough to be concerned about.

Another “feature” of this particular implementation is that it can easily be built to make use of eight additional parallel I/O lines on those RCMs that do not have the A/D.  For those RCMs Parallel Port D is available on the pins that would otherwise have the A/D inputs on them.  You will have to exercise some additional caution if you decide to make use of this feature.

When using a pin as an input you have several options and one limitation.  The options deal with how you use the voltage divider circuit on each of the eight pins.  If you leave the resistors as they are defined on the schematic then you will be able to sense digital inputs of 13.2V (4 x 3.3V).  You can adjust the ratio of the resistors to meet whatever input voltage range you require – just remember that the trip point of the typical CMOS circuit is typically VDD/2.   Whatever you choose you should always have a resistor from the input pin of the Rabbit to ground (or power) so that the input is not left floating.

The limitation is, since there are no clamp diodes, you need to insure that the voltage applied to the Rabbit I/O pin does not exceed 3.3 volts.

To use a pin as an output you will probably want to use a zero ohm jumper in place of the series resistor if you are going to use the CMOS signal directly.  Or, if you want to drive a transistor, you may want to put in a current limiting resistor whose value is determined by the circuit it is driving.  You should be aware that the maximum current per output pin is 8ma sinking or sourcing.

Something to keep in mind if you decide to use this feature is that the Port D pins can be individually configured as either input or output.  This allows you to easily extend the functionality of the board without redesigning the PCB.

A Few Conclusions

• It was fun finding out that it is not very difficult to develop a minimal function SBC
• It did not take very long – the initial design and layout took about 4 hours
• The result is a small, useful and inexpensive (two board) SBC.

Third Schematic (Click for high resolution)

Notice the RS232 chip added on the lower left side

Adding RS-232:

For many applications the board described in the previous article will be adequate.  If you need to communicate with the SBC you can easily use one of the Ethernet or RF enabled Rabbit Core Modules instead of the RCM41xx.

However, there are many cases where RS232 connectivity is required.  The second board design, see schematic 2 and Fig. 2, adds a dual RS232 to CMOS convertor.  This circuit can be used as two independent 3-wire circuits or a single 5-wire circuit.  The board is the same size as the board described above and can still be implemented with two layers.

There is one additional 5 pin header for the RS232 I/O.  There are quite a few dual channel convertors available so I just picked the one that is typically used by Rabbit.  It is relatively inexpensive and easy to implement.  All it needs are the four external 0.1uf capacitors.

Updated Schematic (Click for high resolution)

Just in case some of you are interested I am posting a few pictures of my garage at home, most of which I have turned into my workshop.

Main operating location for my computer and Ham Radio - an ICOM 728. I operate mostly PSK31.

Just to the right showing more stuff. The radio is an ICOM R7000 and receives from 25 to 2000MHz.

My workbench showing the Tek 465 'scope I repaired.

More of my workbench showing tools and, on the shelf, several of the projects which became magazine articles.