After a couple
day cold snap, my best friend called me up to discuss the upcoming
weekend. It turned out that he and his fiancée had spent a
full day defrosting one of the zones on their radiant hot water
heat. To make matters worse, the pipe managed to burst in three
places, requiring repair. Needless to say, I felt very sorry for
his ordeal and wanted to help him prevent it from happening
There were several factors that led to the problem. First, he
owns an older farmhouse, which has known insulation problems from the
time he purchased it a year previous. The pipes for this
particular zone run through a crawlspace that is rather drafty.
Second, the night that this problem happened, the temperature dropped
well below zero (Farenheit) with a stiff wind, allowing the cold to
penetrate this crawlspace more easily. Third, they had been using
the woodstove to provide heat in the living room, which essentially
relieved this heating zone of its duties for the evening. This
inactivity was only extended when the thermostat automatically set back
the temperature at night, keeping the zone off for even longer.
So, what do you do? Ultimately, you don't want to allow the pipes
to cool off to a temperature that will allow the water to freeze.
The solution in this case was to simply cycle the heating loop on
periodically to keep it warm.
Click here to view the schematic
Basically what I needed was a timing circuit that would just switch on
the solenoid from time to time to send 160°F water through the
loop. I could do this easily by just closing the connection that
the thermostat normally uses to call for heat. I also wanted to
add a design feature that would active the timer only at temperatures
where it was needed. This would alleviate my friend from having
to remember to switch the unit on when needed and off when not. I
accomplished this through the use of a 10k Ohm thermistor available at
Radio Shack (Catalog number 271-110) as my temperature sensor. I also
chose to utilize the 24VAC source used to control the servo valves so
that I wouldn't have plug the unit into a wall outlet.
The first portion of the circuit is nothing more than my power
supply. A bridge rectifier converts the 24 volt AC line supply to
DC. A 2200µF Electroyltic capacitor does the primary
smoothing of the resulting waveform into a more stable DC supply.
I used an LM317 adjustable voltage regulator to drop the 32
Volts DC to about 6.5 Volts DC (remember that 24 Volts AC, is 24 volts
RMS, which works out to a peak voltage of about 33.6 volts, then
subtract about 1.4 volts lost in the rectifier). One point
to note: The schematic does not show a 1µF capacitor that I added
across the output of the regulator to ground. This is recommended
by the manufacturer to improve transient performance.
The second portion of the circuit involves the activation temperature
control. I employed the use of the venerable 741
amplifier to perform a voltage comparision between the setpoint and
voltage divider created by the resistors and the thermistor. Some
technical notes here; While you need to watch out that the
resistor values aren't too large that the amplifier's bias currents
don't become a significant factor, it's just as important that the
current flowing through the thermistor remain as small as
possible. Remember what a resistor does, it dissipates power in
the form of heat. So, if the current flowing through the
thermistor is significant, it will begin to heat up, introducing
significant error in the temperature being measured.
The pot in the voltage divider sets the threshold voltage where the
amplifier changes states. This voltage subsequently corresponds
to a particular temperature threshold as determined by the thermistor's
characteristics. To allow for fine tuning, I used a 15 turn pot
so that the voltage can be adjusted at very minute steps.
When the thermistor cools off to where the voltage drops below
the setpoint, the amplifier output goes high, switching on the
transistor to power the timer circuit.
Now, there's a switch leading into the timer circuit. This
switch has a center off position, as well as its two on
positions. One position is the automatic mode, and allows the
temperature detection circuit to determine when to activate the timing
circuit. The other position is the manual mode, where the timing
circuit is always activated. A small technical point here.
The voltage sent to the timer from the temperature control circuit is
about a volt and a half less than when the switch is in manual
mode. This is due to the headroom required by the 741 op amp and
the voltage drop across the transistor. Reflecting back upon the
design of this circuit, I could have probably gotten away with powering
the timer directly from the output of the op amp, but wanted to make
sure that I wasn't creating a short condition. Originally, I was
considering a different switch arrangement that would
have called for an open collector type design to prevent this from
occuring, thus the transistor. Regardless, this
voltage difference is negated in the way the timer works. Since
the timer works with voltage ratios to the input voltage, the timing
remains essentially the same duration at different input voltages as
long as the input voltage remains steady and constant during timing.
The third part of the circuit is the timer itself. Here, I used
another very popular IC, the 555
timer. The timer is basically connected in an astable
multivibrator fashion so that it cycles on and off. It does so
REALLY SLOWLY in this case. The diode is added so that the
timer's on duty cycle could be set below 50%. To aid in setup, I
connected a pushbutton across the capacitor so that it could be
momentarily shorted out, essentially resetting the timing cycle. Now,
the 555 works on the charging and discharging RC time constants of the
capacitor and the resistors between the voltage input and the
discharge. Normally, the circuit oscillates at a frequency much
faster than what I have setup here. In such cases, many resistive
losses can be ignored. However, in this case, such losses
actually become somewhat significant. On the discharge side, I
have a 4.7M Ohm resistor. Using the formula, T = 0.7*(RC), this
should give me an off time of about 54 minutes. In actuality, I
have an off time closer to 40 minutes. This can be accounted to
resistive losses within the capacitor, as well as possible resistive
losses across the printed circuit board (remember, a megohm makes a BIG
difference in this case).
The on time is adjustable through the 500K pot. A 10k Ohm
resistor was placed in series so that the minimum on time was at least
The output drives an indicator LED as well as a small reed relay.
This relay creates a closure in the same way the thermostat would do
when it calls for heat, turning the heat zone on. Note that I
have a recoil diode across the relay. This is important
when driving a relay from an IC, as the relay coil can kick back some
nasty transient voltage spikes when it is deenergized.
Also, if a larger relay is used, you might consider placing a
transistor in between to help provide a stronger drive for the relay,
as well as adding a buffer between the relay coil and the IC.
With this arrangement, I have the circuit setup to turn on the heating
zone for about 2.5 minutes every 40 minutes to prevent it from cooling
off to the point at which it will freeze up.