residential thermostat hack help needed

[Rayleigh]

Hi all,

I'd like to use a programmable residential thermostat to provide temperature control for a thermoelectric element. I am currently using a nice DC motor control kit from All Electronics to control the heating/cooling of the element. An external tiny thermocouple is used to measure the temperature of the thermoelectric and the biological sample being tested. Typical temperature range is 10C to 50C. However, I'd like to integrate the temperature measurement and heating/cooling control into one device and hope it is possible to hack a residential thermostat for this purpose.

A few things I've thought of so far:
1) I suspect I will need to use the output of the thermostat to activate a relay which will power the thermoelectric element. Would I need two relays to switch the current direction for the heating or cooling? Also, the DC motor kit uses "pulse width modification." The temperature of the thermoelectric element stays relatively constant with this method (which is good!). Will the thermostat/relay setup cause the temperature to wildly fluctuate as the relay is kicking on and off full power to the thermoelectric element as the thermostat tries to maintain a set point temperature?

2) I know thermocouples are used in furnaces (and some actually power the thermostat itself). We use tiny k-type thermocouples. In the thermostat wiring where are the thermocouple wires? I've seen the color codes for the wires from the furnace but these don't say what they are - is there another circuit between the thermocouple and the thermostat?

3) I could use tiny thermistors as the temperature sensing device. They just need to be very small. The LM335 solid state temperature sensor is way too big.

Thoughts/ideas/suggestions are welcome and appreciated!

- Rayleigh

[haklesup]

Residential controls generally do not operate in heat and cool modes simultaniously, you have to manually select one or the other. If you didn't when the ambient temp is close to the setpoint, you would get the two systems fighting each other for control. There may be some models that can automatically switch but their price would approach the alternative

The alternative is to use an industrial style temp controller module. With one of these you would get two useful features. One is PWM control of the element whereas a residential control would be just on or off the PWM would rapidly pulse the element to maintain better ramp control and lower power than full on or off. A second feature is PID control which lets you define the ramp and overshoot characteristics to better match the thermal load you are connected to. These units also generally allow the use of a variety of temp sensors.

I can see several "PID" or "temperature controller" for under $60 on eBay right now. Be sure to look up the manual/datasheet for the one you want to bid on to make sure its up to your task and that the instructions have enough detail for it to be useful to you. If you can't find the exact same model # instructions, try a similar one from the same manufacturer.

You may also need an "H-Bridge" which is a sort of amplifier that goes between the PID and the heater (or TEC if you needed cooling) to allow you to use a higher current supply than the circuits in the PID can usually handle. Whether you need one depends on the switching specs of the PID and the heating element.

[Rayleigh]

Excellent suggestion about the industrial temperature controller! I'll look into it. Thanks!

[Rayleigh]

We use a 12V lead acid battery to make the unit easy to take into the field. The programmable residential thermostats I've seen run off batteries (a few volts and low current) that I could grab from the 12VDC battery. The industrial temperature controllers seem to need wall power or much higher DC voltages since they can drive a higher load. Hmmm....

[MrAl]

Hi, and welcome to the forum,


Thermostatic controls are surprisingly simple to build from scratch.
A thermistor, a couple comparators, a few pots for set points, and you have
a custom thermostat that can be adjusted to fit just about any need.

When you do heating and cooling you usually leave a dead band between, but
again this is also very simple to do with comparators.

You would also need some type of driver depending on what you are going to
control. This may be transistors or a relay.

If you would like to see a schematic i could probably draw one up.

[Rayleigh]

Yes. A schematic would be very helpful, thanks!

[MrAl]

Hi again,


Here is a schematic of a basic controller and associated waveforms.
Although this circuit is very basic, it would work as is.

Let me point out a few things about this circuit...

1. The pots on the schematic are shown as two resistors in series with a center tap. This is to
show the setting of the pot that produced the waveforms.
2. The flip flop may be best off being a CMOS type RS flip flop.
3. The voltage source V2 is not really part of the circuit, but that was used to simulate the
thermistor changing resistance and producing a change in voltage at the center tap of that
thermistor and the resistor in series with it.
4. The thermistor resistance is 10k at 25 deg C, so the upper resistor is made 10k also.
5. What isnt shown are scaling resistors that would be used to scale the pots into approximate
range so that they act more as fine tuning the temperature setting. For even finer tuning,
smaller value pots can be used in series with the larger value pot or else 10 turn pots can be used.
6. The pot made from R5 and R6 is for the 'turn on' point when the circuit is used for cooling, and
the other pot would be for the 'turn off' point setting.
If the circuit is used for heating, the pots have the opposite effect. Thus, if two circuit are built
one can be used for heating and the other for cooling.
7. Note that the hysteresis is controlled by making the two pot settings farther away from each
other.

[Rayleigh]

Nice, thanks! It will take awhile to gather together the parts and build it. A few questions:

1) I have 1 kOhm thermistors (normal sized and ~0.3mm diameter). Can these be used with a 1kOhm resistor in place of R2 and R1?

2) One of the nice things about the residential thermostat is that it displays the temperature as well as the set temperature. Is there a way to get the set temperatures and the actual temperature to display to a panel meter?

[MrAl]

Nice, thanks! It will take awhile to gather together the parts and build it. A few questions:

1) I have 1 kOhm thermistors (normal sized and ~0.3mm diameter). Can these be used with a 1kOhm resistor in place of R2 and R1?

2) One of the nice things about the residential thermostat is that it displays the temperature as well as the set temperature. Is there a way to get the set temperatures and the actual temperature to display to a panel meter?

Hi again,

I think you can use 1k thermistors if you replace the upper series resistors with 1k fixed also.
I dont think self heating would be too bad, but you would have to check with the
data sheet for the actual thermistor.
The main idea is to get 2.5v at the junction when T=25 deg C, but other basic bias points
would work too if the pots are adjusted accordingly.

To get inside temperature (actual) you could use a store bought digital thermometer.
To display the two set points (turn on and turn off) you would probably have to use a
microcontroller, or possibly use two voltmeters that measure the voltage of each pot
set point and do the calculation in your head. Another idea would be to simply
calibrate the two pots where the text behind the knobs reads off in degrees C or F.

A microcontroller, if you were to use one, would then also be able to be used for the
entire circuit. It would sense the thermistor voltage and do some calculations.
The set points would be converted to temperatures and displayed on say an LCD.
I dont know if you are into microcontrollers though.

[Rayleigh]

I dont know if you are into microcontrollers though.

Not yet - but I see I need to learn!

[MrAl]

Hello again,

He he, well there is a little bit of a learning curve, and i dont know how much of a hurry you
are in right now.

Oh yeah, also, what kind of accuracy are you after? In other words, if you set it to
25.0 deg C and it actually works at 25.5 deg C would that be a problem, or would
you really really need 25.1 deg C say?
I guess what i really mean to ask is, if you set it to 25.0 deg C and it actually holds
constant between 24.5 and 25.5 would that be ok? By this i mean it would actually
vary a bit rather than stay perfectly constant for all time.
Keep in mind that air drafts cause local spot cooling/heating problems too.

[Rayleigh]

Oh yeah, also, what kind of accuracy are you after? In other words, if you set it to
25.0 deg C and it actually works at 25.5 deg C would that be a problem, or would
you really really need 25.1 deg C say?

Ideally I'd like the temperature to stay +/- 0.1 degree C of the set point.

[Bob Scott]

Ideally I'd like the temperature to stay +/- 0.1 degree C of the set point.

That might be ideal but I think that if you live in reality, you will be lucky to get +/-1.0 degree C. Parts have tolerances, thermal sensors have tolerances and drift, your heating/cooling devices must vary from that temperature in order to move heat, and the air in a room can have various temperatures in different locations.

Wow, this thread has drifted from modding a cheap residential thermostat with ~2-5 degrees of hysteresis to a device with +/- 1 degree C.

[haklesup]

I'm suprized it hasn't come up yet but for any project related to temperature control and measurement a must stop on the web is Omega http://www.omega.com/temperature/tsc.html . Even if you don't buy their stuff, their app notes, manuals and selection guides are really useful. And yes, they do have 6 or more models of temp controllers that run on low voltage like 12VDC. At least you can scope out the models that might work and look around for better deals elsewhere. If you're in for $200, there are several below that retail. You'll also get schooled in the essential options and features you may need for your application. If you have a spare bookshelf, consider ordering their printed catalogs. They tend to have lots of good information inside.

I was all set to discuss how to use a residential thermostat but your precision and range seem to preclude that. If you were to pick up a decent digital wall controller at a hardware store, there should be instructions to help you decide which terminals to use. However just about all thermostats should support 2 wire operation where a simple relay closes when it wants the heater to turn on. Controllers for heat and cooling will have additional relays and terminals to separately control the A/C, Heat and recirculating fans. If you expect to operate from 10C to 50C with ambient at 25C I think you will be disappointed in a household thermostat as you pass from cool to heat mode but if you only need to heat above ambient or cool below it, you might be able to kluge a lower accuracy solution.

I noticed that many consumer digital thermometers will have a tenths place on the display but have specs with +/- 2 degree accuracy. The accuracy in this situation is on the absolute scale, relative readings (difference between now and later) may in fact have 0.1deg resolution. So if your measureing the difference between two temps it should be quite accurate but not necessarily spot on WRT the real temp.

[Rayleigh]

Thanks! I'll check Omega more closely.