Hello all,

I am very new to this forum and pretty much an electronics newbie. I've done some soldering and basic circuit design/wiring, but nothing too complicated - so I would like to get some feedback/ideas from the more experienced on this website.

The problem: I am building a climate-controlled growth chamber for alpine plants. I am using a 12V, 8A Peltier square along with 2, 100W CPU heatsinks to accomplish the cooling (cold air blowing into the insulated enclosure, hot air venting outside... pretty standard). I would like to not just control the temperature (I've built single-setpoint temperature controllers using the STC-1000 before), but ideally be able to control the night and day temperature independently to accomplish a daily temperature drop (dual setpoint).

The idea: I haven't seen any tutorials or advice on how to do this and it seems like most of the pre-made solutions to this problem are for commercial applications and cost in the $100's-$1,000's. So, I am setting out to try and find a solution that is not terribly complicated and also will not break the bank. Again, I am very new to electronics and wiring, so the simpler the better. One thing I did come across was a ton of people using a very cheap 12V DC thermostat (link to an example one here) to control Peltier units in homemade coolers and fridges. Basically, these consist of a thermocouple wired to a 20 A relay and only cost ~$5. My understanding is that they can be set to heat OR cool, but do not do both at the same time So I thought that a good solution might be to get two of them, put them in cool-only mode, and wire them in parallel to the Peltier unit with a 12V DC timer wired in to switch off the lower-temperature thermostat during the daytime (quick diagram attached).

The logic being that the daytime setpoint will be like 22C and the night setpoint will be like 12C (or at least always 10 or so colder then the daytime setpoint. During the day period, the 12V DC timer will switch off the night-control thermostat and the day control thermostat will regulate the temperature in the enclosure. During the night period, the DC timer will switch on the night-control thermostat which will cool the enclosure to the lower setpoint. The day control thermostat will be powered-on, but passive, during this period since the temperature in the enclosure will always be well below the day-control setpoint, so the relay will just be closed. Once night is over, the timer will switch off the night control thermostat and the temperature in the enclosure will slowly rise up to the day temperature, where it will be controlled by the day controller. And so on and so forth.

Now, I understand this is probably not the most elegant way of doing of this and there are a bunch of people here who could not only make this a much better controller, but also have it give you a wake-up call and brew you a hot coffee in the morning. However, like I said, I am looking for something relatively simple (that I can actually understand how to design and build). Any thoughts as to why this wouldn't/couldn't work? Any advice on ways to make it better if it can work? It's just an idea for now, and it seems like it should work as a cheap, quick, solution... but then, if I knew what I was doing, I would just do it and wouldn't be here asking for help .

Thanks in advance for any feedback!

 

1) You don't need two TECs. Simply reverse the current using an H-bridge.

2) It is easier to drive the TEC H-bridge with a PWM digital signal (pulse-width modulation).

3) For stable and accurate temperature control, us a PID controller (proportional-integral-derivative).

4) You will not get much out of a 100W TEC. What are your attempting to heat/cool?

 

Sorry if I was unclear, I don't have two TECs, just one. I don't need to heat AND cool, just cool. What I have are 2 thermostats (see diagram) to control the set temperature at a different setpoint depending on the time of day (day setpoint and night setpoint). I looked into PWM a bit, but could only find reference to that being used to control for a single setpoint, so it doesn't solve my problem of needing to switch setpoints automatically depending on the time (night or day).

Also, I am only looking to cool a very well insulated small space (~1.5 cubic ft) to maybe 10 or at most 20 degrees below ambient (so I don't need too much power).

 

Peltier's are extremely difficult to control because the thermal time constant is really really long. I've done the thermoelectric cooler air to air before. Big heatsinks on both sides. Large area. Fan on the outside.

They are pretty good at maintaining a "delta t" with a constant input.

If temperature control isn't that critical, then I might suggest something similar to http://www.jameco.com/z/K8004-Velleman-DC-to-Pulse-Width-Modulator-Kit-Control-DC-Motors_120539.html This isn't going to work, as is. It's also only good for 6.5A.

Use two different setpoints. i.e switch in two resistors. It would essentially be open loop control, but the outside of the tank temperature may just change a little, so operating at effectively a "constant voltage". In this case, it's a constant duty cycle.

An "H-bridge" (term to look up) is a way of changing the polarity (heat/cool) modes with a single supply.

Thermistors are typically used in Peltier device controllers.

Ideally your looking for a ramp/soak (heating/colling) controller with LIKELY an analog output to drive a PWM module. There are constants like P, I and D that basically make the setpoint be the controlled temperature. Specific Peltier controllers would be ideal.

Software PID control from scratch is likely way beyond your abilities. I've done it a few times.

BANG/BANG control of a Peltier WON'T work! You could use a snap thermostat as a limit type of switch. i.e change the setpoint.

 

Thanks for all the feedback!

I'll look into the parts you linked. I assume there is documentation to be found somewhere on what resistance you would have to use to achieve a certain set-point, or how would I determine that? The ambient temperature outside the tank can be assumed to be roughly constant (it will only change slowly over a narrow range since the tank will be stored indoors and out of direct sunlight).

I guess I should also clarify that I am not trying to necessarily control the exact temperature of the peltier unit directly. I don't necessarily care what the surface temp of the unit is, I just want the peltier to switch on and off as necessary to cool the enclosure to a certain set point (so what matters is that ambient temperature of the enclosure, not necessarily the surface temperature of the peltier. That said, I understand that with the really long thermal time constant this means that the temperature inside the enclosure will not be super stable, but I was thinking that as long as I could keep it +/- 2-3 degrees of the setpoint, that would be good enough (I don't need control to fractions of a degree, a few degrees resolution is fine). Would the type of BANG/BANG control I proposed be good enough for that, or will it just not work at all?

I did google H-bridge after reading MrChips response, since this is a term that I hadn't heard before (learn something new everyday). That wouldn't be an issue (beyond the scope of this project). I don't need to reverse the polarity and heat. I just need to keep it the same for cooling. I will also look into the other options you proposed (the ramp/soak guy). At least now I know what to look for (there are so many types of controllers and it's hard to wrap your head around all of them and their uses). Thanks again for all the advice and leads!

You are correct that scratch software PID control is way outside the scope of what I am currently capable of doing.

 

Bang/bang control just won't work. The Peltier heats and cools really slowly in this situation. In one application, I routinely used Peltiers with running ground water on one side, a 300 W lamp on the other controlling at 25C. They don;t heat up like a resistor.

If the current is exceeded, they melt. More precisely, if delta t is exceeded they easily die. 40 to 60 degrees delta T is typical.

I did air-air (closed container (about 1 cubic ft) with purge), surface-water, air-vacuum(glass plate, light source, (45 cubic inches).
The first one had a surface of about 6" x 8" with forced air cooling on one side with a large heat sink. When I did air to air, a few large heatsinks was attached to the plate to transfer heat to the container.

They do exist commercially: https://www.eicsolutions.com/product-category/thermoelectric-air-conditioners/ Just note that they are large,

 

Fair enough. Like I said, I am ultimately here because I do not really know what I am doing . After some googling, it actually looks like a ramp/soak controller is exactly what I want (cycle between a high and a low on a daily basis). Also not terribly expensive. Now I just need to figure out how to properly wire one to a peltier and also how to tune my setpoints. Time for some more googling I guess. Thanks again.

 

A few tidbits that may be hard to find:


there is sometimes a flag called REVERSE-ACTING. Output goes up and it cools.

You need a controller with some sort of analog output: Typicals are 0-5V, 1-5V. 0-10V, 4-20 mA and 0-20 mA.

Then some other magic that won't be apparent. If you put a 250 ohm (usually 249) on a 0-20 mA out, you get 0-5 V.
And when a 4-20 mA out is impressed across 250 ohm, you get 4 mA.

"Current" outputs generally help to eliminate ground loops.

Your end controller needs to be PWM, i.e. 0 to 100% duty cycle, preferably with a current limit;

H-bridges with (PWM,and heat/cool logic input) with a heat/cool controller heats and cools.

An H-bridge is used a lot in motor control. Terms like CW, CCW, brake, coast are common. A "SIMPLE" H-bridge may have two inputs, NOT ENABLE and DIRECTION.

DIRECTION is the heat/cool mode thing
NOT ENABLE is the PWM thing. It's usually NOT ENABLE because, when you disconnect it, outputs float high, so the thing will be off.

 

Thanks for all the advice. I did some more digging around and it seems like although ramp/soak controllers would be a good option they are also fairly large (ideally I would like to keep this kind of compact). Additionally, I was having trouble locating one that has an analog output (or at least specifically says anything about an analog output in it's datasheet... unless there is some other language I should be looking for here).

I did stumble across these (link) last night though - it is a very small PWM TEC driver from an optics company (0-15V, up to 20 A). It will only let you adjust to a single setpoint, but it seems like I could use the strategy I proposed above and use something like this instead of using the ON/OFF thermostats that I had originally proposed. A ramp soak controller would be better, but more complicated to source and set up. The wiring on these would still be very simple since they only require that you give them power, the temp sensor (in this case a thermistor), and the TEC wires. Then, the TEC would be driven and controlled by a PWM signal rather than ON/OFF. At ~$30 apiece, getting two of them would run about the same cost as getting a lower end ramp/soak controller and I could use the same strategy of putting the lower setpoint controller on a timer. It seems like originally the main concern was with trying to use an ON/OFF controller to regulate temperature with a TEC, would this work better (ramping would be really nice, but is not a necessity)?

 

Actually, I just realized that wouldn't work unless I had some way to switch between the two outputs such that the peltier would only "see" one signal at a time. Perhaps for now I will just use one of these to set the desired night temperature and then turn it off completely during the day while I think about/put some more research into how to control the day temperature too.

 

That module sets the temperature by a trimmer. It could very well be a two terminal device (one terminal shorted) or a three terminal device.

If it's a two terminal device, you can switch in multiple potentiometers with one side common.

A three terminal device COULD be a reference (or fixed voltage) on one side and ground on the other. The center tap is just a variable voltage.
So, it might be able to be dealt with too. Some reverse engineering might be necessary. Sometimes, you can ask for the schematic of circuit pieces. e.g. The circuit surrounding the potentiometer because you want to have multiple setpoints or a variable setpoint.

 

Sweet. I will shoot them an email, explain the problem, and see if I can get the schematic to work off. Tiny soldering doesn't scare me, so although trying to design a PWM controller for the peltier from scratch intimidates me, modifying an already existing one is well within my capabilities if I have the schematic to work from.

Alternatively, if I can't get the schematic, I was thinking about it some more and could probably get an SPDT relay and wire this to the 12 V DC timer to swap between the two signals from two controllers at different set points based on time.

I'm feeling good about this now! Once I get some more answers I will post again with any questions that crop up. Thanks again.

 

I did a quick search and ended up with this http://www.omega.com/manuals/manualpdf/M4437A.pdf Omega re-brands other people's stuff.
So, it can have analog outputs of 0 to 10 V or 4 to 20 mA, so it MAY not be QUITE right, but could be coerced.

0-10 V is easy to convert to 0-5, for instance. Very easy with a dual power supply. Zero is hard to achieve with a single supply.
The IMPLIES function https://en.wikipedia.org/wiki/Material_conditional might work. T&T (In heat and cool at the same time)would never happen and F & F meaning not heat and not cool would never happen either. So, that logic would switch the output 1 and output 2. OPTOMOS relays would work.

So, it's probably not hard.

For fun and giggles, here's a 4-20 mA to 0-5 V converter.

Then you would create a circuit that would look at the output of cool and heat e.g. >10 mV and create a heat/cool signal that could drive an H-bridge. This http://www.canakit.com/Media/Manuals/UK1122.pdf is an example of an H-bridge, but one that won't work.

 

You can share attachments via Conversations (The forum's name for PM (Private messages), so you don't have to share it publicly.

 

You don;t even have to do fancy logic. You can effectively use some output like heat (analog out) > some small number and set the mode to heating, otherwise cooling. Using the IMPLIES logic is only slightly better.

 

Here's http://www.robotshop.com/en/devantech-md03-50v-20a-h-bridge-driver.html an H bridge that's more complicated than you need, but it has a 0-5V in and direction. This does PWM at a fixed frequency of 15 kHz (slightly above hearing range) and a fixed current limit of 20 Amps. A bit pricey because of the other options. Not ideal for a number of reasons (price and non-adjustable current limit)

See: http://www.robotshop.com/en/brushed-motor-controllers.html

 

Interesting. I hadn't thought about re-purposing a DC motor speed controller; although it looks like buying one rated for the correct amperage (to be safe I probably need at least a 10 A rating since my TEC is 8 A) would probably be about the same cost as just buying the TEC controller.

Adding an H-bridge seems like an unnecessary complication, the device should only ever have to cool (the set points will always be lower than ambient, never higher). So having it set to do heating as well seems counterproductive .

 

Also... a quick, somewhat stupid question/clarification. All of my previous exercises with building temperature controllers used thermocouples for sensing the temperature, so I am new to thermistors. In this case, the thermistor would be acting like the thermocouple and sensing the temperature of the environment to be controlled, right? So the thermistor should go inside the enclosure and then be wired back to controller (which will ideally be outside the enclosure), correct?

 

I know, i went round-a-bout and covered heat/cool controllers, Open loop control at all, ramp/soak, PID, PWM for Peltier.

PWM frequency usually doesn't matter except, you don't want it talking (Audio range).

How to convert 0-20 mA to 0-5 V or even 4-20 mA to 0 to 5V.

I missed talking about sensors: Thermister's are good, RTD is good and Type T thermocouples are good for this temperature range. K Thermocouples are common.

Bang/bang is out. So, you need some sort of analog output controller that can control some sort of (PWM device/Peltier controller)

Just two temperatures and cooling only brings up other ideas.

 

Also... a quick, somewhat stupid question/clarification. All of my previous exercises with building temperature controllers used thermocouples for sensing the temperature, so I am new to thermistors. In this case, the thermistor would be acting like the thermocouple and sensing the temperature of the environment to be controlled, right? So the thermistor should go inside the enclosure and then be wired back to controller (which will ideally be outside the enclosure), correct?

Thermisters are non-linear and they have different curves too. They can also increase or decrease their resistance with increasing T. e.g. NTC and PTC types. One example is a meat thermometer.

The tiny TEC controllers that you are seeing may use the thermister in either a bridge or voltage/divider circuit. I used one in an aquarium heater controller may years ago.

So, they are sensors and would go into the environment being controlled.

RTD's, specifically an RTD-100 is 2 wire device requiring a 3 or 4 leads. These are very well calibrated. I used them from -80 to 200 C with 0.1 degrees resolution,