Peltier Temperature Controller

[russ_watters]

Yesterday, I tested a peltier thermoelectric cooler ghett-rigged onto my telescope CCD camera (pic attached) and was pleased with the results, so I want to make a more permanent/user friendly solution. Besides some obvious mechanical and thermodynamic needs, I also need a controller. I did some very quick gooling and found that there are temperature controllers in the $300 range that would work, but I don't want to pay that much. Dew heater controllers could be useful for the Peltier control itself (or not...se below), but cost $110.
http://www.thousandoaksoptical.com/dew.html

I don't have a data sheet for the peltier, but I *think* it is rated at 60W at 12V, like this one: https://www.amazon.com/dp/B002UQQ3Q2/?tag=pfamazon01-20

If I ever use a car battery or other not-really 12V source, I could get 13.5V and about 5.5A, so I need something that is rated for at least 6A.

I'm thinking that automatic control would be too expensive to buy and too difficult to make (but I'm open to it...), besides which I'm always monitoring my stuff anyway, so I'm ok with just watching temps and controlling output with potentiometers.

So. Fan control should be easy - I can just do that with a potentiometer, I assume (not sure what size...). Or just buy a computer fan speed controller for <$10.

Peltier control is more difficult. Many I've seen use pluse width modulation and I've seen fan controllers work that way too. How hard is it to make a circuit to do that? Can I easily make one that can handle 6A? Could I make it do active temperature control? I have a dew heater that uses pulse width modulation, but when I tested it on a fan, the pulses were much too wide for my taste. I can't have the fan going zero to full speed and back for 5 seconds at a time, every 10 seconds. And I would think the peltier would be better off doing a constant voltage instead of pulses (though I don't really know that).

One other possibility: I'm usng a PC power supply for this, so I have both 12V and 5V available. If life gets too difficult, I can just do a high/low setting on the peltier and make finer adjustments with the fan speed control.

Opinions?

 

[dlgoff]

Hey Russ,

I'm not sure how much they cost, but Watlow makes nice PID controllers.
http://www.watlow.com/products/controllers/index.cfm"
And they have application assistance.
http://www.watlow.com/products/forms/tempcont_spec.cfm"

 

[berkeman]

Love the rubber bands in your "ghett-rigged" prototype!

Could you use something as simple as a mechanical thermostat to control the temp? How tightly do you need to control the delta-Temp?

 

[chemisttree]

Like http://mcshaneinc.com/html/5C7-350.html" Russ?

 

[sophiecentaur]

Love the rubber bands in your "ghett-rigged" prototype!

Could you use something as simple as a mechanical thermostat to control the temp? How tightly do you need to control the delta-Temp?

Hot Dang
You got there first!

 

[mgb_phys]

PID is probably a bit over the top for a peltier, they don't have very fast responses - with enough thermal mass on the cold side and some insulation it will hold it's temperature very well.
As a sensor I used an Analog Devices AD590 - it's a temperature sensitive current source that gives 1microA/kelvin. Just put it through an appropriate resistor (250kR = 250kelvin), use this to control an op-amp then a power FET that controls the current to the peltier.
http://www.analog.com/static/imported-files/data_sheets/ad590.pdf

The only thing to be careful of with large peltiers is that they take a lot of current when you first turn them on and you can end up heating them with resistive power faster than they can cool themselves.

 

[waht]

Here is a schematic of a simple op-amp peltier temperature regulator

http://kroegersworld.com/cookbook/peltier.gif

It's probably not that accurate, but could still maintain a steady temperature.

 

[dlgoff]

PID is probably a bit over the top for a peltier, they don't have very fast responses...

Well that depends on how stable he wants the temp to be. Do you mean the PID is slow or the Peltier? I don't see a problem here.

Frost point control with a Peltier is very common, using PID, when doing Relative Humidity sensor calibrations.

Edit: I stand corrected.

Analog temperature control of the surface of the mirror produces
oscillations, particularly at low frost point levels. It is
difficult to apply self-tuning or PID constants to analog
control loops which are overdamped or underdamped.
Digiloop utilizes digital sampling and feed-forward control
by taking time-based samples when the dew point is within
a predetermined proportional band. By recording the
oscillations, the digital control effectively predicts the
magnitude of change and adjusts the current to the Peltier,
which steps the mirror temperature in 0.05°C increments.
This results in a significantly improved control and dew
point precision.

http://www.gesensing.com/products/resources/application_notes_new/CalLab_Soleyn.pdf"

 

[mgb_phys]

Sorry I meant if you have a few hundred grams of CCD mount heatsink and a few*10W of cooling power then you aren't going to need a very high controller bandwidth to maintain temperature to a degree or so.

 

[russ_watters]

Sorry, guys, my ambition here is tempered by my needs (not very precise), budget and frankly, attention span. Here's what I'm dealing with:

A CCD camera picks up the glow of its own heat when you take long exposure photos of dim objects. To counter that, you subtract a "dark" frame from every image. A dark frame is just an image of the same exposure, taken with the lens cap on. Since dark signal is a function of temperature, my telescope CCD camera has an internal temperature sensor to record the temperature each photo is taken at so it can be matched with a pre-recorded dark frame. It needs to be within about 1C for good results when it is warm outside, more than 2C, the software won't match at all.

So when I go out to take pictures, I'll take dark frames before I start to take pictures of objects. This is a long process, since I take the average of 4 frames of 6 minutes each, for a total of 24 minutes. After that, I might take pictures for several hours using that dark frame. But in the early evening, the temperature might drop by 5C in a matter of a couple of hours. I've been pointing a desk fan at the camera and dropping the speed as the evening progresses and that helps, but I'd like to do better and be able to start lower (since dark subtraction isn't perfect, colder is better).

Also, while I can take my darks while it is still getting dark outside, I don't always use the same exposure and the dropping temperature means I'll waste time redoing my darks later in the evening (unless I happen to have one from a previous night that is at the right temp - but I don't want to keep them too long because it isn't completely consistent over time).

So what I'd like to do is this: when I start for the evening, I'll turn the peltier and fan on full, turn on the camera, and let everything get to equilibrium (a good 10 minutes). Then I'll take my darks. All of them for whatever exposures I'm going to use that night. This could take a good hour. Then as it gets cooler, reduce the cooling to stay at the target temp.

Now 1-2C might sound like a pretty tight tolerance and tough to do manually, but once the system reaches equilibrium, it is highly stable. My desk fan works ok, but not great, since the delta-T I'm dealing with is small and so a very small breeze gives virtually all of the extra cooling a fan alone can provide. Without the fan, the heat sink surface is 3.0C above ambient and with the fan on low it is 1.2C (CCD: 4.1C and 2.6C above ambient). With the setup you're looking at, I achieved a CCD temp 4.4C below ambient with a heat sink temp (away from the peltier) of 9.4C below ambient.

 

[russ_watters]

So here's what I'm actually thinking right now. Since I have two different voltages available, I may just use a double-pole switch to switch the peltier from high to low and a fan speed controller from a computer store for the fan...unless I see a good alternative...

...chemisttree, that one is a very good possibility. I'll have to look into it more.

 

[chemisttree]

I wouldn't experiment with controlling the fan speed. I'd either leave it on or abandon the project. The torque of turning the thing on/off is going to cause a reaction force... a wobble. There goes your pic! How are you going to know if a slower speed is cooling the hot side correctly without introducing another circuit to control that? And what data are you going to use program it? Nice development project but a little too much overhead unless you are severely limited on the power available.

 

[russ_watters]

I wouldn't experiment with controlling the fan speed. I'd either leave it on or abandon the project. The torque of turning the thing on/off is going to cause a reaction force... a wobble. There goes your pic!

Good point - what I didn't tell you before is that I first tested it with the fan on full and got a vibration that showed-up in the pics, so the fan was always on low speed (5V). While imaging, it would definitely be a good idea to leave it on at a constant speed.

That thermostat you linked would therefore just be set up to control the peltier and the fan could stay at constant speed. With the fan on low and the peltier on high, the cold side of the peltier reached -4.4C while the surface of the heat sink was 37C (not mounted on the CCD, just open to room temp air). That's acceptable...barely. I'm considering getting a better heat sink (that was just one I had laying around), but may not need to if I sand-smooth and insulate where it connects to the camera's heat sink.

How are you going to know if a slower speed is cooling the hot side correctly without introducing another circuit to control that?

I'd do static testing and/or just monitor the temperaturre of the hot and cold sides manually. I have a couple of spare PC thermocouples with displays.

And what data are you going to use program it?

Not sure what you mean. If the thermostat can do temperature control, I'll just start with the Peltier on full, see what temperature it comes to at equilibrium and then adjust the thermostat to control to that temperature.

Nice development project but a little too much overhead unless you are severely limited on the power available.

I'm not really limited by power since I'm using a PC power supply. It is conceivable I could use a car battery if I went on the road with it, but the car battery would always be connected to an actual car, so I'd keep it charged. I was actually thinking of just doing a 3 or 4 position switch for the peltier and making a voltage divider to give me, say, 9V from a 12V feed. I didn't do the math yet, but I'd think I'd be dissipating a good 20W or so that way and would need a resistor (with its own heatsink) that could handle that.

 

[russ_watters]

Update:

Like http://mcshaneinc.com/html/5C7-350.html" Russ?

So I bought that temperature controller. I have it, I haven't tested it yet. I bought other parts and have some nylon screws on order to secure the whole setup to the camera. I'll do that before testing it (I have to tune the PI control and need to have everything installed first for it to be accurate).

I have a plastic project boxes from Radio Shack and I'm going to mount the controller in it, along with some switches, temperature sensor displays and fan rheostats. I'm making it more complicated than I should, but that's me.

Question, though: I'd like to use a headphones' cable for simplicity and ease of use in supplying the power. I'll need a max of 5A at 12V. Can a headphones cable support that? Might there be an issue with high frequency impedance? This is where my EE knowledge gets pretty thin...

Also, I'm sure I'm going to be making a mess of cutting holes in the project box to mount all my switches and rheostats. Any suggestions on an easy/neat way to do that?

 

[vk6kro]

Question, though: I'd like to use a headphones' cable for simplicity and ease of use in supplying the power. I'll need a max of 5A at 12V. Can a headphones cable support that? Might there be an issue with high frequency impedance? This is where my EE knowledge gets pretty thin...

Also, I'm sure I'm going to be making a mess of cutting holes in the project box to mount all my switches and rheostats. Any suggestions on an easy/neat way to do that?

You would need the sort of cable used to supply mains power for small appliances. Headphone cable would probably have too much resistance.

Round holes can just be drilled with a drill press (bench drill). If the plastic box is too flexible, put a piece of scrap wood under the plastic to support it. You can drill from underneath the panel if this makes it easier.
Put masking tape on the plastic and mark on this before drilling. Measure exactly then make a dent with a center punch.
Drill a smaller "pilot" hole first, then the bigger hole.

If the holes are an oddball shape, you can mount a small router bit in a bench drill and cut sideways like that, setting up barriers so that the cut is straight. This allows you to use the router bit at a lower RPM than in a router, so you don't melt the plastic.

 

[russ_watters]

Yeah, I was thinking I may need to buy a router...

 

[vk6kro]

Go for it. Even a small router is handy to have. I have a 20 year old Ryobi.

I got a cheap set of router bits for it and they are great even for putting a finished edge on wood.

Usually, the speed isn't variable. They have one speed and that is FAST. So, you have to watch out for burning wood or melting plastic.
But, I have used router bits in a drill press to avoid some of the high RPM problems. Even putting slots in aluminum becomes possible.

Good to see you want to get your project looking professional. I hate to see something someone knocked up using a hand drill and guesswork to place the holes. Even if it works well, it looks like junk.

 

[dlgoff]

I like using a nibbler tool like this
[PLAIN]http://lh4.googleusercontent.com/qtlZe-ysFsgaYY0JwtOMNmomlmTtGZNRM4P5v7jyCZ5bUE6i4NolsdvqSffVWEiKO6-TU6dNPi1cNdEh8N0fXWRSqDFNtjG5t9fMU9rdmPEUgQNRFF-JTvVxYMkZs533wO0meusG0Yy4oBM01huFFCKcENl8MyHoL_SCrCY9w-Zsh9oLyvtDsucARM_cD-kpErR1
for making square holes, slots, etc. It's probably best for metal boxes but it would work on plastic if it's not too thick.

 

[chemisttree]

Update: So I bought that temperature controller. I have it, I haven't tested it yet. I bought other parts and have some nylon screws on order to secure the whole setup to the camera...

I was just thinking that the controller might be best on the table next to your computer but I'm certainly no expert, especially about any impedance effects that a long cable between the controller and the peltier might cause. Just thinking about minimizing the load on the nosepiece of the camera and your focuser... especially if the tube is racked out somewhat. I know you will probably be imaging with your C11 and that won't really be a problem but if you want to use your 80mm to widefield image and guide with the C11, it might be that your 2" tube on the crayford is racked out pretty far. Less mass, less flexure.

Love to see the results!

Clear skies!

 

[russ_watters]

I like using a nibbler tool like this
[PLAIN]http://lh4.googleusercontent.com/qtlZe-ysFsgaYY0JwtOMNmomlmTtGZNRM4P5v7jyCZ5bUE6i4NolsdvqSffVWEiKO6-TU6dNPi1cNdEh8N0fXWRSqDFNtjG5t9fMU9rdmPEUgQNRFF-JTvVxYMkZs533wO0meusG0Yy4oBM01huFFCKcENl8MyHoL_SCrCY9w-Zsh9oLyvtDsucARM_cD-kpErR1
for making square holes, slots, etc. It's probably best for metal boxes but it would work on plastic if it's not too thick.

I have one of those, but I don't think it would work for the plastic box I'm using. I think I did ok...

Unnecessarily complicated controller box pic is attached. It has on/off switches on the right side for the fan and peltier, thermocouple readouts of the hot and cold temps, selectable fan speed controllers (5V and 12V), and the temperature controller pot. Everything except main power can be disconnected from the box - I'm going to gather all the cords together into one cable.

Still to do:
-Get a better wire for the peltier power and get a couple more connectors to finish the connection at the peltier.
-Pretty-up the cords by gathering them together.
-Permanently attach the peltier/fan assembly to the camera (insulated with foam rubber/weather stripping).
-Labels for the box and wires.

But it is all tested and it works. Temperature stability at the cold side is about +-.5F. How much of a difference it makes when I'm imaging remains to be seen, but making the project is half the fun ;)

 

[russ_watters]

I was just thinking that the controller might be best on the table next to your computer but I'm certainly no expert, especially about any impedance effects that a long cable between the controller and the peltier might cause. Just thinking about minimizing the load on the nosepiece of the camera and your focuser... especially if the tube is racked out somewhat. I know you will probably be imaging with your C11 and that won't really be a problem but if you want to use your 80mm to widefield image and guide with the C11, it might be that your 2" tube on the crayford is racked out pretty far. Less mass, less flexure.

I'm going to use that extra strong velcro type stuff to secure the controller box to the end of the C11 OTA, within about 18" of where the camera will end up.

The main goal here is temperature stability, so I shouldn't need to - and don't want to - adjust it at all once it is set. So I don't need to be near it and the most important temperature - the CCD chip itself - has a readout on my pc.

 

[dlgoff]

I have one of those, but I don't think it would work for the plastic box I'm using. I think I did ok...

Looks good russ. Can't wait to see it mounted then hear how it's working when you're imaging.

 

[russ_watters]

Ok, here it is, finished except for finding a better peltier power cord and straightening-up the external wiring.

FYI, the high temp thermocouple is stuck between the fins of the "hot" heatsink and the low temp thermocouple is kinda floating between the peltier and camera heatsink. I'll have to see if I can find a better place for it, maybe taped to the side of the camera heatsink.

I'm not going to do a full-fledged test today - it is really humid out and I don't have my AC on. I don't want any puddles in my living room.

 

[waht]

Very nice system Russ.

 

[russ_watters]

Update:
I did a full-fledged "static" test on the floor of my living room and it works, but there are a few issues:

1. The halves of the camera cover conduct heat to each other. This reduces the cooling effectiveness, increases the power requirement and increases the risk of condensation. I'm going to put a couple of layers of electrical tape between the halves where they meet to try to reduce the heat flow.
2. When I turn the fan speed up, the cold side temperature goes up, implying I'm blowing some air past my insulation. I'll seal it against the heatsink better with hot glue or silicone caulk (do they make black?).
3. The controller case gets too hot to touch. This one needs to be dealt with before I try again so I don't burn-out the controller. I can add a heatsink to the controller (cut a hole in the case and there is a metal plate on the controller a few mm behind it) and/or add a cooling fan to blow some air through the case.

Anyway, with the dial on the controller set to 5C and the fan on medium, the hot side was 36.7C, the cold side was 9.9C and the CCD temp was 13.5C. Ambient was 23.0C (dewpoint ~13C). I know it will do even better (especially with the above issues dealt with), but didn't want to push it due to issue #3, plus I was starting to get condensation on the heatsink.

 

[arjun000]

Peltier Temperature Controlled box

Guys i am trying to design a box 3 X 2 X 2 feet approx which will be temperature controlled between 2 - 8 degree range using peltier system with ambient temperature around 20 degree c...i am not very tech savvy...you suppose the peltier system will be effective in maintaining this range?

 

[vk6kro]

That is the size of a medium size refigerator (340 litres). Why don't you get a refrigerator from the classifieds and just use that?

Or you could butcher it to get the working parts and then adapt that to your box.

You could add some electronics to get good temperature control.


Peltier devices are low powered, so you would need a lot of them ...and they are not cheap so the project would be expensive.

 

[Proton Soup]

if fan vibration were still a problem, i think i'd try a bigger heatsink and go with passive cooling. if you don't mind the semipermanence of it, expanding urethane foam blown into a throwaway containment box should work.

 

[angeline_1210]

hi..
we're doing a project that needs temp. maintained from about 2-8 degrees Celsius.. we don't know where to buy and how we'll we know if the device can achieve the temp. within the needed range?
the size we're planning for our project is at least 1x0.5x1 ft..
our project is to maintain the temperature for vaccines which is 2-8 deg C..
is it possible? and where can we buy the device?and what should be the deltaT of the peltier device? tnx!:)

 

[angeline_1210]

its not 1x0.5x1 ft, I'm sorry. it's maybe around 1x.7x.7 ft.. tnx :)