DIY Mini-chiller

Keyoke

New member
So first, a little background on this project;

I've been having a problem with my BC29 overheating after I retrofit a 150W MH into the hood. Temps in the unit shot up to 87 at one point (yikes!).


Click for full size

So I decided to do a low-tech approach and install 4 80mm fans into the hood in the "doors" of the BC29:

Door #1 in the front:

Click for full size

Door #2:

Click for full size

And it helped, quite a bit. I'm peaking out at about 81 degrees now, which is still too high for my taste.

So it's time to do something a little more drastic I think. Now chillers IMO are awfully expensive. I started looking around a things like the Ice Probe, and the mini chillers from JBJ, etc. But none of those were really what I wanted or could afford.

So being the geek that I am I started doing some research on how the guys over at nanocustoms/nanotuners put together the ICA Nanochiller, which doesn't seem to be for sale anymore.

The ICA (and the iceprobe) both use TEC or Peltier units for cooling the water. If you're not familiar with the TEC/Peltier concept, check out the Wikipedia entries on Thermoelectric Cooling and the Peltier Effect.

Essentially the TEC is a little device that you pass electricity through and you end up with a cold side, and a hot side. The cold side can get REALLY cold too. These little units are used for cooling lasers, lab equipment, popsicles, etc.

I did quite a bit of research on these. My biggest hurdle was figuring out how to pass water over the TEC. The Nanocustoms guys were using cpu waterblocks, and an annodized aluminum "back plate" that the TEC cooled. I even found a post from one of the nanotuners guys in a corrosion engineering forum on the effects of saltwater against annodized 6061 aluminum.

Most CPU waterblocks are made of a cast plastic, or annodized aluminum, with a copper "cold plate". The problem here, is that if the annodization isn't PERFECT, you can wind up with little exposed bits of AL, when salt water is passed over these bits of AL, it corrodes. AL comes in various formats, but nearly ALL of them have some copper in it. And copper is bad to get "dissolved" into your tank!

Most of the oceanic corrosion people I spoke with (one is an uncle-in-law who's done some consulting for the Shedd) suggested I use Titanium as a cold plate.

Wow. Titanium. TI is EXPENSIVE stuff, and it's awfully hard to work with w/o the proper tools (of which I have none) or the proper skills (of which I have very little).

If all these terms are getting confusing, here's a pic to help explain how these are normally put together:

Click for full size

So my compromise for the cold plate area was to use Acrylic. Now Acrylic doesn't transfer heat very well. It's actually a fairly good insulator. So I'm going to need to be able to put some serious cold against this thing to make much of a difference in water temprature. The NC guys are using a 78W unit for the ICA, I need something bigger.

After doing some additional digging, I found some 138W units for sale on ebay that are the same size as the 78W unit over at NC. 40mmx40mm. This allows me to use the ICA waterblock to handle the water.

As for the heatsink, I didn't have anything on hand, but I did have some store credit with Fry's. I stopped over and found a cheap chipset cooler from Thermaltake called the "Crystal Orb", this little unit has a fan built into it. I'm not sure how well it'll work to be honest, but at $9.00 what the heck. I later found some old CPU heatsnks from a Pentium that I can use if the orb can't keep up.

Anyway, here's what I've built so far:

Side shot:

Click for full size

What your looking at here, is the completed unit. There are two peices of acrylic 1/8" glued together with weld-on #3. the piece closest to the heatsink has a hole cut in it large enough for the peltier to fit through, and the bottom piece has been routed down to about 1/16" thick around the peltier unit to get it as close to the water as possible.

Top:

Click for full size

Bottom:

Click for full size

The heatsink is held on with some thermal grease, and zip ties. I'll find something more permanent once I figure out if the Orb can handle the heat produced by the TEC.

The plan is to mount this in my hood above the return chamber, and use a small pump to push water through it.

I've still got some work to do. I'm performing a leak test this afternoon, and I plan on powering it up with a couple of 12V 1A "wall wart" style transformers. The TEC is rated at 12V/10A max current draw, so my little 1A unit isn't going to cut it for long, but it should be able to provide me with a proof-of-concept run and cool some luke-warm water. :)

One of the things that the ICA and IceProbe units lack are temprature controllers. The units are always "on". I don't really care for that much, especially since the current draw on these is pretty high. A little more research netted me a nifty little microcontroller called an "Arduino". This is like, the coolest thing ever. Easy to use, easy to program, and cheap. Just the way I like it.

I picked up an Arduino, a monochrome LCD, and thermistor "SparkFun". I also picked up some relays and other misc components from Mouser. I plan on embedding the thermistor in some airline tubing and epoxy and sinking that in the water. From there I should be able to sense the temp of the water and toggle on/off the TEC and heatsink fan, and the pump that will push the water through the waterblock.

I'm still waiting on the parts for the microcontoller. But I wanted to get this first part out there while it's still fresh. I'll post more once the microcontroller shows up.
 
Cool project, and good job getting it this far.

I am curious to see how well the 1/16" acrylic works as heat transfer medium.


I think peltiers are an awesome way to chill/heat a tank. The biggest downside to peltiers is that they do not scale especially well for larger water volumes. Usually you just end up needing more peltiers all hooked up in series (just look at the ocean geotronic peltier chillers).

If your TEC is wanting 10a, I wonder if 1amp to start with will even be noticable, especially since you used acrylic. But who knows 1/16" might be thin enough.


Also, I suggest that you use an independant temperature controller for your chiller, like a single stage Ranco. I dont think any kind of peltier has automatic temperature control, but alot of commercial units come with temperature regulation installed onboard.


Very promising, tagging along!!
 
Yea, I'm not sure if the Acrylic is going to do the job or not. I'm just awfully leery of any sort of easily accessable metal coldplates - Annodized AL, Stainless, etc. Perhaps some sort of ceramic might work, but again as a hobbyist, I don't have access to the materials or fabrication tools to create something like this.

There are other plastics out there that might work, someone (I can't remember where) pointed out that you could mix Dow 411 resin with barium powder to create a mouldable clay that when hardened would work well. The problem with that though, is that I can't seem to source Dow 411 resin, OR barium powder in any sort of small quantities.

I'm hoping that the larger TEC will do the trick. I'm a little concerned about condensation on the "dry" side of the acrylic as well.

Power wise I plan on using an old 350W PC power supply I have. It's pretty small (designed to fit in a 1.75" high chassis), so this should work out pretty well.

Thermostat control will be done by the Arduino and a Thermistor/relay control. I hope that stuff will show up in the next couple of days. I've already written a "sketch" that I think should work, but w/o the board here to test with I can't be sure.
 
I wouldn't use a thermistor. There are better temperature sensors on the market. Thermistors are great for measuring higher temperatures somewhat accurately, but not too good giving accurate cold temperatures. Not to mention that their output is not always linear.

I'm doing something similar, but not for an aquarium. This is the sensor I'm using: LM34. I have it connected to a 12-bit ADC and it is remarkably accurate and extremely easy to use.

The Arduino will work fine. Personally, I'd rather design my own board with my choice of microcontroller, which is exactly what I did:
TemperatureControlleredit.jpg


And the breadboard/prototype:
DSCF0629.jpg


Good luck with your project.
 
Very nice.

Thermistor wise, I was looking at the LM35's and Dallas makes a One-Wire unit as well. One of my other hobbies is building audio amplifiers and speakers (I did mention i was a geek, right?). I had a non-Microcontroller circuit that I was going to use originally based on a OpAmp (LM339N), a thermistor, and Transistor (2N3904), but I wanted to have the temp set functions, and not mess around with pots, calibration, etc.

Enter the Microcontroller.

I was looking into other MCUs (PICs, PSoC, etc). I actually stumbled across the Ardunio while looking for information on the PIC16F84 i think it was. The Ard was mentioned on some site and.. well, the little IDE is pretty easy, and I already know C/C++ (well, I can fumble through C/C++) from my day-job. And being able to program it right there thru the USB port is a big plus.

Since this will be my first microcontroller project, and I figured the Ard is the quickest way to start w/ the least amount of startup cost. I may just take one of the ATMega chips, slap the bits and pecices ona board needed and just use the Ard as a programming platform later on.. I have some other fun things I wanna do with this device as well.
 
Oh, and I should mention that the leak test passed. I used a higher-rated pump than I normally would have to "stress" the plumbing a bit. A Rio 6HF has been pushing water through the unit for about 3 hours now w/o a leak.

Sweet.

I think I fried the TEC though. The crystal orb wasn't anywhere near enough heatsink, so it's back to the drawing board to see if I can't fit the bigger unit on there somehow. One of the reasons I chose to go ahead and try it with the Orb was that the nuts that hold the waterblock on stick up about 1/64th of an inch too high for a "regular" heatsink and I can't make contact with the TEC. i may have to countersink those down and flip the screw head/nut combinations around.
 
I'm not trying to steer you away from the Arduino. The Arduino works great and has quite a following.

Personally, I prefer the Parallax Propeller. It requires only two decoupling caps, a 5 MHz crystal, EEPROM and a 3.3v supply. With that you get 8 32-bit processors running in parallel, each at up to 80 MHz. You can do some powerful stuff with that chip.
 
Last edited:
No, no worries there. I looked at the propeller, but it's a bit on the heavy side for this little project IMO. not sure what I'd put the other cores to... though I'm having these funny dreams of a Aquacontroller/Solarius/Geotherm .. monstrosity that keeps nagging at me...

<excited ramble>
One of my other projects is to DIY the Solaris. A couple of Luxeon K2's with the coliminator lens assembly, and away you go, right? Course I got this microcontroller bug now (man why didn't I use these before?) and I found this EZ-LED color mixing chip from Cypress.. takes an RGB LED array, compensates for the Bin values and temp differences and powers those little dim-able Buck/Boost LED constant current drivers, and mixes it to whatever color or temp that your little heart desires, and it works on an I2C bus.

And the aquacontroller bit.. well that'd be pretty easy too I think. Who needs X10 when you can just relay control the AC sockets? I/O & Datalogging back to a PC either over Ethernet, USB, or.. ZigBee.. HAH! I could build a mini-itx system.. I've always wanted an excuse to play with those fanless micro-PCs...

Tie it all together... Oooh god. *drool*
</excited ramble>

*ahem* My geek got the best of me there for a second. :)
 
Those little heatsinks arent going to work for a tec.Also you could use a block of delrin for the water to pass thru and attach the peltier directly to the block of delrin.

I have mostly all the parts to do this,but on a little larger scale that I bought like 6 yrs ago.Your actually the second person to bring this up this past week.I think I may need to gather up the last few parts that I need,and do a detailed how to for the nano guys.

I actually purchased everything with the intent to do a detailed write up in the nano forum for everybody.Just never got around to doing it.
 
<a href=showthread.php?s=&postid=12846065#post12846065 target=_blank>Originally posted</a> by Keyoke
No, no worries there. I looked at the propeller, but it's a bit on the heavy side for this little project IMO. not sure what I'd put the other cores to... though I'm having these funny dreams of a Aquacontroller/Solarius/Geotherm .. monstrosity that keeps nagging at me...

Actually, I'm using the chip modestly. One core will be running the main program, reading the sensor and activating the relay when necessary, one core will control the menu and keypad interface and one core will be acting as the display controller. Of course this could all fit on a single 8-bit AVR or PIC, but I can say my project is running on an 8 core CPU. :D

<a href=showthread.php?s=&postid=12846065#post12846065 target=_blank>Originally posted</a> by Keyoke
<excited ramble>
One of my other projects is to DIY the Solaris. A couple of Luxeon K2's with the coliminator lens assembly, and away you go, right? Course I got this microcontroller bug now (man why didn't I use these before?) and I found this EZ-LED color mixing chip from Cypress.. takes an RGB LED array, compensates for the Bin values and temp differences and powers those little dim-able Buck/Boost LED constant current drivers, and mixes it to whatever color or temp that your little heart desires, and it works on an I2C bus.

I discovered that chip myself, but I was planning on using it to make an 8' tall Tetris board. Still may do so...
 
I also forgot to mention that you don't want to use a relay to regulate the temperature of the Peltier anyway. You need to look into proportional control. Basically, you send a PWM pulse to the Peltier and vary the duty cycle depending on the current temperature and set point. With this method, you'll simply need a MOSFET and a 1K resistor between the controller and the Peltier. However, the software gets more complex.
 
<a href=showthread.php?s=&postid=12846137#post12846137 target=_blank>Originally posted</a> by alpine
Those little heatsinks arent going to work for a tec.Also you could use a block of delrin for the water to pass thru and attach the peltier directly to the block of delrin.

Yeah, the waterblock peice I have is actually derlin. I just wasn't sure about running water directly across the peltier's "cold" side. One of my worries is that the el-cheapo TECs I bought are potted with RTV sillicone, and I couldn't figure out how to attach the derlin to the peltier.. I suppose I'd need a larger TEC than the derlin waterblock?.
 
To be honest you might want to try a larger waterblock instead of a larger peltier.Keep in mind peltiers get below zero.I am going to pull my stuff out of storage tomorrow.I believe the delrin block I have is 10" x 12" x 2 1/2".The peltiers are 230 watts each and I have a few of them.To be honest that is way overrated,but I wanted to see if I could keep a 75 gallon system cool.I might just pick up some smaller peltiers and try them first.

Hopefully I will find my heatsink.Its completely copper,and about the same size as the delrin block I have with about 300 fins on it.If I cant find it I will have to find an alternative on ebay.
 
Keyoke, try using a copper water block with a small titanium loop in the aquarium. Run a water/antifreeze mix through the loop. In other words, make this a closed loop. I remember reading somewhere that the hot side heat sink needs to be somewhere around 90% larger than the cold side. You're going to need a big heat sink.
 
FWIW I started playing with PICs but never really got into them. I hate assembly and I hate C even worse... I was going to start using PICBASIC when I stumbled upon BASCOM-AVR. I will never use another PIC again (I have a drawer full of em too!). The AVR is just so much easier to work with. I am right at home with BASCOM-AVR (I used to do a lot of VB work).

Anyway... I also would recomend coupling the TEC directly or through a metal thermocouple. Stainless 316 will be fine or even a very thin sheet of Ti from McMaster.com

You can thermal epoxy the Ti to the TEC and then gasket the waterblock to the Ti/Tec assembly (or epoxy it if you wish).

The even better choice may be to drill a small block of Ti or SS and make your own waterblock. A small drill press will be suitable.

You can sometimes find Ss and even Ti block on eBay dirt cheap.
 
<a href=showthread.php?s=&postid=12847127#post12847127 target=_blank>Originally posted</a> by alpine
To be honest you might want to try a larger waterblock instead of a larger peltier.

I was thinking dimension wise, larger peltier, not wattage. I think I can find some larger.. I'm not sure how to go about attaching the peltier to the waterblock though. I somehow think if I drill through these, it's bad. :)

I managed to crack the acrylic plate I'd fashioned this evening trying to widen out the screw holes a bit. I'll have to fashion another one.

Keep in mind peltiers get below zero.I am going to pull my stuff out of storage tomorrow.

Hmm, good point. I wonder if the acrylic will get too brittle exposed to sub-zero temps. I'm not to worried about it cracking unless I've overtightened the waterblock.. to the acrylic.

I have a MUCH larger heatsink/fan assembly. It's a big sucker - looks like one of these. Copper fins and base, with an aluminum shroud. I tested the TEC with this and the darn thing got awfully cold.

Originally posted by MarkS
Keyoke, try using a copper water block with a small titanium loop in the aquarium. Run a water/antifreeze mix through the loop. In other words, make this a closed loop. I remember reading somewhere that the hot side heat sink needs to be somewhere around 90% larger than the cold side. You're going to need a big heat sink.

sort of like the chillers with the drop-in titanium probe, right? The problem is that TI is awfully expensive. Someone else here is using 316L stainless for a chiller tube. I commented that it might not be saltwater safe, but apparently it's being used elsewhere. Considering it's poor resistance to warm saltwater, i think i'll stay away myself. More info here.

I did look at some liquid chillers, but i think it starts to get more complicated than it needs to be. really i'm only looking for a 5-10 degree drop on my 30 gallon tank...
 
I responded in the other thread with regard to the 316L suitability so will not repeat it here.

30 Gallons is 250 pounds. So a 10 degree pulldown would equate to 2,500 BTU. Lets say that you need to do that over a 10 hour period. That is 250 BTU/h you need to be able to remove from the system. If we take into account the very inefficent heat exchanger (lets give it the benefit of the doubt and call it 60%) we are looking at a device that needs to be capable of around 625 BTU/h. to get the 10 degree pulldown during the hottest part of the day.

Of course, my numbers could be WAY off if you don't need that much cooling or need it that fast.
 
Well I decided to go ahead and do my little project after all these years.I went into my storage room to only come out with the block of delrin :(.Seems like after 3 moves the copper heatsink and my peltiers have disappeared.So since it looks like I have to start from scratch,I figured I would go a different route and try to help you in the process.

This is what I am going to use for a heatsink/fan.

http://www.xtremetek.com/reviews/index.php?id=53

I bought that a while ago,and realized it was the wrong socket for my cpu when I got it.Thats what happens when you impulse buy.Anyway gonna pick up a lower wattage peltier off ebay and use a smaller block of pvc.

Already got the tec on my ebay watch list,and I am searching for a 4" or so square block of pvc.Once I get all the parts in(next week sometime) I'll put up some pics if you dont mind.

Also not sure if you can get a peltier larger in size and the wattage not go up.I may be dead wromg though.
 
i think you can get different sizes.. they even make them with holes and stuff in them. Check out TETech http://www.tetech.com/Peltier-Thermoelectric-Cooler-Modules/Standard.html

actually.. they have a cold plate assembly that might work, but it looks like it's machined aluminum. littel pricey tho @ $140...


@Beananimal
No i think your math is about right, that's what I came up with as well. that works out to about ~180 watts or so. I might not get 10, but i should fall somewhere in the desired range. I'm thinking about this as a MkI version at this point. proof of concept. i'll make it all pretty and enlarge it some if I don't like the performance i'm getting out of it.
 
for a 138w pelt, you will need a MUUCH larger heat sink I mean really big. I would also seriously check out what that water block is made of. most of them are Al or copper which are not coated inside.

Also for a temp controller, just use a ranco, you can get unwired ones for $40 on ebay.

If you do it right yes it can work and very well, but I tried many different ways to get a similar project to work on my old 14g and they all failed miserably.

I later just went and got a 1/10hp chiller that worked ok for the system. it actually was a TEC based chiller and when I took the cover off the heat sink was 6x6x3 with a 120mm fan on it and that was for a sub 100w TEC.
 
Back
Top